The Normal Object Scheme Forest with Respect to Conflict-Free Dependencies

合集下载

Computer-Vision计算机视觉英文ppt

At the same time, AI MIT laboratory has attracted many famous scholars from the world to participate in the research of machine vision,which included the theory of machine vision , algorithm and design of system .
Its mainstream research is divided into three stages:
Stage 1: Research on the visual basic method ,which take the model world as the main object;
Stage 2: Research on visual model ,which is based on the computational theory;
the other is to rebuild the three dimensional object according to the two-dimensional projection images .
History of computer vision
1950s: in this period , statistical pattern recognition is most applied in computer vision , it mainly focuse on the analysis and identification of two-dimensional image,such as: optical character recognition, the surface of the workpiece, the analysis and interpretation of the aerial image.

Autodesk Nastran 2023 参考手册说明书

DATINFILE1 ........................................................................................................................................................... 9
FILESPEC ............................................................................................................................................................ 13
DISPFILE ............................................................................................................................................................. 11
File Management Directives – Output File Specifications: .............................................................................. 5
BULKDATAFILE .................................................................................................................................................... 7

离散数学英中名词对照表

离散数学英中名词对照表英文Abel categoryAbel group (commutative group) Abel semigroup Abelian groupabsorption property accessibility relation acyclicaddition principleadequate set of connectives adjacentadjacent matrixadjugateadjunctionaffine planealgebraic closed field algebraic element algebraic extensionalphabetalternating groupannihilatorantecedentanti symmetryanti-isomorphismarc setargumentarityarrangement problem associateassociativeassociative algebraassociatorasymmetricatomatomic formulaaugmenting pigeon hole principle augmenting path automorphism automorphism group of graph auxiliary symbol A 离散数学英文—中文名词axiom of choiceaxiom of equalityaxiom of extensionalityaxiom of infinityaxiom of pairsaxiom of regularityaxiom of replacement for the formulaaxiom of the empty setaxiom of unionB balanced imcomplete block designbarber paradoxbase (base 2 exponential function)base (logarithm function to the base 2)Bell numberBernoulli numberBerry paradoxbiconditionalbijection (one-to-one correspondence)bi-mdulebinary relationbinary operationbinary symmetric channel (BSC)binary treebinomial coefficientbinomial theorembinomial transform bipartite graphblockblockblock codeblock designBondy theoremBoolean algebra Boolean expression Boolean functionBoole homomorophism Boole latticeBoolean matrixBoolean productbound occurrencebound variablebounded latticeBruijn theorem Burnside lemmaC cagecancellation property canonical epimorphism Cantor conjecture Cantor diagonal method Cantor paradoxcapacitycardinal number cardinalityCartesion product of graph Catalan numbercatenationCayley graphCayley theoremceiling functioncell (block)centercertain eventchain (walk) characteristic function characteristic of ring characteristic polynomial check digitsChinese postman problem chromatic number chromatic polynomial circuitcirculant graph circumferenceclassclassical completeness classical consistent cliqueclique numberclose with respect to closed termclosureclosure of graphcode elementcode lengthcode wordcoefficientcoimageco-kernalcoloringcoloring problemcombinationcombination numbercombination with repetationcommon divisorcommon factorcommutativecommutative diagramcommutative ringcommutative seimgroupcomparablecompatible withcomplementcomplement elementcomplement of B with respect to A complementary relation complemented latticecomplete bipartite graphcomplete graphcomplete k-partite graphcomplete latticecomplete matchcomplete n-treecompositecomposite operationcomposition (molecular proposition) composition of graph (lexicographic product) compound statementconcatenation (juxtaposition) concatenation graphconditional statement (implication) congruence relationcongruent toconjectureconjunctive normal form connected component connective connectivityconnectivity relation consecutively consequence (conclusion) conservation of flow consistent (non-contradiction) constructive proofcontain (in)contingencycontinuumcontraction of graph contradiction contravariant functor contrapositiveconversecoproductcorankcorresponding universal map countable (uncountable) countably infinite set counter examplecountingcovariant functorcoveringcovering numbercrossing number of graph cosetcotreecutcut edgecut vertexcyclecycle basiscycle matrixcycle rankcycle spacecycle vectorcyclic groupcyclic indexcyclic permutation cyclic semigroupD De Morgan's law decision procedure decoding table deduction theorem degreedegree sequence derivation algebra Descartes product descendant designated truth value deterministic diagonal functor diagonal matrix diameterdigraphdilemmadirect consequence direct limitdirect sumdirected by inclutiondisconnecteddiscrete Fourier transform discrete graph (null graph) disjoint setdisjunctiondisjunctive normal form disjunctive syllogism distancedistance transitive graph distinguished element distributivedistributive lattice divisibilitydivision subringdivison ringdivisor (factor) dodecahedrondomaindual categorydual formdual graphdual principledual statementdummy variableE eccentricityedge chromatic number edge coloringedge connectivityedge coveringedge covering numberedge cutedge setedge-independence number eigenvalue of graph element (entry) elementary divisor ideal elementary product elementary sumempty graphempty relationempty set endomorphismendpointentry (element) enumeration function epimorphismequipotentequivalenceequivalent category equivalent class equivalent matrix equivalent object equivalent relationerror functionerror patternEuclid algorithmEuclid domainEuler characteristicEuler circuitEuler functionEuler graphEuler numberEuler pathEuler polyhedron formula Euler tourEuler traileven permutationeventeverywhere defined excess capacity existence proof existential generalization existential quantification existential quantifier existential specification explicitextended Fibonacci number extended Lucas number extensionextension field extension graphexterior algebraF facefactorfactorablefactotialfactorizationfaithful (full) functor Ferrers graphFibonacci numberfieldfilterfinite dimensional associative division algebra finite extensionfinite field (Galois field )finite groupfinite setfinitely generated modulefirst order theory with equalityfive-color theoremfive-time-repetitionfixed pointfloor functionflowforestforgetful functorfour-color theorem (conjecture)F-reduced productfree elementfree monoidfree occurrencefree R-modulefree variablefree-Ω-algebrafull n-treefunction schemeG Galileo paradoxGauss coefficientGBN (G?del-Bernays-von Neumann system) GCD (Greatest Common Divisor) generalized Petersen graphgenerating functiongenerating proceduregeneratorgenerator matrixgeneric elementgenusgirthG?del completeness theoremgolden section numbergraceful graphgraceful tree conjecturegraphgraph of first class for edge coloring graph of second class for edge coloring graph rankgraph sequencegreatest common factorgreatest elementgreedy algorithmGrelling paradoxGr?tzsch graphgroupgroup codegroup of graphgrowth of functionHajós conjectureHamilton cycleHamilton graphHamilton pathHarary graphhash functionHasse diagramHeawood graphheightHerschel graphhom functorhomemorphism homomorphism homomorphism image homomorphism of graph hyperoctahedronhypothelical syllogism hypothesis (premise)idealidempotentidentityidentity functionidentity natural transformation imageimbeddingimmediate predcessor immediate successorimpossible eventincidentincident axiomincident matrixinclusion and exclusion principle inclusion relationindegreeindependentindependent number independent setindependent transcendental element indexindirected method H Iindividual variableinduced subgraphinfinite extensioninfinite groupinfinite setinitial endpointinitial objectinjectioninjection functorinjective (one to one mapping) inner faceinner neighbour setinorder searchintegral domainintegral subdomaininternal direct sum intersectionintersection of graph intersection operation intervalinvariant factorinvariant factor idealinverseinverse limitinverse morphisminverse natural transformation inverse operationinverse relationinversioninvertableinvolution property irreflexiveisolated vertexisomorphic categoryisomorphismisomorphism of graphjoinjoin of graphJ Jordan algebraJordan product (anti-commutator)Jordan sieve formulaj-skewjuxtapositionk-chromatic graphk-connected graphk-critical graphk-edge chromatic graphk-edge-connected graphk-edge-critical graph Kanaugh mapkernelKirkman schoolgirl problem Klein 4 groupKonisberge Brudge problem Kruskal's algorithm Kuratowski theoremlabeled graphLah numberLatin rectangleLatin squarelatticelattice homomorphismlawLCM (Least Common Multiple) leader cosetleast elementleafleast upper boundleft (right) identityleft (right) invertible element left (right) moduleleft (right) zeroleft (right) zero divisorleft adjoint functorleft cancellableleft cosetlengthlexicographic orderlLie algebraline- grouplinear array (list)linear graphlinear order (total order)K Llinear order set (chain)logical connective logical followlogically equivanlent logically implies logically valid loopLucas numbermagicmany valued proposition logic map coloring problem matchingmathematical structure matrix representation maximal element maximal idealmaximal outerplanar graph maximal planar graph maximum flow maximum matching maxtermmaxterm normal form (conjunctive normal form)McGee graph meetMenger theorem Meredith graph message word mini term minimal -connected graph minimal polynomial minimal spanning tree Minimanoff paradox minimum distance Minkowski summinterm (fundamental conjunctive form)minterm normal form (disjunctive normal form)M?bius function M?bius ladder M?bius transform (inversion)modal logic modelmodule homomorphismMkmoduler latticemodulusmodus ponensmodus tollensmodule isomorphismmonic morphismmonoidmonomorphismmorphism (arrow)M?bius functionM?bius ladderM?bius transform (inversion)multigraphmultinomial coefficientmultinomial expansion theoremmultiple-error-correcting codemultiplication principlemutually exclusivemultiplication tablemutually orthogonal Latin squareN n-ary operationn-ary productn-ary tree (n-tree)n-tuplenatural deduction systemnatural homomorphismnatural isomorphismnatural transformationnearest neighbernegationneighbour setnext state transition functionnon-associative algebranon-standard logicNorlund formulanormal formnormal modelnormal subgroup (invariant subgroup)n-relationnull graph (discrete graph)null objectnullary operationobjectodd permutationoffspringone to oneone-to-one correspondence (bijection) onto optimal solutionorbitorderorder (lower order,same order) order ideal order relationordered pairOre conditionorientationorthogonal Latin square orthogonal layoutoutarcoutdegreeouter faceouter neighbourouterneighbour setouterplanar graphpancycle graphparallelismparallelism classparentparity-check codeparity-check equationparity-check machineparity-check matrixpartial functionpartial ordering (partial relation) partial order relation partial order set (poset)partitionpartition number of integerpartition number of setPascal formulapathperfect code O Pperfect t-error-correcting code perfect graph permutationpermutation grouppermutation with repetation Petersen graphp-graphPierce arrowpigeonhole principleplanar graphplane graphPolish formPólya theorempolynomailpolynomial codepolynomial representation polynomial ring positional treepossible worldpostorder searchpower functorpower of graphpower setpredicateprenex normal formpreorder searchpre-ordered setprimary cycle modulePRIM's algorithmprimeprime fieldprime to each otherprimitive connectiveprimitive elementprimitive polynomialprincipal idealprincipal ideal domainprinciple of dualityprinciple of mathematical induction principle of redundancy probabilisticprobability (theory)productproduct categoryproduct partial orderproduct-sum formproof (deduction)proof by contraditionproper coloringproper factorproper filterproper subgroupproperly inclusive relationproposition (statement)propositional constantpropositional formula (well-formed formula,wff) propositional functionpropositional variablepseudocodepullbackpushoutquantification theoryquantifierquasi order relationquaternionquotient (difference) algebraquotient algebraquotient field (field of fraction)quotient groupquotient modulequotient ring (difference ring , residue ring) quotient set Ramsey graph Ramsey number Ramsey theorem rangerankreachability reconstruction conjecture recursive redundant digits reflexiveregular expression regular graph R Qregular representationrelation matrixrelative setremainderreplacement theoremrepresentationrepresentation functorrestricted proposition formrestrictionretractionreverse Polish formRichard paradoxright adjoint functorright cancellableright factorright zero divisonringring of endomorphismring with unity elementR-linear independencerooted treeroot fieldrule of inferenceRussell paradoxS sample spacesatisfiablesaturatedscopesearchingsectionself-complement graphsemantical completenesssemantical consistentsemigroupseparable elementseparable extensionsequencesequentsequentialSheffer strokesiblingssimple algebraic extensionsimple cyclesimple extensionsimple graphsimple pathsimple proposition (atomic proposition) simple transcental extension simplicationsinkslopesmall categorysmallest element Socrates argument soundness (validity) theorem sourcespanning subgraph spanning treespectra of graphspetral radiussplitting fieldsquare matrixstandard modelstandard monomil statement (proposition) Steiner tripleStirling numberStirling transformstrong induction subalgebrasubcategorysubdirect product subdivison of graph subfieldsubformulasubdivision of graph subgraphsubgroupsub-modulesubmonoidsublatticesubrelationsubringsub-semigroup subscript。

