Mixed parsing of tree insertion and tree adjoining grammars

合集下载

More than 80R2R3MYB regulatory genes in the genome of Arabidopsis

The Plant Journal (1998)14(3),273–284More than 80R2R3-MYB regulatory genes in the genome of Arabidopsis thalianaI.Romero 1,A.Fuertes 1,M.J.Benito 1,J.M.Malpica 2,A.Leyva 1and J.Paz-Ares 1,*1Centro Nacional de Biotecnologı´a-CSIC,Campus de Cantoblanco,28049-Madrid,Spain,and2Instituto Nacional de Investigaciones Agrarias,ctra.de La Corun ˜a,Km.7,528040-Madrid,SpainSummaryTranscription factors belonging to the R2R3-MYB family contain the related helix-turn-helix repeats R2and R3.The authors isolated partial cDNA and/or genomic clones of 78R2R3-MYB genes from Arabidopsis thaliana and found accessions corresponding to 31Arabidopsis genes of this class in databanks,seven of which were not represented in the authors’collection.Therefore,there are at least 85,and probably more than 100,R2R3-MYB genes present in the Arabidopsis thaliana genome,representing the largest regulatory gene family currently known in plants.In contrast,no more than three R2R3-MYB genes have been reported in any organism from other phyla.DNA-binding studies showed that there are differences but also frequent overlaps in binding speciﬁcity among plant R2R3-MYB proteins,in line with the distinct but often related functions that are beginning to be recognized for these proteins.This large-sized gene family may contribute to the regulatory ﬂexibility underlying the developmental and metabolic plasticity displayed by plants.IntroductionTranscription factors play a central role in the regulation of developmental and metabolic programs.Despite the large differences in these programs,existing among organisms from different eukaryotic phyla,their transcrip-tion factors are quite conserved and most of them can be grouped into a few families according to the structural features of the DNA-binding domain they contain.One of these families is that of the R2R3-MYB proteins,whose complexity in plants is addressed in this study.The prototype of this family is the product of the animal c-MYB proto-oncogene,whose DNA-binding domain consists of three related helix-turn-helix motifs of about 50amino acid residues,the so-called R1,R2andReceived 18August 1997;revised 26January 1998;accepted 28January 1998.*For correspondence (fax ϩ33415854506;e-mail jpazares@cnb.uam.es).©1998Blackwell Science Ltd273R3repeats.The repeat most proximal to the N-terminus(R1)does not affect DNA-binding speciﬁcity and is missing in oncogenic variants of c-MYB,such as v-MYB,and in the known plant R2R3-MYB proteins (Graf,1992;Lipsick,1996;Lu ¨scher and Eisenman,1990;Martin and Paz-Ares,1997;Thompson and Ramsay,1995).R2R3-MYB proteins belong to the MYB superfamily,which also includes proteins with two or three more distantly related repeats (e.g.of the R1/2type,the progenitor of the R1and R2repeats),and proteins with one repeat,either of the R1/2type (Feldbru ¨gge et al .,1997)or of the R3type (Bilaud et al .,1996;Kirik and Ba ¨umlein,1996).Genes of the MYB superfamily have been found in all eukaryotic organisms in which their presence has been investigated.However,the R2R3-type is not present in Saccharomyces cerevisiae and only 1–3copies of R2R3-MYB genes per haploid genome have been described in organisms from protists and animals (Graf,1992;Lipsick,1996;Lu ¨scher and Eisenman,1990;Thompson and Ramsay,1995).In contrast,preliminary evidence suggest that plants contain a much larger number of these genes (Avila et al .,1993;Jackson et al .,1991;Marocco et al .,1989;Oppenheimer et al .,1991).Little is known about the function of most plant R2R3-MYB genes although,in those few cases in which functions are known,these are different from those of their animal counterparts,which are mostly associated with the control of cell proliferation,prevention of apoptosis,and commit-ment to development (Graf,1992;Lipsick,1996;Lu ¨scher and Eisenman,1990;Martin and Paz-Ares,1997;Taylor et al .,1996;Thompson and Ramsay,1995;Toscani et al .,1997).Thus,most members of the plant R2R3-MYB family with known functions have been implicated in the regula-tion of the synthesis of different phenylpropanoids (Cone et al .,1993;Franken et al .,1994;Grotewold et al .,1994;Moyano et al .,1996;Paz-Ares et al .,1987;Quattrocchio et al .,1993;Quattrocchio,1994;Sablowski et al .,1994;Solano et al .,1995a).Phenylpropanoids are a large class of chemically different metabolites originating from phenylalanine,which includes ﬂavonoids,coumarins and cinnamyl alcohols among others (Hahlbrock and Scheel,1989).Despite their chemical diversity,these compounds are biosynthetically related as their synthesis does include common enzymatic steps.Other functions associated with members of the plant R2R3-MYB gene family include the control of cell differentiation (Noda et al .,1994;Oppenheimer et al .,1991)and the mediation of responses to signalling molecules such as salicylic acid and the phytohormones abscisic acid (ABA)and giberellic acid274I.Romero et al.(GA)(Gubler et al .,1995;Urao et al .,1993;Yang and Klessig,1996).Sequence speciﬁc DNA-binding has been demonstrated for several R2R3-MYB proteins,in agreement with their role in transcriptional control (Biedenkapp et al .,1988;Grotewold et al .,1994;Gubler et al .,1995;Howe and Watson,1991;Li and Parish,1995;Moyano et al .,1996;Sablowski et al .,1994;Sainz et al .,1997;Solano et al .,1995a;Solano et al .,1997;Stober-Gra ¨sser et al .,1992;Urao et al .,1993;Watson et al .,1993;Yang and Klessig,1996).The information available indicates that these proteins bind to one or more of the following types of site:I,CNGTTR;II,GKTWGTTR;and IIG,GKTWGGTR (where N indicates A,G,C or T;K,G or T;R,A or G;W,A or T).For instance,animal R2R3-MYB proteins recognize type I sequences (Biedenkapp et al .,1988;Howe and Watson,1991;Stober-Gra ¨sser et al .,1992;Watson et al .,1993),the ZmMYBP (also known as P)proteins bind to type IIG sequences,the ZmMYBC1(also known as C1)and AmMYB305proteins bind to both type II and type IIG,and the PhMYB3protein can bind to types I and II (Grotewold et al .,1994;Sablowski et al .,1994;Sainz et al .,1997;Solano et al .,1995a;Solano et al .,1997).Recent studies with protein PhMYB3from Petunia,including molecular modelling based on the solved structure of the mouse c-MYB protein (MmMYB),have highlighted the importance of residues Lys67,Leu71,Lys121and Asn122in determining recognition speciﬁcity (Ogata et al .,1994;Solano et al .,1997).These residues are fully conserved in all known plant R2R3-MYB proteins.In contrast,protein AtMYBCDC5,which has two R1/2-type repeats and does not conserve these residues,has a completely different speci-ﬁcity (CTCAGCG,Hirayama and Shinokazi,1996).To evaluate the number of R2R3-MYB genes in plants,and as a ﬁrst step towards determining the full range of functions associated with these genes using a reverse genetic approach,we have carried out a PCR-based systematic search for R2R3-MYB genes in the model species Arabidopsis thaliana .We estimate that it contains at least 85,and probably more than 100R2R3-MYB genes,representing the largest gene family of regulatory genes described thus far in any plant species.In addition,we have investigated the DNA-binding speciﬁcity of representative R2R3-MYB proteins and have shown that there may be differences but also considerable similarities in binding speciﬁcitiy between R2R3-MYB proteins,particularly among members of the same phylogenetic group,which is in agreement with the recognizable functional relationships between the members of the R2R3-MYB family.ResultsIsolation of R2R3-MYB clonesAll known plant R2R3-MYB proteins contain highly con-served stretches of amino acid residues within the©Blackwell Science Ltd,The Plant Journal ,(1998),14,273–284Figure 1.Consensus amino acid sequence of the two repeats comprising the DNA binding domain of plant R2R3-MYB proteins as described by A´vila et al .(1993),and oligonucleotide mixtures used in the isolation of the R2R3-MYB genes (N1–6and C1–3).Upper case indicates residues fully conserved in all proteins used to derive the consensus.Lower case indicates residues identical in at least 80%of the proteins.Other symbols are:ϩ,basic amino acid;–,acidic amino acid;#,hydrophobic amino acid.New sequences (published since this alignment,see Figure 2)have not altered this consensus sequence in the regions from which the oligonucleotide sequences were derived,with the exception of PhMYBAn2which has a D/A substitution in the region corresponding to oligonucleotide mixtures C1-C3,although they have increased the variability of residues in variable positions.This variability was taken into account in the design of the oligonucleotide mixtures (R ϭA ϩG,Y ϭC ϩT,S ϭG ϩC,D ϭA ϩG ϩT,N ϭA ϩG ϩC ϩT)and so the oligonucleotide mixtures should have recognized all the more recent additions to the R2R3MYB gene family.recognition helices of the R2and R3repeats from which R2R3-MYB -speciﬁc mixtures of oligonucleotides can be derived (Avila et al .,1993;Figure 1).These oligonucleotide mixtures do not recognize the AtMYBCDC5gene encoding a MYB protein with two highly divergent repeats of the R1/2-type (Hirayama and Shinokazi,1996;Lipsick,1996).To search for R2R3-MYB genes,we ﬁrst prepared cDNA and genomic DNA libraries (of 1000and 3000clones,respectively)enriched in these genes using PCR with R2R3-MYB -speciﬁc oligonucleotides.Sequencing of all the different clones present in each of these libraries (for details,see Experimental procedures),revealed that 36and 74different R2R3-MYB genes were represented in the cDNA and genomic DNA libraries,respectively,and that 32were represented in both libraries.A total of 78different R2R3-MYB genes were therefore represented in our collec-tion.A computer search revealed that there were 31R2R3-MYB genes from Arabidopsis described in databanks,of which seven were not represented in the set of 78isolated in this study.There are,therefore,at least 85(78ϩ7),and probably more than 100(78ϫ31/24,see Experimental procedures)R2R3-MYB genes in the Arabidopsis thaliana genome.More than half of the R2R3-MYB genes identiﬁed in this study were characterized only at the genomic DNA level,raising the possibility that many of these R2R3-MYBThe R2R3-MYB gene family in Arabidopsis275genomic sequences might represent pseudogenes rather than active genes.However,in no case was the reading frame of the exonic sequences(represented in the genomic clones)prematurely terminated.In addition,the numberof fully conserved residues in plant R2R3-MYB proteins is the same independently of whether those protein sequences from the R2R3-MYB genes characterized onlyat the genomic DNA level are considered in the estimation. On the other hand,pseudogenes usually show higher rates of non-synonymous substitutions(Kns)relative to synonymous substitutions(Ks)than active genes(Satta, 1993).We calculated the Kns/Ks ratio for all possible pairsof R2R3-MYB genes in this population and these ratios were compared to those in the population of R2R3-MYB genes known to be expressed(i.e.those for which a cDNA clone was available),using the method of Nei and Gojobori (1986).The Kns/Ks values in the two populations(Kns/Ksin genomic DNA population:0.393Ϯ0.016;Kns/Ks in cDNA population:0.392Ϯ0.115)were not signiﬁcantly differentin a t-test(Pϭ0.83ജ0.10).Collectively,these data are in agreement with the conclusion that most,if not all,plantR2R3-MYB sequences represent active genes.Phylogenetic analysis of R2R3-MYB proteinsA phylogram of R2R3-MYB proteins was constructed with the neighbor-joining method(Saitou and Nei,1987)using the sequences of the proteins in Figure2(except HvMYB33, LeMYB1,AtMYB67,AtMYB41and AtMYB45;Figure3). Three major groups were distinguished in the phylogram, A,B and C(Figure3).The bootstrap support for the node corresponding to groupC was not very high(30%),perhaps due to the short size of the sequences used.However, when the analysis was made using the whole R2R3-MYB domain from the proteins for which this sequence was available,the bootstrap support of this node was more than75%(see Figure3).In addition,the existence of the three groups was also supported by the tree constructed using parsimony(Eck and Dayhoff,1966)(not shown)andby the different intron/exon structure of the genes encoding the proteins of each group,with the exception of AtMYB67 (see Figure3).Group A(accounting for about10%of the A.thaliana proteins),which also includes the animal and protist R2R3-MYB proteins,represents genes with no intronin the region sequenced,with the exception of AtMYB1 which has an intron at position1.Group B(5%of the A. thaliana proteins)represents proteins encoded by genes with an intron at position3.Finally,group C(85%of A. thaliana proteins)contains genes with an intron at position 2.As shown below(see Discussion),this classiﬁcation is also in agreement with the data on DNA-binding speciﬁcityof R2R3-MYB proteins,as similarities in this property were usually higher between proteins belonging to the same group than between proteins belonging to different groups.©Blackwell Science Ltd,The Plant Journal,(1998),14,273–284Each group,particularly group C,can be further subdivided into subgroups of more closely related members.Many of these subgroups contain R2R3-MYB proteins from other plant species(although the search for this type of MYB genes in these species has not been exhaustive),consistent with the high functional similarity of regulatory systems among plants(Benfey and Chua,1989).DNA-binding speciﬁcity of representative R2R3-MYB proteinsTo evaluate the degree of similarity in DNA binding speciﬁcity between different Arabidopsis R2R3-MYB proteins,we isolated cDNA clones containing the entire coding region of four representative R2R3-MYB proteins, AtMYB15,AtMYB77,AtMYB84and AtMYBGl1(see Methods).Full length and deletion derivatives of these proteins were produced by in vitro transcription and translation.To determine their DNA-binding speciﬁcity, an EMSA(electrophoretic mobility shift assay)-based random-site selection procedure was used(Blackwell and Weintraub,1990;Solano et al.,1995a).Selection experi-ments were performed with two oligonucleotide mixtures, OI and OII,which had a partially random core sequence representing the three types of sites deﬁned for R2R3-MYB proteins:OI,type I;OII,types II and IIG(Biedenkapp et al., 1988;Grotewold et al.,1994;Gubler et al.,1995;Howe and Watson,1991;Li and Parish,1995;Moyano et al.,1996; Sablowski et al.,1994;Sainz et al.,1997;Solano et al., 1995a;Solano et al.,1997;Stober-Gra¨sser et al.,1992;Urao et al.,1993;Watson et al.,1993;Yang and Klessig,1996; Figure4;see Introduction).In fact,the nucleotides(or their counterparts in the complementary strand)present in the non-randomized positions(–2,ϩ1andϩ3)are contacted by residues fully conserved in all plant R2R3-MYB proteins (Leu71,Lys121and Asn122,respectively,in PhMYB3;the G in the complementary strand of position–2in type I targets is contacted by another fully conserved residue, Lys67(Solano et al.,1997).AtMYB15and AtMYB84bound the partially randomized oligonucleotide mixture OII and,to a lesser extent,the OI oligonucleotide mixture,and the reciprocal was true with a carboxy-terminal deletion derivative of AtMYB77 (AtMYB77∆C1)which bound better to OI(data not shown). AtMYB77∆C1was used because the full size protein had lower binding afﬁnity,as is the case with other R2R3-MYB proteins(PhMYB3and MmMYB)(Ramsay et al.,1992; Solano et al.,1995a).In contrast,neither AtMYBGl1nor its carboxy-terminal deletion derivatives showed detectable binding to either of these oligonucleotide mixtures(not shown).A similar result was obtained with an increased amount of probe and/or a decreased amount of non-speciﬁc competitor DNA,independently of the type of probe used,the partially randomized oligonucleotide276I.Romero et al.mixtures OI and OII,or a fully randomized mixture (O,data not shown).Protein phosphatase treatments,which have been shown to increase binding afﬁnity of one R2R3-MYB protein (Moyano et al .,1996),were also ineffective (not shown).Collectively,these data suggest limited in vitro DNA-binding afﬁnity for this protein.It is possible that low DNA-binding afﬁnity is an intrinsic property of AtMYBGl1and that it might be increased in vivo after interaction(s)with other protein(s).For example,there is evidence that maize C1protein (ZmMYBC1),which also shows low bind-ing afﬁnity in vitro (Sainz et al .,1997),requires an inter-action with a second protein (the MYC protein R,Goff et al .,1992)to activate ﬂavonoid biosynthetic genes.A similar interaction is possibly necessary for the activity of AtMYBGl1in vivo (Lloyd et al .,1992).©Blackwell Science Ltd,The Plant Journal ,(1998),14,273–284After four cycles of enrichment,oligonucleotides selected by the R2R3-MYB proteins were cloned and sequenced.In all instances,despite using two target oligonucleotide mixtures,only one type of sequence was recovered for each protein,indicating strong preference for one of the types of sequences (Figure 4).For instance,in the case of protein AtMYB77∆C1,which preferred type I sequences,the sequences selected from oligonucleotide OII were also of type I (generated in variable positions of OII,not shown)and the reciprocal was true for proteins AtMYB15and AtMYB84(not shown).These results argue against a bias in the binding site selection experiments due to the use of partially degenerated oligonucleotide mixtures,although this possibility cannot be fully excluded.Next,we used oligonucleotides representing the deﬁnedThe R2R3-MYB gene family in Arabidopsis 277optimal target sites and mutants of these sites in binding experiments with each of the above Arabidopsis proteins and with carboxy-terminal deletion derivatives of PhMYB3(PhMYB3∆C1),AmMYB305(AmMYB305∆C1)and MmMYB (MmMYB ∆C2R1;Solano et al .,1997)as controls (Figure 5a).The results of these experiments agreed with those from site selection experiments,but revealed that AtMYB77∆C1also recognised certain type II sequences,although with reduced afﬁnity compared to that for type I sequences.In addition,they also showed speciﬁc DNA binding afﬁnity for AtMYBGl1,as it could weakly bind to oligonucleotide II-1.In an apparent discrepancy with binding site selection experiments,protein AtMYB77∆C1bound better to the oligonucleotide containing one of the optimal binding sites of PhMYB3(MBSI,oligonucleotide I-1;Solano et al .,1995a)than to that containing its deduced optimal binding sequence (oligonucleotide I-2).Discrepancies between a binding site selection derived sequence with the optimal binding site have also been reported for MADS box proteins (Riechmann et al .,1996).A difference between the two oligonucleotides (I-1and I-2)is that I-1is ﬂanked by three extra As,which would increase its ability to bend,a property known to greatly inﬂuence binding by DNA-distorting/bending proteins,such as R2R3-MYB proteins and MADS proteins (Parvin et al .,1995;Riechmann et al .,1996;Solano et al .,1995b;Thanos and Maniatis,1992).To test whether this difference could be the cause of the preference of AtMYB77∆C1for oligonucleotide I-1versus I-2,DNA binding experiments were conducted with new oligonucleotides in which the three extra As of oligonucleo-tide I-1had been removed.The binding by AtMYB77∆C1to this deletion version of I-1(I-1∆)was similar to that obtained for the oligonucleotide derived from binding siteFigure 2.Deduced amino acid sequences of Arabidopsis R2R3-MYB proteins.For comparison,the sequences of R2R3-MYB proteins from other plant species and from representative organisms of other phyla are also given.The region shown is that ﬂanked by the sequences used to derive the oligonucleotide mixtures shown in Figure 1.The clones corresponding to AtMYB41and to AtMYB45did not encode the carboxy-terminal part of their sequence due to mispriming events.For protein (and gene)names,a standardized nomenclature has been used (Martin and Paz-Ares,1997)whereby the name of each protein includes a two-letter preﬁx as species identiﬁer,the term MYB,and then a term describing the particular family member.The codes for the species identiﬁer are:Am,Antirrhinum majus ;At,Arabidopsis thaliana ;Cp,Craterostigma plantagineum ;Dd,Dictyostelium discoideum ;Dm,Drosophila melanogaster ;Gh,Gossypium hirsutum ;Hv,Hordeum vulgare ;Le,Lycopersicon esculentum ;Mm,Mus musculus ;Nt,Nicotiana tabacum ;Os,Oryza sativa ;Ph,Petunia hybrida ;Pm,Picea mariana ;Pp,Physcomitrella patens ;Ps,Pisum sativum ;Xl,Xenopus laevis ;Zm ,Zea mays.As family member identiﬁer we have always used a number except where the previously given name was based on functional information,such as the phenotype of mutants (e.g.the Gl1(Glabrous1)protein from Arabidopsis is named AtMYBGl1).Thus,all the genes identiﬁed in this study have been given a standardized number independent of whether a different non-standardized name has been given by other authors.This has occurred in the following cases:AtMYB13,also named AtMYBlfgn (accession number Z50869);AtMYB15,also named Y19(X90384);AtMYB16,also named AtMIXTA (X99809);AtMYB23,also named AtMYBrtf (Z68158);AtMYB31,also named Y13(X90387);AtMYB44,also named AtMYBR1(Z54136);AtMYB77,also named AtMYBR2(Z54137).In addition,the following R2R3-MYB genes,which were not identiﬁed in this study,were renamed (with the agreement of the authors who ﬁrst described them):AtMYB101(M1);AtMYB102(M4).AtMYB90is described in the EMBL databank as an anonymous EST (H76020).The column on the right of the amino acid sequence gives the accession number from which the sequences were derived.The accession numbers of the cDNAs encoding the full-size proteins AtMYB15,AtMYB77and AtMYB84are Y14207,Y14208and Y14209,respectively.In case of PhMYBAn2,the sequence was copied directly from Quattrochio (1994).The second column shows the position of the intron interrupting that part of coding sequence represented in the ﬁgure:–,unknown;0,no intron;the localization of introns 1,2and 3is shown relative to the consensus sequence.The third column shows the type of clone isolated in this study:a,cDNA clone;b,genomic clone.Other letters in this column indicate that the sequence shown in the ﬁgure was previously described in databanks or published (c)or that only part of the sequence shown was previously described (d).Two additional sequences (accession numbers H36793and T42245),each corresponding to a novel Arabidopsis R2R3-MYB gene,were found in the EST databank,but are not represented in the ﬁgure because they were incomplete.These sequences were,however,used for the estimation of the size of the R2R3-MYB gene family.Asterisks indicate proteins for which the sequence of the whole R2R3-MYB domain is known.Symbols in the consensus sequence are as in Figure 1.©Blackwell Science Ltd,The Plant Journal ,(1998),14,273–284selection experiments (Figure 5b).This result underscored the importance of DNA conformational properties in bind-ing by transcriptional factors.DiscussionGenes of the R2R3-MYB family are quite widespread in eukaryotes,with the exception of yeast,and in plants the number of these genes is especially high.Whereas no more than three R2R3-MYB genes have been described in any organisms from other eukaryotic phyla,here we isol-ated partial cDNA and/or genomic clones corresponding to 78different R2R3-MYB genes from Arabidopsis and estimated that there are probably more than 100R2R3-MYB genes in this species.The different size of regulatory gene families in different groups of eukaryotes,a situation which is not exclusive for R2R3-MYB genes (for instance,see the case of MADS box proteins;Theissen et al .,1996),might reﬂect major differences in developmental and meta-bolic programs generated during evolution of these groups,which largely involved a different use of pre-existing regu-latory systems rather than the generation of new systems (Martin and Paz-Ares,1997).According to recent estimates on the number of genes in Arabidopsis (16000–43000;Gibson and Sommerville,1993),members of the R2R3-MYB family would represent at least 0.2–0.6%of the total Arabidopsis genes,the largest proportion of genes thus far assigned to a single regulatory gene family (and even to a gene family encoding any type of protein)in plants.In other types of eukaryotes there are families of equal,or even larger,size;for instance,it is estimated that genes encoding zinc-ﬁnger proteins represent about 1%of the human genes (Hoovers278I.Romero et al.et al .,1992)and,in Caenorhabditis elegans ,about 0.4%of its genes contain homeoboxes (Bu ¨rglin,1995).However,in these families overall sequence conservation is very low and variability in DNA-binding speciﬁcity is high (Klug and Schwabe,1995;Treisman et al .,1992).In contrast,members of the plant R2R3-MYB family share higher amino acid sequence similarity,particularly in their recogni-tion helices (Figure 1)and display considerable DNA-recognition similarities (Figures 3and 5).These similarities in recognition speciﬁcity are par-ticularly noticeable between members of thesame ©Blackwell Science Ltd,The Plant Journal ,(1998),14,273–284phylogenetic group,although in some cases overlaps in binding speciﬁcity between members belonging to differ-ent groups have been observed (Figures 3and 5).Thus,in the cases studied here or elsewhere (Biedenkapp et al .,1988;Grotewold et al .,1994;Gubler et al .,1995;Howe and Watson,1991;Li and Parish,1995;Moyano et al .,1996;Sablowski et al .,1994;Sainz et al .,1997;Solano et al .,1995a,1997;Stober-Gra ¨sser et al .,1992;Urao et al .,1993;Watson et al .,1993;Yang and Klessig,1996)members from group A (including both those from plants and from organisms from other phyla)prefer (or bindThe R2R3-MYB gene family in Arabidopsis 279to)a type I sequence,members of group B bind equally well to both type I and type II,and most members of group C prefer (or bind to)a type IIG.Possible exceptions are the proteins from group C AtMYB2,reported to bind type I sequences (Urao et al .,1993),and GLABROUS1(AtMYBGl1),which only bound to a type II sequence (Figure 5)although,in the ﬁrst case,binding to IIG sequences was not studied and,in the second case,binding site selection experiments failed to provide information on its optimal binding site (see Results).However,it is striking that the only sequence bound by AtMYBGl1(AAAGTTAGTTA)perfectly conforms to the sequence of gibberellic acid responsive elements,and gibberellic acid is known to affect the AtMYBGl1-controlled trait trichome formation (Oppenheimer et al .,1991;Telfer et al .,1997).In line with these similarities in binding speciﬁcity,and despite the fact that target selectivity is usually also inﬂuenced by interactions with other factors,most of the R2R3-MYB proteins studied so far,which are scattered throughout groups B and C,have been implicated in the control of phenylpropanoid biosynthetic genes (Cone et al .,1993;Franken et al .,1994;Grotewold et al .,1994;Moyano et al .,1996;Paz-Ares et al .,1987;Quattrocchio et al .,1993,1994;Sablowski et al .,1994;Solano et al .,1995a;Figure 3).Nevertheless,there are some R2R3-MYB proteins that have been implicated in other functions,including the control of cell differentiation and the mediation of plant responses to several signal molecules (Gubler et al .,1995;Noda et al .,1994;Oppenheimer et al .,1991;Urao et al .,1993;Yang and Klessig,1996).TargetFigure 3.Phylogenetic tree of the R2R3-MYB family using the neighbor-joining method (Saitou and Nei,1987).The phylogram shown was constructed with the sequences given in Figure 2,except HvMYB33,LeMYB1,AtMYB67,AtMYB41and AtMYB45.The ﬁrst two were excluded because they were the only ones out of the 57known complete-MYB-domain sequences which grouped differently (with bootstrap support Ͼ50%)depending on whether the complete MYB domains or the portion characterized in this study was used in the calculations.Protein AtMYB67was the only one which was not grouped with the other proteins encoded by genes with the same intron/exon structure.Proteins AtMYB41and AtMYB45were not used because only partial sequence data were available,although their probable position in the phylogram,inferred from a tree constructed also using their incomplete sequences (not shown),is indicated in the tree with dashed lines.Exclusion of these ﬁve proteins increased the bootstrap support of the major nodes (not shown).Names of R2R3-MYB proteins from non-plant species are shown in red.The three major nodes,A,B and C,are denoted.Numbers (0,1,2or 3)in some branches indicate the type of intron in the cloned portion of the genes encoding proteins originating from the respective branch,as far as the genes for which this information is available are concerned (Figure 2).Nodes with high bootstrap support are indicated (empty symbols,bootstraps Ͼ50%;ﬁlled symbols,bootstraps Ͼ75%).Circles refer to bootstraps data corresponding to the represented tree.Squares refer to bootstraps data corresponding to the tree constructed with the sequence of the whole MYB domain of the proteins for which this information was available (Figure 2).The known functions associated with some plant R2R3-MYB proteins are indicated:Ph,regulation of phenylpropanoid biosynthetic genes (proteins ZmMYBC1,ZmMYBPl,ZmMYBP ,ZmMYB38,ZmMYB1,AmMYB305,AmMYB340,PhMYBAn2;PhMYB3,Cone et al .,1993;Franken et al .,1994;Grotewold et al .,1994;Moyano et al .,1996;Paz-Ares et al .,1987;Quattrocchio et al .,1993;Quattrocchio,1994;Sablowski et al .,1994;Solano et al .,1995a);CD,control of cell differentiation (proteins AtMYBGl1and AmMYBMx,Noda et al .,1994;Oppenheimer et al .,1991);SA,GA and ABA,involved in signal transduction pathway,respectively,salicylic acid (gene NtMYB1;Yang and Klessig,1996),gibberellic acid (proteins HvMYBGa,Gubler et al .,1995)and abscisic acid (proteins AtMYB2and ZmMYBC1;Hattori et al .,1992;Urao et al .,1993).Capital letters are used when the functions associated are based on genetic evidence (i.e.analysis of mutants).Also indicated is the available information on DNA-binding speciﬁcity of some of the R2R3-MYB proteins,(arrowheads indicate the proteins examined in this study):I,CNGTTR (proteins MmMYB,MmMYBA,MmMYBB,DdMYB,AtMYB1,AtMYB2,AtMYB77,PhMYB3,HvMYBGa,NtMYB1;Biedenkapp et al .,1988;Howe and Watson,1991;Solano et al .,1995a;Stober-Gra ¨sser et al .,1992;Urao et al .,1993;Watson et al .,1993);II,GTTWGTTR (proteins PhMYB3,HvMYBGa,AmMYB305,ZmMYBC1,AtMYBGl1;Gubler et al .,1995;Sainz et al .,1997;Solano et al .,1995a;Solano et al .,1997);IIG,GKTWGGTR (proteins AmMYB305,AmMYB340,ZmMYBP ,ZmMYBC1,AtMYB6,AtMYB7,AtMYB15,AtMYB84,NtMYB1;Grotewold et al .,1994;Li and Parish,1995;Moyano et al .,1996;Sablowski et al .,1994;Sainz et al .,1997;Solano et al .,1995a;Yang and Klessig,1996)(where N indicates A or G or C or T;K,G or T;R,A or G;W,A or T).Capital letters are used in those cases in which the sequences are known to be the optimal binding site.When a given protein is able to bind to more than one type of site,the size of the letter reﬂects the relative binding afﬁnity for these sites.©Blackwell Science Ltd,The Plant Journal ,(1998),14,273–284genes of these latter R2R3-MYB genes are mostly unknown,thus precluding deﬁnite conclusions about whether they are functionally related between themselves or indeed with the R2R3-MYB genes regulating phenyl-propanoid biosynthetic genes.However,the signal molecules salicylic acid,ABA and GA inﬂuence,among others,the expression of phenylpropanoid biosynthetic genes,in several instances through cis -acting elements resembling R2R3-MYB binding sites (Dixon and Paiva,1995;Hahlbrock and Scheel,1989;Hattori et al .,1992;Sablowski et al .,1994;Shirasu et al .,1997;Weiss et al .,1990,1992).In addition,GA also affects trichome forma-tion,another trait under the control of an R2R3-MYB gene,AtMYBGl1(Telfer et al .,1997).Moreover,the MIXTA gene (AmMYBMx )controls the specialized shape of inner epidermal petal cells of Antirrhinum ﬂowers,and these changes in cell shape correlate with changes in the cell wall,a structure containing phenylpropanoid derivatives (Noda et al .,1994).The number of R2R3-MYB genes with distinct but related functions might therefore be extraordinarily high,particularly with regard to the regulation of different phenylpropanoid biosynthetic genes,although some of these genes could also (or alternatively)act on other types of targets (e.g.the barley gibberellic acid induced α-amy gene is a likely target of HvMYBGa,Gubler et al .,1995).In any case,the broad (phylogenetic)distribution of the R2R3-MYB genes for which there is evidence of their involvement in the regulation of phenylpropanoid metabolism,suggests that a very early plant-speciﬁc R2R3-MYB ancestor already had this function,and that。