随机森林GLS包使用指南说明书

How to use RandomForestsGLSThe package RandomForestsGLSﬁts non-linear regression models on dependent data with Generalised Least Square(GLS)based Random Forest(RF-GLS)detailed in Saha,Basu and Datta(2020)https: ///abs/2007.15421.We will start by loading the RandomForestsGLS R package.library(RandomForestsGLS)Next,we discuss how the RandomForestsGLS package can be used for estimation and prediction in a non-linear regression setup under correlated errors in diﬀerent scenarios.1.Spatial DataWe consider spatial point referenced data with the following model:y i=m(x i)+w(s i)+ i;where,y i,x i respectively denotes the observed response and the covariate corresponding to the i th observed location s i.m(x i)denotes the covariate eﬀect,spatial random eﬀect,w(s)accounts for spatial dependence beyond covariates,and accounts for the independent and identically distributed random Gaussian noise. In the spatial mixture model setting,the package RandomForestsGLS allows forﬁtting m(.)using RF-GLS. Spatial random eﬀects are modeled using Gaussian Process as is the practice.For modelﬁtting,we use the computationally convenient Nearest Neighbor Gaussian Process(NNGP)(Datta,Banerjee,Finley,and Gelfand(2016)).Along with prediction of the covariate eﬀect(mean function)m(.)we also oﬀer kriging based prediction of spatial responses at new location.IllustrationWe simulate a data from the following model:y i=10sin(πx i)+w(s i)+ i; ∼N(0,τ2I),τ2=0.1;w∼exponential GP;σ2=10;φ=1. Here,the mean function is E(Y)=10sin(πX);w accounts for the spatial correlation,which is generated as a exponential Gaussian process with spatial varianceσ2=10and spatial correlation decayφ=1;and is the i.i.d random noise with varianceτ2=0.1,which is also called the nugget in spatial literature.For illustration purposes,we simulate with n=200:rmvn<-function(n,mu=0,V=matrix(1)){p<-length(mu)if(any(is.na(match(dim(V),p))))stop("Dimension not right!")D<-chol(V)t(matrix(rnorm(n*p),ncol=p)%*%D+rep(mu,rep(n,p)))}set.seed(5)n<-200coords<-cbind(runif(n,0,1),runif(n,0,1))set.seed(2)x<-as.matrix(runif(n),n,1)sigma.sq=10phi=1tau.sq=0.1D<-as.matrix(dist(coords))R<-exp(-phi*D)w<-rmvn(1,rep(0,n),sigma.sq*R)y<-rnorm(n,10*sin(pi*x)+w,sqrt(tau.sq))ModelﬁttingIn the package RandomForestsGLS,the working precision matrix used in the GLS-loss are NNGP approxima-tions of precision matrices corresponding to Matérn covariance function.In order toﬁt the model,the code requires:•Coordinates(coords):an n×2matrix of2-dimensional locations.•Response(y):an n length vector of response at the observed coordinates.•Covariates(X):an n×p matrix of the covariates in the observation coordinates.•Covariates for estimation(Xtest):an ntest×p matrix of the covariates where we want to estimate the function.Must have identical variables as that of X.Default is X.•Minimum size of leaf nodes(nthsize):We recommend not setting this value too small,as that will lead to very deep trees that takes a lot of time to be built and can produce unstable estimates.Default value is20.•The parameters corresponding to the covariance function(detailed afterwards).For the details on choice of other parameters,please refer to the helpﬁle of the code RFGLS_estimate_spatial, which can be accessed with?RFGLS_estimate_spatial.Known Covariance ParametersIf the covariance parameters are known,we set param_estimate=FALSE(default value);the code additionally requires the following:•Covariance Model(cov.model):Supported keywords are:“exponential”,“matern”,“spherical”,and “gaussian”for exponential,Matérn,spherical and Gaussian covariance function respectively.Default value is“exponential”.•σ2(sigma.sq):The spatial variance.Default value is1.•τ2(tau.sq):The nugget.Default value is0.01.•φ(phi):The spatial correlation decay parameter.Default value is5.•ν(nu):The smoothing parameter corresponding to the Matérn covariance function.Default value is0.5.We canﬁt the model as follows:set.seed(1)est_known<-RFGLS_estimate_spatial(coords,y,x,ntree=50,cov.model="exponential",nthsize=20,sigma.sq=sigma.sq,tau.sq=tau.sq,phi=phi)The estimate of the function at the covariates Xtest is given in estimation_reult$predicted.For inter-pretation of the rest of the outputs,please see the helpﬁle of the code RFGLS_estimate_ing covariance models other than exponential model are in beta testing stage.Unknown Covariance ParametersIf the covariance parameters are not known we set param_estimate=TRUE;the code additionally requires the covariance model(cov.model)to be used for parameter estimation prior to RF-GLSﬁtting.Weﬁt the model with unknown covariance parameters as follows.set.seed(1)est_unknown<-RFGLS_estimate_spatial(coords,y,x,ntree=50,cov.model="exponential",nthsize=20,param_estimate=TRUE)Prediction of mean functionGiven aﬁtted model using RFGLS_estimate_spatial,we can estimate the mean function at new covariate values as follows:Xtest<-matrix(seq(0,1,by=1/10000),10001,1)RFGLS_predict_known<-RFGLS_predict(est_known,Xtest)Performance comparisonWe obtain the Mean Integrated Squared Error(MISE)of the estimateˆm from RF-GLS on[0,1]and compare it with that corresponding to the classical Random Forest(RF)obtained using package randomForest(with similar minimum nodesize,nodesize=20,as default nodesize performs worse).We see that our method has a signiﬁcantly smaller MISE.Additionally,we show that the MISE obtained with unknown parameters in RF-GLS is comparable to that of the MISE obtained with known covariance parameters.library(randomForest)set.seed(1)RF_est<-randomForest(x,y,nodesize=20)RF_predict<-predict(RF_est,Xtest)#RF MISEmean((RF_predict-10*sin(pi*Xtest))^2)#>[1]8.36778#RF-GLS MISEmean((RFGLS_predict_known$predicted-10*sin(pi*Xtest))^2)#>[1]0.150152RFGLS_predict_unknown<-RFGLS_predict(est_unknown,Xtest)#RF-GLS unknown MISEmean((RFGLS_predict_unknown$predicted-10*sin(pi*Xtest))^2)#>[1]0.1851928We plot the true m(x)=10sin(πx)along with the loess-smoothed version of estimatedˆm(.)obtained from RF-GLS and RF where we show that RF-GLS estimate approximates m(x)better than that corresponding to RF.rfgls_loess_10<-loess(RFGLS_predict_known$predicted~c(1:length(Xtest)),span=0.1)rfgls_smoothed10<-predict(rfgls_loess_10)rf_loess_10<-loess(RF_predict~c(1:length(RF_predict)),span=0.1)rf_smoothed10<-predict(rf_loess_10)xval<-c(10*sin(pi*Xtest),rf_smoothed10,rfgls_smoothed10)xval_tag<-c(rep("Truth",length(10*sin(pi*Xtest))),rep("RF",length(rf_smoothed10)), rep("RF-GLS",length(rfgls_smoothed10)))plot_data<-as.data.frame(xval)plot_data$Methods<-xval_tagcoval<-c(rep(seq(0,1,by=1/10000),3))plot_data$Covariate<-covallibrary(ggplot2)ggplot(plot_data,aes(x=Covariate,y=xval,color=Methods))+geom_point()+labs(x="x")+labs(y="f(x)")Prediction of spatial responseGiven aﬁtted model using RFGLS_estimate_spatial,we can predict the spatial response/outcome at new locations provided the covariates at that location.This approach performs kriging at a new location using the mean function estimates at the corresponding covariate values.Here we partition the simulated data into training and test sets in4:1ratio.Next we perform prediction on the test set using a modelﬁtted on the training set.est_known_short<-RFGLS_estimate_spatial(coords[1:160,],y[1:160],matrix(x[1:160,],160,1),ntree=50,cov.model="exponential",nthsize=20,param_estimate=TRUE)RFGLS_predict_spatial<-RFGLS_predict_spatial(est_known_short,coords[161:200,],matrix(x[161:200,],40,1))pred_mat<-as.data.frame(cbind(RFGLS_predict_spatial$prediction,y[161:200]))colnames(pred_mat)<-c("Predicted","Observed")ggplot(pred_mat,aes(x=Observed,y=Predicted))+geom_point()+geom_abline(intercept=0,slope=1,color="blue")+ylim(0,16)+xlim(0,16)Misspeciﬁcation in covariance modelThe following example considers a setting when the parameters are estimated from a misspeciﬁed covariance model.We simulate the spatial correlation from a Matérn covariance function with smoothing parameter ν=1.5.Whileﬁtting the RF-GLS,we estimate the covariance parameters using an exponential covariance model(ν=0.5)and show that the obtained MISE can compare favorably to that of classical RF.#Data simulation from matern with nu=1.5nu=3/2R1<-(D*phi)^nu/(2^(nu-1)*gamma(nu))*besselK(x=D*phi,nu=nu)diag(R1)<-1set.seed(2)w<-rmvn(1,rep(0,n),sigma.sq*R1)y<-rnorm(n,10*sin(pi*x)+w,sqrt(tau.sq))#RF-GLS with exponential covarianceset.seed(3)est_misspec<-RFGLS_estimate_spatial(coords,y,x,ntree=50,cov.model="exponential",nthsize=20,param_estimate=TRUE)RFGLS_predict_misspec<-RFGLS_predict(est_misspec,Xtest)#RFset.seed(4)RF_est<-randomForest(x,y,nodesize=20)RF_predict<-predict(RF_est,Xtest)#RF-GLS MISEmean((RFGLS_predict_misspec$predicted-10*sin(pi*Xtest))^2)#>[1]0.1380569#RF MISEmean((RF_predict-10*sin(pi*Xtest))^2)#>[1]2.2956392.Autoregressive Time Series DataRF-GLS can also be used for function estimation in a time series setting under autoregressive errors.We consider time series data with errors from an AR(q)process as follows:y t=m(x t)+e t;e t=qi=1ρi e t−i+ηtwhere,y i,x i denotes the response and the covariate corresponding to the t th time point,e t is an AR(q) pprocess,ηt denotes the i.i.d.white noise and(ρ1,···,ρq)are the model parameters that captures the dependence of e t on(e t−1,···,e t−q).In the AR time series scenario,the package RandomForestsGLS allows forﬁtting m(.)using RF-GLS.RF-GLS exploits the sparsity of the closed form precision matrix of the AR process for modelﬁtting and prediction of mean function m(.).IllustrationHere,we simulate from the AR(1)process as follows:y=10sin(πx)+e;e t=ρe t−1+ηt;ηt∼N(0,σ2);e1=η1;ρ=0.9;σ2=10.Here,E(Y)=10sin(πX);e which is an AR(1)process,accounts for the temporal correlation,σ2denotes the variance of white noise part of the AR(1)process andρcaptures the degree of dependence of e t on e t−1. For illustration purposes,we simulate with n=200:rho<-0.9set.seed(1)b<-rhos<-sqrt(sigma.sq)eps=arima.sim(list(order=c(1,0,0),ar=b),n=n,rand.gen=rnorm,sd=s)y<-c(eps+10*sin(pi*x))ModelﬁttingIn case of time series data,the code requires:•Response(y):an n length vector of response at the observed time points.•Covariates(X):an n×p matrix of the covariates in the observation time points.•Covariates for estimation(Xtest):an ntest×p matrix of the covariates where we want to estimate the function.Must have identical variables as that of X.Default is X.•Minimum size of leaf nodes(nthsize):We recommend not setting this value too small,as that will lead to very deep trees that takes a lot of time to be built and can produce unstable estimates.Default value is20.•The parameters corresponding to the AR process(detailed afterwards).For the details on choice of other parameters,please refer to the helpﬁle of the code RFGLS_estimate_timeseries, which can be accessed with?RFGLS_estimate_timeseries.Known AR process ParametersIf the AR process parameters are known we set param_estimate=FALSE(default value);the code additionally requires lag_params=c(ρ1,···,ρq).We canﬁt the model as follows:set.seed(1)est_temp_known<-RFGLS_estimate_timeseries(y,x,ntree=50,lag_params=rho,nthsize=20) Unknown AR process ParametersIf the AR process parameters are not known,we set param_estimate=TRUE;the code requires the orderof the AR process,which is obtained from the length of the lag_params input vector.Hence if we want to estimate the parameters from a AR(q)process,lag_params should be any vector of length q.Here weﬁt the model with q=1set.seed(1)est_temp_unknown<-RFGLS_estimate_timeseries(y,x,ntree=50,lag_params=rho,nthsize=20,param_estimate=TRUE) Prediction of mean functionThis part of time series data analysis is identical to that corresponding to the spatial data.Xtest<-matrix(seq(0,1,by=1/10000),10001,1)RFGLS_predict_temp_known<-RFGLS_predict(est_temp_known,Xtest)Here also,similar to the spatial data scenario,RF-GLS outperforms classical RF in terms of MISE both with true and estimated AR process parameters.library(randomForest)set.seed(1)RF_est_temp<-randomForest(x,y,nodesize=20)RF_predict_temp<-predict(RF_est_temp,Xtest)#RF MISEmean((RF_predict_temp-10*sin(pi*Xtest))^2)#>[1]7.912517#RF-GLS MISEmean((RFGLS_predict_temp_known$predicted-10*sin(pi*Xtest))^2)#>[1]2.471876RFGLS_predict_temp_unknown<-RFGLS_predict(est_temp_unknown,Xtest)#RF-GLS unknown MISEmean((RFGLS_predict_temp_unknown$predicted-10*sin(pi*Xtest))^2)#>[1]0.8791857Misspeciﬁcation in AR process orderWe consider a scenario where the order of autoregression used for RF-GLS modelﬁtting is mis-speciﬁed.We simulate the AR errors from an AR(2)process andﬁt RF-GLS with an AR(1)process.#Simulation from AR(2)processrho1<-0.7rho2<-0.2set.seed(2)b<-c(rho1,rho2)s<-sqrt(sigma.sq)eps=arima.sim(list(order=c(2,0,0),ar=b),n=n,rand.gen=rnorm,sd=s)y<-c(eps+10*sin(pi*x))#RF-GLS with AR(1)set.seed(3)est_misspec_temp<-RFGLS_estimate_timeseries(y,x,ntree=50,lag_params=0,nthsize=20,param_estimate=TRUE)RFGLS_predict_misspec_temp<-RFGLS_predict(est_misspec_temp,Xtest)#RFset.seed(4)RF_est_temp<-randomForest(x,y,nodesize=20)RF_predict_temp<-predict(RF_est_temp,Xtest)#RF-GLS MISEmean((RFGLS_predict_misspec_temp$predicted-10*sin(pi*Xtest))^2)#>[1]1.723218#RF MISEmean((RF_predict_temp-10*sin(pi*Xtest))^2)#>[1]3.735003ParallelizationFor RFGLS_estimate_spatial,RFGLS_estimate_timeseries,RFGLS_predict and RFGLS_predict_spatial one can also take the advantage of parallelization,contingent upon the availability of multiple cores.The component h in all the functions determines the number of cores to be used.Here we demonstrate an example with h=2.#simulation from exponential distributionset.seed(5)n<-200coords<-cbind(runif(n,0,1),runif(n,0,1))set.seed(2)x<-as.matrix(runif(n),n,1)sigma.sq=10phi=1tau.sq=0.1nu=0.5D<-as.matrix(dist(coords))R<-exp(-phi*D)w<-rmvn(1,rep(0,n),sigma.sq*R)y<-rnorm(n,10*sin(pi*x)+w,sqrt(tau.sq))#RF-GLS model fitting and prediction with parallel computationset.seed(1)est_known_pl<-RFGLS_estimate_spatial(coords,y,x,ntree=50,cov.model="exponential",nthsize=20,sigma.sq=sigma.sq,tau.sq=tau.sq,phi=phi,h=2)RFGLS_predict_known_pl<-RFGLS_predict(est_known_pl,Xtest,h=2)#MISE from single coremean((RFGLS_predict_known$predicted-10*sin(pi*Xtest))^2)#>[1]0.150152#MISE from parallel computationmean((RFGLS_predict_known_pl$predicted-10*sin(pi*Xtest))^2)#>[1]0.150152For RFGLS_estimate_spatial with very small dataset(n)and small number of trees(ntree),communi-cation overhead between the nodes for parallelization outweighs the beneﬁts of the parallel computing hence it is recommended to parallelize only for moderately large n and/or ntree.It is strongly recommended that the max value of h is kept strictly less than the number of total cores available. Parallelization for RFGLS_estimate_timeseries can be addressed identically.For RFGLS_predict and RFGLS_predict_spatial,even for large dataset,single core performance is very fast,hence unless ntest and ntree are very high,we do not recommend using parallelization for RFGLS_predict and RFGLS_predict_spatial.。