九年级地理环境保护英语阅读理解20题

九年级地理环境保护英语阅读理解20题1<背景文章>Global warming is one of the most serious environmental issues facing our planet. It refers to the long-term increase in the average temperature of the Earth's climate system. The phenomenon of global warming is caused by a variety of factors. One of the main causes is the increase in greenhouse gas emissions, such as carbon dioxide, methane, and nitrous oxide. These gases trap heat in the atmosphere and cause the Earth's temperature to rise.The effects of global warming are far-reaching. Rising sea levels threaten coastal areas and low-lying islands. More frequent and intense extreme weather events, such as hurricanes, floods, and droughts, can cause significant damage to human lives and property. Changes in precipitation patterns can also affect agriculture and water resources.To address global warming, we need to take action on multiple fronts. We can reduce greenhouse gas emissions by using renewable energy sources, such as solar and wind power. We can also improve energy efficiency in buildings, transportation, and industry. In addition, we can protect and restore forests, which absorb carbon dioxide from the atmosphere.1. What is global warming?A. A short-term increase in temperature.B. A long-term increase in the average temperature of the Earth's climate system.C. A decrease in temperature.D. No change in temperature.答案：B。

Lightnin MagMixer MBE系列混合器产品说明书

A Standard for Cleanability, Durability and Performance - MagMixer ® MBEMAG M I X E R M B E S E R I E S The big advantage of Lightnin’s magnetic agitator is the complete separation of the interior of the tank from the outside. In contrast to conventional agitators, there is no shaft penetrating the tank and therefore no mechanical seal. This eliminates the risks of leaks and microbial contamination and the need for special maintenance that are associated withconventional agitators. We have developed our magnetic agitators with special emphasis on optimizing their cleanability, which is essential for sterile processes. The MBE Series conforms to EC 1935 and AMSE BPE confirming the design of these agitators are qualified for such applications.Bottom-mounted magnetic agitators are state of the art for low-viscosity liquids in pharmaceutical and biotechnology production. The compact design, low maintenance and high reliability guarantee trouble-free production. Using a bottom-mounted agitator also frees up space on the tank lid for sensors, valves and sight glasses.Based in Charlotte, North Carolina, SPX FLOW is a leading global supplier of highly engineered flow components, process equipment and turn-key systems, along with the related aftermarket parts and services, into the food and beverage, power and energy and industrial end markets. SPX FLOW has more than $2 billion in annual revenues and approximately 8,000 employees with operations in over 35 countries and sales in over 150 countries around the world. T o learn more about SPX FLOW, please visit our website at A World Leader in Industrial Mixing since 1923, Lightnin has over 90 years of unrivaled experience in industrial mixing technology, process knowledge, and technological innovation. Lightnin enjoys a global reputation for durable, long-lasting mixers, agitators, aerators, and flocculators for fluid process systems. We offer a full spectrum of impeller designs for diverse applications. In addition, we offer a worldwide service network, mixer repair, gearbox repair, and replacement parts programs. Look to Lightnin for knowledge,technology, and service excellence.About SPX FLOWAbout LightninDairy, Food & Beverage P harmaceutical Fine ChemicalsM B E: OPE N D E S I G N & OPTI MAL FLOW TH R OU G H R OTOROpen design with excellent cleaning: hub and magnetic rotor are connected only by the impeller blades.The ceramic bearings are oversized (in diameter and height), product-lubricated and consist of outer (silicon carbide) and inner (zirconium oxide) bearing. This results in exceptional stability, good emergency running properties and particle generation below detectable levels.Ease of maintenance - ceramic bearing parts can be replaced by users on site; no spare rotor needed.A very large gap between the rotor and the containment shell maximizes flow through the gap and minimizes shear stress.CFD-engineered mixing: fluid is drawn from above and pumped radially. Perfect for mixing solid powders into liquids; rapid breakdown of temperature and concentration gradients, ensures good heating and cooling.M B E: MAG N ETI C I M PE LLE R SNew and stronger magnetic materials enable us to reach a higher transferable torque for the same geometry of the drive and rotor. Lightnin offers a wide range of drive sizes with operating torques from 30 Ncm to 300 Nm (2.65lbf.in to 2655lbf.in) to suit a wide range of applications.Lightnin MBE magnetic impellers ensure that the optimal formulation of the product is achieved throughout the whole volume. When the product is being transferred out of the vessel, the homogeneity of the mixed product is maintained reliably down to the last drop.The open design enables easy cleaning (CIP) and sterilization (SIP). Lightnin experts will help you select the best impeller variant for your process, the best drive unit and the most suitable installation option (welded or Plug In containment shell).Magnetic ImpellersDesign Features & Functions of Bottom-Mounted Magnetic AgitatorsB EAR I N G S• Zirconium oxide inner bearing: less risk of breakage, resistantto damage by sudden settling of the rotor• Silicon carbide outer bearing: with channels in the face side for better lubrication of the bottom contact surface andenhanced cleaning (CIP)• Run-dry capability with the rotor type MBE: Magnetic lifting of the impeller reduces the load on the bearing surface so thatthe agitator can be kept running while the vessel is emptied(mixing down to the last drop)LOW E R I N G D EVI CEAgitators size from MBE200 and upwards are supplied with aspecial lowering mechanism that withdraws the drive magnet outof the containment shell.B E N E FITS OF TH ER ETRACTAB LE D R IVE MAG N ETControlled removal and safe insertion of the impeller due to withdrawl of the drive magnet. The drive unit remains in positionwhile the magnetic drive rotor is lowered out of the containment shell. Avoids damage to the ceramic bearings. Improved safety: The device protects against the crushing hazard involved in placing theagitator head on the bearing, and thus meets the demand of theEC Machinery Directive for designed protection against injury.R E M OVAB LE CONTAI N M E NT S H E LL (PLU G-I N)As an alternative to a version with a flange for welding into the vessel, the agitator can be supplied with a removable containment shell. This can facilitate maintenance. This practical Plug In solutionis becoming more popular and makes it easier to switch fromshaft-driven agitators with mechanical seals to magnetic agitators.E LE CTR OP OLI S H I N G OFSTAI N LE S S STE E L S U R FACE SAs well as mechanical polishing to two levels, we also electropolishas standard so that we can meet the growing demand for thehighest possible surface quality.E XTRA LAR G E PLU G I N FLAN G EThe Plug In flange is also available with a large-diameter flange which allows the whole agitator head to be extracted from the vessel through the bottom opening.I M PE LLE R S PE E D M ON ITOR I N GSometimes a problem inside the tank or an operating error can lead to forces on the agitator head that exceed the maximum tranmissible torque, so that the magnetic coupling decouples. In this situation the agitator head stops turning although the drive is still running. T o monitor this issue, we offer an optional contact-free rpm sensor for the agitator head.ATTACH M E NT OF TH ED R IVE S H E LL BY TR I CLAM PAttaching the drive with a TriClamp fitting enables it to be removed quickly without tools, eg: when using the agitator with an autoclavable tank. As another alternative, bayonet fittings are also possible.E XTRACTOR TOOLAn optional tool for the removable containment shell (Plug In). The tool enables the containment shell to be removed from the tank easily and gently.Reliable and easy to maintainFor safety reasons, the speed sensor is included as standard in agitators for use in ATEXZone 0 (directive 2014/34/EU)5T echnical SpecificationsW ETTE D MATE RIALSStandard Max. Temp.: 280F (138°C)Standard Dry Running for 15 minutes at reduced speedStandard pH range: 1-14 OptionalImpeller removal toolOther Drive Options Stainless SteelTri-Clamp connectionTAN K PLATE / FLAN G E M OU NTI N G OPTI ON SWelded in Tank Plate - T ank plate welded to vessel by customer, Mixer bolts to plate, impeller sits on tank plate bearing postPlug in Tank Plate - Modified tank plate welded to vessel by customer, Mixer with coupling and bearing inserts into tankplate, Impeller sits on bearing D R IVE U N ITS7MBE 12D 80/3.150.090.1211701000140/37MBE 25D 105/4.130.180.241170950350/92MBE 25130/5.120.180.241110550500/132MBE 50165/6.490.370.4961105201,200/317MBE 75190/7.480.550.738854602,500/660MBE 100250/9.840.75 1.0068534014,500/3,830MBE 200300/11.81 1.50 2.0128530028,000/7,397MBE 300350/13.78 2.20 2.958528531,000/8,189MBE 550400/15.74 4.00 5.3648527025,000/6,604MBE 750450/17.72 5.507.3768527040,000/10,567MBE 1000500/19.697.5010.068523660,000/15,850MAG M I X E R M B E S E R I ESD R IVE PE D E STAL 316SS StandardG EAR E D M OTORWhite FDA Compliant Epoxy Paint IP55, IE3 Motor - Voltage to suit ATEX option available IP65, IP66, IP67 availableStainless Steel drive option availableW E LD PLATE Welds to vessel wall(Removable Plug In option available)AG ITATOR I M PE LLE R with integrated NeFeB magnets B EAR I N G SFemale bearing SiC (Silicon Carbide)Male bearing Zr02 (Zirconium Oxide)Options availableD R IVE MAG N ETSO-R I N GR ETAI N I N G B OLT with sealing O-ringD R IVE U N ITSGlobal LocationsS PX FLOW - LI G HTN I N & PLE NTY M I X E R S135 Mt. Read Blvd.Rochester, NY 14611P: +1 (888) 649-2378 (MIX-BEST) (US and Canada) or +1 (585) 436-5550F: +1 (585) 436-5589E:********************•/lightninSPX FLOW, Inc. reserves the right to incorporate our latest design and material changes without notice or obligation. Design features, materials of construction and dimen-sional data, as described in this bulletin, are provided for your information only and should not be relied upon unless confirmed in writing. Please contact your local sales representative for product availability in your region. For more information visit .The green “ ” and “ ” are trademarks of SPX FLOW, Inc.LIG-B-971 Version: 04/2018COPYRIGHT © 2018 SPX FLOW, Inc.AM E R I CAS S PX FLOW135 Mt. Read Blvd.Rochester, NY 14611P: +1 585 436 5550S PX FLOW B RA Z I LRua João Daprat231 São Bernardo do Campo SP 09600-010, Brazil P: +55 11 2127 8278S PX FLOW CH I LESPX FLOW Chile Limitada Av. Ricardo Lyon 222 of. 503Providencia - Santiago, Chile P: +56 2 8969 320S PX FLOW M E X I COAPV Soluciones Integrales, SA de CV Amargura # 60 Primer Piso Col. Lomas de La Herradura 52785 Huixquilucan, Edo de Mexico P: +52 55 5293 9048E M EA S PX FLOWOcean House, T owers Business Park Didsbury, Manchester M20 2YY, UK P: +44 161 249 1170S PX FLOW AFR I CAUnit 12BGrowthpoint Office Park, T onetti Street Midrand, South Africa P: +27 11 207 3700S PX FLOW M I D D LE EAST FZ EP.O Box 299745, Downtown Jebel Ali The Galleries 4Dubai, U.A.E.P: +971 4 8143400APACS PX FLOW PTY LTDSuite 2.3, Quad 28 Parkview DriveHomebush Bay NSW 2127, Australia P: +61 2 9763 4900S PX FLOW CH I NA7F, Nanfung T ower 1568 Hua Shan Road Shanghai 200052, China P: +86 (21) 2208 5888S PX FLOW S I N GAP OR E20 Pioneer Crescent, #06-01West Park BizCentral Singapore 628555P: +65 6264 4366。

碳纤维掺杂酚醛树脂

Wear272 (2011) 43–49Contents lists available at ScienceDirectWearj o u r n a l h o m e p a g e:w w w.e l s e v i e r.c o m/l o c a t e/w e arEffect of the addition of carbonaceousﬁbers on the tribological behavior of a phenolic resin sliding against cast ironS.Betancourt a,∗,L.J.Cruz a,A.Toro ba New Materials Research Group,School of Engineering,Pontiﬁcia Bolivariana University,Circular1a N.70–01,Medellín,Colombiab Tribology and Surfaces Group,School of Materials Engineering,National University of Colombia,Cra80N.65-223,Medellín,Colombiaa r t i c l e i n f oArticle history:Received20January2011Received in revised form12July2011 Accepted22July2011Available online 30 July 2011Keywords:CarbonaceousﬁbersPhenolic resinCompositesSliding contactCast iron a b s t r a c tThe tribological behavior of novolac phenolic resin matrix composites reinforced with three kinds of carbonaceousﬁbers was studied in sliding contact against cast iron.Slow pyrolysis was used to obtain carbonaceousﬁbers from Colombian plantainﬁber bundles(crops residues from Urabáregion).After the carbonization process the samples were heated up to either1200or1400◦C ensuring that many morphological aspects of the naturalﬁbers were retained.Then,novolac phenolic resin with HMTA as curing agent and the carbonaceousﬁbers were used to obtain a composite material by compression molding process.Samples with different type and volume fraction of carbonaceousﬁbers were prepared and tested in sliding contact against cast iron in a pin-on-disc wear testing machine.At the end of the tests,the worn surfaces and the debris were analyzed by SEM.A decrease in both friction coefﬁcient and wear of composites was observed with the increase inﬁber volume fraction,which was associated to a beneﬁcial effect of the detachment of carbonaceous material from the worn surface.Under the tested conditions,this material remains at the interface between the composite and the cast iron and helps reduce the shear resistance of the interface.On the other hand, surface fatigue and adhesion wear was identiﬁed as the dominant wear mechanism of the phenolic resin matrix.© 2011 Elsevier B.V. All rights reserved.1.IntroductionPyrolysis process transforms lignocellulosic materials into char-coal and slow heating rate pyrolysis gives rise to a porous carbon frame with the morphology derived from its precursor[1–3],which can be used to ceramic synthesis such as silicon carbide,alu-mina and titania[3].Typical lignocellulosic precursors of charcoal materials that retain morphological aspects of precursor are wood resources.Although non-wood plants are used to get carbon pow-ders and activated carbon,ﬁbrous wastes from agricultural residues of edibles fruits from Musa species such as commercial plantain (Musa AAB,cv“Dominico Harton”)produced in Colombia have been used to get carbonaceousﬁbers[4–7].The sliding friction coefﬁcients of some carbon-based materi-als,either amorphous or crystalline,are among the lowest for any solids.They are important ingredients of brake composite materi-als including different types such as graphite,coke,carbon black, and carbonﬁber[8].Charcoal and carbonaceous substances,for instance,have good frictional characteristics even though they do not exhibit the basal slip properties of graphitic structures.∗Corresponding author.Tel.:+5744488388;fax:+5743544532.E-mail address:santiago.betancourt@.co(S.Betancourt).On the other hand,novolac phenolic resin is a common binder for resin-based friction materials[9].Tribological applications of phenolic resins are usually limited due to their relatively poor stability and wear resistance.Therefore,it is imperative to incorpo-rate various reinforcing andﬁlling constituents such as reinforcing ﬁbers,abrasives,binders,ﬁllers,and friction modiﬁers(solid lubri-cants)into phenolic resin-based friction composites with the purpose of increasing their stability and wear resistance[10–12].The type and relative amount of solid lubricants and abrasives in brake friction materials signiﬁcantly affect the brake performance [13].Solid lubricants are added in relatively small amounts but they strongly affect wear resistance,stopping distance,friction stability and torque variation.Graphite and MoS2are frequently used in commercial brake linings and other chalcogenide compounds such as Sb2S3,ZnS,PbS,and Cu2S are often added for better brake per-formance.However,a few reports focused on the effect of solid lubricants on tribological properties of solidﬁlms coated on the metal substrates are available[14,15].Charcoal in particular has been reported as a potential replace-ment for graphite into composites for antifriction and antiwear applications as aluminum alloys applications[16,17],wood ceram-ics and in brake friction materials using carbonized coconut char powders[18].Nevertheless,it is rarely discussed in the literature the effect of charcoal materials on wear mechanisms acting on the0043-1648/$–see front matter© 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.wear.2011.07.01444S.Betancourt et al./Wear272 (2011) 43–49surfaces in contact.In this work,novolac phenolic resin matrix composites reinforced with three kinds of carbonaceousﬁbers and different volume fraction of reinforcement were prepared.Chem-ical composition and structural characteristics of carbonaceous ﬁbers were studied.Tribological behavior of composites was tested in sliding contact against cast iron in a pin-on-disc wear testing machine.Worn surfaces were studied by SEM and wear mecha-nisms were identiﬁed.2.Materials and experimental methods2.1.MaterialsCarbonaceousﬁbers were obtained by pyrolysis of plantain ﬁbers from leaf sheaths from Urabá,Colombia region.A mechani-cal extraction process for plantainﬁbers was used according to the method described in[19].The samples were further air-dried for at least24h and milled with a RETSCH SM100(Haan,Germany)to obtain a particle size smaller than5mm.Three types of carbonaceousﬁbers(T1000,T1200,T1400)were synthesized by slow pyrolysis performed in an electrical tube fur-nace MTI GSL1600-80X.T1000samples were heated from room temperature up to1000◦C,held at this temperature for90min and then cooled down to room temperature in N2atmosphere with99.9999%purity.T1200and T1400samples were initially treated under the same conditions used for T1000samples,and then submitted to a second thermal process at1200◦C and1400◦C, respectively.The heating(5◦C min−1)and cooling(−5◦C min−1) rates,as well as the nitrogenﬂow rate(200ml min−1),were the same for all the samples.Phenolic resin powder containing hexamethylenetetramine as curing agent was used as the matrix.Resin and carbonaceousﬁbers were molded by compression molding process.Previously,both components were mixed for5min by stirring them in a blender so as to improve the dispersion of the mixture.Contents of car-bonaceousﬁbers were2.5,5.0,7.5,10and12.5%(v/v)and phenolic resin was the balance.Fully mixed raw material was loaded into a 140mm diameter and3.5mm high steel mold and hot pressed at 200bar,165◦C for15min in a thermo hydraulic pressing machine.2.2.ExperimentalThe carbon,hydrogen,nitrogen and sulphur content of the car-bonaceousﬁber bundles were determined using a LECO-CHNS-932 microanalyzer.Oxygen content was determined by means of a LECO-VTF-900furnace coupled to the microanalyzer.Ash content and moisture were determined for each carbon content in accor-dance with UNE32001and32002standards,respectively.For XRD analysis,carbonaceousﬁbers were ground to aﬁne powder and the measurements were carried out in X’Pert Pro MPD,Panalytical, diffractometer operating at25mA and40kV,using Cu-Ka radiation.Table2Pin-on-disc testing conditions.Tribological test conditionsEnvironmental temperature25±2◦C Atmosphere Air–Relative speed1m s−1 Load15.11NTest distance1000m2.3.Tribological behaviorFriction and wear behavior of the phenolic resin and composites were evaluated by using a pin-on-disc tribometer.A cast iron brake disc(220mm in diameter and8mm in thickness)was adapted for the tests and the surfaceﬁnishing wasﬁxed to circa Ra=0.5␮m in all the samples.Table1presents some of the most relevant metallurgical and chemical characteristics of cast iron disc.Phenolic and composite pins were glued to a metallic pin holder with conventional instant glue and machined by lathe turning to get theﬁnal dimensions shown in Fig.1a.The pin wasﬁxed to a rigid arm(Fig.1b)and put in contact with the surface of the disk in movement,while a normal load was applied by dead weights (Fig.1c).A general view of the equipment is shown in Fig.1d.All samples were tested under the same test conditions summarized in Table2and three replicas were obtained for each experimen-tal condition.Friction force was registered10times per second with the aid of a load cell connected to a data acquisition card and software Labview5.1provided by National Instruments under an educational contract.After each friction test,the pins were ultra-sonically cleaned in alcohol for5min,dried in air and weighed in an analytical balance Sartorius CPA225D with resolving power of 0.01mg.2.4.Surface examinationThe worn surfaces of disks and pins were examined by stere-omicroscopy(Nikon SMZ1500)and scanning electron microscopy (JEOL5910LV)in order to identify the main wear mechanisms act-ing during the tests.3.Results and discussionElemental composition ofﬁbers after pyrolysis process and sec-ond thermal treatment are shown in Table3.All samples have high amount of carbon as a consequence of carbonization of ligno-cellulosic compounds.A second thermal treatment increases the carbon content and reduces the amount of oxygen and hydrogen by volatilization.These changes are considered typical in carbon materials due to exposure to high temperature[20].Carbonaceuosﬁbers produced from plantainﬁbers bundles were crystallographically characterized by means of X-ray diffrac-Table1Metallurgical and chemical characteristics of cast iron disc.Nominal chemical composition(wt.%)C 3.65–3.85Sn Max.0.10 Si 2.15–2.795Cu Max.0.60 S Max.0.15Mo Max.0.10 P Max.0.10Ni Max.0.20 Mn0.5–0.9Fe Balance Cr Max.0.25Microstructure Graphite type:1–1A(sheets) Grain size:ASTM3–4 Pearlite:90%min.Ferrite:5%max. Cementite:5%max.Hardness170–217HBS.Betancourt et al./Wear272 (2011) 43–4945Fig.1.Details of positioning of the samples(a–c)and general aspect of pin-on-disc tribometer(d). tion as seen in Fig.2.All samples presented crystalline andamorphous phases associated to carbonaceous material and crys-tals of inorganic substances.Identiﬁcation of diffraction angles andmineral substance is summarized in Table4as a function of car-bonization temperature.Table3Chemical composition of carbonaceousﬁbers measured by elemental analysis.Element(wt.%)SampleT1000T1200T1400Carbon66.6775.9180.12Hydrogen 1.37 1.21 1.29Nitrogen0.520.530.33Sulphur0.120.090.12Oxygen14.059.568.08Table4Crystallographic analysis of carbonaceousﬁbers(refer to Fig.2).Sample Peak2Â(◦)Mineral/chemical formulaT 1000a29.50Calcite/CaCO3b31.95Butschliite/K2Ca(CO3)2 c33.30d43.64T 1200e31.16Butschliite/K2Ca(CO3)2 f33.32g42.92T 1400h26.35Calciumcarbide/CaC2i32.65T1000samples show peaks corresponding to calcium and potas-sium crystals such as Calcite(CaCO3)and Butschliite(K2Ca2(CO3)3),the latter being a non-conventional phase found in wood ashesaccording to Winbo et al.[20].Amorphous region showsdiffuseFig.2.X-ray diffractograms of samples obtained by different processing parameters.Diffuse bands around2Â=26◦and44◦correspond to(002)and(100)planes of agraphite-like carbon structure.Peaks a to i are identiﬁed in Table4.46S.Betancourt et al./Wear 272 (2011) 43–49Fig.3.Friction force vs.testing time curves.Non-reinforced phenolic resin (a)and samples with 7.5%reinforcement content:T 1000(b),T 1200(c)and T 1400(d).bands at 2Â=26.5◦and 42.4◦,which correspond to typical crys-talline planes of non-graphitic carbon materials (002and 100,respectively)[21–23].Structural changes are observed in T 1200and T 1400samples due to the changes in thermal history.Calcite inorganic mineralpeak disappears and a slight shift of Butschliite peak is observed.Also,T 1400samples show a new crystalline phase identiﬁed as cal-cium carbide,which is not observed in T 1000and T 1200samples.Intensity of carbon 002and 100peaks increased for both sam-ples.Packing of basic structural unities (graphenes)along c axisisFig.4.Coefﬁcient of friction (a)and wear rate (b)of phenolic resin and carbonaceous composites in pin-on-disc tests.S.Betancourt et al./Wear272 (2011) 43–4947Fig.5.Aspect of the worn surface of reinforcement-free resin.SEM.The arrows represent the sliding direction.a thermal activated process which occurs at temperatures higher to1000◦C[24].3.1.Friction coefﬁcient and wear resistanceAfter a running-in period that lasted around100s for all the samples,a stable friction force was measured during testing of both phenolic and composite samples,as can be veriﬁed,for instance,in Fig.3where the friction force vs.testing time curves for non-reinforced phenolic resin samples and T1000,T1200and T1400 samples with7.5%of carbonaceousﬁber content are shown.Results of the measurements of stable friction coefﬁcient dur-ing pin-on-disc tests are shown in Fig.4.All composites exhibit an important reduction of friction coefﬁcient as the carbon content is increased.The maximum relative reduction(near to40%)was observed for12.5%carbon content in T1200samples(Fig.4a). Fig.6.Aspect of the worn surface of composite pin.SEM.The arrows represent the sliding direction.48S.Betancourt et al./Wear 272 (2011) 43–49Fig.7.Aspect of the worn surface of grey iron disc SEM.Arrow represents sliding direction.Darker zones correspond to polymer adhered to the disc’s surface.The reduction in wear rate with the addition of carbonaceous ﬁbers was much more dramatic,with relative variations up to 85%with respect to pure phenolic resin,as seen in Fig.4b.The results did not reveal statistical differences among the wear rates of samples T 1000with 2.5,5.0,and 7.5%of carbonaceous ﬁbers.Similar results were obtained for T 1200composites with the same reinforcement content.According to friction coefﬁcient and wear rate results,the addition of carbonaceous ﬁbers to phenolic resin led to an improve-ment in tribological behavior due to signiﬁcant reduction in wear rates while the friction coefﬁcient values remained quite stable.This behavior can be attributed to the presence of graphite-like crystalline layers into the amorphous structure of carbonaceous material,which were detected by means of X-ray diffraction (see diffuse bands in diffractograms of Fig.2),as well as to the beneﬁ-cial effect of the detachment of carbonaceous ﬁbers that remain trapped between the composite and the cast iron during the tests.3.2.Examination of worn surfacesThe typical aspect of the worn surface of ﬁber-free phenolic resin samples is shown in Fig.5.Plastic deformation is observed up to some extent (Fig.5a and b)and evidences of fatigue-related wear mechanisms are clearly identiﬁed in Fig.5c where a parti-cle of material is about to be detached as a result of the joining of cracks that grew parallel to the contact surface.The character-istic pattern consisting of regularly spaced cracks perpendicular to the sliding direction (Fig.5d)has been previously observed in thermoset polymers sliding against smooth surfaces [25,26].These marks were systematically observed at the worn surface of the ﬁber-free samples and effectively denote the occurrence of fatigue-related phenomena.Crazing evidences can also be observed in Fig.5d,which are normally expected in polymers submitted to sliding wear conditions.In regards to composite samples,worn surface examination revealed a random distribution of carbonaceous ﬁbers,with nopreferred orientation with respect to the sliding direction.Fig.6a shows a low magniﬁcation image of the surface of a typical sam-ple,where a number of features such as holes and incomplete carbonaceous ﬁbers can be observed.Typical vascular plant mor-phology can be observed in carbonaceous ﬁbers,Fig.6b and c,in a region where the ﬁbers were broken and the inner part was exposed.Similarly to what was found in ﬁber-free pheno-lic resin samples,crazing and surface fatigue of the polymer are the dominant wear mechanisms,although some cracks were observed along the ﬁbers as well.A fragment of broken ﬁbers clearly shows the morphology of carbonaceous ﬁbers which retain cellular features from precursor and evidences the fracture of carbonaceous material as one of the active wear mechanisms,Fig.6d.Regarding the analysis of the disc’s surface,evidences of adhe-sion of phenolic resin on the disc were found by SEM as shown in Fig.7.In Fig.7a,a slightly darker region that corresponds to the mark left by the fragments of matrix attached to the asperities of the surface of the disc during sliding contact can be observed.Accord-ing to the detailed examination of the worn surfaces,it can be said that initially the resin partially ﬁlls the valleys of the surface rough-ness of the disc and next it continues to accumulate in the whole contact area (Fig.7b–d).Therefore,two main wear mechanisms are active:adhesive wear and surface fatigue.A sequence of events occurring during sliding of composite sam-ples against cast iron is proposed according to the analysis of worn surfaces (Fig.8):once pin and disc surfaces are put in contact,stresses are produced by effect of the normal load and tangential forces during sliding.The main wear mechanisms of phenolic resin are adhesion and surface fatigue.On the other hand,once the car-bonaceous ﬁbers get in contact with cast iron they crack and suffer brittle fracture.Some fragments remain at the interface between pin and disk so the carbonaceous material is crushed and a ﬁne powder is formed.Then,a ﬁne distribution of carbonaceous parti-cles is spread over the contact interface acting as a solid lubricant that provides the conditions for the stable friction coefﬁcient and reduced wear rate observed in the experiments.S.Betancourt et al./Wear272 (2011) 43–4949Fig.8.Schematic of the sequence of events leading to wear of phenolic matrix composites reinforced with carbonaceousﬁbers in contact with cast iron.4.ConclusionsThe tribological behavior of phenolic resin composites rein-forced with carbonaceousﬁbers sliding against cast iron was studied.A reduction in wear rate of composites was observed with the increase in carbonaceousﬁber volume fraction.Friction coef-ﬁcient also showed a trend to reduce with the increase ofﬁber volume fraction,although the measured values were close to those found in pure phenolic resin samples.Adhesion,surface fatigue and crazing were identiﬁed as the dominant wear mechanisms of the phenolic matrix while brittle fracture was the main cause of detachment of carbonaceousﬁbers.The analysis of the worn surfaces revealed that a signiﬁ-cant amount of carbonaceous material removed from the matrix remained at the interface between the composite pins and metallic discs,which was associated to prevention of adhesive wear of the phenolic matrix and a possible function of carbonaceous material as a solid lubricant.References[1]C.Byrne,D.Nagle,Carbonization of wood for advanced materials applications,Carbon35(1997)259–266.[2]P.Greil,Biomorphous ceramics from lignocellulosics,J.Am.Ceram.Soc.21(2001)105–118.[3]H.Sieber,C.Hoffmann,A.Kaindl,P.Greil,Biomorphic cellular ceramics,Adv.Eng.Mater.2(2000)105–109.[4]S.Betancourt,Desarrollo de materiales compuestos de matriz fenólica conmateriales carbonosos sintetizados a partir de lasﬁbras vegetales obtenidas de la vena central de la hoja de plátano.Ph.D.Thesis.Universidad Pontiﬁcia Bolivariana,Medellín,2010.[5]C.R.Rambo,Materiais avanc¸ados inspirados na natureza:estruturas celulares,ﬁbras e compósitos,Exacta9(2006)5–103.[6]C.Rambo,T.Andrade,T.Fey,H.Sieber,A.E.Martinelli,P.Greil,MicrocellularAl2O3ceramics from wood forﬁlter applications,J.Am.Ceram.Soc.91(2008) 852–859.[7]Z.Weiss,J.C.Crelling,G.Simha Martynkova,M.Valaskova,P.Filip,Identiﬁca-tion of carbon forms and other phases in automotive brake composites using multiple analytical techniques,Carbon44(2006)792–798.[8]Y.Gewen,Y.Fengyuan,Mechanical and tribological properties of phenolicresin-based friction compositesﬁlled with several inorganicﬁllers,Wear262 (2007)121–129.[9]S.K.Rhee,M.G.Jacko,P.H.S.Tsang,The role of frictionﬁlm in friction,wear andnoise of automotive brakes,Wear146(1991)89–97.[10]S.J.Kim,H.Jang,Friction and wear of friction materials containing two differentphenolic resins reinforced with aramid pulp,Tribol.Int.146(2000)477–484.[11]P.Gopal,L.R.Dharani,F.D.Blum,Fade and wear characteristics of a glassﬁberreinforced phenolic friction material,Wear174(1994)119–127.[12]M.H.Cho,J.J.S.J.Kim,H.Jang,Tribological properties of solid lubricants(graphite,Sb2S3,MoS2)for automotive brake friction materials,Wear260 (2006)855–860.[13]W.J.Bartz,J.Xu,Wear behavior and failure mechanism of bonded solid lubri-cants,Lub.Eng.43(1987)514–521.[14]M.N.Gardos,Synergistic effects of graphite on the friction and wear of MoS2ﬁlms in air,Tribol.Trans.31(1988)214–227.[15]T.Wang,T.Fan,D.Zhang,G.Zhang,Fabrication and the wear behaviors ofthe carbon/aluminum composites based on wood templates,Carbon44(2006) 900–906.[16]T.P.Murali,S.V.Prasad,M.K.Surappa,P.K.Rohatgi,K.Gopinath,Friction andwear behaviour of aluminium alloy coconut shell char particulate composites, Wear80(1982)149–158.[17]M.Sutikno,P.Marwoto,S.Rustad,The mechanical properties of carbonizedcoconut char powder-based friction materials,Carbon48(2010)3616–3620.[18]D.Joseph,A.Oberlin,Oxidation of carbonaceous Matter I Elemental analysis(C,H,O)and IR spectrometry,Carbon21(1983)559–564.[19]P.Ganán,R.Zuluaga,A.Restrepo,bidi,Mondragon plantainﬁber bun-dles isolated from Colombian agro-industrial residues,Bioresour.Technol.99 (2008)486–491.[20]C.Winbo,E.Rosen,M.Heimb,Thermal analytical study of the decompositionof K2Ca2(CO3)3,Acta Chem.Scand.(1998)431–434.[21]R.Franklin,Crystallite growth in graphitizing and non-graphitizing carbons,Proc.R.Soc.Lond.(1951)196–218.[22]X.Bourrat,in:H.Marsh,F.Rodríguez-Reinoso(Eds.),Science of Carbon Mate-rials,Ed.Universidad de Alicante,Alicante,2000,pp.1–90.[23]F.G.Emmerich,Evolution with heat treatment of crystallinity in carbons,Car-bon33(1995)1709–1715.[24]G.A.Zickler,T.Schoumlberl,O.Paris,Mechanical properties of pyrolysed wood:a nanoindentation study,Philos.Mag.86(2006)1373–2138.[25]B.Bhushan,Introduction to Tribology,John Wiley&Sons,New York,2002,p.406.[26]M.Terheci,Microscopic investigation on the origin of wear by surface fatiguein dry sliding,Mater.Charact.45(2000)1–15.。

树种调查英文专业术语

树种调查英文专业术语Conducting a Tree Species Investigation: A Comprehensive ExplorationThe world around us is teeming with diverse and fascinating flora, each species uniquely adapted to its environment. Among the most captivating of these natural wonders are the myriad tree species that grace our landscapes, providing not only aesthetic beauty but also vital ecological functions. As professionals in the field of environmental science, it is our responsibility to delve deeper into the intricacies of these arboreal marvels, uncovering their taxonomic classifications, physiological characteristics, and roles within their respective ecosystems.At the heart of this endeavor lies the tree species investigation, a meticulous process of identifying, categorizing, and studying the trees that populate a given region. This endeavor requires a comprehensive understanding of botanical terminology, as well as a keen eye for the subtle differences that distinguish one species from another. By mastering the English language lexicon associated with tree taxonomy, we can unlock a wealth of knowledge and unlock the doors to a deeper appreciation of the natural world.To begin our investigation, we must first familiarize ourselves with the fundamental taxonomic hierarchy that governs the classification of trees. At the broadest level, we have the kingdom Plantae, which encompasses all photosynthetic organisms, including both flowering and non-flowering plants. Within this kingdom, trees are classified under the division Tracheophyta, also known as vascular plants, which possess specialized tissues for the transport of water and nutrients.Delving deeper, we find that trees belong to the class Magnoliopsida, or dicotyledons, characterized by the presence of two seed leaves, or cotyledons, upon germination. This class is further divided into numerous orders, each with its own distinctive features and adaptations. For instance, the order Fagales includes the familiar oak, beech, and chestnut trees, while the order Pinales encompasses the coniferous species, such as pines, firs, and spruces.At the species level, the specific epithet, or scientific name, serves as the primary means of identification. This binomial nomenclature system, pioneered by the renowned Swedish naturalist Carl Linnaeus, assigns each plant a unique two-part name, consisting of the genus and species. For example, the common oak tree is known scientifically as Quercus rubra, where Quercus represents the genus and rubra denotes the specific species.In addition to taxonomic classification, the tree species investigation also delves into the morphological characteristics that distinguish one species from another. These include the shape and arrangement of leaves, the texture and pattern of bark, the structure of the branching pattern, and the overall form and size of the tree. By carefully observing and documenting these features, we can build a comprehensive understanding of the trees within a given ecosystem.Beyond the external features, the physiological attributes of trees also play a crucial role in their identification and classification. Factors such as the presence or absence of needle-like leaves, the arrangement and structure of the vascular system, and the unique chemical compounds produced by the tree can all provide valuable insights into its identity and evolutionary relationships.In the context of a tree species investigation, the study of leaf morphology is particularly important. Characteristics such as leaf shape, venation patterns, and the presence of trichomes (hair-like structures) can be used to differentiate between species. For instance, the maple tree (Acer) is known for its distinctive palmate leaves, while the oak tree (Quercus) typically displays lobed or deeply indented leaf blades.Bark texture and pattern also serve as important identificationmarkers, as the outer layer of a tree's trunk can exhibit a wide range of characteristics, from the smooth, silvery bark of the birch (Betula) to the deeply furrowed, scaly bark of the shagbark hickory (Carya ovata). By familiarizing ourselves with the diverse range of bark textures and patterns, we can enhance our ability to accurately identify tree species in the field.Another crucial aspect of the tree species investigation is the study of the overall tree form, including the branching pattern, the shape of the crown, and the overall height and diameter of the trunk. These characteristics can provide valuable insights into the tree's adaptation to its environment, as well as its stage of growth and developmental maturity. For example, the towering, columnar shape of the bald cypress (Taxodium distichum) is a testament to its ability to thrive in wetland environments, while the broad, spreading canopy of the American sycamore (Platanus occidentalis) is an adaptation to life in open, sunny landscapes.As we delve deeper into the world of tree species, we must also consider the unique chemical compounds and secondary metabolites produced by these remarkable organisms. From the fragrant terpenes of the pine (Pinus) to the medicinal alkaloids found in the bark of the willow (Salix), these chemical signatures can serve as valuable tools in the identification and classification of trees. By understanding the role of these compounds in the tree's overallphysiology and ecological interactions, we can gain a more holistic appreciation for the complex and interconnected nature of the natural world.In the course of our tree species investigation, we may also encounter instances of hybridization, where two distinct species interbreed to produce offspring with characteristics intermediate between the parent plants. This phenomenon can present a unique challenge, as the resulting trees may exhibit a blend of features that do not neatly fit into established taxonomic categories. In such cases, a thorough understanding of genetic analysis and molecular techniques may be necessary to accurately identify the hybrid and its parent species.Throughout the tree species investigation, it is essential to maintain meticulous records and documentation. This includes the creation of detailed field notes, the collection of herbarium specimens, and the compilation of comprehensive photographic documentation. By building a robust database of information, we can not only aid in the identification and classification of tree species, but also contribute to the broader scientific understanding of the natural world.As we embark on this journey of tree species investigation, we must remember that our work extends far beyond the mere cataloging of these remarkable organisms. By deepening our knowledge andappreciation of the trees that surround us, we can unlock insights into the complex web of ecological relationships that sustain our planet. From the vital role of trees in the carbon cycle to their importance as habitats for countless other species, our understanding of these arboreal wonders can inform and shape our efforts to protect and preserve the natural environments that we call home.In conclusion, the tree species investigation is a multifaceted and captivating endeavor that requires a deep understanding of botanical terminology, morphological characteristics, and physiological adaptations. By mastering the English language lexicon associated with tree taxonomy, we can unlock a wealth of knowledge and become more effective stewards of the natural world. Through our continued efforts to identify, classify, and study the diverse tree species that grace our landscapes, we can contribute to the greater scientific understanding of the living world and inspire others to appreciate the beauty and complexity of the natural environment.。