风云四号红外高光谱GIIRS中波通道亮温偏差订正

文章编号：1672-8785(2021)05-0039-06风云四号红外高光谱GIIRS中波温王根12陈娇1戴娟3王悦1$.安徽省气象台大气科学与卫星遥感安徽省重点实验室，安徽合肥230031；2.中亚大气科学研究中心，新疆乌鲁木齐830002；3.气，230031)摘要：变分同化风云四号干涉式大气垂直探测仪(Geostationary InterferometricInfrared Sounder,GIIRS)中波通道亮温偏差高，需进行GI-IRS资料偏差&在Harns B A等人提出的“离线”法的,了基于随机(Random Forest,RF)的GIIRS偏差法&在行过程中，基于风云四号多通道扫描成像辐射计(Advanced Geosynchronous Radiation Imager，AGRI)云产品对GIIRS资料进行了测&，经过偏差的GI-IRS亮温偏差高的。

与“离线”法，RF法的效好。

关键词：高光谱GIIRS；偏差订正；“离线”法；随机森林；云检测中图分类号：P407文献标志码：A DOI：10.3969/j.issn.1672-8785.2021.05.007BiasC0rrecti0n0fBrightne s Temperaturesin Medium WaveChannelof FY-4A Infrared Hyperspectral GIIRSWANG Gen GH，CHEN Jiao'，DAI Juan3，WANG Yue1(9.Anhui Key Lab of Atmospheric Science and Satellite Remote Sensing，Anhui MeteorologicalObservatrry，Hefei230031，China；2.Center of Central Asia Atmospheric ScienceResearch，Urumqi830002，China；3.Anhui Climate Center，Hefei230031，China) Abstract：The brightne s temperature bias of the medium wave channel of the variational a s imilation geostationary interferometrc infrared sounder(GIIRS)of F5-4is required to meet the Gaussian clistnbution，so thebias correction of GIIRS data is necessary.Based on Harns B A and Kelly Gs"off-line"method，a method forG I RSbiasco r ectionbasedontherandomforestisdevelopedinthispaper.Inthespecificimplementationproce s theclouddetectionofG I RSdataisca r iedoutbasedontheadvancedgeosynchronousradiationima-ger(AGRI)cloudproductsofFY-4.Theexperimentalresultsshowthatthebrightne s temperaturebiasofG I RS satisfies the assumption of Gau s ian distribution after the bias co r pared with"o f-line"method randomforestmethodhasabe t erco r ectione f ect.收稿日期：2021-01-07基金项目：国家自然科学基金项目(41805080)；中亚大气科学研究基金项目(CAAS202003)；安徽省气象台自立项目(AHMO202007；AHMO202004)作者简介：王根(1983-)，男，江苏泰州人，高级工程师，博士，主要从事卫星资料同化、正则化反问题与人工智能应用等方面的研究&E-mail：203wanggen@Key words:hyperspectral GIIRS；bias correction；"offline"method；random forest；cloud detection数值天气预报是一个初/边值问题&星载高光谱红外探测器通道主要覆盖CO2和HQ 光谱区域&CO?和HQ吸收带提供的温度和湿度值预报的模式变量。

白雪公主英文版

A long time ago, and far away there lived a King and a Queen. They were very happy, except for one thing -- they both longed for a child.The Queen sat sewing by her window one winters day when she suddenly pricked her finger. A drop of blood fell on the snow by the window."Oh, I wish I had a daughter with skin as white as snow, hair as black as ebony and red, red lips," she sighed.Happily the Queens wish came true. In the autumn that year a baby girl was born. Sadly, the Queen died soon after.The little Princess was called Snow White and she grew to be a lovely girl.After many years the King married again. His new wife was beautiful and proud. She liked to use magic and had a magic mirror. She would say:"Mirror, mirror on the wall, who is the fairest of them all"And the mirror would answer:"You, Queen, are the fairest in the land."The Queen was very proud.One day the Queen spoke to the mirror as usual, but this time the answer was different."You, Queen, are fair, but the fairest of all is Snow White."The Queen was very angry to hear this. She plotted and schemed and decided to kill Snow White."Huntsman," she said, one day. " I want you to take Snow White into the forest and kill her. Bring back her heart to prove she is dead.""Your Majesty," said the huntsman, bowing. He was horrified. He didnt want to kill the Princess."Come," he said to Snow White. "We are going into the forest." He took Snow White deeper and deeper into the forest, then he spoke to her."Beware of you stepmother," he said. "She wants you killed. Those were my orders. But I cant kill you. Go deeper into the forest and one day you will be able to return. But be careful. She may find out that youre alive."The huntsman then killed a small deer and took its heart back to the queen. The Queen was very happy. She was now the fairest in the land.Meanwhile, Snow White ran deeper and deeper into the forest. Finally, when it was growing dark she came to a small cottage.She knocked but there was no answer."I wonder who lives here," she said. She pushed the door open.What a sight met her eyes! The little house was very untidy, plates and cups were waiting to be washed up and the table was still laid for breakfast."What a mess," said Snow White, and she bustled around and cleaned the house from top to bottom. She was very surprised to find that there was seven of everything: seven cups, seven plates, seven knives, seven forks and seven beds and chairs.Then Snow White had some bread and cheese. She was very tired, so she went upstairs to the newly cleaned bedroom, and fell asleep on one of the beds.Later on, when the owners of the cottage returned, they couldnt believe their eyes! Their little home was spotless."Who can have done this" they asked.They looked around. Dinner was cooking in the oven and the table was laid."Its a girl," called one, from the top of the stairs. "Shes asleep in our room."Snow White woke up to find seven little men standing by the bed. They were seven dwarfs and they worked in a mine in the nearby hills."Dont be scared," they told her. "How did you find this place Were a long way into the forest."Snow White told her story. The dwarfs were horrified."You must stay here," they said. "Youll be safe here.""Thank you," said Snow White.Snow White was very happy in the forest, and the dwarfs were delighted to look after her. For many days only six dwarfs went to work, because one would stay behind with Snow White, but after a while Snow White convinced them that she would be all right."Promise us that you will keep the door locked," said the dwarfs. "The Queen may find out that youre with us and try to harm you.""The door will stay shut all the time," said Snow White. "I wont open it to anyone."The dwarfs were happy with this and they all went off to their work in the mine.Meanwhile, the Queen, Snow Whites stepmother, was happy in the knowledge that she was the fairest in the land, but one day she thought shed just make sure."Mirror, mirror on the wall, who is the fairest of them all" she asked."You, Queen, are fair, but Snow White is the fairest of all." The mirror showed a picture of Snow White deep in the forest, with the seven dwarfs.The Queen was furious. The huntsman had lied to her."Ill find her myself," muttered the Queen. She used a little magic to make herself look very old. Then she dressed herself in peasant clothes and made her way to the forest.When she was there, she waited for the seven dwarfs to go off to the mine. Snow White stayed inside the house as shed promised the dwarfs."Combs for sale! Combs and ribbons for sale!" called an old peasant woman. "Oh, how lovely," said Snow White. "Please may I see your combs.""Come out into the lovely sunshine," said the peasant woman."I cant, I promised the dwarfs," said Snow White. "Can I look at them over the windowsill""Of course," said the woman. "Heres a pretty one. It will suit your lovely dark hair.""I have no mirror," said Snow White. "Can you put it in for me""Of course," said the woman. "Turn around."Snow White turned and the wicked Queen pushed the comb hard into Snow Whites hair. It was poisoned and Snow White fell to the floor."No more pretty Snow White," said the Queen, changing back to her normal self. "Goodbye my dear."The dwarfs returned home early that night to find Snow White on the floor, looking quite dead. They rushed to her pulling her upright."We should never have left her," said one."It must have been the wicked Queen," said another."Look!" one suddenly said. "Shes waking up." The comb had been dislodged and the poison no longer worked.Snow White woke up to find herself surrounded by the dwarfs."You must be very careful," they told her. "This is the work of the wicked Queen."The Queen had arrived back at the palace and went at one to the mirror.She was very angry with the answer the mirror gave her. How could Snow White still be alive"This time she will die," said the wicked Queen. She spent many days preparing the poison she would use.Snow White was feeling much better and promised she would not talk to anyone. Several weeks passed and everyone felt safe.The dwarfs went to work, and into the forest walked an old woman collecting sticks."Shes selling nothing," said Snow White,"but I will still be careful."A knock at the door took Snow White to the window."May I have a drink" asked the old woman."Theres no harm in that," said Snow White. "Wait there." She handed a glass of water to the woman."Would you like an apple" asked the old woman."No thank you," said Snow White."Theyre lovely and ripe," said the woman. "Try some of mine." She cut a piece from the apple she was eating.It must be safe, thought Snow White. Shes eating the apple too.The Queen had very cleverly poisoned just one half of the apple -- the half she cut of for Snow White.Snow White took one bite and fell, as if dead, onto the floor.The old woman turned back into the wicked Queen."This time your friends wont be able to help you," laughed the Queen."Good-bye!"The seven dwarfs returned from the mine to find Snow White lying on the floor. This time they couldnt revive her. Snow White was dead.They were very sad and they couldnt bear the thought of burying her in the ground, so they made a glass coffin and filled it with sweet-smelling flowers.They put it in a sunny glade in the forest, keeping guard over it day after day. Snow White seemed to be sleeping rather than dead.The Queen had spoken to her mirror and was once again the fairest in the land.One day a young Prince rode through the forest. He reached the glade and saw Snow White in her glass coffin."Shes beautiful," he said and he could think of nothing else but Snow White."She is so beautiful," he said to the dwarfs. "May I take her with me. I promise I will always look after her."The dwarfs could see that the Prince meant it, and agreed. Just as the servants were lifting the glass coffin one of them tripped and jolted Snow White. It was enough to make the piece of apple fall out of her mouth. Within moments she was waking up.The Prince and the seven dwarfs were delighted."Will you marry me" asked the Prince. They prepared to return to his kingdom.Meanwhile, the wicked Queen was looking in her mirror."You, Queen, are fair, but Snow White with her Prince is the fairest of all."The mirror showed the Queen a picture of the Prince with Snow White.This time the Queen was so angry that the magic inside her boiled up and killed her. Snow White would never have to worry about her again.。