A survey and comparison of tree generation algorithms

George Mason University /∼sean/Liviu PanaitGeorge Mason University /∼lpanait/AbstractThis paper discusses and comparesﬁve majortree-generation algorithms for genetic program-ming,and their effects onﬁtness:RAMPEDHALF-AND-HALF,PTC1,PTC2,RANDOM-BRANCH,and UNIFORM.The paper comparesthe performance of these algorithms on three ge-netic programming problems(11-Boolean Multi-plexer,Artiﬁcial Ant,and Symbolic Regression),and discovers that the algorithms do not have asigniﬁcant impact onﬁtness.Additional experi-mentation shows that tree size does have an im-portant impact onﬁtness,and further that theideal initial tree size is very different from thatused in traditional GP.1INTRODUCTIONThe issue of population initialization has received surpris-ingly little attention in the genetic programming literature. [Koza,1992]established the GROW,FULL,and RAMPED HALF-AND-HALF algorithms,only a few papers have ap-peared on the subject,and the community by and large still uses the original Koza algorithms.Some early work was concerned with algorithms simi-lar to GROW but which operated on derivation grammars. [Whigham,1995a,b,1996]analyzed biases due to popula-tion initialization,among other factors,in grammatically-based genetic programming.[Geyer-Schulz,1995]also de-vised similar techniques for dealing with tree grammars. Theﬁrst approximately uniform tree generation algorithm was RAND-TREE[Iba,1996],which used Dyck words to choose uniformly from all possible tree structures of a given arity set and tree size.Afterwards the tree structure would be populated with nodes.[Bohm and Geyer-Schulz, 1996]then presented an exact uniform algorithm for choos-ing among all possible trees of a given function set.Recent tree generation algorithms have focused on speed. [Chellapilla,1997]devised RANDOMBRANCH,a simple al-gorithm which generated trees approximating a requested tree size.After demonstrating problems with the GROW al-gorithm,[Luke,2000b]modifed GROW to produce PTC1 which guaranteed that generated trees would appear around an expected tree size.[Luke,2000b]also presented PTC2 which randomly expanded the tree horizon to produce trees of approximately the requested size.All three of these al-gorithms are linear in tree size.Both[Iba,1996]and[Bohm and Geyer-Schulz,1996]ar-gued for the superiority of their algorithms over the Koza standard algorithms.[Whigham,1995b]showed that bias-ing a grammar-based tree-generation algorithm could dra-matically improve(or hurt)the success rate of genetic pro-gramming at solving a given domain,though such bias must be hand-tuned for the domain in question.In contrast,this paper examines several algorithms to see if any of the existing algorithms appears to make much of a difference,or if tree size and other factors might be more signiﬁcant.2THE ALGORITHMSThis paper comparesﬁve tree generation algorithms from the literature.These algorithms were chosen for their widely differing approaches to tree creation.The chief al-gorithm not in this comparison is RAND-TREE[Iba,1996]. This algorithm has been to some degree subsumed by a more recent algorithm[Bohm and Geyer-Schulz,1996], which generates trees from a truly uniform distribution(the original unachieved goal of RAND-TREE).The algorithms discussed in this paper are:2.1Ramped Half-And-Half and Related Algorithms RAMPED HALF-AND-HALF is the traditional tree gener-ation algorithm for genetic programming,popularized by[Koza,1992].RAMPED HALF-AND-HALF takes a tree depth range(commonly2to6–for this and future refer-ences,we deﬁne“depth”in terms of number of nodes,not number of edges).In other respects,the user has no control over the size or shape of the trees generated.RAMPED HALF-AND-HALFﬁrst picks a random value within the depth range.Then with1/2probability it uses the GROW algorithm to generate the tree,passing it the cho-sen depth;otherwise it uses the FULL algorithm with the chosen depth.GROW is very simple:GROW(depth d,max depth D)Returns:a tree of depth≤D−dIf d=D,return a random terminalElse◦Choose a random function or terminal fIf f is a terminal,return fElseFor each argument a of f,Fill a with GROW(d+1,D)Return f withﬁlled argumentsGROW is started by passing in0for d,and the requested depth for D.FULL differs from GROW only in the line marked with a◦.On this line,FULL chooses a nonter-minal function only,never a terminal.Thus FULL only creates full trees,and always of the requested depth. Unlike other algorithms,because it does not have a size pa-rameter,RAMPED HALF-AND-HALF does not have well-deﬁned computational complexity in terms of size.FULL always generates trees up to the depth bound provided.As [Luke,2000b]has shown,GROW without a depth bound may,depending on the function set,have an expected tree size of inﬁnity.2.2PTC1PTC1[Luke,2000b]is a modiﬁcation of the GROW algo-rithm which is guaranteed to produce trees around aﬁnite expected tree size.A simple version of PTC1is described here.PTC1takes a requested expected tree size and a max-imum legal depth.PTC1begins by computing p,the prob-ability of choosing a nonterminal over a terminal in order to maintain the expected tree size E as:p=1−1n∈N1Problem Domain Algorithm Parameter Avg.Tree Size11-Boolean Multiplexer RAMPED HALF-AND-HALF(No Parameter)21.211-Boolean Multiplexer RANDOMBRANCH Max Size:4520.011-Boolean Multiplexer PTC1Expected Size:920.911-Boolean Multiplexer PTC2Max Size:4021.411-Boolean Multiplexer UNIFORM-even Max Size:4221.811-Boolean Multiplexer UNIFORM-true Max Size:2120.9Artiﬁcial Ant RAMPED HALF-AND-HALF(No Parameter)36.9Artiﬁcial Ant RANDOMBRANCH Max Size:9033.7Artiﬁcial Ant PTC1Expected Size:1238.5Artiﬁcial Ant PTC2Max Size:6735.3Artiﬁcial Ant UNIFORM-even Max Size:6533.9Artiﬁcial Ant UNIFORM-true Max Size:3736.8Symbolic Regression RAMPED HALF-AND-HALF(No Parameter)11.6Symbolic Regression RANDOMBRANCH Max Size:2111.4Symbolic Regression PTC1Expected Size:410.9Symbolic Regression PTC2Max Size:1811.1Symbolic Regression UNIFORM-even Max Size:1911.2Symbolic Regression UNIFORM-true Max Size:1110.8Table1:Tree Generation Parameters and Resultant Sizes RANDOMBRANCH(requested size s)Returns:a tree of size≤sIf a nonterminal with arity≤s does not existReturn a random terminalElseChoose a random nonterminal n of arity≤sLet b n be the arity of nFor each argument a of n,Fill a with RANDOMBRANCH( sUNIFORM-even,described later).It is our opinion that the“uniformity”of sampling among theﬁve algorithms presented is approximately in the following order(from most uniform to least):UNIFORM(of course),PTC2, RAMPED HALF-AND-HALF,PTC1,RANDOMBRANCH. The comparisons were done over three canonical genetic programming problem domains,11-Boolean Multiplexer, Artiﬁcial Ant,and Symbolic Regression.Except for the tree generation algorithm used,these domains followed the parameters deﬁned in[Koza,1992],using tournament se-lection of size7.The goal of11-Boolean Multiplexer is to evolve a boolean function on eleven inputs which per-forms multiplexing on eight of those inputs with regard to the other three.The goal of the Artiﬁcial Ant problem is to evolve a simple robot ant algorithm which follows a trail of pellets,eating as many pellets as possible before time runs out.Symbolic Regression tries to evolve a symbolic math-ematical expression which bestﬁts a training set of data points.To perform this experiment,we did50independent runs for each domain using the RAMPED HALF-AND-HALF algo-rithm to generate initial trees.From there we measured the mean initial tree size and calibrated the other algorithms to generate trees of approximately that size.This calibra-tion is not as simple as it would seem atﬁrst.For example, PTC1can be simply set to the mean value,and it should produce trees around that mean.However,an additional complicating factor is involved:duplicate u-ally genetic programming rejects duplicate copies of the same individual,in order to guarantee that every initial in-dividual is unique.Since there are fewer small trees than large ones,the likelihood of a small tree being a duplicate is correspondingly much larger.As a result,these algo-rithms will tend to produce signiﬁcantly larger trees than would appear atﬁrst glance if,as was the case in this exper-iment,duplicate rejection is part of the mix.Hence some trial and error was necessary to establish the parameters re-quired to produce individuals of approximately the same mean size as RAMPED HALF-AND-HALF.Those param-eters are shown in Table1.In the PTC1algorithm,the parameter of consequence is the expected mean tree size.For the other algorithms,the parameter is the“maximum tree size”.For PTC2,RAN-DOMBRANCH,and UNIFORM-even,a tree is created by ﬁrst selecting an integer from the range1to the maximum tree size inclusive.This integer is selected uniformly from this range.In UNIFORM-true however,the integer is se-lected according to a probability distribution deﬁned by the number of trees of each size in the range.Since there are far more trees of size10than of1for example,10is chosen much more often than1.For each remaining algorithm,50 independent runs were performed with both problem do-Fisher LSD Algorithm TukeyUNIFORM-trueRANDOMBRANCHFigures7,14,and21show the results of this experiment. The light gray dots represent each run.The dark gray dots represent the means of the30runs for each maximum-size value.Because of duplicate rejection,runs typically have mean initial tree sizes somewhat different from the values predicted by the provided maximum-size.Also note the ap-parent horizontal lines in the11-Boolean Multiplexer data: this problem domain has the feature that certain discreteﬁt-ness values(multiples of32)are much more common than others.These graphs suggest that the optimal initial tree size for UNIFORM-even for both domains is somewhere around pare this to the standard tree sizes which occur due to RAMPED HALF-AND-HALF:21.2for11-Boolean Multiplexer and36.9for Artiﬁcial Ant!5CONCLUSIONThe tree generation algorithms presented provide a variety of advantages for GP researchers.But the evidence in this paper suggests that improvedﬁtness results is probably not one of those advantages.Why then pick an algorithm over RAMPED HALF-AND-HALF then?There are several rea-sons.First,most new algorithms permit the user to specify the size desired.For certain applications,this may be a cru-cial feature,not the least because it allows the user to cre-ate a size distribution more likely to generate good initial individuals.Fighting bloat in subtree mutation also makes size-speciﬁcation a desirable trait.Second,some algorithms have special features which may be useful in different circumstances.For example,PTC1 and PTC2have additional probabilistic features not de-scribed in the simpliﬁed forms in this paper.Both algo-rithms permit users to hand-tune exactly the likelihood of appearance of a given function in the population,for exam-ple.The results in this paper were surprising.Uniformity ap-pears to have little consequence in improvingﬁtness.Cer-tainly this area deserves more attention to see what addi-tional features,besides mean tree size,do give evolution that extra push during the initialization stly,while this paper discussed effects onﬁtness,it did not delve into the effects of these algorithms on tree growth,another crit-ical element in the GP puzzle,and a worthwhile study in its own right.AcknowledgementsThe authors wish to thank Ken DeJong,Paul Wiegand,and Jeff Bassett for their considerable help and insight.ReferencesWalter Bohm and Andreas Geyer-Schulz.Exact uniform initialization for genetic programming.In Richard K. Belew and Michael V ose,editors,Foundations of Ge-netic Algorithms IV,pages379–407,University of San Diego,CA,USA,3–5August1996.Morgan Kaufmann. Kumar Chellapilla.Evolving computer programs without subtree crossover.IEEE Transactions on Evolutionary Computation,1(3):209–216,September1997. Andreas Geyer-Schulz.Fuzzy Rule-Based Expert Systems and Genetic Machine Learning,volume3of Studies in Fuzziness.Physica-Verlag,Heidelberg,1995.Hitoshi Iba.Random tree generation for genetic program-ming.In Hans-Michael V oigt,Werner Ebeling,Ingo Rechenberg,and Hans-Paul Schwefel,editors,Parallel Problem Solving from Nature IV,Proceedings of the In-ternational Conference on Evolutionary Computation, volume1141of LNCS,pages144–153,Berlin,Germany, 22-26September1996.Springer Verlag.John R.Koza.Genetic Programming:On the Program-ming of Computers by Means of Natural Selection.MIT Press,Cambridge,MA,USA,1992.Sean Luke.ECJ:A Java-based evolutionary compu-tation and genetic programming system.Available at /projects/plus/ecj/,2000a.Sean Luke.Two fast tree-creation algorithms for genetic programming.IEEE Transactions in Evolutionary Com-putation,4(3),2000b.P.A.Whigham.Grammatically-based genetic program-ming.In Justinian P.Rosca,editor,Proceedings of the Workshop on Genetic Programming:From Theory to Real-World Applications,pages33–41,Tahoe City,Cal-ifornia,USA,9July1995a.P.A.Whigham.Inductive bias and genetic programming. In A.M.S.Zalzala,editor,First International Con-ference on Genetic Algorithms in Engineering Systems: Innovations and Applications,GALESIA,volume414, pages461–466,Shefﬁeld,UK,12-14September1995b. IEE.P.A.Whigham.Search bias,language bias,and genetic programming.In John R.Koza,David E.Goldberg, David B.Fogel,and Rick L.Riolo,editors,Genetic Pro-gramming1996:Proceedings of the First Annual Con-ference,pages230–237,Stanford University,CA,USA, 28–31July1996.MIT Press.1020304050Generation200400600800F i t n e s sFigure 1:Generation vs.Fitness,RAMPED HALF-AND-HALF ,11-Boolean Multiplexer Domain1020304050Generation2004006008001000F i t n e s sFigure 2:Generation vs.Fitness,PTC1,11-Boolean Mul-tiplexer Domain01020304050Generation2004006008001000F i t n e s sFigure 3:Generation vs.Fitness,PTC2,11-Boolean Mul-tiplexer Domain01020304050Generation2004006008001000F i t n e s sFigure 4:Generation vs.Fitness,RANDOMBRANCH ,11-Boolean Multiplexer Domain01020304050Generation2004006008001000F i t n e s sFigure 5:Generation vs.Fitness,UNIFORM-even ,11-Boolean Multiplexer Domain1020304050Generation2004006008001000F i t n e s sFigure 6:Generation vs.Fitness,UNIFORM-true ,11-Boolean Multiplexer Domain1020304050Mean Initial Tree Size400500600700800B e s t F i t n e s s o f R u nFigure 7:Mean Initial Tree Size vs.Fitness at Generation 8,11-Boolean Multiplexer Domain1020304050Generation10203040506070F i t n e s sFigure 8:Generation vs.Fitness,RAMPED HALF-AND-HALF ,Artiﬁcial Ant Domain1020304050Generation1020304050607080F i t n e s sFigure 9:Generation vs.Fitness,PTC1,Artiﬁcial Ant Do-main1020304050Generation1020304050607080F i t n e s sFigure 10:Generation vs.Fitness,PTC2,Artiﬁcial Ant Domain1020304050Generation1020304050607080F i t n e s sFigure 11:Generation vs.Fitness,RANDOMBRANCH ,Arti-ﬁcial Ant Domain1020304050Generation1020304050607080F i t n e s sFigure 12:Generation vs.Fitness,UNIFORM-even ,Arti-ﬁcial Ant Domain1020304050Generation1020304050607080F i t n e s sFigure 13:Generation vs.Fitness,UNIFORM-true ,Arti-ﬁcial Ant Domain1020304050Mean Initial Tree Size1020304050607080B e s t F i t n e s s o f R u nFigure 14:Mean Initial Tree Size vs.Fitness at Generation 8,Artiﬁcial Ant Domain1020304050Generation0.511.522.533.5F i t n e s sFigure 15:Generation vs.Fitness,RAMPED HALF-AND-HALF ,Symbolic Regression Domain1020304050Generation0.511.522.533.54F i t n e s sFigure 16:Generation vs.Fitness,PTC1,Symbolic Re-gression Domain1020304050Generation0.511.522.533.54F i t n e s sFigure 17:Generation vs.Fitness,PTC2,Symbolic Re-gression Domain1020304050Generation0.511.522.533.54F i t n e s sFigure 18:Generation vs.Fitness,RANDOMBRANCH ,Symbolic Regression Domain1020304050Generation0.511.522.533.54F i t n e s sFigure 19:Generation vs.Fitness,UNIFORM-even ,Symbolic Regression Domain1020304050Generation0.511.522.533.54F i t n e s sFigure 20:Generation vs.Fitness,UNIFORM-true ,Symbolic Regression DomainMean Initial Tree Size0.511.522.53M e a n F i t n e s s o f R u nFigure 21:Mean Initial Tree Size vs.Fitness at Generation 8,Symbolic Regression Domain。

IC设计flow简介

1. Architectural and electrical specification.2. RTL(Register Transfer Level) coding in HDL(Hardware Description Language).3. DFT(Design For Test) memory BIST(Built In Self Test) insertion, for designs containing memory elements.4. Exhaustive dynamic simulation of the design, in order to verify the functionality of the design.5. Design environment setting. This includes the technology library to be used, along with other environmental attributes.6. Constraining and synthesizing the design with scan insertion (and optional JTAG) using Design Compiler.7. Block level static timing analysis, using Design Compiler’s built-in static timing analysis engine.8. Formal verification of the design. RTL compared against the synthesized netlist, using Formality.9. Pre-layout static timing analysis on the full design through PrimeTime.10. Forward annotation of timing constraints to the layout tool.11. Initial floorplanning with timing driven placement of cells, clock tree insertion and global routing12. Transfer of clock tree to the original design (netlist) residing in Design Compiler.13. In-place optimization of the design in Design Compiler.14. Formal verification between the synthesized netlist and clock tree inserted netlist, using Formality.15. Extraction of estimated timing delays from the layout after the global routing step (step 11).16. Back annotation of estimated timing data from the global routed design, to PrimeTime.17. Static timing analysis in PrimeTime, using the estimated delays extracted after performing global route.18. Detailed routing of the design.19. Extraction of real timing delays from the detailed routed design.20. Back annotation of the real extracted timing data to PrimeTime.21. Post-layout static timing analysis using PrimeTime.22. Functional gate-level simulation of the design with post-layout timing (if desired).23. Tape out after LVS(Layout Versus Schematic) and DRC(Design Rule Checking) verification.。

《农业科学研究》2023_年总目次

Effect of exogenous rapeseed lactone on photosynthetic characteristics and fruit quality of Merlot grape
……………… Liu Yan，Qiao Zichun，Yin Mengting，Guo Xueliang，Wang Yuening，He Yan，Dai Hongjun，Wang Zhenping（1，33）
辉，梁晓珊，王雪妍，高
瑞，谢玉杰，许立华（2，1）
施钾量对滴灌水肥一体化下春玉米钾吸收及产量的影响
…………………………………………………………………王晓苹，康建宏，田仲红，王
磷钾肥配施对卷丹百合鳞茎活性成分积累的影响 …………卜虎柏，王云霞，张
萍，杨
佳，慕瑞瑞，徐
英，王
涵，靳
灿（2，6）
磊（2，12）
植物生长调节剂对切花小菊瓶外生根及外观品质的影响
…………………………………………………………罗
艳，
锁
岚，
虎
瑛，
马蓉蓉，
，61）

专论与综述

农民多层次幸福感的测度与分解研究——基于甘肃和宁夏的社会调查数据
…………………………………………………………………………………………… 李宝军，陈秋霖，王博（1，57）
Correlation analysis between agronomic traits and yield per plant of asparagus lettuce in Liupanshan region
……………………………………………………… Wu Lixiao，Cao Shaona，Zhang Jianhu，Wang Kexiong，Guan Yaobing（1，33）

小学上册第9次英语第5单元综合卷(有答案)

小学上册英语第5单元综合卷(有答案)英语试题一、综合题(本题有100小题，每小题1分，共100分.每小题不选、错误，均不给分)1.What do we call the process of a caterpillar turning into a butterfly?A. MetamorphosisB. EvolutionC. TransformationD. Development 答案: A. Metamorphosis2.Helium was first discovered in the ______ spectrum.3.The capital of Indonesia is _______.4. A ____ has large, flapping ears and can hear very well.5.What do we call the stars and planets in the sky?A. UniverseB. Solar SystemC. GalaxyD. Atmosphere答案: A6.The _______ (The fall of the Berlin Wall) marked the end of Communist control in Eastern Europe.7.My friend is very ________.8.When it snows, I enjoy making __________ with my friends. (雪人)9.What is the main purpose of a refrigerator?A. To heat foodB. To cool foodC. To cook foodD. To freeze food答案: B10. A _____ (植物研究合作) can lead to groundbreaking discoveries.11.The __________ is a natural wonder located in the United States. (黄石公园)12.Turtles can live for a ______ (很长的时间).13.My brother is __________ (富有想象力).14. A ____(mixed-use development) combines residential and commercial spaces.15.What is the name of the famous ancient ruins in Mexico?A. TeotihuacanB. Machu PicchuC. Angkor WatD. Petra答案: A16.We visit the ______ (自然史博物馆) to learn about fossils.17.The discovery of ________ changed the course of history.18. A dolphin leaps gracefully out of the _______ and splashes down again.19.I enjoy playing ________ with my family.20.I like to ___ (play/watch) games.21.What do we call a young female goat?A. KidB. CalfC. LambD. Foal答案:A.Kid22.My friend is __________ (聪明绝顶).23.The _______ can change its shape with the seasons.24.The _____ (养分) in the soil is vital for plant health.25.What is the term for a young goat?A. CalfB. KidC. LambD. Foal答案: B26.An electric motor converts electrical energy into _______ energy.27.Animals that have scales are typically __________.28.The capital of Bonaire is __________.29.My favorite animal is a ______ (dolphin).30. A __________ is a reaction that involves a change in temperature.31.The first successful cloning of a mammal was of _____.32.I like to go ________ (爬山) with my friends.33.The ______ (小鸟) builds a nest for its eggs.34.My _____ (仓鼠) runs on its wheel.35.The ______ helps us learn about communication.36.The painting is very ___ (colorful).37.I often visit my ____.38.I can see a ______ in the sky. (bird)39. A strong acid has a pH less than ______.40.The atomic number of an element tells you the number of _____ (protons) it has.41.What do we call the part of the brain that controls balance?A. CerebellumB. CerebrumC. BrainstemD. Cortex答案:A42.The __________ is a famous natural landmark in the United States. (黄石公园)43.The capital of Ecuador is __________.44.The iguana is often seen basking in the ______ (阳光).45.The __________ (农业) is important for our economy.46.The ______ (小龙) is a mythical creature often found in ______ (故事).47.What is the term for a baby capybara?A. PupB. KitC. CalfD. Hatchling答案:c48.The fish swims in the ___. (water)49.The chemical formula for calcium chloride is ______.50.The ancient Romans practiced ________ (宗教多元).51.I want to _____ (go/stay) at home.52.The speed of light is very ______.53.What do we call a baby dog?A. KittenB. PuppyC. CalfD. Chick答案:B54.The chemical formula for yttrium oxide is _____.55.The Earth's surface is shaped by both climatic and ______ factors.56.Understanding plant _____ (结构) helps in gardening.57.The _____ (spoon) is shiny.58.The _____ (温带雨林) hosts a variety of plant species.59.The balloon is ______ (floating) in the air.60.The river is ______ (calm) and clear.61. A solution with a pH of contains more ______ than a solution with a pH of .62. A ____ is a large animal that can be trained to work.ets are made of ice, dust, and ______.64.__________ are used in the beauty industry for skincare.65.The _____ is a phenomenon where the moon blocks the sun.66.My cat enjoys the warmth of the _______ (阳光).67.The __________ is important for keeping bones strong.68.The __________ is the area of land between two rivers.69.The __________ (历史的深度剖析) reveals nuances.70.Certain plants can ______ (提供) habitat for endangered species.71. A _______ can measure the amount of energy consumed by a device.72.The ________ was a significant treaty that fostered diplomatic relations.73.The chemical symbol for silver is ________.74.I like to draw pictures of my ________ (玩具名) and imagine their adventures.75.I share my toys with my ______. (我和我的______分享玩具。

Carbon Sequestration in Forests

Carbon Sequestration in ForestsRoss W. GorteSpecialist in Natural Resources PolicyAugust 6, 2009Congressional Research Service7-5700RL31432 CRS Report for CongressSummaryWidespread concern about global climate change has led to interest in reducing emissions of carbon dioxide (CO2) and, under certain circumstances, in counting additional carbon absorbed in soils and vegetation as part of the emissions reductions. Congress may consider options to increase the carbon stored (sequestered) in forests as it debates this and related issues.Forests are a significant part of the global carbon cycle. Plants use sunlight to convert CO2, water, and nutrients into sugars and carbohydrates, which accumulate in leaves, twigs, stems, and roots. Plants also respire, releasing CO2. Plants eventually die, releasing their stored carbon to the atmosphere quickly or to the soil where it decomposes slowly and increases soil carbon levels. However, little information exists on the processes and diverse rates of soil carbon change.How to account for changes in forest carbon has been contentious. Land use changes—especially afforestation and deforestation—can have major impacts on carbon storage. Foresters often cut some vegetation to enhance growth of desired trees. Enhanced growth stores more carbon, but the cut vegetation releases CO2; the net effect depends on many factors, such as prior and subsequent growth rates and the quantity and disposal of cut vegetation. Rising atmospheric CO2 may stimulate tree growth, but limited availability of other nutrients may constrain that growth.In this context, timber harvesting is an especially controversial forestry practice. Some argue that the carbon released by cutting exceeds the carbon stored in wood products and in tree growth by new forests. Others counter that old-growth forests store little or no additional carbon, and that new forest growth and efficient wood use can increase net carbon storage. The impacts vary widely, and depend on many factors, including soil impacts, treatment of residual forest biomass, proportion of carbon removed from the site, and duration and disposal of the products. To date, the quantitative relationships between these factors and net carbon storage have not been established.Some observers are concerned that “leakage” will undermine any U.S. efforts to sequester carbon by protecting domestic forests. By leakage, they mean that wood supply might shift to other sites, including other countries, exacerbating global climate change and causing other environmental problems, or that wood products might be replaced by other products that use more energy to manufacture (thus releasing more CO2). Others counter that the “leakage” arguments ignore the enormous disparity in ecological systems and product preferences, and discount possible technological solutions.Several federal government programs affect forestry practices and thus carbon sequestration. Activities in federal forests affect carbon storage and release; timber harvesting is the most controversial such activity. Federal programs also provide technical and financial help for managing and protecting private forests, and tax provisions affect private forest management.V arious federal programs can also affect the extent of forested area, by supporting development (which may cause deforestation) or encouraging tree planting in open areas, such as pastures.ContentsBackground: Congressional Interest in Carbon Sequestration (1)Carbon Cycling in Forests (2)The Forest Cycle (3)Forest Types (4)Tropical Forests (5)Temperate Forests (6)Boreal Forests (6)Measuring and Altering Forest Carbon Levels (6)Forest Carbon Accounting (7)Land Use Changes (8)Forestry Events and Management Activities (9)Vegetation and Soil Carbon (9)Forest Events—Wildfires (11)Forestry Practices (12)Wood Energy (16)Leakage (17)Land Use Leakage (17)Product Demand Leakage (18)Federal Government Programs (19)Federal Forests (19)Federal Assistance for State and Private Forestry (20)Federal Tax Expenditures (20)Federal Programs Affecting Land Use (21)Accounting for Forest Carbon Sequestration (22)Conclusions (22)TablesTable 1. Average Carbon Stocks for V arious Biomes (5)ContactsAuthor Contact Information (23)Global climate change is a widespread and growing concern that has led to extensiveinternational discussions and negotiations.1 Responses to this concern have focused onreducing emissions of greenhouse gases, especially carbon dioxide, and on measuring carbon absorbed by and stored in forests, soils, and oceans. One option for slowing the rise of greenhouse gas concentrations in the atmosphere, and thus possible climate change, is to increase the amount of carbon removed by and stored in forests. As Congress debates climate change and options for addressing the issue, ideas for increasing carbon sequestration in forests are likely to be discussed.This report examines basic questions concerning carbon sequestration in forests. The first section provides a brief background on congressional interest in forest carbon sequestration. The second describes the basic carbon cycle in forests, with an overview of how carbon cycling and storage vary among different types of forests. The third section then addresses how forest carbon is considered in the global climate change debate. This third section begins with an overview of accounting for forest carbon, then discusses the carbon consequences of forest management practices, the effects of changes in land use, and “leakage.” The section then concludes with a summary of existing federal programs that could affect forest carbon sequestration.2Background: Congressional Interest inCarbon SequestrationThe widespread and growing concern over global climate change has led to extensive international negotiations. In 1992, at the Earth Summit in Rio de Janeiro, the United Nations Framework Convention on Climate Change (which included voluntary pledges to reduce greenhouse gas emissions) was opened for signature. President George H. W. Bush signed this treaty, which was then ratified by the U.S. Senate.Subsequent negotiations led to the 1997 Kyoto Protocol, under which the developed nations agreed to specified reductions in their emissions of greenhouse gases. President Clinton signed the Kyoto Protocol, but did not submit it to the Senate for ratification. Early in 2001, the George W. Bush Administration decided to reject the Kyoto Protocol, and withdrew from active participation in negotiations on the issues that remain to be resolved.3 Although enough other parties have ratified the Protocol to bring it into force, the lack of U.S. involvement means that the United States will not participate in the emissions trading or other elements of the Kyoto Protocol activities that might relate to carbon sequestration.The most voluminous greenhouse gas produced by humans is carbon dioxide (CO2). In calculating overall carbon emissions, the Protocol allows certain removals of carbon by a nation’s forests and soils—“carbon sinks”—to be counted and deducted from emissions. Thus, one option1 This report does not address underlying questions of whether global warming is occurring or of the possible human role. See CRS Report RL33849, Climate Change: Science and Policy Implications, by Jane A. Leggett.2 This report does not address the impacts of climate change on forests, although this is also an important scientific issue. For a discussion of these impacts, see the series of articles in “A Special Section on Climate Change and Forest Ecosystems,” Bioscience, v. 51, no. 9 (Sept. 2001): 720-779.3 See CRS Report RL33826, Climate Change: The Kyoto Protocol, Bali “Action Plan,” and International Actions, by Jane A. Leggett.for mitigating greenhouse gas emissions—and thus possible climate change—is to increase the amount of carbon stored in forests.Carbon sequestration, and the extent to which it can be counted as a reduction in a nation’s carbon emissions, have been the focus of substantial controversy in international negotiations subsequent to the Kyoto Protocol.4 The United States, with its extensive forests, argued that the carbon absorbed by them should be allowed to offset emissions, with no quantitative limit to the amounts that can be counted in this way. The European Union argued strongly in negotiations prior to 2001 that there should be fairly strict limits on how much carbon absorbed by “sinks” such as forests could be counted against emissions. In final negotiations during 2001 on rules to implement the Kyoto Protocol, after the United States had withdrawn from the negotiations, a compromise was reached that allows significant credit for carbon sinks (removals and storage of carbon). The Members of Congress attending the negotiations prior to 2001 followed this debate with interest, and were aware of the potential impacts of the various possible results of the negotiations. In particular, if emissions trading were to begin under the Kyoto Protocol, forest owners and managers in countries that were parties to the treaty might be able to receive credits and participate in the trading regime.Administration and congressional interest in carbon sequestration continues, but U.S. participation in the Kyoto process is moot at this time. It is not clear whether a domestic forest carbon program might be established, although options have been discussed in legislative proposals. Protecting forests in developing countries, which might earn credits under the Kyoto Protocol, is already supported under the Tropical Forest Conservation Act (P.L. 105-214; 22U.S.C. §§2341, et seq.).5Mitigating climate change by enhancing forest carbon sequestration may be a relatively low-cost option and would likely yield other environmental benefits. However, forest carbon sequestration faces challenges: measuring the additional carbon stored (over and above what would occur naturally); monitoring and verifying the results; and preventing leakage. Numerous issues regarding the carbon cycle in forests, monitoring the levels and changes in forest carbon, and the scientific uncertainties about the relationships among forests, carbon, and climate change are likely to be the subject of ongoing federal research efforts, with funding and oversight by the Congress.Carbon Cycling in ForestsPhotosynthesis is the chemical process by which plants use sunlight to convert nutrients into sugars and carbohydrates. Carbon dioxide (CO2) is one of the nutrients essential to building the organic chemicals that comprise leaves, roots, and stems. All parts of a plant—the stem, limbs and leaves, and roots—contain carbon, but the proportion in each part varies enormously, depending on the plant species and the individual specimen’s age and growth pattern. Nonetheless, as more photosynthesis occurs, more CO2 is converted into biomass, reducing carbon in the atmosphere and sequestering (storing) it in plant tissue (vegetation) above and below ground.4 See CRS Report RS22806, The Bali Agreements and Forests, by Ross W. Gorte and Pervaze A. Sheikh.5 See CRS Report RL31286, Debt-for-Nature Initiatives and the Tropical Forest Conservation Act: Status and Implementation, by Pervaze A. Sheikh.Plants also respire, using oxygen to maintain life and emitting CO2 in the process. At times (e.g., at night and during winter seasons in non-tropical climates), living, growing forests are net emitters of CO2, although they are generally net carbon sinks over the life of the forest.When vegetation dies, carbon is released to the atmosphere. This can occur quickly, as in a fire,6 or slowly, as fallen trees, leaves, and other detritus decompose. For herbaceous plants, the above-ground biomass dies annually and begins to decompose right away, but for woody plants, some of the above-ground biomass continues to store carbon until the plant dies and decomposes. This is the essence of the carbon cycle in forests—net carbon accumulation (sequestration) with vegetative growth, and release of carbon when the vegetation dies. Thus, the amount of carbon sequestered in a forest is constantly changing with growth, death, and decomposition of vegetation.In addition to being sequestered in vegetation, carbon is also sequestered in forest soils. Carbon is the organic content of the soil, generally in the partially decomposed vegetation (humus) on the surface and in the upper soil layers, in the organisms that decompose vegetation (decomposers), and in the fine roots.7 The amount of carbon in soils varies widely, depending on the environment and the history of the site. Soil carbon accumulates as dead vegetation is added to the surface and decomposers respond. Carbon is also “injected” into the soil as roots grow (root biomass increases). Soil carbon is also slowly released to the atmosphere as the vegetation decomposes. Scientific understanding of the rates of soil carbon accumulation and decomposition is currently not sufficient for predicting changes in the amount of carbon sequestered in forest soils.The Forest CycleForests generally go through cycles of growth and death, sequestering and releasing carbon. Some forests begin on spacious sites, with little or no existing vegetation, that may have been cleared by a natural disaster (most commonly wildfire) or by human activities (e.g., for agriculture). Other forests are relatively continuous, with natural clearings typically limited to the area occupied by one or a few large trees killed by lightning, disease, and such. Regardless of the size or origin of a clearing, most forests begin from essentially bare land, with some carbon stored in the soil (how much depends on the environment and history of the site, especially the last clearing process).As trees and other woody plants become established, carbon stored on the site increases as woody biomass increases and as annual vegetation (e.g., tree leaves and herbaceous plants) typically grows faster than it decomposes. Productivity for commercially usable wood generally follows an S-shaped curve, with the volume growing at an increasing rate for many years, to a point known to foresters as the culmination of mean annual increment (generally taking 20 to 100 years or more, depending on the fertility of the site and the tree species), and then growing at a decreasing rate for many more years. In theory, forests can eventually become “over-mature,” where the loss of commercial volume through tree mortality equals or exceeds the additional growth on the remaining trees. However, one study has shown that some old-growth (“over-mature”) forests continue to accumulate carbon in their soils.86 Fire is a self-sustaining chemical process that breaks organic chemicals down into minerals, water, and CO.27 Roots less than 2 millimeters in diameter are generally considered to be part of the soil, not part of the plant that grew them.8 Guoyi Zhou et al., “Old-Growth Forests Can Accumulate Carbon in Soils,” Science, v. 314 (Dec. 1, 2006): 1417. (continued...)The relationship between commercially usable wood produced and carbon sequestered varies substantially in three ways. First, the proportion of carbon in a tree’s commercial wood (compared to the noncommercial biomass in bark, limbs, roots, and leaves or needles) varies among species; some (e.g., pines and other conifers) have a greater proportion of their total carbon in commercial wood.9 Second, the proportion of carbon in a tree’s commercial wood undoubtedly changes over time; while a temporal graph of carbon storage is probably also S-shaped (as for commercial wood productivity), the changes in timing and rates of increase (that cause the characteristic S shape) almost certainly differ. Finally, a significant portion of the vegetative carbon sequestered in a forest is in other plants—noncommercial species of trees, shrubs, grasses, and other herbaceous plants. The amount of carbon stored in this other (noncommercial) growth varies widely among forests. Thus, although many research studies assume a fixed relationship between commercial wood inventories and the amount of carbon stored,10 the traditional measures of commercial wood production might not be very accurate for estimating carbon sequestration in forests.Eventually, trees die. They may be cut down, burned in a wildfire, blown over or snapped off in a wind or ice storm, or killed by insects or diseases. Death can happen to a single tree in a forest, creating a small opening, or to all or most trees in an area. How quickly the carbon is released to the atmosphere depends on the cause of tree death, on whether it is harvested for use, and on various environmental factors. As noted above, fires quickly break down biomass and release an enormous amount of CO2 into the atmosphere. Natural death and decay may require several weeks to several decades to completely decompose the biomass (depending on site conditions), putting some carbon into the soil and some directly into the atmosphere. Timber harvesting can store some vegetative carbon for very long periods in solid wood products with long-term uses (e.g., construction lumber in houses), while tree tops and limbs and noncommercial species are left to decay or to be burned. These possibilities are discussed in more depth below, under “Forestry Events and Management Activities.”Forest TypesCarbon sequestration and release vary substantially by forest. Nonetheless, some generalizations are possible, because of the relative similarity of forests in specific “biomes”11—tropical,(...continued)Hereafter referred to as Zhou et al., “Soil Carbon in Old-Growth Forests.”9 This is one reason why these species are preferred for timber plantations—a greater proportion of total biomass production goes into commercial wood, and is not “wasted” on noncommercial biomass.10 One study—Paul Schroeder, “Can Intensive Management Increase Carbon Storage in Forests?” Environmental Management, v. 15, no. 4 (1991): 475-481, hereafter referred to as Schroeder, “Intensive Management for Carbon Storage”—assumed a “biomass expansion factor” of 1.6; that is, it assumed that total biomass was 60% greater than (1.6 times) the commercial wood biomass. Another study—Jack K. Winjum, Sandra Brown, and Bernhard Schlamadinger, “Forest Harvests and Wood Products: Sources and Sinks of Atmospheric Carbon Dioxide,” Forest Science, v. 44, no. 2 (1998): 272-284—assumed biomass expansion factors of 1.3 for conifer forests and 2.0 for non-conifer forests. A third study—Robert J. Moulton and Kenneth R. Richards, Costs of Sequestering Carbon Through Tree Planting and Forest Management in the United States, USDA Forest Service, Gen. Tech. Rept. WO-58, Washington, DC, Dec. 1990—used biomass expansion factors ranging from less than 2.0 to more than 8.0 just for different forests within the United States.11 A “biome” is defined as a “[r]egional land-based ecosystem type ... characterized by consistent plant forms and ... found over a large climatic area.” Examples include tropical rainforests, tundra, temperate grasslands, deserts, etc. From The Dictionary of Ecology and Environmental Science, Henry W. Art, ed., Henry Holt & Co., New York, NY, 1993, p. 65.temperate, and boreal forests. Table 1 shows average carbon levels sequestered in vegetation andsoils for several major biomes, and the weighted average for all biomes.12T able 1. Average Carbon Stocks for Various Biomes(in tons per acre)TotalSoilBiome PlantsTropical forests 54 55 109Temperate forests 25 43 68Boreal forests 29 153 1825760Tundra 33637Croplands 1Tropical savannas 13 52 65Temp. grasslands 3 105 10820Desert/semidesert 119306287Wetlands 19Weighted Average 14 59 73Source: Adapted from Intergovernmental Panel on Climate Change, “Table 1: Global carbon stocks invegetation and carbon pools down to a depth of 1 m [meter],” Summary for Policymakers: Land Use, Land-UseChange, and Forestry. A Special Report of the Intergovernmental Panel on Climate Change, at http://www.ipcc.ch/pub/srlulucf-e.pdf, p. 4.Tropical ForestsTropical forests are generally defined by their location—between the Tropic of Cancer and theTropic of Capricorn (23° north and south of the Equator, respectively). Some tropical forests arerelatively dry, open woodlands, but many receive heavy rains and are called moist or humidtropical forests; these are the classic rainforests, or “jungles.” Tropical forests contain anenormous diversity of “hardwood” tree species,13 and are difficult to categorize into “foresttypes,” because of this breadth of species diversity.Moist tropical forests are important for carbon sequestration, because they typically have highcarbon contents—averaging nearly 110 tons per acre. (See Table 1.) About half of the carbon in12 Data on carbon stocks presented in this section are CRS calculations from data in Intergovernmental Panel onClimate Change, “Table 1: Global carbon stocks in vegetation and carbon pools down to a depth of 1 m [meter],”Summary for Policymakers: Land Use, Land-Use Change, and Forestry. A Special Report of the IntergovernmentalPanel on Climate Change, at http://www.ipcc.ch/pub/srlulucf-e.pdf. Hereafter referred to as IPCC, Special Report.13 “Hardwoods” is a term commonly used for trees that are angiosperms—flowering plants—because the dominantflowering tree species of temperate climates (oaks and maples) are harder (more dense) than the major “softwood”species (pines, firs, and spruces), trees of the order Coniferales (conifers). However, some “hardwood” species (e.g.,aspen and poplar) are much softer (less dense) than many “softwoods.” In temperate areas, most hardwoods are alsodeciduous (losing all their leaves annually), while most conifers are evergreen (retaining their needles for more thanone year), leading to common use of “deciduous” and “evergreen” as synonyms for angiosperms and conifers.However, certain conifers (notably larches) are deciduous, while many hardwoods in subtropical climates and most intropical climates are evergreen. For this report, “hardwood” is used to indicate angiosperms, while “softwood” (orconifer) is used for coniferous species.moist tropical forests is contained in the vegetation, a higher percentage and a much higher quantity than in any other biome. The remaining carbon is in tropical forest soils. Tropical forest soils have only modest carbon levels (compared with other biomes), because the dead biomass rapidly decomposes in the warm, humid conditions and the minerals rapidly leach out of tropical forest soils.Temperate ForestsTemperate forests typically occur in the mid-latitudes—generally to about 50° north and south of the Equator (a little farther north in Europe, because of the continental warming from the Gulf Stream). There are a large variety of temperate forests, including hardwood types (e.g., oak-hickory and maple-beech-birch), softwood types (e.g., southern pine, Douglas-fir, and lodgepole pine), and a few mixed types (e.g., oak-pine). However, within each forest type, and overall, temperate forests have much lower tree species diversity than tropical forests.Temperate forests generally contain less carbon than tropical forests, averaging nearly 70 tons per acre. More than one-third of the carbon is stored in the vegetation, and nearly two-thirds in the soil. The higher proportion (but lower level) in the temperate forest soils (compared to tropical forest soils) is because of slower decomposition rates. Many of these forests are managed to produce commercial wood products, and the management practices used in temperate forests can thus have a significant impact on carbon sequestration.Boreal ForestsBoreal forests generally occur north of temperate forests, mostly in Alaska, Canada, Scandinavia, and Russia. (The only boreal forests in the Southern Hemisphere are on mountaintops in New Zealand and high in the Andes Mountains of South America.) Boreal forests are dominated by conifers—mostly spruce, fir, and larch, with scattered birch and aspen stands.Boreal forests generally contain more carbon than temperate or tropical forests, averaging more than 180 tons per acre. Less than one-sixth of boreal forest carbon is in vegetation. The rest, 84%, is in boreal forest soils—about three times the amount in temperate and tropical forests, and far higher than any other biome except wetlands.14 Carbon accumulates to high levels in boreal forest soils because of the very slow decomposition rates, owing to the short summers and high acidity of conifer forest soils, both of which inhibit decomposition. The high boreal forest soil carbon level is important for carbon cycling, because many believe that management activities that disturb boreal forest soils can increase their release of carbon.Measuring and Altering Forest Carbon LevelsAside from the questions of whether climate change is occurring and whether human activities are the cause, the role of forestry and land use in mitigating climate change has been quite controversial. The disputes are largely the result of the scientific uncertainties in measuring14 Soil carbon levels in wetlands are nearly double the level in boreal forests, because the frequent standing water prevents aerobic decomposition, and anaerobic (without oxygen) decomposition processes are much slower than aerobic processes.changing carbon levels in forests, changing land uses, and changing demand for products. This section summarizes forest carbon accounting concerns, possible consequences of changes in landuse and of forest management events and practices, “leakage,” and existing federal programsrelated to these concerns.Forest Carbon AccountingDifferent countries have various views on how to count carbon sequestered or released from forests. In general, countries with extensive and expanding forests (e.g., Russia, Canada, Brazil,and the United States) prefer a full accounting, while countries with less forestland (e.g., manyEuropean countries) are concerned about the potential to overstate the carbon benefits of forestry management practices and land use changes that enhance carbon sequestration. Countries withnet deforestation rates are also concerned about counting forest sequestration, because it couldeffectively increase their net emission rated under international agreements.Article 3.3 of the Kyoto Protocol allows counting the carbon effects (both sequestration andrelease) of afforestation, reforestation, deforestation, and other forestry and land use changes thathave occurred since 1990, if the change in carbon stock is verified. Verification requires a systemfor estimating the carbon effects—because a census of carbon changes on every forested acre is infeasible—and for reporting the carbon effects.For countries with carbon commitments (rather than for projects), the surest, easiest system forverifying the change in carbon levels is to measure the change in the levels from the beginning to the end of the relevant time period—1990 (the baseline) and 2008-2012 (the Kyoto Protocolcommitment period); however, this is a very slow and expensive approach.15 A variety of modelscan be used for estimating carbon level changes. The two basic approaches are: • a “land-based” approach, which begins by identifying the acceptable activities for sequestering carbon and estimating the carbon consequences of thoseactivities, and then monitors the lands to determine the extent to which thoseactivities occur; and•an “activity-based” approach, which also begins by identifying the acceptable activities for sequestering carbon and estimating the carbon consequences ofthose activities, and then monitors the activities to determine the extent to whichthose activities occur.16The approach taken affects the intensity and spatial scale of the monitoring required, and different models impose different requirements for data, boundary conditions, carbon stocks, and more. However, the Intergovernmental Panel on Climate Change contends that, regardless of the approach and model:1715 Richard Birdsey, Ralph Alig, and Darius Adams, “Chapter 8. Mitigation Activities in the Forest Sector to Reduce Emissions and Enhance Sinks of Greenhouse Gases,” The Impact of Climate Change on America’s Forests: A Technical Document Supporting the 2000 USDA Forest Service RPA Assessment, USDA Forest Service, Rocky Mountain Research Station, Gen. Tech. Rept. RMRS-GTR-59, Fort Collins, CO, 2000, p. 116. Hereafter referred to as Birdsey et al., “Forest Mitigation.”16 IPCC, Special Report, “4. Carbon Accounting.”17 IPCC, Special Report, paragraph 31.。

2010-2014MCMProblems建模竞赛美赛题目重点

2010-2014MCMProblems建模竞赛美赛题目重点2010 MCM ProblemsPROBLEM A: The Sweet SpotExplain the “sweet spot” on a baseball bat.Every hitter knows that there is a spot on the fat part of a baseball bat where maximum power is transferred to the ball when hit. Why isn’t this spot at the end of t he bat? A simple explanation based on torque might seem to identify the end of the bat as the sweet spot, but this is known to be empirically incorrect. Develop a model that helps explain this empirical finding.Some players believe that “corking” a bat (h ollowing out a cylinder in the head of the bat and filling it with cork or rubber, then replacing a wood cap enhances the “sweet spot” effect. Augment your model to confirm or deny this effect. Does this explain why Major League Baseball prohibits “corking”?Does the material out of which the bat is constructed matter? That is, does this model predict different behavior for wood (usually ash or metal (usually aluminum bats? Is this why Major League Baseball prohibits metal bats?MCM 2010 A题：解释棒球棒上的“最佳击球点”每一个棒球手都知道在棒球棒比较粗的部分有一个击球点，这里可以把打击球的力量最大程度地转移到球上。

explanatory sequential mixed method

explanatory sequential mixed method Explanatory Sequential Mixed MethodsIntroduction:Research methodology plays a crucial role in informing decision-making and understanding complex phenomena. Mixed methods research designs have gained popularity in recent years due to their ability to provide a comprehensive and holistic understanding of a research problem. One such design is the explanatory sequential mixed methods approach, which incorporates both quantitative and qualitative components in a sequential manner. This article aims to explain the explanatory sequential mixed methods design, its components, advantages, and limitations, with examples from various research studies.Components of the Explanatory Sequential Mixed Methods Design:The explanatory sequential mixed methods design consists of two distinct phases, namely the quantitative phase and the qualitative phase. In this design, the quantitative component is conducted first and is followed by the qualitative component. The purpose of the quantitative phase is to explore relationships and identify patterns in the data, while the qualitative phase aims to provide a deeper understanding of these relationships and patterns.Quantitative Phase:During the quantitative phase, researchers collect and analyze quantitative data using structured surveys, experiments, orsecondary data sources. The goal is to generate numerical data that can be analyzed using statistical techniques to test hypotheses or patterns. The findings from the quantitative analysis inform the selection of participants and the focus of the qualitative phase.Qualitative Phase:The qualitative phase involves collecting and analyzing qualitative data, such as interviews, observations, or document analysis. The purpose of this phase is to understand the underlying reasons and processes behind the quantitative findings. Qualitative data collection methods allow researchers to gather rich and detailed information that provides context and meaning to the statistical results. The qualitative analysis involves coding, categorizing, and interpreting the data to identify themes and patterns.Integration of Findings:The integration of findings is a crucial step in the explanatory sequential mixed methods design. During this stage, researchers compare and contrast the findings from both the quantitative and qualitative phases to develop a comprehensive understanding of the research problem. This integration can occur in different ways, such as comparing the results side by side, using statistical findings to interpret qualitative data, or using qualitative findings to explain statistical patterns. The aim is to provide a richer and more nuanced understanding of the research topic than could be achieved through a single method or phase.Advantages of the Explanatory Sequential Mixed Methods Design:The explanatory sequential mixed methods design offers several advantages over traditional quantitative or qualitative approaches. Firstly, it provides a more comprehensive understanding of the research problem by combining quantitative and qualitative data. The sequential nature of the design allows researchers to build on the findings from the quantitative phase and explore them in more depth during the qualitative phase. This approach strengthens the validity and reliability of the research findings.Secondly, the design allows researchers to address research questions that require both numerical data and contextual information. Some phenomena cannot be fully understood or explained by numbers alone, and qualitative data can provide valuable insights into the underlying reasons and processes.Thirdly, the design enhances triangulation, which refers to the use of multiple data sources or methods to validate findings. By combining quantitative and qualitative data, researchers can compare and contrast the different perspectives and identify converging or conflicting evidence. This strengthens the overall validity and trustworthiness of the research.Limitations of the Explanatory Sequential Mixed Methods Design:Despite its advantages, the explanatory sequential mixed methods design also has some limitations. Firstly, it requires time and resources to implement both quantitative and qualitative components. Researchers need to consider the feasibility of conducting both phases and ensure that they have the necessaryskills and expertise in both quantitative and qualitative methods.Secondly, the design may face challenges in terms of data integration and interpretation. Combining quantitative and qualitative findings can be complex and may require expertise in both types of data analysis. Researchers need to carefully consider how to integrate the findings in a meaningful and coherent manner.Example Studies:To illustrate the application of the explanatory sequential mixed methods design, three example studies are presented below:1. A study on the effectiveness of a health intervention program uses a quantitative survey to measure participants' health outcomes and satisfaction levels. The qualitative phase involves in-depth interviews with a sub-sample of participants to explore their experiences and perceptions of the program.2. A study on the impact of a teacher training program uses a quantitative pre-test and post-test design to measure changes in students' academic performance. The qualitative phase involves focus group discussions with teachers to understand their perspectives on the program's effectiveness and challenges.3. A study on the factors influencing employee satisfaction and retention uses a quantitative survey to measure employee satisfaction levels. The qualitative phase involves semi-structured interviews with a subset of employees to explore the underlying reasons for their satisfaction or dissatisfaction.Conclusion:The explanatory sequential mixed methods design offers a powerful approach to research that combines the strengths of quantitative and qualitative methods. By integrating numerical data with contextual information, this design provides a comprehensive and holistic understanding of research problems. Despite its limitations, the design has gained popularity due to its ability to address complex research questions and enhance the validity and reliability of findings. Researchers should consider the feasibility and appropriateness of this design for their specific research objectives and resources.。