纹理物体缺陷的视觉检测算法研究--优秀毕业论文

摘要
在竞争激烈的工业自动化生产过程中，机器视觉对产品质量的把关起着举足轻重的作用，机器视觉在缺陷检测技术方面的应用也逐渐普遍起来。与常规的检测技术相比，自动化的视觉检测系统更加经济、快捷、高效与安全。纹理物体在工业生产中广泛存在，像用于半导体装配和封装底板和发光二极管，现代化电子系统中的印制电路板，以及纺织行业中的布匹和织物等都可认为是含有纹理特征的物体。本论文主要致力于纹理物体的缺陷检测技术研究，为纹理物体的自动化检测提供高效而可靠的检测算法。纹理是描述图像内容的重要特征，纹理分析也已经被成功的应用与纹理分割和纹理分类当中。本研究提出了一种基于纹理分析技术和参考比较方式的缺陷检测算法。这种算法能容忍物体变形引起的图像配准误差，对纹理的影响也具有鲁棒性。本算法旨在为检测出的缺陷区域提供丰富而重要的物理意义，如缺陷区域的大小、形状、亮度对比度及空间分布等。同时，在参考图像可行的情况下，本算法可用于同质纹理物体和非同质纹理物体的检测，对非纹理物体的检测也可取得不错的效果。在整个检测过程中，我们采用了可调控金字塔的纹理分析和重构技术。与传统的小波纹理分析技术不同，我们在小波域中加入处理物体变形和纹理影响的容忍度控制算法，来实现容忍物体变形和对纹理影响鲁棒的目的。最后可调控金字塔的重构保证了缺陷区域物理意义恢复的准确性。实验阶段，我们检测了一系列具有实际应用价值的图像。实验结果表明本文提出的纹理物体缺陷检测算法具有高效性和易于实现性。关键字: 缺陷检测；纹理；物体变形；可调控金字塔；重构
Keywords: defect detection, texture, object distortion, steerable pyramid, reconstruction
II

2021年美国大学生数学建模竞赛题目A--真菌范文六篇(含Matlab源代码)

with the rate of decomposition, several questions may arise to include: Using these two traits, how
do the different fungi interact and decompose ground litter in a fixed patch of land in different
?12122?1142?1???1???11112206?12?2242?2???1???2222110102025wheremoisturetoleranceiswelldeterminedby
2021年美国大学生数学建模竞赛题目A--真菌范文六
篇（含Matlab源代码）
问题A：真菌……………………………………………………………………2
Include predictions about the relative advantages and disadvantages for each species and
combinations of species likely to persist, and do so for different environments including arid,
(difference of each isolate’s competitive ranking and their
moisture niche width, both scaled to [0,1]) of various fungi
and the resulting wood decomposition rate (% mass loss
Your complete solution.

离散数学中英文名词对照表

离散数学中英文名词对照表外文中文AAbel category Abel 范畴Abel group (commutative group) Abel 群(交换群)Abel semigroup Abel 半群accessibility relation 可达关系action 作用addition principle 加法原理adequate set of connectives 联结词的功能完备（全）集adjacent 相邻（邻接）adjacent matrix 邻接矩阵adjugate 伴随adjunction 接合affine plane 仿射平面algebraic closed field 代数闭域algebraic element 代数元素algebraic extension 代数扩域（代数扩张）almost equivalent 几乎相等的alternating group 三次交代群annihilator 零化子antecedent 前件anti symmetry 反对称性anti-isomorphism 反同构arboricity 荫度arc set 弧集arity 元数arrangement problem 布置问题associate 相伴元associative algebra 结合代数associator 结合子asymmetric 不对称的(非对称的)atom 原子atomic formula 原子公式augmenting digeon hole principle 加强的鸽子笼原理augmenting path 可增路automorphism 自同构automorphism group of graph 图的自同构群auxiliary symbol 辅助符号axiom of choice 选择公理axiom of equality 相等公理axiom of extensionality 外延公式axiom of infinity 无穷公理axiom of pairs 配对公理axiom of regularity 正则公理axiom of replacement for the formula Ф关于公式Ф的替换公式axiom of the empty set 空集存在公理axiom of union 并集公理Bbalanced imcomplete block design 平衡不完全区组设计barber paradox 理发师悖论base 基Bell number Bell 数Bernoulli number Bernoulli 数Berry paradox Berry 悖论bijective 双射bi-mdule 双模binary relation 二元关系binary symmetric channel 二进制对称信道binomial coefficient 二项式系数binomial theorem 二项式定理binomial transform 二项式变换bipartite graph 二分图block 块block 块图（区组）block code 分组码block design 区组设计Bondy theorem Bondy 定理Boole algebra Boole 代数Boole function Boole 函数Boole homomorophism Boole 同态Boole lattice Boole 格bound occurrence 约束出现bound variable 约束变量bounded lattice 有界格bridge 桥Bruijn theorem Bruijn 定理Burali-Forti paradox Burali-Forti 悖论Burnside lemma Burnside 引理Ccage 笼canonical epimorphism 标准满态射Cantor conjecture Cantor 猜想Cantor diagonal method Cantor 对角线法Cantor paradox Cantor 悖论cardinal number 基数Cartesion product of graph 图的笛卡儿积Catalan number Catalan 数category 范畴Cayley graph Cayley 图Cayley theorem Cayley 定理center 中心characteristic function 特征函数characteristic of ring 环的特征characteristic polynomial 特征多项式check digits 校验位Chinese postman problem 中国邮递员问题chromatic number 色数chromatic polynomial 色多项式circuit 回路circulant graph 循环图circumference 周长class 类classical completeness 古典完全的classical consistent 古典相容的clique 团clique number 团数closed term 闭项closure 闭包closure of graph 图的闭包code 码code element 码元code length 码长code rate 码率code word 码字coefficient 系数coimage 上象co-kernal 上核coloring 着色coloring problem 着色问题combination number 组合数combination with repetation 可重组合common factor 公因子commutative diagram 交换图commutative ring 交换环commutative seimgroup 交换半群complement 补图(子图的余) complement element 补元complemented lattice 有补格complete bipartite graph 完全二分图complete graph 完全图complete k-partite graph 完全k-分图complete lattice 完全格composite 复合composite operation 复合运算composition (molecular proposition) 复合（分子）命题composition of graph (lexicographic product)图的合成（字典积）concatenation (juxtaposition) 邻接运算concatenation graph 连通图congruence relation 同余关系conjunctive normal form 正则合取范式connected component 连通分支connective 连接的connectivity 连通度consequence 推论（后承）consistent (non-contradiction) 相容性（无矛盾性）continuum 连续统contraction of graph 图的收缩contradiction 矛盾式（永假式）contravariant functor 反变函子coproduct 上积corank 余秩correct error 纠正错误corresponding universal map 对应的通用映射countably infinite set 可列无限集（可列集）covariant functor (共变)函子covering 覆盖covering number 覆盖数Coxeter graph Coxeter 图crossing number of graph 图的叉数cuset 陪集cotree 余树cut edge 割边cut vertex 割点cycle 圈cycle basis 圈基cycle matrix 圈矩阵cycle rank 圈秩cycle space 圈空间cycle vector 圈向量cyclic group 循环群cyclic index 循环（轮转）指标cyclic monoid 循环单元半群cyclic permutation 圆圈排列cyclic semigroup 循环半群DDe Morgan law De Morgan 律decision procedure 判决过程decoding table 译码表deduction theorem 演绎定理degree 次数,次(度)degree sequence 次(度)序列derivation algebra 微分代数Descartes product Descartes 积designated truth value 特指真值detect errer 检验错误deterministic 确定的diagonal functor 对角线函子diameter 直径digraph 有向图dilemma 二难推理direct consequence 直接推论（直接后承）direct limit 正向极限direct sum 直和directed by inclution 被包含关系定向discrete Fourier transform 离散 Fourier 变换disjunctive normal form 正则析取范式disjunctive syllogism 选言三段论distance 距离distance transitive graph 距离传递图distinguished element 特异元distributive lattice 分配格divisibility 整除division subring 子除环divison ring 除环divisor (factor) 因子domain 定义域Driac condition Dirac 条件dual category 对偶范畴dual form 对偶式dual graph 对偶图dual principle 对偶原则(对偶原理) dual statement 对偶命题dummy variable 哑变量（哑变元）Eeccentricity 离心率edge chromatic number 边色数edge coloring 边着色edge connectivity 边连通度edge covering 边覆盖edge covering number 边覆盖数edge cut 边割集edge set 边集edge-independence number 边独立数eigenvalue of graph 图的特征值elementary divisor ideal 初等因子理想elementary product 初等积elementary sum 初等和empty graph 空图empty relation 空关系empty set 空集endomorphism 自同态endpoint 端点enumeration function 计数函数epimorphism 满态射equipotent 等势equivalent category 等价范畴equivalent class 等价类equivalent matrix 等价矩阵equivalent object 等价对象equivalent relation 等价关系error function 错误函数error pattern 错误模式Euclid algorithm 欧几里德算法Euclid domain 欧氏整环Euler characteristic Euler 特征Euler function Euler 函数Euler graph Euler 图Euler number Euler 数Euler polyhedron formula Euler 多面体公式Euler tour Euler 闭迹Euler trail Euler 迹existential generalization 存在推广规则existential quantifier 存在量词existential specification 存在特指规则extended Fibonacci number 广义 Fibonacci 数extended Lucas number 广义Lucas 数extension 扩充（扩张）extension field 扩域extension graph 扩图exterior algebra 外代数Fface 面factor 因子factorable 可因子化的factorization 因子分解faithful (full) functor 忠实（完满）函子Ferrers graph Ferrers 图Fibonacci number Fibonacci 数field 域filter 滤子finite extension 有限扩域finite field (Galois field ) 有限域（Galois 域）finite dimensional associative division algebra有限维结合可除代数finite set 有限（穷）集finitely generated module 有限生成模first order theory with equality 带符号的一阶系统five-color theorem 五色定理five-time-repetition 五倍重复码fixed point 不动点forest 森林forgetful functor 忘却函子four-color theorem(conjecture) 四色定理（猜想）F-reduced product F-归纳积free element 自由元free monoid 自由单元半群free occurrence 自由出现free R-module 自由R-模free variable 自由变元free-Ω-algebra 自由Ω代数function scheme 映射格式GGalileo paradox Galileo 悖论Gauss coefficient Gauss 系数GBN (Gödel-Bernays-von Neumann system)GBN系统generalized petersen graph 广义 petersen 图generating function 生成函数generating procedure 生成过程generator 生成子（生成元）generator matrix 生成矩阵genus 亏格girth （腰）围长Gödel completeness theorem Gödel 完全性定理golden section number 黄金分割数（黄金分割率）graceful graph 优美图graceful tree conjecture 优美树猜想graph 图graph of first class for edge coloring 第一类边色图graph of second class for edge coloring 第二类边色图graph rank 图秩graph sequence 图序列greatest common factor 最大公因子greatest element 最大元（素）Grelling paradox Grelling 悖论Grötzsch graph Grötzsch 图group 群group code 群码group of graph 图的群HHajós conjecture Hajós 猜想Hamilton cycle Hamilton 圈Hamilton graph Hamilton 图Hamilton path Hamilton 路Harary graph Harary 图Hasse graph Hasse 图Heawood graph Heawood 图Herschel graph Herschel 图hom functor hom 函子homemorphism 图的同胚homomorphism 同态（同态映射）homomorphism of graph 图的同态hyperoctahedron 超八面体图hypothelical syllogism 假言三段论hypothese (premise) 假设(前提)Iideal 理想identity 单位元identity natural transformation 恒等自然变换imbedding 嵌入immediate predcessor 直接先行immediate successor 直接后继incident 关联incident axiom 关联公理incident matrix 关联矩阵inclusion and exclusion principle 包含与排斥原理inclusion relation 包含关系indegree 入次（入度）independent 独立的independent number 独立数independent set 独立集independent transcendental element 独立超越元素index 指数individual variable 个体变元induced subgraph 导出子图infinite extension 无限扩域infinite group 无限群infinite set 无限（穷）集initial endpoint 始端initial object 初始对象injection 单射injection functor 单射函子injective (one to one mapping) 单射（内射）inner face 内面inner neighbour set 内（入）邻集integral domain 整环integral subdomain 子整环internal direct sum 内直和intersection 交集intersection of graph 图的交intersection operation 交运算interval 区间invariant factor 不变因子invariant factor ideal 不变因子理想inverse limit 逆向极限inverse morphism 逆态射inverse natural transformation 逆自然变换inverse operation 逆运算inverse relation 逆关系inversion 反演isomorphic category 同构范畴isomorphism 同构态射isomorphism of graph 图的同构join of graph 图的联JJordan algebra Jordan 代数Jordan product (anti-commutator) Jordan乘积（反交换子）Jordan sieve formula Jordan 筛法公式j-skew j-斜元juxtaposition 邻接乘法Kk-chromatic graph k-色图k-connected graph k-连通图k-critical graph k-色临界图k-edge chromatic graph k-边色图k-edge-connected graph k-边连通图k-edge-critical graph k-边临界图kernel 核Kirkman schoolgirl problem Kirkman 女生问题Kuratowski