A monothetic clustering method

A monothetic clustering method∗Marie Chavent(*)(**)(*)INRIA Rocquencourt,Action SODAS,Domaine de Voluceau,B.P.105,78153Le Chesnay cedex,France(**)Universit´e de Paris IX Dauphine,Lise Ceremade,Place du Mar´e chal De Lattre de Tassigny,75775Paris cedex16,Francee-mail:Marie.Chavent@inria.frAbstract:The proposed divisive clustering method performs simultaneously a hierarchy of a set of objects and a monothetic characterization of each cluster of the hierarchy.A division is performed according to the within-cluster inertia criterion which is minimized among the bipartitions induced by a set of binary questions.In order to improve the clustering,the algorithm revises at each step the division which has induced the cluster chosen for division.Key Words:Hierarchical clustering methods,Monothetic cluster,Inertia criterion1.IntroductionThe objective of cluster analysis is to group a setΩof N objects into clusters having the property that objects in the same cluster are similar to another and diﬀerent from objects of other clusters.In the pattern recognition literature(Duda and Hart,1973)this type of problem is referred to as unsupervised pattern recognition.The most common clustering methods are partitioning,hierarchical agglomerative and hierarchical divisive ones.A partition ofΩis a list(C1,...,C K)of clusters verifying C1∪...∪C K=Ωand C k∩C k =∅for all k=k .The essence of partitioning is the optimization an objective function measuring the homogeneity within the clusters and/or the separation between the clusters.Algorithms of the exchange type are frequently used toﬁnd a local optimum of the objective function, because of the complexity of the exact algorithms.Well-known partitioning procedures are the Forgy’s k-means and the isodata methods,described in Anderberg(1973),and the dynamical clustering method(Diday,1974).Agglomerative and divisive hierarchical clustering methods are diﬀerent,in the type of structure they are searching,from partitioning.Indeed,a hierarchy ofΩis a family H of clusters satisfying Ω∈H,{ω}∈H for allω∈Ωand A∩B∈{∅,A,B}for all A,B∈H.A hierarchy can be represented in the form of a tree or dendogram,that shows how the clusters are hierarchically organized.The general algorithm for agglomerative clustering starts with N clusters,each consisting of ∗Pattern Recognition Letters19(1998)989-9961one element ofΩ,and merges successively two clusters on the basis of a similarity measure. Well-known agglomerative hierarchical methods are described in Everitt(1974).Divisive hierarchical clustering reverses the process of agglomerative hierarchical clustering,by starting with all objects in one cluster,and dividing successively each cluster into smaller ones. Those methods are usually iterative and determine at each iteration the cluster to be divided and the subdivision of this cluster.This process is continued until suitable stopping rule arrests further division.There is a variety of divisive clustering methods(Kaufman and Rousseeuw,1990).A natural approach of dividing a cluster C of n objects into two non-empty subsets would be to consider all the possible bipartitions.In this,Edward and Cavalli-Sforza(1965)choose among the2n−1−1 possible bipartitions of C,the one having the smallest within-cluster sum of squares.It is clear that such complete enumeration procedure provides a global optimum but is computationally prohibitive.Neverless,it is possible to construct divisive clustering methods that does not consider all bipar-titions.MacNaughton-Smith(1964)proposed an iterative divisive procedure using an average dissimilarity between an object and a group of objects.Chidananda Gowda and Krishna(1978) proposed a disaggregative clustering method based on the concept of mutual nearest neighbor-hood.Other methods taking as input a dissimilarity matrix are based on the optimization of criterions like the split or the diameter of the bipartition(Gu´e noche,Hansen and Jaumard,1991; Wang,Yan and Sriskandarajah,1996).Probabilistic validation approach for divisive clustering has also been proposed(Har-even and Brailovsky,1995).Another family of divisive clustering methods is monothetic.A cluster is called monothetic if a conjunction of logical properties is both necessary and suﬃcient for membership in the cluster (Sneath and Sokal,1973).Indeed,each division is carried out using a single variable and by separating objects possessing some speciﬁed values of this variable from those lacking them. Monothetic divisive clustering methods haveﬁrst been proposed in the particular case of binary data(Williams and Lambert,1959;Lance and Williams,1968).Since then,monothetic cluster-ing methods have mostly been developed in theﬁeld of unsupervised learning and are known as descendant conceptual clustering methods(Michalski,Diday and Stepp,1981;Michalski and Stepp,1983).In theﬁeld of discriminant analysis,monothetic divisive methods have also been widely devel-oped.However,those methods are diﬀerent from clustering in which the clusters are inferred from data.Indeed,a partition ofΩis pre-deﬁned and the problem concerns the construction of a systematic way of predicting the class membership of a new object.In the pattern recognition literature,this type of classiﬁcation is referred to as supervised pattern recognition.Divisive methods of this type are usually known as tree structured classiﬁer like cart(Breiman,Fried-man,Olshen and Stone,1984)or id3(Quinlan,1986).Recently,Ciampi(1994)insisted on the idea that trees oﬀer a natural approach for both class formation(clustering)and development of classiﬁcation rules(discrimination).The clustering method proposed in this paper was developed in the framework of symbolic data analysis(Diday,1995),which aims at bringing together data analysis and machine learning. More precisely,we propose a monothetic hierarchical clustering method performed in the spirit of cart from an unsupervised point of view.We have restricted the presentation of this method to the particular case of quantitative data.At each stage,the division of a cluster is performed according to the within-cluster inertia criterion(section??).This criterion is minimized among bipartitions induced by a set of binary questions(section??).Moreover,clusters are not sys-2tematically divided but one of them is chosen according to a speciﬁc criterion(section??).The divisions are stopped after a number of iterations given as input by the user,usually interested in few clusters partitions.The output of this divisive clustering method is an indexed hierarchy. It is also a decision tree(section??).The Ruspini’s data are given as aﬁrst illustration of this method(section??).We propose a modiﬁcation of the algorithm in order to soften the property shared by both agglomerative and divisive hierarchical methods,that eﬃcient early partition cannot be corrected at a later stage.It consists in revising,after the division of a cluster,the previous division which has induced the cluster itself(section??).Before the conclusion(section ??),the method is performed on Fisher’s iris dataset(section??).2.The inertia criterionLet N be the number of objects inΩ.Each object is described on p real variables Y1,...,Y p by a vector x i∈R p and weighted by a real value p i(i=1,...,N).Indeed,the analyst will prefer sometimes to weight the objects diﬀerently.For instance,countries could be weighted according to the size of their population.But usually,the weights are equal to1or equal to1n.The inertia I of a cluster C k is an homogeneity measure equal to:I(C k)=xi ∈C kp i d2M(x i,x k)(1)where d M is the Euclidean distance(M is a symmetric matrix positively deﬁned):∀x,y∈R p,d2M(x,y)=(x−y)t M(x−y)(2) and x k is the center of gravity of the cluster C k:x k=1µkxi∈C kp i x i(3)µk=xi ∈C kp i(4)The within-cluster inertia W of a K-clusters-partition P K=(C1,...,C K)is equal to:W(P K)=Kk=1I(C k)(5)According to the Huygens Theorem,minimizing the within-cluster inertia of a partition(e.g. the homogeneity within the clusters)is equivalent to maximizing the between-cluster inertia (e.g.the separation between the clusters).This equals to:B(W K)=Kk=1µk d2M(x k,x)(6)3.Bipartitioning a clusterLet C be a set of n objects.We want toﬁnd a bipartition(C1,C2)of C such that the within-cluster inertia is minimum.In the Edward and Cavalli-Sforza method(1965)one chooses the optimal bipartition(C1,C2)among the2n−1−1possible bipartitions.It is clear that the amount of calculation needed when n is large will be prohibitive.3In our approach,to reduce the complexity,we divide C according to a binary question(Breiman, Friedman,Olshen and Stone,1984)of the form“Y i≤c?”where Y i:Ω→R is a real variable and c∈R is called the cut point.The bipartition(C1,C2)induced by the binary question is deﬁned as follows.Letωbe an object in C.If Y i(ω)≤c thenω∈C1elseω∈C2.Those objects in C answering“yes”go to the left descendant cluster and those answering“no”to the right descendant cluster(Fig.??).Figure1:“Is height≤172?”For each variable Y i,there will be at most n−1diﬀerent bipartitions(C1,C2)induced by the above procedure.Indeed,whatever the cut point c between two consecutive observations Y i(ω) may be,the bipartition induced is the same.In order to ask only n−1questions to generate all these bipartitions,we decide to use the n−1cut points c,chosen as the middle of two consecutive observations Y i(ω)∈R.Indeed,if the n observations Y i(ω)are diﬀerent,there are n−1cut points on Y i.If there are p variables,we choose among the p(n−1)corresponding bipartitions(C1,C2),the bipartition having the smallest within-cluster inertia.4.Choice of the clusterLet P K=(C1,...,C K)be a K-clusters-partition ofΩ.At each stage,a new(K+1)-clusters-partition is obtained by dividing a cluster C k∈P K into two new clusters C1k and C2k.Thepurpose is to choose the cluster C k∈P K so that the new partition,P K+1=P K∪{C1k,C2k}−{C k}has minimum within-cluster inertia.We know that:W(P K+1)=W(P K)−I(C k)+I(C1k)+I(C2k)In this,minimizing W(P K+1)is equivalent to choosing the cluster C k∈P K so that the diﬀerencebetween the inertia of C k and the within-cluster inertia of its bipartition(C1k ,C2k)is maximum.The criterion used to determine the cluster that will be divided is then equal to:∆(C k)=I(C k)−I(C1k)−I(C2k)(7) Of course,it means that the bipartitions of all the clusters of the partition P K have been deﬁnedpreviously.At each stage,the bipartitions of the two new clusters C1k and C2kare deﬁned andused in the next stage.5.The stopping rule and the outputThe divisions are stopped after a number L of iterations and L is given as input by the user, usually interested in few clusters partitions.Indeed,the last partition obtained in the last iteration is a L+1-clusters-partition.The issue of stopping the divisions before obtaining the4total hierarchy(L=N)is to ensure that the partitions of smallest within-cluster inertia of the total hierarchy are still in the hierarchy obtained after L iterations.This property is veriﬁed because the clusters are not systematically divided but one cluster is chosen according to the criterion∆given in(??)which ensures that the partition induced by this division has minimum within-cluster inertia.However,this stopping rule doesn’t solve the issue of determining the number of clusters in the dataset(Milligan and Cooper,1985).The output of this divisive clustering method is a hierarchy H which singletons are the L+1 clusters of the partition obtained in the last iteration of the algorithm.Each cluster C k∈H is indexed by∆(C k).Because∆is a non-decreasing mapping,C k⊂C k ⇒∆(C k)≤∆(C k )(8)there will be no inversions in the dendogram of the hierarchy.This hierarchy is also a decision tree.The L clusters are the leaves and the nodes are the binary questions selected by the algorithm.Each cluster is characterized by a rule deﬁned according to the binary questions leading from the root to the corresponding leaves.6.A simple exampleThe dataset is75points of R2(Ruspini,1970).Weﬁnd successively a partition in2,3and4 clusters(L=3).At theﬁrst stage,the method induces2(75−1)=148bipartitions.We choose among the 148bipartitions(C1,C2),the one of smallest within-cluster inertia.It has been induced by the binary question“Is Y1≤75.5?”.Notice that the number of subdivisions has been reduced from 275−1=3,77×1022to148.At the second stage,we have to choose whether we divide C1or C2.Here,we choose the cluster C1and its bipartition(C11,C21)because∆(C1)>∆(C2).The binary question is“Is Y2≤54?”. At the third stage,we choose the cluster C2and its bipartition(C12,C22).The binary question is“Is Y2≤75.5?”.Finally,the divisive algorithm gives the4clusters represented Fig.??.Figure2:The4-clusters partitionAccording to the dendogram of the hierarchy givenﬁgure??,the four clusters are characterized by four rules.For instance cluster C11is characterized by the following rule:If[Y1(ω)≤75,5]and[Y2(ω)≤54]thenω∈C11.5Figure3:The dendogram of the indexed hierarchyThis dendogram can be read as a decision tree and the rules can be read as classiﬁcation rules of new objects to one of the four clusters.7.Revising a binary questionThe purpose is to enable the analyst to revise at each division of a cluster the binary question which has induced the cluster itself.Let C be a cluster which has been divided in two clusters C and C according to the binary question“Is Y1≤c1?”.Then C is chosen to be divided in two clusters C1and C2according to the binary question“Is Y2≤c2?”.Figure4:Revising a binary questionAt this stage,the binary question“Is Y1≤c1?”is revised by modifying the cut point c1.We choose a new cut point c among all possible cut points on Y1,such that the3-clusters-partition (C 1,C 2,C )induced by“Is Y1≤c ?”and“Is Y2≤c2?”has minimum within-cluster inertia (ﬁgure??).For instance,ﬁgure??gives the3-clusters-partition of320points of R2simulated from four 2-dimensional Gaussian distributions.The points have been dividedﬁrst according to the binary question“Is Y2≤10,9?”and then according to the binary question“Is Y1≤8?”.Theﬁrst cut point10.9is then modiﬁed in order toﬁnd,with the second binary question “Is Y1≤8?”,the3-clusters-partition of minimum within-cluster inertia.The new cut point is 12.1(ﬁgure??).8.The Fisher’s iris datasetThe above clustering method has been examined with the well-known Fisher’s iris dataset.The length and breadth of both petals and sepals were measured on150ﬂowers.There are three varieties of iris:Setoa,Versicolor and Virginia.There are50iris of each variety.6Figure5:The two3-clusters-partitionsOf course,the knowledge of this pre-deﬁned3-clusters-partition is not used in our unsupervised clustering procedure which is performed only with four quantitative variables:the petal width (PeWi),the petal length(PeLe),the sepal width(SeWi)and the sepal length(SeLe).First,we have used the Euclidean distance d M,with M=I,the identity matrix.Figure?? gives the dendogram of the hierarchy and the3-clusters-partition(C1,C2,C3)obtained after two divisions of the dataset.Theﬁrst cluster is composed of53iris including50Setoa,3Versicolor and no Virginia.Wholly,the3-clusters-partition contains19iris misclassiﬁed.Theﬁrst binary question“Is PeLe≤3.4”is then revised in order to improve the within-cluster inertia of the3-clusters-partition.Figure??gives the dendogram of the hierarchy obtained with the revised binary question“Is PeLe≤2.45”.We can notice that the mis-classiﬁcations have been reduced to16.Indeed,the50Versicolor are all in C2.Figure6:Before the revision Figure7:After the revisionDynamical clustering and Ward agglomerative hierarchical clustering methods have also been performed on the same dataset.The same distance was used.The partitions obtained with the two clustering methods contained the same number of mis-classiﬁcations since16iris were misclassiﬁed.where U i is Secondly,we have used the normalized Euclidean distance d M,with M=D1/U2ithe length between the maximum and the minimum value for the variable Y i.Theﬁgure?? gives the dendogram of the hierarchy obtained with this distance and we notice a reduction of the number of mis-classiﬁcations from16to10iris.It conﬁrms the inﬂuence of the choice of the distance in the result of a clustering.Then,before the second division,we have normalized the Euclidean distance,according to the four length U i computed locally in the cluster which7Figure8:Global normalization Figure9:Local normalizationwas divided.Theﬁgure??gives the dendogram of the hierarchy obtained with the locally normalized Euclidean distance.We can notice that the number of mis-classiﬁcations is now reduced to6.It corresponds to an error rate of0.04.In their comparative study of the performance of diﬀerent classiﬁers with Fisher’s iris dataset, Weiss&Kulikowski(1991)give for the cart decision tree an error rate equal to0.04.In this,we obtain with the Fisher’s iris dataset comparable results with both unsupervised and supervised approaches.However,the goal of the proposed clustering method and the cart algorithm are diﬀerent since we aim at inferring clusters from the data and cart algorithm aims at discovering classiﬁcation rules.9.ConclusionThe proposed clustering method has the advantages to be simple and to give simultaneously a hierarchy and a simple interpretation of its cluster.Moreover,it deals easily with very large datasets.Indeed,is possible to construct the hierarchy on a sample of the dataset,and to use the classiﬁcation rules to assign the rest of the objects.This method has also given good results on the Fisher’s iris dataset and on other real applications where it has been compared with the dynamical clustering method and the Ward agglomerative hierarchical method(Chavent,1997). However,dividing a cluster according to a single variable can also be a deﬁciency in some situa-tions.As for cart algorithm,in situations where the cluster structure depends on combinations of variables,the divisive method will do poorly at discovering the structure.A perspective would be on the one hand to use a local stopping rule(Milligan and Cooper,1985; Har-even and Brailovsky,1995)for deciding if a cluster should be divided into two subclusters and on the other hand to divide a cluster according to a metric locally deﬁned in the cluster itself.ReferencesAnderberg,M.R.(1973).Cluster analysis for applications.Academic Press,New York. Breiman,L.,J.H.Friedman,R.A.Olshen and C.J.Stone(1984).Classiﬁcation and regression Trees.C.A:Wadsworth.8Chavent,M.(1997).Analyse des Donn´e es Symboliques.Une m´e those divisive de classiﬁcation.PhD Thesis,Universit´e Paris-IX Dauphine,France.Chidananda Gowda,K.and G.Krishna(1978).Disaggregative Clustering Using the Concept of Mutual Nearest Neighborhood.ieee Transactions on Systems,Man,and Cybernetics 8,888-895.Ciampi,A.(1994).Classiﬁcation and Discrimination:the recpam Approach.In proc.of compstat’94,129-147.Diday,E.(1974).Optimization in non-hierarchical clustering.Pattern Recognition6,17-33. Diday,E.(1995).Probabilist,possibilist and belief objects for knowledge analysis.Annals of Operations Research55,227-276.Duda,R.O.and Hart,P.E.(1973).Pattern Classiﬁcation and Scene Analysis.Wiley,New York.Edwards,A.W.F.and L.L.Cavalli-Sforza(1965).A method for cluster analysis.Biometrics 21,362-375.Everitt,B.(1974).Cluster Analysis.Social Sciences Research Council,Heineman Educational Books.Gu´e noche,A.,P.Hansen and B.Jaumard(1991).Eﬃcient algorithms for divisive hierarchical clustering.Journal of Classiﬁcation8,5-30.Har-even,M.and V.L.Brailvosky(1995).Probabilistic validation approach for clustering.Pattern Recognition16,1189-1196.Kaufman,L.and P.J.Rousseeuw(1990).Finding groups in data.Wiley,New York. Lance,G.N.and W.T.Williams(1968).Note on a new information statistic classiﬁcation program.The Computer Journal11,195-197.MacNaughton-Smith,P.(1964).Dissimilarity analysis:A new technique of hierarchical sub-division.Nature202,1034-1035.Michalski,R.S.and E.Diday,R.Stepp(1981).A recent advance in data analysis:Clustering objects into classes characterized by conjunctive concepts.Progress in Pattern Recognition, L.N.Kanal and A.Rosenfeld(eds),North Holland,33-56.Michalski,R.S.and R.Stepp(1983).Learning from observations:Conceptual clustering.Machine Learning:An Artiﬁcial Intelligence Approach,R.S.Michalsky,J.Carbonell and T.Mitchell(eds),163-190.Milligan,G.W.and M.C.Cooper(1985).An examination of procedures for determining the number of clusters in a data set.Psychometrika50,159-179.Quinlan,J.R.(1986).Induction of decision trees.Machine Learning1,81-106.Ruspini,E.M.(1970).Numerical Methods for Fuzzy rmation Science2,319-350.Sneath,P.H.and R.R.Sokal(1973).Numerical Taxonomy.Freeman and company,San Fran-cisco.Weiss,S.M.and C.A.Kulikowski(1991).Computer systems that learn:Classiﬁcation and prediction methods from statistics,neural network,machine learning,and expert systems.San Mateo,Calif:Morgan Kaufmann.Williams,W.T.and mbert(1959).Multivariate methods in plant ecology.Journal of Ecology47,83-101.Wang,Y.and H.Yan,C.Sriskandarajah(1996).The weighted Sum of Split and Diameter Clustering.Journal of Classiﬁcation13,231-248.9List of Figures10。