theorem Kuratowski 定理Llabeled graph 有标号图Lah number Lah 数Latin rectangle Latin 矩形Latin square Latin 方lattice 格lattice homomorphism 格同态law 规律leader cuset 陪集头least element 最小元least upper bound 上确界（最小上界）left (right) identity 左（右）单位元left (right) invertible element 左（右）可逆元left (right) module 左（右）模left (right) zero 左（右）零元left (right) zero divisor 左（右）零因子left adjoint functor 左伴随函子left cancellable 左可消的left coset 左陪集length 长度Lie algebra Lie 代数line- group 图的线群logically equivanlent 逻辑等价logically implies 逻辑蕴涵logically valid 逻辑有效的（普效的）loop 环Lucas number Lucas 数Mmagic 幻方many valued proposition logic 多值命题逻辑matching 匹配mathematical structure 数学结构matrix representation 矩阵表示maximal element 极大元maximal ideal 极大理想maximal outerplanar graph 极大外平面图maximal planar graph 极大平面图maximum matching 最大匹配maxterm 极大项（基本析取式）maxterm normal form(conjunctive normal form) 极大项范式（合取范式）McGee graph McGee 图meet 交Menger theorem Menger 定理Meredith graph Meredith 图message word 信息字mini term 极小项minimal κ-connected graph 极小κ-连通图minimal polynomial 极小多项式Minimanoff paradox Minimanoff 悖论minimum distance 最小距离Minkowski sum Minkowski 和minterm (fundamental conjunctive form) 极小项（基本合取式）minterm normal form(disjunctive normal form)极小项范式（析取范式）Möbius function Möbius 函数Möbius ladder Möbius 梯Möbius transform (inversion) Möbius 变换（反演）modal logic 模态逻辑model 模型module homomorphism 模同态(R-同态)modus ponens 分离规则modus tollens 否定后件式module isomorphism 模同构monic morphism 单同态monoid 单元半群monomorphism 单态射morphism (arrow) 态射（箭）Möbius function Möbius 函数Möbius ladder Möbius 梯Möbius transform (inversion) Möbius 变换（反演）multigraph 多重图multinomial coefficient 多项式系数multinomial expansion theorem 多项式展开定理multiple-error-correcting code 纠多错码multiplication principle 乘法原理mutually orthogonal Latin square 相互正交拉丁方Nn-ary operation n-元运算n-ary product n-元积natural deduction system 自然推理系统natural isomorphism 自然同构natural transformation 自然变换neighbour set 邻集next state 下一个状态next state transition function 状态转移函数non-associative algebra 非结合代数non-standard logic 非标准逻辑Norlund formula Norlund 公式normal form 正规形normal model 标准模型normal subgroup (invariant subgroup) 正规子群（不变子群）n-relation n-元关系null object 零对象nullary operation 零元运算Oobject 对象orbit 轨道order 阶order ideal 阶理想Ore condition Ore 条件orientation 定向orthogonal Latin square 正交拉丁方orthogonal layout 正交表outarc 出弧outdegree 出次(出度)outer face 外面outer neighbour 外（出）邻集outerneighbour set 出(外)邻集outerplanar graph 外平面图Ppancycle graph 泛圈图parallelism 平行parallelism class 平行类parity-check code 奇偶校验码parity-check equation 奇偶校验方程parity-check machine 奇偶校验器parity-check matrix 奇偶校验矩阵partial function 偏函数partial ordering (partial relation) 偏序关系partial order relation 偏序关系partial order set (poset) 偏序集partition 划分,分划,分拆partition number of integer 整数的分拆数partition number of set 集合的划分数Pascal formula Pascal 公式path 路perfect code 完全码perfect t-error-correcting code 完全纠-错码perfect graph 完美图permutation 排列（置换）permutation group 置换群permutation with repetation 可重排列Petersen graph Petersen 图p-graph p-图Pierce arrow Pierce 箭pigeonhole principle 鸽子笼原理planar graph （可）平面图plane graph 平面图Pólya theorem Pólya 定理polynomail 多项式polynomial code 多项式码polynomial representation 多项式表示法polynomial ring 多项式环possible world 可能世界power functor 幂函子power of graph 图的幂power set 幂集predicate 谓词prenex normal form 前束范式pre-ordered set 拟序集primary cycle module 准素循环模prime field 素域prime to each other 互素primitive connective 初始联结词primitive element 本原元primitive polynomial 本原多项式principal ideal 主理想principal ideal domain 主理想整环principal of duality 对偶原理principal of redundancy 冗余性原则product 积product category 积范畴product-sum form 积和式proof (deduction) 证明（演绎）proper coloring 正常着色proper factor 真正因子proper filter 真滤子proper subgroup 真子群properly inclusive relation 真包含关系proposition 命题propositional constant 命题常量propositional formula(well-formed formula,wff)命题形式（合式公式）propositional function 命题函数propositional variable 命题变量pullback 拉回(回拖) pushout 推出Qquantification theory 量词理论quantifier 量词quasi order relation 拟序关系quaternion 四元数quotient (difference) algebra 商（差）代数quotient algebra 商代数quotient field (field of fraction) 商域（分式域）quotient group 商群quotient module 商模quotient ring (difference ring , residue ring) 商环（差环，同余类环）quotient set 商集RRamsey graph Ramsey 图Ramsey number Ramsey 数Ramsey theorem Ramsey 定理range 值域rank 秩reconstruction conjecture 重构猜想redundant digits 冗余位reflexive 自反的regular graph 正则图regular representation 正则表示relation matrix 关系矩阵replacement theorem 替换定理representation 表示representation functor 可表示函子restricted proposition form 受限命题形式restriction 限制retraction 收缩Richard paradox Richard 悖论right adjoint functor 右伴随函子right cancellable 右可消的right factor 右因子right zero divison 右零因子ring 环ring of endomorphism 自同态环ring with unity element 有单元的环R-linear independence R-线性无关root field 根域rule of inference 推理规则Russell paradox Russell 悖论Ssatisfiable 可满足的saturated 饱和的scope 辖域section 截口self-complement graph 自补图semantical completeness 语义完全的（弱完全的）semantical consistent 语义相容semigroup 半群separable element 可分元separable extension 可分扩域sequent 矢列式sequential 序列的Sheffer stroke Sheffer 竖（谢弗竖）simple algebraic extension 单代数扩域simple extension 单扩域simple graph 简单图simple proposition (atomic proposition) 简单（原子）命题simple transcental extension 单超越扩域simplication 简化规则slope 斜率small category 小范畴smallest element 最小元（素）Socrates argument Socrates 论断（苏格拉底论断）soundness (validity) theorem 可靠性（有效性）定理spanning subgraph 生成子图spanning tree 生成树spectra of graph 图的谱spetral radius 谱半径splitting field 分裂域standard model 标准模型standard monomil 标准单项式Steiner triple Steiner 三元系大集Stirling number Stirling 数Stirling transform Stirling 变换subalgebra 子代数subcategory 子范畴subdirect product 子直积subdivison of graph 图的细分subfield 子域subformula 子公式subdivision of graph 图的细分subgraph 子图subgroup 子群sub-module 子模subrelation 子关系subring 子环sub-semigroup 子半群subset 子集substitution theorem 代入定理substraction 差集substraction operation 差运算succedent 后件surjection (surjective) 满射switching-network 开关网络Sylvester formula Sylvester公式symmetric 对称的symmetric difference 对称差symmetric graph 对称图symmetric group 对称群syndrome 校验子syntactical completeness 语法完全的（强完全的）Syntactical consistent 语法相容system Ł3 , Łn , Łא0 , Łא系统Ł3 , Łn , Łא0 , Łאsystem L 公理系统 Lsystem Ł公理系统Łsystem L1 公理系统 L1system L2 公理系统 L2system L3 公理系统 L3system L4 公理系统 L4system L5 公理系统 L5system L6 公理系统 L6system Łn 公理系统Łnsystem of modal prepositional logic 模态命题逻辑系统system Pm 系统 Pmsystem S1 公理系统 S1system T (system M) 公理系统 T(系统M)Ttautology 重言式（永真公式）technique of truth table 真值表技术term 项terminal endpoint 终端terminal object 终结对象t-error-correcing BCH code 纠 t -错BCH码theorem (provable formal) 定理（可证公式）thickess 厚度timed sequence 时间序列torsion 扭元torsion module 扭模total chromatic number 全色数total chromatic number conjecture 全色数猜想total coloring 全着色total graph 全图total matrix ring 全方阵环total order set 全序集total permutation 全排列total relation 全关系tournament 竞赛图trace (trail) 迹tranformation group 变换群transcendental element 超越元素transitive 传递的tranverse design 横截设计traveling saleman problem 旅行商问题tree 树triple system 三元系triple-repetition code 三倍重复码trivial graph 平凡图trivial subgroup 平凡子群true in an interpretation 解释真truth table 真值表truth value function 真值函数Turán graph Turán 图Turán theorem Turán 定理Tutte graph Tutte 图Tutte theorem Tutte 定理Tutte-coxeter graph Tutte-coxeter 图UUlam conjecture Ulam 猜想ultrafilter 超滤子ultrapower 超幂ultraproduct 超积unary operation 一元运算unary relation 一元关系underlying graph 基础图undesignated truth value 非特指值undirected graph 无向图union 并(并集)union of graph 图的并union operation 并运算unique factorization 唯一分解unique factorization domain (Gauss domain) 唯一分解整域unique k-colorable graph 唯一k着色unit ideal 单位理想unity element 单元universal 全集universal algebra 泛代数（Ω代数）universal closure 全称闭包universal construction 通用结构universal enveloping algebra 通用包络代数universal generalization 全称推广规则universal quantifier 全称量词universal specification 全称特指规则universal upper bound 泛上界unlabeled graph 无标号图untorsion 无扭模upper (lower) bound 上（下）界useful equivalent 常用等值式useless code 废码字Vvalence 价valuation 赋值Vandermonde formula Vandermonde 公式variery 簇Venn graph Venn 图vertex cover 点覆盖vertex set 点割集vertex transitive graph 点传递图Vizing theorem Vizing 定理Wwalk 通道weakly antisymmetric 弱反对称的weight 重（权）weighted form for Burnside lemma 带权形式的Burnside引理well-formed formula (wff) 合式公式(wff)word 字Zzero divison 零因子zero element (universal lower bound) 零元（泛下界）ZFC (Zermelo-Fraenkel-Cohen) system ZFC系统form)normal(Skolemformnormalprenex-存在正则前束范式(Skolem 正则范式)3-value proposition logic 三值命题逻辑。