(完整版)环境科学交叉关系学科课后题答案第十一、十二章

CHAPTER 11 three ways that humans directly alter ecosystems.1)In the past, prehistoric men used human-induced fire to capture game animals or clean landfor agriculture, thus destroying climax communities. The harvesting of tropical forests today works in the same way.2)The conversion of natural land into agricultural land and even urban land in cities , whichcontinues today, has greatly undermined the biodiversity.3)The overexploitation of fishery resources, coupled with the introduction of exotic species,have spawned a series of problems that alter the local water environment.2.Why is the impact of humans greater today than at any time in the past?As the technology advanced, the ability of people to modify their surroundings has increased significantly, the agricultural revolution and modern agricultural technologies, for example, have efficiently turned large parts of the earth into agricultural land.The growing number of human population contributed to the draining of natural resources, as well as the extinctions of many species.3. Describe three factors that influence the genetic diversity of a population.Several things can influence the genetic diversity of a population.1)Mutations are changes in the genetic information of an organism, which introduce newgenetic information into a population by modifying genes that are already present. The DDT-tolerance of insects and the evolution of human resistance to antibiotic medication are such examples.2)Migration of individuals of a species from one place to another is also an important way. Itresults in the reduction of genetic information in the former population and the addition in the new population. This can have a significant effect on both populations if the migrating individual possess rare characteristics.3)Sexual reproduction is another process that influence genetic diversity. Rather than creatingnew genetic information, it tends to generate new genetic combinations when genetic information from two individuals mixes during fertilization, forming a unique individual, which may have a combination to out-compete its peer by being more successful in producing offspring, thus influence the genetic diversity.5.What are the major causes of loss of biodiversity in marine ecosystems?1)Habitat loss is a problem in marine ecosystems, as much of the harvest is restricted toshallow parts of the ocean where bottom dwelling fish can be easily harvested. It involved the use of trawls which are nets that can be dragged along the bottom. The trawls can disturb the sea floor and create conditions that make it harder for the fish population to recover. It captures various other species (25%) that are not commercially valuable and often left dead on board. Their removal further alter the ecological nature of the seafloor. 2)Overexploitation has driven some species to extinction and threaten many others.Organisms can be harvested for various reasons, food, ornaments or other aesthetic uses, and uncrupulous people often poach the already endangered species for quick profit. It isalso common in marine fisheries, and efforts are made to develop aquaculture methods and market new fish species which levitate the problem.3)Climate change has a great effect on the survival of species with limited physiologicaltolerance, such as corals in oceans and amphibians. The warming of water are leading to the declining of coral reefs.7.What is desertification? What causes it?Desertification is the process of converting arid and semiarid land to desert because of improper use by humans.Rangelands are too dry to support crops and grazing of domesticated animals is the only viable solution. But in areas where human population pressure is great, overgrazing is seemingly unavoidable, as people graze too many animals and cut down more trees for firewood. These would expose soil to wind erosion and lead to loss in soil fertility. Cutting down legumes that fix nitrogen would worsen the case. The land would gradually turn into a desert-like ecosystem.9.List six techniques utilized by wildlife managers.Habitat management are modifications to the habitat to enhance their survival and reproduction. The first step is to understand the habitat need of target species, and identify the critical habitat requirements of it. Then, they can alter the habitat and improve the success of the species. Population assessment and management also requires careful planning and the techniques involved included:Population census to keep the numbers of animals in check.Regulating hunting seasons, i.e. In fall so as to take the surplus animals, can ensure adequate and sustainable reproduction of animals.Artificially introduction of certain species when their population is below the desired number or extinct from the local area.Refuges for waterfowls can be built to provide resting places, food and protection from hunting. Transboundary parks can accommodate the movements of migratory animals across different countries.11.What is extinction? Why does it occur?Extinction is the death of a species, the elimination of all individuals of a particular kind. Extinction is a natural and common process through out the evolution, yet human activity has sped up its rate by a factor of 1000-10000. Some species with low population density and low reproductive rate, in specialized niche are prone to extinction. As technology advanced, human populations grew, we have increasingly huge influence on our surroundings. Consequently, many species have gone extinct.15.List three actions that be taken to prevent extinctions.IUCN lists over 19000 species as threatened with extinction in the Red List of Threatened Species, this can encourage countries to protect the related species and to build natural reserves.The Convention on Biological Diversity are adopted by many countries to preserve the biological diversity.The Endangered Species Act demands all government agencies to do whatever necessary to preserve the endangered species and the following amendment in 1978 saw the “god squad”to exempt some projects from the Act.16.Describe the role of the red list of threatened species in species preservation.IUCN is a highly visible international preservation organization, but has very little power to effect change. It generally seeks to protect species in danger by encouraging countries to complete inventories of plants and animals within their borders and encourage the training of plant and animal biologists within countries involved and the establishment of preservers to protect species in danger of extinction.CHAPTER 121. List three reasons why land-use planning is necessary.1)In modern world, significant amounts of land is covered with buildings, streets and otherproducts of society. But in many cases, cities are established before there is an understanding of the challenges presented by the location, when these cities grew and technology and society changed, the shortcomings of the location become apparent.Therefore, we should understand that each piece of land has its specific qualities based on its location and physical make-up.2)The land should be considered a nonrenewable resource nowadays as the land and theresources it supports(soil, vegetation and watersheds, etc) are not being created today. We need to plot carefully about the use of it. Once it is converted from natural ecosystems or agriculture to intensive human use, it is generally unavailable for other purposes.3)As the human population continues to boom, competition for the use of land wouldundoubtedly increase and systematic land-use planning would be important. Furthermore, as the population becomes more urbanized and cities grow, urban planning becomes critical.3. List three factors that encourage people to move from rural farms to cities in 1800s.4)First, the Industrial Revolution led to improvements in agriculture that required less farmlabor at the same time industrial jobs became available in the city, leading to the rural-to-urban migration.5)Then, the second factor that affected the growth of cities was the influx of immigrants fromEurope. They settled in towns and cities.6) A third reason for the growth was that they offered a greater variety of cultural, social, andartistic opportunities than did rural communities.Thus they were attractive for cultural as well as economic reasons.5.List three physical and three social consequences of urban sprawl.Physical:1)The automobile based society in US can cause serious traffic congestion for those who workin cities but live in the suburbs.2)The new housing or commercial development in suburbs would require the municipalservices to be extended to such areas, which is way more costly than supplying services to areas already in the city. The same is of energy costs because of low energy efficiency.3)Air pollution is also significant due to the reliance on automobiles as primary method oftransportation, and the infrastructure that support automobile travel is impervious to water, and the runoff are channeled directly into local water sources, bringing pollutants(oil, coolant and rubber pieces) into local streams.Social:1)The death of central city occurred as more people move to the suburbs and quality ofservices in urban center drops which starts a downward spiral of decay. This can deprive the remaining residents of basic services. It has a particular hard hit on the poor and elderly.2)Open fields, parks boulevards and similar land uses allow people to visually escape from thecongestion of the city. However, the urban sprawl have deprived a lot of land that could have been used as open space.3)Unpleasant odors, disagreeable tastes, annoying sounds and offensive sights are aggravating,and may be deemed harmful from an aesthetic point of view. Yet this are often the case of unplanned development in suburbs.7.What is a megapolis?As suburbs continued to grow, cities began to merge, and it became difficult to tell were one city ended and another began. This type of growth led to the development of regional cities. Although their cities maintain their individual names, they are really just part of one large urban area called a megalopolis.9.State three consequences of the dominance of the automobile as a means of transport in urban areas.1)The reliance on the automobile has required the constant building of new highways andaccording to DOT it costs 1 trillion per year on maintaining and building new ones.2)The average person in US travels about 260 kilometers per week in car and a person inmetropolitan area spends more than 40 hours per year stuck in traffic delays.3)It is hard to divert funding to establish mass transit besides the dispersed nature of suburbs.11.What characteristics of suburbs contribute to high infrastructure and high energy costs.1)Infrastructure include all physical, social and economic elements needed to support thepopulation, and it is often costly to extend it to the newly developed suburbs as everything need to be built from scratch.2)Energy costs are high due to low energy efficiency, and there are several reasons for this: Firstly, the automobiles are the least energy-efficient means of transporting people. Secondly, the separation of blocks of home from business and shopping areas require greater distance driven to meet basic needs.Thirdly, congested traffic routes result in hours being spent in stop-and-go traffic and wasting much fuel.Finally, the single-family homes require more energy for heating and cooling than multifamilydwellings.13. What land uses are suitable on floodplains?Floodplains are low areas near rivers that are subject to periodic floods, and it is often used for residential or commercial purposes due to its flat character. But flood-control structures need to be built which have detrimental downstream effects and could pose threats during floods.A better use of floodplains is for open space or recreation or agriculture.15. Why is a understanding of the geology and resources base of an area important in land use planning?1)The geologic status of an area must be considered in land-use decisions to prevent possibledisasters and hazards(i.e. Near volcanoes or earthquake-prone faults) or the lack of water, which will inevitably worsen as cities grow. To understand to resource base can lead to wise planning.2)Some land has unique features that should be preserved because of their special value tosociety( Grand Canyon and Yellowstone, etc) and should take precedence over other uses.17.What role do state and regional planning, purchasing of land, and use restrictions play in implementing land-use plans?1)State and regional planning is often more effective than local land-use planning since manyimportant geographic, geological and habitat characteristics cross local political boundaries.2)In addition, a regional approach is likely to prevent duplication of facilities and lead togreater efficiency.3)State or regional planning bodies are also more likely to have the financial resources to hireprofessional planners to assist in the planning process.4)Purchasing of land is the easiest way to protect them, and many environment organizationsopt to purchase lands with special historic, scenic or environmental value. In some cases, the landowners may sell the right to develop the land or place restrictions on the future uses of land.5)Many kinds of l and-use restrictions involve some form of zoning, that designates specificareas within a community for certain kinds of land use. But it has both positive and negative impacts on good land-use planning. Sometimes it help in preserving important historic or cultural sites while in many ways it also contributed to the segregation found in urban sprawl.19.List ten common smart growth principles.The smart growth approach has the following guidelines:1)Preserve open space. Farmland, natural beauty and critical environmental areas.2)Direct development toward existing urban areas, which encourages the reuse of abandonedor poorly used urban space.3)Take advantage of compact building design, so more people can be housed, and a smallercarbon footprint can be achieved. It can also reduce the need to develop new land4)Create a range of housing opportunities and choices, to accommodate people with differentlifestyles, desires and income levels.5)Foster distinctive, attractive communities with a strong sense of place. Pay attention to thedesign of buildings and their relationship with open space and cultural attractions to createa pleasing urban setting.6)Mix land uses, so that people need not drive somewhere to fulfill their basic needs.7)Create walkable neighborhoods, provide pedestrian walks to separate vehicle traffic frompedestrians.8)Provide a variety of transportation choices.9)Encourage community and stakeholder collaboration in development decisions.10)Make development decisions predictable, fair and cost-effective.21.Give examples of conflict over the use of federally owned property.One of the major conflict is between those who prefer to use motorized vehicles and those who prefer to use muscle power over the outdoor recreation activities. They both paid taxes and wish the land can be used as they wish.Conflicts also arise between business interests and recreational users of public lands. The grazing and skiing are hard to reconciliate, and as the regulatory agencies are often understaffed, the ranchers tend to overgraze the land.A particular sensitive issue is the designation of certain areas as wilderness areas. Many people argue it’s unfair because they are paying tax but their access to the wilderness is restricted. While others fear too many people would destroy the charm and unique character.。

2020年智慧树知道网课《数据结构(全英文)》课后章节测试满分答案

第一章测试1【单选题】(10分) ThealgorithmandflowchartcanhelpustoA.TostorethedataB.ToknowthememorycapacityC.IdentifythedatatypeofavariableD.Specifytheproblemcompletelyandclearly2【单选题】(10分) TherhombusordiamondshapeinflowchartingdenotesA.DecisionB.InputC.InitializationD.Output3【单选题】(10分) Whichofthefollowingisnotanadvantageofaflowchart?A.EfficientcodingB.BettercommunicationC.SystematictestingD.Improperdocumentation4【单选题】(10分) Theflowchartsymbolsusedforstartandstopoperationsarecalledas_______.A.decisionB.processingC.terminalsD.connectors5【单选题】(10分)TheformulaF n=F n-1+F n-2willproduceA.FibonacciNumberB.RamanujanNumberC.PrimeNumberD.EulerNumber6【单选题】(10分) ThemainmeasuresfortheefficiencyofanalgorithmareA.ComplexityandcapacityB.ProcessorandmemoryC.TimeandspaceD.Dataandspace7【单选题】(10分) WhichoneofthefollowingistheconstanttimecomplexityintermsofBig-OhnotationA.O(1)B.O(n2)C.O(n3)D.O(n)8【单选题】(10分)Whatisthetimecomplexityofthefollowingcode?inta=0;for(i=0;i<n;i++){for(j=n;j>i;j--){a=a+i+j;}}A.O(nlog n)B.O(n)C.O(n2)D.O(1)9【单选题】(10分) Whichoneofthefollowingisanexampleforexponentialtimecomplexity?A.O(n2)B.O(2n)C.O(n)D.O(1)10【单选题】(10分)Forlargervaluesof n,whichonerepresentstheslowesttime?A.O(n2)B.O(2n)C.O(n)D.O(n!)第二章测试1【单选题】(10分) Deletionofanelementfromthearrayreducesthesizeofarrayby___________.A.threeB.twoC.zeroD.one2【单选题】(10分)Assumingthatint isof4bytes,whatisthesizeof intarr[10];?A.30B.10C.40D.3【单选题】(10分) Twodimensionalarraysareusefulwhentheelementsbeingprocessedaretobearran gedintheformof___________.A.NoneoftheaboveB.Both(a)and(b)C.rowsD.columns4【单选题】(10分)Inthepolynomial,A(x)=3x2+2x+4,thedegreeofthispolynomialisA.3B.1C.D.5【单选题】(10分)Inthepolynomial,A(x)=3x2+2x+4,coefficientoffirsttermisA.2B.1C.D.36【单选题】(10分) Amatrixhavingalargernumberofelementswithzerovaluesthanthenumberofnon-zeroelem entsissaidtobea_____________.A.triangularmatrixB.zeromatrixC.diagonalmatrixD.sparsematrix7【单选题】(10分)WhilerepresentingthesparsematrixA(m×n)withtnon-zerotermsin3-tuplesform,the sizeofthematrixbecomesA.t×nB.m×nC.3×tD.(t+1)×38【单选题】(10分)Consideringasparseof m×n matrixwith t non-zeroterms,in FAST_TRANSPOSE algorithm,thesi zeofone-dimensionalarray(SorT)isequalto:A.n+tB.mC.nt9【单选题】(10分)Consideringasparseof m×n matrixwith t non-zeroterms,thetimecomplexityof TRANS POSE algorithmis:A.O(n*t)B.O(n+t)C.O(n t)D.O(n-t)10【单选题】(10分)Whichofthefollowingstatementistrueregarding TRANSPOSE and FAST_TRANSPOSE algorit hms.A.NoneoftheaboveB.The TRANSPOSE algorithmisslowerthan FAST_TRANSPOSEC.TheTRANSPOSEalgorithmisfasterthanFAST_TRANSPOSETimecomplexitiesofTRANSPOSEandFAST_TRANSPOSEaresame第三章测试1【单选题】(10分) Theelementisinsertedfirstandwillberemovedlastin_____________.A.queueB.stackC.noneoftheaboveD.linkedlist2【单选题】(10分)Theexpression1*2^3*4^5*6isevaluatedas(^isforpower,asin a^b=a b):A.49152B.173458C.162^30D.32^303【单选题】(10分) Thedatastructurerequiredtocheckwhetheranexpressioncontainsbalancedparenthesisis?A.TreeB.ArrayC.QueueD.Stack4【单选题】(10分)Thepostfixformof A*B+C/D is?A.AB*CD/+B.ABCD+/*C.A*BC+/DD.5【单选题】(10分) Whichdatastructureisneededtoconvertinfixnotationtopostfixnotation?A.StackB.BranchC.QueueD.Tree6【单选题】(10分) Transformthefollowinginfixexpressiontoprefixform.((C*2)+1)/(A+B)A./+*C21+ABB.AB+12C*+/C.NoneoftheaboveD.7【单选题】(10分)Transformthefollowinginfixexpressiontopostfixform.(A+B)*(C-D)/EA.AB+CD-*E/B.AB*C+D/-C.AB+CD*-/ED.ABC*CD/-+8【单选题】(10分) Astackisadatastructureinwhichallinsertionsanddeletionsaremaderespectivelyat:A.atanypositionB.boththeendsC.inthemiddleD.oneend9【单选题】(10分) Whichofthefollowingapplicationsmayuseastack?:A.AlloftheaboveB.SyntaxanalyzerforacompilerC.AparenthesisbalancingprogramD.Keepingtrackoflocalvariablesatruntime10【单选题】(10分) Whichofthefollowingstatementiscorrect.A.NoneoftheaboveB. ApostfixexpressionismerelythereverseoftheprefixexpressionC.PostfixandprefixexpressionsuseparenthesisD. Apostfixexpressionisnotthereverseoftheprefixexpression第四章测试1【单选题】(10分) Aqueueisadatastructureinwhichallinsertionsanddeletionsaremaderespectivelyat:A.rearandfrontB.frontandrearC.rearandrearD.frontandfront2【单选题】(10分) Thefollowingdatastructureisusedforschedulingofjobsduringbatchprocessingincomputer s.A.stackB.queueC.linkedlistD.tree3【单选题】(10分) Inaqueuethedeletionsaretakeplaceat_________.A.NoneoftheaboveB.topC.frontD.rear4【单选题】(10分) Inaqueuetheinsertionsaretakeplaceat_________.A.rearB.topC.NoneoftheaboveD.front5【单选题】(10分)Incircularqueue,thefrontwillalwayspointtooneposition__________fromthefirstelementint hequeue.A.leftB.clockwiseC.counterclockwiseD.right6【单选题】(10分)Whichofthefollowingisnotthetypeofqueue.A.priorityqueueB.doubleendedqueueC.circularqueueD.singleendedqueue7【单选题】(10分)Oneoftheadvantageofcircularqueueis_____________.A.NoneoftheaboveB.effectiveuseofmemoryC.easiercomputationsD.deletingelementsbasedonpriority8【单选题】(10分) Whatisthetimecomplexityofalinearqueuehaving n elements?A.O(nlogn)B.O(logn)C.O(1)D.O(n)9【单选题】(10分)Whatisadequeue?A.AqueueimplementedwithadoublylinkedlistB.Aqueuewithinsert/deletedefinedforfrontendofthequeueC.Aqueuewithinsert/deletedefinedforbothfrontandrearendsofthequeueD. Aqueueimplementedwithbothsinglyanddoublylinkedlist10【单选题】(10分) Onedifferencebetweenaqueueandastackis:A.Queuesrequiredynamicmemory,butstacksdonot.B.Stacksrequiredynamicmemory,butqueuesdonot.C.Stacksusetwoendsforaddinganddeleting,butqueuesuseone.D.Queuesusetwoendsforaddinganddeleting,butstacksuseone.第五章测试1【单选题】(10分) Alinearlistofdataelementswhereeachelementcallednodeisgivenbymeansofpointeriscalle dA.nodelistB.linkedlistC.queueD.stack2【单选题】(10分)Consideranimplementationofunsortedsinglylinkedlist.Supposeithasrepresentationwhich aheadpointeronly.Giventherepresentation,whichofthefollowingoperationcanbeimpleme ntedinO(1)time?(I).Insertionatthefrontofthelinkedlist.(II).Insertionattheendofthelinkedlist.(III).Deletionofthefrontnodeofthelinkedlist.(IV).Deletionofthelastnodeofthelinkedlist.A.IandIIIB.I,II,andIIIC.I,II,andIVD.IandII3【单选题】(10分) Whatisthetimecomplexitytocountthenumberofelementsinthelinkedlist?A.O(1)B.O(n2)C.O(logn)D.O(n)4【单选题】(10分) InwhichofthefollowinglinkedliststherearenoNULLlinks?A.DoublylinkedlistB.NoneoftheaboveC.SinglylinkedlistD.Circularlinkedlist5【单选题】(10分)Indoublylinkedlists,traversalcanbeperformed?A.OnlyinforwarddirectionB.InbothdirectionsC.NoneD.Onlyinreversedirection6【单选题】(10分)Whatkindoflistisbesttoanswerquestionssuchas:“Whatistheitematposition n?”A.Singly-linkedlistsB.NoneoftheaboveC.Doubly-linkedlistsD.Listimplementedwithanarray7【单选题】(10分) Inasinglylinkedlistwhichoperationdependsonthelengthofthelist.A.DeletethelastelementofthelistB.AddanelementbeforethefirstelementofthelistC.DeletethefirstelementofthelistD.Interchangethefirsttwoelementsofthelist8【单选题】(10分)Thelinkfieldinanodecontains:A.dataofcurrentnodeB.addressofthenextnodeC.dataofnextnodeD.dataofpreviousnode9【单选题】(10分)Linkedlistdatastructureoffersconsiderablesavingin:A.SpaceutilizationB.ComputationaltimeC.SpaceutilizationandcomputationaltimeD.Noneoftheabove10【单选题】(10分) Alinearlistinwhicheachnodehaspointerstopointtothepredecessorandsuccessorsnodesis calledas:A.CircularlinkedlistB.Singly-linkedlistsC.Doubly-linkedlistsD.Linearlinkedlist第六章测试1【单选题】(10分) Torepresenthierarchicalrelationshipbetweenelements,whichdatastructureissuitable?A.treeB.arrayC.stackD.queue2【单选题】(10分) Whatisthemaximumnumberchildrenthatabinarytreenodecanhave?A.1B.C.2D.33【单选题】(10分) TheinordertraversaloftreewillyieldasortedlistingofelementsoftreeinA.NoneoftheaboveB.BinarysearchtreesC.BinarytreesD.Heaps4【单选题】(10分) Ifwestorethenodesofabinarytreeinanarraywithindexstartingfromzero,therightchil dofanodehavingindex n canbeobtainedat:A.2n+2B.n+1C.(n-1)/2D.2n+15【单选题】(10分) WhichofthefollowingtraversaloutputsthedatainsortedorderinaBST?A.InorderB.PostorderC.PreorderD.Levelorder6【单选题】(10分)Toobtainaprefixexpression,whichofthefollowingtraversalsisused?A.LevelorderB.InorderC.PostorderD.Preorder7【单选题】(10分) Themaximumnumberofnodesinatreeforwhichpostorderandpreordertraversalsmaybeequ altois_______.A.B.3C.2D.18【单选题】(10分)Supposethenumbers7,5,1,8,3,6,0,9,4,2areinsertedinthatorderintoaninitiallyempty BinarySearchTree.TheBinarySearchTreeusestheusualorderingonnaturalnumbers.What istheinordertraversalsequenceoftheresultanttree?A.024*******B.7510324689C.0123456789D.98642301579【单选题】(10分)Afullbinarytreeisatreewhere________________.A.eachnodehasexactlyzeroortwochildren.B.eachnodehasexactlytwochildrenC.alltheleavesareatthesamelevel.D.eachnodehasexactlyoneortwochildren.10【单选题】(10分) Acompletebinarytreeisatreewhere________________.A. everylevelofthetreeiscompletelyfilledexceptthelastlevelB.eachnodehasexactlytwochildrenC. eachnodehasexactlyzeroortwochildrenD. eachnodehasexactlyoneortwochildren。

森林生态学碳固定关键科学问题

森林生态学碳固定关键科学问题英文回答：Key scientific questions in forest ecology carbon sequestration:1. How does forest structure and composition influence carbon sequestration? Forests are composed of various tree species with different growth rates and carbon storage capacities. Understanding how these factors interact and influence carbon sequestration is crucial. For example, a mixed-species forest with a diverse range of tree sizes and ages may have higher carbon sequestration rates compared to a monoculture forest.2. What are the effects of disturbance events on carbon sequestration in forests? Forests are subject to various disturbances such as wildfires, insect outbreaks, and logging activities. These disturbances can release large amounts of carbon stored in vegetation and soils. Studyingthe recovery process and the long-term effects of disturbances on carbon sequestration can provide insights into forest management strategies. For instance, a study on the impact of wildfire on carbon sequestration in a pine forest may reveal the importance of post-fire reforestation efforts.3. How do climate change and atmospheric CO2 concentrations affect forest carbon sequestration? Rising atmospheric CO2 levels and changing climatic conditions can influence forest productivity and carbon sequestration. Increased CO2 concentrations may enhance photosynthesis and promote tree growth, leading to higher carbon sequestration rates. However, the effects of climate change on forest carbon dynamics can be complex and vary across different forest types and regions. For example, a study on the response of a tropical rainforest to changing precipitation patterns and CO2 levels may reveal the potential impacts on carbon sequestration in these ecosystems.4. What is the role of soil carbon in forest carbon sequestration? Soils play a crucial role in carbonsequestration, as they can store large amounts of organic carbon. Understanding the processes that control soil carbon storage and turnover in forests is essential. For instance, a study on the effects of land management practices, such as forest thinning or organic matter additions, on soil carbon sequestration can provideinsights into sustainable forest management strategies.5. How do forest disturbances and management practices interact with carbon sequestration? Forest management practices, such as timber harvesting or afforestation, can influence carbon sequestration rates. Additionally, disturbances like fire or insect outbreaks can affect the success of management interventions. Understanding the interactions between forest disturbances and management practices is crucial for optimizing carbon sequestration efforts. For example, a study on the combined effects of timber harvesting and wildfire on carbon sequestration in a managed forest can inform sustainable forest management strategies.中文回答：森林生态学碳固定的关键科学问题：1. 森林结构和组成如何影响碳固定？森林由不同生长速度和碳储存能力的树种组成。

林木杂交育种名词解释

林木杂交育种名词解释英文回答：Hybrid breeding of forest trees refers to the process of crossing two different tree species to create offspring with desirable traits from both parents. This method is commonly used in forestry to improve characteristics such as growth rate, disease resistance, and wood quality.For example, let's say we want to develop a tree that grows faster and produces higher quality timber. We might cross a species known for its fast growth with another species known for its superior wood properties. Byselecting the best individuals from the resulting offspring and continuing to breed them over several generations, we can eventually create a hybrid tree that combines the desired traits.Hybrid breeding can also help increase geneticdiversity within tree populations, making them moreresilient to environmental stresses such as pests, diseases, and climate change. This can ultimately lead to healthier forests and more sustainable timber production.Overall, hybrid breeding of forest trees is a valuable tool for foresters looking to improve the performance and adaptability of tree species in commercial plantations and natural forests.中文回答：林木杂交育种是指将两种不同的树种进行交配，以创造具有双亲理想特征的后代的过程。

高一英语植物种群分布单选题50题

高一英语植物种群分布单选题50题1. In a forest, we find oak trees. If the oak trees are very close to each other, which term best describes this situation?A. High population densityB. Low population densityC. Random distributionD. Uniform distribution答案：A。

解析：种群密度（population density）是指单位面积或单位体积内的个体数量。

在森林里橡树彼此很靠近，这意味着单位面积内橡树的数量多，这就是高种群密度（High population density）。

B 选项低种群密度表示单位面积内个体数量少，不符合题意。

C选项随机分布（Random distribution）是指个体在空间的分布是随机的，没有特定规律，这里说树木彼此靠近不是随机分布。

D选项均匀分布Uniform distribution）是指个体等距离分布，这里没有提到等距离，所以也不符合。

2. There is a meadow with daisies. The daisies seem to be spread out evenly across the meadow. What kind of distribution is this?A. Clumped distributionB. High population densityC. Uniform distributionD. Random distribution答案：C。

解析：均匀分布（Uniform distribution）是指个体等距离分布。

在草地里雏菊均匀地分布在草地上，这符合均匀分布的特点。

A选项集群分布 Clumped distribution）是指个体成群、成簇地分布在一起，与均匀分布相反。

1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。

Mixed Parsing of Tree Insertion and TreeAdjoining GrammarsMiguel A.Alonso1,Vicente Carrillo2,and V´ıctor J.D´ıaz21Departamento de Computaci´o n,Universidade da Coru˜n aCampus de Elvi˜n a s/n,15071La Coru˜n a(Spain)alonso@udc.es2Departamento de Lenguajes y Sistemas Inform´a ticos,Universidad de SevillaAvda.Reina Mercedes s/n,41012Sevilla(Spain){carrillo,vjdiaz}@.esAbstract.Adjunction is a powerful operation that makes Tree Adjoin-ing Grammar(TAG)useful for describing the syntactic structure of natu-ral languages.In practice,a large part of wide coverage grammars writtenfollowing the TAG formalism is formed by trees that can be combinedby means of the simpler kind of adjunction deﬁned for Tree InsertionGrammar.In this paper,we describe a parsing algorithm that makes useof this characteristic to reduce the practical complexity of TAG parsing:the expensive standard adjunction operation is only considered in thosecases in which the simpler cubic-time adjunction cannot be applied.1IntroductionTree Adjoining Grammar(TAG)[4]and Tree Insertion Grammar(TIG)[6]are grammatical formalisms that make use of a tree-based operation called adjunc-tion.However,adjunctions are more restricted in the case of TIG than in the case of TAG,which has important consequences with respect to the set of languages generated and the worst-case complexity of parsing algorithms:–TAG generates tree adjoining languages,a strict superset of context-free languages,and the complexity of parsing algorithms is in O(n6)for time and in O(n4)for space with respect to the length n of the input string.–TIG generates context-free languages and can be parsed in O(n3)for time and in O(n2)for space.Albeit the powerful adjunction provided by TAG makes it useful for describ-ing the syntax of natural languages,most of the trees involved in wide coverage grammars like XTAG[3]do not make use of such operation,and so a large por-tion of XTAG is in fact a TIG[6].As the full power of a TAG parser is only put into practice in adjunctions involving a given set of trees,to apply a parser work-ing in O(n6)time complexity when most of the work can be done by a O(n3) parser seems to be a waste of computing resources.In this paper,we propose a mixed parser that takes the best of both worlds:those parts of the grammar thatcorrespond to a TIG are managed in O(n3)time and O(n2)space complexity, and only those parts of the grammar involving the full kind of adjunction present in TAG are managed in O(n6)time and O(n4)space complexity.1.1Tree Adjoining GrammarsFormally,a TAG is a5-tuple G=(V N,V T,S,I,A),where V N is aﬁnite set of non-terminal symbols,V T aﬁnite set of terminal symbols,S the axiom of the grammar,I aﬁnite set of initial trees and A aﬁnite set of auxiliary trees.I∪A is the set of elementary trees.Internal nodes are labeled by non-terminals and leaf nodes by terminals or the empty stringε,except for just one leaf per auxiliary tree(the foot)which is labeled by the same non-terminal used as the label of its root node.The path in an elementary tree from the root node to the foot node is called the spine of the tree.Fig.1.Adjunction operationNew trees are derived by adjunction:letγbe a tree containing a node Nγlabeled by A and letβbe an auxiliary tree whose root and foot nodes are also labeled by A.Then,the adjunction ofβat the adjunction node Nγis obtained by excising the subtree ofγwith root Nγ,attachingβto Nγand attaching the excised subtree to the foot ofβ.We illustrate the adjunction operation in Fig.1, where we show a simple TAG with two elementary trees:an initial tree rooted S and an auxiliary tree rooted VP.The derived tree obtained after adjoining the VP auxiliary tree on the node labeled by VP located in the initial tree is also shown.We useβ∈adj(Nγ)to denote that a treeβmay be adjoined at node Nγof the elementary treeγ.If adjunction is not mandatory at Nγthen nil∈adj(Nγ)where nil/∈I∪A is a dummy symbol.If adjunction is not allowed at Nγthen {nil}=adj(Nγ).1.2Tree Insertion GrammarsWe can consider the set A as formed by the union of the sets A L,containing left auxiliary trees in which every nonempty frontier node is to the left of the foot node,A R,containing right auxiliary trees in which every nonempty frontier node is to the right of the foot node,and A W,containing wrapping auxiliary trees in which nonempty frontier nodes are placed both to the left and to the right of the foot node.Given an auxiliary tree,we call spine nodes to those nodes placed on the spine and left nodes(resp.right nodes)to those nodes placed to the left(resp.right)of the spine.The set A SL⊆A L(resp.A SR⊆A R)of strongly left(resp.strongly right)auxiliary trees is formed by trees in which no adjunction is permitted on right(resp.left)nodes and only strongly left(resp. right)auxiliary trees are allowed to adjoin on spine nodes.Figure2shows three derived trees resulting from the adjunction of a wrapping,left and right auxiliary tree,respectively.In essence,a TIG is a restricted TAG where auxiliary trees must be either strongly left or strongly right and adjunctions are not allowed in root and foot nodes of auxiliary trees.Wrapping auxiliary treeLeftauxiliary treeRightauxiliary treeFig.2.TAG vs.TIG adjunction operation1.3Notation for Parsing AlgorithmsWe will describe parsing algorithms using Parsing Schemata,a framework for high-level descriptions of parsing algorithms[8].A parsing system for a grammar G and string a1...a n is a triple I,H,D ,with I a set of items which repre-sent intermediate parse results,H an initial set of items called hypothesis that encodes the sentence to be parsed,and D a set of deduction steps that allow new items to be derived from already known items.Deduction steps are of theformη1,...,ηk cond,meaning that if all antecedentsηi of a deduction step are present and the conditions cond are satisﬁed,then the consequentξshould be generated by the parser.A set F⊆I ofﬁnal items represent the recognition of a sentence.A parsing schema is a parsing system parameterized by a grammar and a sentence.In order to describe the parsing algorithms for tree-based formalisms,we must be able to represent the partial recognition of elementary trees.Parsing algorithms for context-free grammars usually denote partial recognition of pro-ductions by dotted productions.We can extend this approach to the case of tree-based grammars by considering each elementary treeγas formed by a set of context-free productions P(γ):a node Nγand its children Nγ1...Nγg are rep-resented by a production Nγ→Nγ1...Nγg.Thus,the position of the dot in the tree is indicated by the position of the dot in a production in P(γ).The elements of the productions are the nodes of the tree.To simplify the description of parsing algorithms we consider an additional production →Rαfor eachα∈I and the two additional productions →Rβand Fβ→⊥for eachβ∈A,where Rβand Fβcorrespond to the root node and the foot node ofβ,respectively.After disabling and⊥as adjunction nodes the generative capability of the grammars remains intact.We introduce also the following notation:given two pairs(p,q)and(i,j)of integers,(p,q)≤(i,j)is satisﬁed if i≤p and q≤j and given two integers p and q we deﬁne p∪q as p if q is undeﬁned and as q if p is undeﬁned,being undeﬁned in other case.2A Mixed Parser for TIG and TAGIn this section we deﬁne a parsing system Mix= I Mix,H Mix,D Mix correspond-ing to a mixed parsing algorithm for TAG and TIG in which the adjunction of strongly left and strongly right auxiliary trees1will be managed by specialized deduction steps,the rest of adjunctions will be managed with the classical de-duction steps included in most of TAG parsers[1].For Mix,we consider a set of items I Mix=I(a)Mix ∪I(b)Mix∪I(c)Mix formed bythe union of the following subsets:–A subset I(a)Mix with items of the form[Nγ→δ•ν,i,j|p,q|adj]such thatNγ→δν∈P(γ),γ∈I∪A,0≤i≤j,(p,q)=(−,−)or(p,q)≤(i,j), and adj∈{true,false}.The two indices with respect to the input string i and j indicate the portion of the input string that has been spanned from δ(seeﬁgure3).Ifγ∈A,p and q are two indices with respect to the input string that indicate that part of the input string recognized by the foot 1Given the set A of a TAG,we can determine the set A SL as follows:ﬁrstly,we determine the set A L examining the frontier of the trees in A and we set A SL:=A L; secondly,we eliminate from A SL those trees that permit adjunctions on nodes to the right of their spine;and thirdly,we iteratively eliminate from A SL those trees that allow adjoining trees in A−A SL on nodes of their spine.A SR is determined in an analogous way.R γFig.3.Graphical representation of itemsnode of γif it is a descendant of δ.In other case p =q =−representing they are undeﬁned.Therefore,this kind of items satisfy one of the following conditions:1.γ∈A −(A SL ∪A SR ),δ= ,(p,q )=(−,−)and δspans the string a i +1...a p F γa q +1...a j2.δ= ,(p,q )=(−,−)and δspans the string a i +1...a j .The last boolean component of items is used to avoid several adjunctions on a node.A value of true indicates that an adjunction has taken place on the node N γand therefore further adjunctions on the same node are forbidden.If adj =true and ν= ,it means that a strongly left auxiliary tree β∈A L has been adjoined at N γ.If adj =true and ν= ,it means that an auxiliary tree has been adjoined at N γ.A value of false indicates that no adjunction was performed on that node.In this case,during future processing this item can play the role of the item recognizing the excised part of an elementary tree to be attached to the foot node of a right auxiliary tree.As a consequence,only one adjunction can take place on a node,as is prescribed by the tree adjoining grammar formalism.–A subset I (b )Mix with items of the form [N γ→•υ,j,j |−,−|false]such that M γ→δν∈P (γ),γ∈I ∪A and 0≤i ≤j .The last boolean component indicates any tree has been adjoined at N γ.–A subset I (b )Mix with items of the form [N γ→•υ,i,j |−,−|true]such that M γ→δν∈P (γ),γ∈I ∪A ,0≤i ≤j and there exists a β∈A SL such that β∈adj(N γ)and R βspans a i +1...a j (i.e.βhas been adjoined at N γ).In this case,i and j indicate the portion of the input string spanned by the left auxiliary tree adjoined at N γ.The hypotheses deﬁned for this parsing system encode the input string in the standard way:H Mix = [a,i −1,i ]|a =a i ,1≤i ≤n .The set of deduction steps is formed by the following subsets:D Mix=D InitMix ∪D ScanMix∪D Mix∪D PredMix∪D CompMix∪D AdjPred Mix ∪D FootPredMix∪D FootCompMix∪D AdjCompMix∪D LAdjPred Mix ∪D LAdjCompMix∪D RAdjPredMix∪D RAdjCompMix∪D LRFootMixThe parsing process starts by creating the items corresponding to productions having the root of an initial tree as left-hand side and the dot in the leftmost position of the right-hand side:D Init Mix =[ →•R,0,0|−,−|false]α∈I∧S=label(Rα)Then,a set of deductive steps in D PredMix and D CompMixtraverse each elementarytree while steps in D ScanMix and D Mix scan input symbols and the empty symbol,respectively:D Pred Mix =[Nγ→δ•Mγν,i,j|p,q|adj][Mγ→•υ,j,j|−,−|false]nil∈adj(Mγ)∨(∃β∈A SL∪A SR,β∈adj(Mγ))D CompMix=[Nγ→δ•Mγν,i,j|p,q|adj],[Mγ→υ•,j,k|p ,q |adj ][Nγ→δMγ•ν,i,k|p∪p ,q∪q |adj]with(nil∈adj(Mγ)∧adj =false)∨(∃β∈A,β∈adj(Mγ)∧adj =true)D ScanMix=[Nγ→δ•Mγν,i,j|p,q|adj],[a,j,j+1][Nγ→δMγ•ν,i,j+1|p,q|adj]a=label(Mγ)D Mix=[Nγ→δ•Mγν,i,j|p,q|adj][Nγ→δMγ•ν,i,j|p,q|adj]=label(Mγ)The rest of steps are in charge of managing adjunction operations.If a strongly left auxiliary treeβ∈A SL can be adjoined at a given node Mγ,a step in D LAdjPredMixstarts the traversal ofβ.Whenβhas been completely traversed,astep in D LAdjCompMix starts the traversal of the subtree corresponding to Mγandsets the last element of the item to true in order to forbid further adjunctions on this node.D LAdjPred Mix =[Mγ→•υ,i,i|−,−|false][ →•R,i,i|−,−|false]β∈adj(Mγ)∧β∈A SLD LAdjComp Mix =[Mγ→•υ,i,i|−,−|false],[ →Rβ•,i,j|−,−|false][M→•υ,i,j|−,−|true]β∈A SL∧β∈adj(Mγ)If a strongly right auxiliary tree β∈A SR can be adjoined at a given node M γ,when the subtree corresponding to this node has been completely traversed,a step in D RAdjPred Mix starts the traversal of the tree β.When βhas been completelytraversed,a step in D RAdjComp Mix updates the input positions spanned by M γtaking into account the part of the input string spanned by β,and sets the last element of the item to true in order to forbid further adjunctions on this node.D RAdjPred Mix =[M γ→υ•,i,j |p,q |false][ →•R ,j,j |−,−|false]β∈A SR ∧β∈adj(M γ)D RAdjComp Mix=[M γ→υ•,i,j |p,q |false],[ →R β•,j,k |−,−|false][M →υ•,i,k |p,q |true]β∈A SR ∧β∈adj(M γ)No special treatment is given to the foot node of strongly left and right auxiliary trees and so,it is simply skipped by a step in the set D LRFoot Mix.D LRFoot Mix =[F β→•⊥,j,j,false][F β→⊥•,j,j,false]β∈A SL ∪A SR A step in D AdjPred Mix predicts the adjunction of an auxiliary tree β∈A −(A SL ∪A SR )in a node of an elementary tree γand starts the traversal of β.Once the foot of βhas been reached,the traversal of βis momentary suspended by a step in D FootPred Mix ,which re-takes the subtree of γwhich must be attached to the foot of β.At this moment,there is no information available about the node in which the adjunction of βhas been performed,so all possible nodes are predicted.When the traversal of a predicted subtree has ﬁnished,a step in D FootComp Mix re-takes the traversal of βcontinuing at the foot node.When the traversal of βis completely ﬁnished,a deduction step in D AdjComp Mix checks if the subtree attached to the foot of βcorresponds with the adjunction node.The adjunction if ﬁnished by a step in D Comp Mix ,taking into account that p and q are instantiated if and only if the adjunction node is on the spine of γ.It is interesting to remark that we follow the approach of [5],splitting the completion of adjunction between D AdjComp Mix and D Comp Mix .D AdjPred Mix =[N γ→δ•M γν,i,j |p,q |adj ]β∈A −(A SL ∪A SR )∧β∈adj(M γ)D FootPred Mix =[F β→•⊥,k,k |−,−|false][M γ→•υ,k,k |−,−|false]β∈A −(A SL ∪A SR )∧β∈adj(M γ)D FootComp Mix =[F β→•⊥,k,k |−,−|false],[M γ→υ•,k,l |p ,q |false][F β→⊥•,k,l |k,l |false]β∈A −(A SL ∪A SR )∧β∈adj(M γ)D AdjComp Mix=[ →R β•,j,m |k,l |false],[M γ→υ•,k,l |p ,q |false][M γ→υ•,j,m |p ,q |true]β∈A −(A SL ∪A SR )∧β∈adj(M γ)The input string belongs to the language deﬁned by the grammar if aﬁnal item in the set F= [ →Rα•,0,n|−,−|false]|α∈I∧S=label(Rα) is generated.3ComplexityThe worst-case space complexity of the algorithm is O(n4),as at most four input positions are stored into items corresponding to auxiliary trees belonging to A−(A SL∪A SR).Initial trees and strongly left and right auxiliary trees contribute O(n2)to theﬁnal result.With respect to the worst-case time complexity:–TIG adjunction,the adjunction of a strongly left or right auxiliary tree ona node of a tree belonging to I∪A SL∪A SR,is managed in O(n3)by stepsin D RAdjCompMix and D CompMix.–Full TAG adjunction is managed in O(n6)by deduction steps in D AdjCompMix ,which are in charge of dealing with auxiliary trees belonging to A−(A SL∪A SR).In fact,O(n6)is only attained when a wrapping auxiliary tree is adjoined on a spine node of a wrapping auxiliary tree.The adjunction of a wrapping auxiliary tree on a right node of a wrapping auxiliary tree ismanaged in O(n5)due to deduction steps in D CompMix .The same complexityis attained by the adjunction of a strongly right auxiliary tree on a spine orright node of a wrapping auxiliary tree,due to deduction steps in D RAdjCompMix .–Other cases of adjunction,e.g.the adjunction of a strongly left or right auxiliary tree on a spine node of a tree belonging to(A L−A SL)∪(A R−A SR),are managed in O(n4).4Experimental ResultsWe have incorporated the parsing algorithms described in this paper into a naiveimplementation in Prolog of the deductive parsing machine presented in[7].As aﬁrst experiment,we have compared the performance of the Earley-like parsing algorithms for TIG[6]and TAG[1]with respect to TIGs.For thispurpose,we have designed two artiﬁcial TIGs G l(with A SR=∅)and G r(withA SL=∅).For a TIG,the time complexity of the adjunction completion step of a TAG parser is O(n4),in contrast with the O(n2)and O(n3)complexities ofleft and right adjunction completion for a TIG parser,respectively.Therefore,we expected the TIG parser to be considerably faster than the TAG parser.Ineﬀect,for G l we have observed that the TIG parser is up to18times faster thanthe TAG parser,but in the case of G r the diﬀerence becomes irrelevant.These results have been corroborated by a second experiment performed onartiﬁcial TAGs with the Mixed(Mix)and the TAG parser:the performance ofthe Mixed parser improves when strongly left auxiliary trees are involved in theanalysis of the input string.Table1.XTAG results,in seconds,for the TAG and Mixed parsersSentence TAG Mixed ReductionSrini bought a book0.610.4919.67%Srini bought Beth a book0.770.717.79% Srini bought a book at the bookstore0.940.93 1.06% he put the book on the table0.830.7114.46%the sun melted the ice0.710.667.04%the ice melted0.440.3813.64%Elmo borrowed a book0.550.4910.91%a book borrowed0.390.3315.38%he hopes Muriel wins0.930.7717.20% he hopes that Muriel wins 1.26 1.167.94% the man who Muriel likes bought a book 2.14 1.4830.84%the man that Muriel likes bought a book 1.21 1.0414.05% the music should have been being played for the president 1.27 1.260.79%Clove caught a frisbee0.550.4910.91%who caught a frisbee0.550.4420.00%what did Clove catch0.600.558.33% the aardvark smells terrible0.440.3813.64% the emu thinks that the aardvark smells terrible 1.48 1.3210.81% who does the emu think smells terrible0.990.7722.22% who did the elephant think the panda heard the emu said smells terrible3.13 2.3624.60%Herbert is more livid than angry0.500.4412.00% Herbert is more livid and furious than angry0.500.500.00% In a third experiment,we have taken a subset of the XTAG grammar[3], consisting of27elementary trees that cover a variety of English constructions: relative clauses,auxiliary verbs,unbounded dependencies,extraction,etc.In order to eliminate the time spent by uniﬁcation,we have not considered the feature structures of elementary trees.Instead,we have simulated the features using local constraints.Every sentence has been parsed without previousﬁltering of elementary trees.Table1shows the results of this experiment.The application of the Mixed parser results in a reduction in time that varies in percentage from 31%to0%,depending on the kind of trees involved in the analysis of each sentence.5ConclusionWe have deﬁned a parsing algorithm which reduces the practical complexity of TAG parsing by taking into account that a large part of actual TAG grammars can be managed as a TIG.This parsing algorithm does not preserve the correct preﬁx property[5].It is possible to obtain a variant satisfying this property by means of the introduction of an additional element h into items,which is used to indicate the position ofthe input string in which the traversal of the elementary tree involved in each item was started.The worst-case space complexity increases to O(n5)but the worst-case time complexity remains O(n6)if we modify steps AdjComp0and Comp as indicated in[5].The performance of the algorithm could be improved by means of the appli-cation of practical optimizations,such as the replacement of the components pand q of items[Nγ→δ•ν,i,j|p,q|adj]∈I(a)Mix by the list of all adjunctionsthat are still under completion on Nγ[2],albeit this modiﬁcation can increase the worst-case complexity of the algorithm.AcknowledgementsSupported in part by Plan Nacional de Investigaci´o n Cient´ıﬁca,Desarrollo e Innovaci´o n Tecnol´o gica(TIC2000-0370-C02-01),Ministerio de Ciencia y Tec-nolog´ıa(HP2001-0044)and Xunta de Galicia(PGIDT01PXI10506PN). References1.Miguel A.Alonso,David Cabrero,Eric de la Clergerie,and Manuel Vilares.Tabularalgorithms for TAG parsing.In Proc.of EACL’99,Ninth Conference of the European Chapter of the Association for Computational Linguistics,pages150–157,Bergen, Norway,June1999.ACL.2.Eric de la Clergerie.Reﬁning tabular parsers for TAGs.In Proceedings of Lan-guage Technologies2001:The Second Meeting of the North American Chapter of the Association for Computational Linguistics(NAACL’01),pages167–174,CMU, Pittsburgh,PA,USA,June2001.3.Christy Doran,Dania Egedi,Beth Ann Hockey,B.Srinivas,and Martin Zaidel.XTAG system—a wide coverage grammar for English.In Proc.of the15th Inter-national Conference on Computational Linguistics(COLING’94),pages922–928, Kyoto,Japan,August1994.4.Aravind K.Joshi and Yves Schabes.Tree-adjoining grammars.In Grzegorz Rozen-berg and Arto Salomaa,editors,Handbook of Formal Languages.Vol3:Beyond Words,chapter2,pages69–123.Springer-Verlag,Berlin/Heidelberg/New York, 1997.5.Mark-Jan Nederhof.The computational complexity of the correct-preﬁx propertyfor putational Linguistics,25(3):345–360,1999.6.Yves Schabes and Richard C.Waters.Tree insertion grammar:A cubic-timeparsable formalism that lexicalizes context-free grammar without changing the trees putational Linguistics,21(4):479–513,December1995.Also as Tech-nical Report TR-94-13,June1994,Mitsubishi Electric Research Laboratories,Cam-bridge,MA,USA.7.Stuart M.Shieber,Yves Schabes,and Fernando C.N.Pereira.Principles andimplementation of deductive parsing.Journal of Logic Programming,24(1–2):3–36, July-August1995.8.Klaas Sikkel.Parsing Schemata—A Framework for Speciﬁcation and Analysis ofParsing Algorithms.Texts in Theoretical Computer Science—An EATCS Series.Springer-Verlag,Berlin/Heidelberg/New York,1997.。