资源环境与城乡规划管理外文资料翻译

资源环境与城乡规划管理外文资料翻译毕业设计外文资料翻译题目陆面过程模式CLM的稳定同位素的季节变化仿真学院资源与环境学院专业资源环境与城乡规划管理班级资源0702班学生包芳学号20072102002指导教师王永森二〇一一年三月二十五日Simulations of seasonal variations of stable water isotopes in land surface process model CLMZHANG XinPing1†, WANG XiaoYun2, YANG ZongLiang3, NIU GuoYue3 & Xie ZiChu11 College of Resources and Environmental Sciences, Hunan Normal University, Changsha 410081, China;2 Qingdao Meteorological Bureau, Qingdao 266003, China;3 Department of Geological Sciences, the University of Texas at Austin, Texas 78721-0254, USAAbstract: In this study, we simulated and analyzed the monthly variations of stable water isotopes in different reservoirs at Manus, Brazil, using the Community Land Model (CLM)that incorporates stable isotopic effects as a diagnostic tool for understanding stable water isotopic processes, filling the observational data gaps and predicting hydro meteorological processes. The simulation results show that the δ18O values in precipitation, vapor and surface runoff have distinct seasonality with the marked negative correlations with correspondingwater amount. Compared with the survey results by the InternationalAtomic Energy Agency (IAEA) in co-operation with the World Meteorological Organization (WMO), the simulations by CLM reveal the similar temporal distributions of theδ18O in precipitation. Moreover, the simulated amount effect between monthlyδ18O and monthly precipitation amount, and MWL (meteoric water line) are all close to the measured values. However, the simulated seasonal difference in the δ18O in precipitation is distinctly smaller than observed one, and the simulated temporal distribution of theδ18O in precipitation displays the ideal bimodal seasonality rather than the observed single one. Thesemismatches are possibly related to the simulation capacity and the veracity in forcing data.Key word : stable water isotope, CLM, simulation, amount effect, seasonal variationThe modeling of the land surface and soil moisture is increasingly seen as an important component in understanding hydrological cycles]1[. Stable water isotopes, for example18O and D, are superlative tracers for the hydrological cycles because their abundance in water reflects the accumulated record of physical phase change]3,2[. Using the features of stable isotopes can accurately determine the partitioning of precipitation into transpiration, evaporation andrunoff]5,4[, which cannot be detected with mass balance alone]6[. Recently, the conducting iPILPS (isotopes in the Project for Intercomparison of Land-surface Parameterization Schemes) incorporates stable water isotopes into land surface parameterization schemes. The iPILPS experiment aims to identify and test different land-surface schemes that incorporate stable water isotopes, appraise the applicability of stable isotopic data in hydro-climatic study and water resources survey, identify observational data gaps required for evaluating the land-surface schemes with isotopes and apply stable water isotopic data to specific prediction of hydro meteorological processes]8,7[.This study, a part of iPILPS, incorporates stable water isotopes in CLM as a diagnostic tool, simulates and analyses variations of stable water isotopes in different reservoirs on monthly time scales at Manaus, Brazil. The simulated behaviors of stable isotopes in precipitation on monthly time scale have good consistency with actual survey result at Manaus station set up byIAEA/WMO, howing that the simulation by the CLM incorporating stable water isotopes is reasonable.1 Model description1.1 CLMEarth’s biosphere is an important part of the Earth’s climate system. Relatedly,the dynamic, thermodynamic and physiological processes of vegetation coverage arethe key factors impacted climate change. These numerical models including physicalprocess parameterizations are called as the land surface process model. Land surfacemodel is composed of different physical processmodes including the parameterization of dynamics characteristics, the longwave and shortwave radiation transfer and rainfall interception, etc. in canopy associated with vegetation shape; photosynthesis, transpiration and evaporation related to plantphysiology; and physical process of water-heat conduction, soil chemical processes,freezing and thawing of permafrost within soil, and so on. The Community LandModel (CLM) is currently one of well-developed and potential land surface models.CLM is developed from the Biosphere-AtmosphereTransfer Scheme (BATS), the Institute of Atmospheric Physics, ChineseAcademy of Sciences land model (IAP94) and the NCAR land surface model (LSM)]9[.The model takes into account ecological differences among vegetation types,hydraulic and thermal differences among soil types, and allows for multiple landcover types within a grid cell. Strictly speaking, CLM is a single point model.According to different physical processes, the model structure can be separated intotwo parts, the biogeophysical processes relating to vegetation cover at surface andphysical processes relating to hydraulic and thermal transfer in soil, mainly includingradiation transfer, turbulence diffusion and thermal conduction in soil and so on.Detailed descriptions about CLM may refer to relational references and technicalnotes]10,9[.1.2 Stable water isotope parameterizationThe stable isotopic ratio incorporated into CLM is noted as- 1 -- 2 -R W =OH O H 162182 (1)The subscription w stands for reservoir water, for example precipitation, runoff or vapor, etc.There are two possible ways of mixing the reservoir water with input, “total mixing” scheme and “partial mixing” scheme:Rw(t) = [N1Rw (t −1) + N2Rinputs (t)] N(2)N = N1 + N2 (3)in a total mixing,R overflow (t ) = Rw (t ) (4)and in a partial mixing, namely asmax(N 1)≥N(5)R overflow (t ) = R inputs (t )(6)where R inputs is the stable isotopic ratio of any inputs, R overflow is the ratio ofoverflow that is the water of exceeding the maximum storage capacity of the reservoir, N 1 is the mass of water in reservoir, N 2 is the mass of input water, N is the total mass after mixing and t is thetime. As phase change is generated, there will appear fractionation effect of stable isotope. As water evaporating, the stable isotopic ratio in residual water isaf Rw t Rw 1)0()( (7)where f = N 1(t ) / N 1(0) is the fraction of residual water in the reservoir afterevaporation event, andα = Rw / Rv is the fractionation factor of stable isotopes between liquid and vapor, Rv is the ratio in evaporated vapor. As vapor condensing,Rd =α Ra , (8) where Rd and Ra are stable isotopic ratios in dew and in atmospheric vapor, respectively, and α is the stable isotopic fractionation factor calculated between liquid and vapor phase.As known, vegetation transpiration does not produce stable isotopic fractionation]6[, thus, the stable isotopic ratio in transpiration equals to that in root region, namelyR tr = R root . (9) 1.3 Experimental schemeThree sites with different geophysical and climatic conditions are selected for iPILPS Phase 1 experiment. They are Munich, Germany (48.08°N, 11.34°E), Tumbarumba, Australia (35.49°S, 148.01°E) located in middle latitudes and Manaus, Brazil (3.08°S, 60.01°W) in tropical rainforest of South America.Manaus, with an equatorial climate characterized by agreeable temperatures but plenty of rain and humidity, is situated in the heart of Amazons, north of Brazil more than 1450 km inland from the Atlantic. According to statistical data, the annual mean precipitation amountis about 2190 mm at Manaus, with the maximal monthly mean precipitation of 308 mm in April and the minimal mean precipitation of 52 mm in August; the annual mean temperature is 26.8℃, with the highest monthly mean temperature of 27.9℃in October and the lowest monthly mean temperature of 26.0℃in March.The survey of stable isotopes in precipitation shows that there is the marked negative correlation between monthly stable isotopic ratios in precipitation and precipitation amount at Manaus]7[. In view of that some variation features of stable isotopes in precipitation at Manaus have comparability with that under monsoon climate in East Asia, the simulation experiment of stable water isotopes was carried out at Manaus.The CLM simulation requires forcing that contains isotopes in precipitation and atmospheric vapor etc. at high resolution (Table 2). These forcing data, commended in iPILPS Phase 1 exclusively, are derived from output of REMOiso (Regional Model with isotopes) at 15-min time step for one ideal year (360 days)]11[. For the details on the forcing data see ref]12[. In this experiment, the model iterates a 1-year calculation until differences between the initial and final values decrease below 0.01 mm/a for water storage and 0.01 mm/a R V-SMOW for isotopic species. The simulation year is defined as equilibrium year.1.4 Stable isotopic balanceWater balance is the base of calculating water amount in land surface scheme. Similarly, stable isotopic balance is the base of stable isotopic simulation.The magnitude of stable isotopic ratio in water is related to that in initial origin, e.g. in atmospheric precipitation or in vapor. By averaging the δ18O in reservoirs- 3 -- 4 - and the specific humidity as well as aggregating the water budgets at 15-min time step ]11[, the daily variations of the δ18O in reservoirs and corresponding water budgets are obtained (Figure 1).，Figure 1 The daily variations of the δ18O inprecipitation (a), vapor (b), with the corresponding water budgets from REMOiso as inputs at Manaus, Brazil.In Figure 1, the δ18O in precipitation and in vapor show all obviousseasonality and the typical isotopic signature in evergreen tropic forest: the heavy rain or the moist atmosphere (great q) is usually depleted in stable isotopes, whereas the light rain or the dry atmosphere (small q) is usually enriched in stable isotopes. Compared with precipitation, vapor is isotopic ally depleted obviously.The soil column is discredited into ten layers with different depths from 0.0175 m to 1.437 m in vertical direction. In this study, the variations of stable water isotopes and water budgets are concerned in super-surface (0―0.0175 m, the first layer in CLM) and root-region (0.0175―3.433 m, the 2nd―10th layer in CLM).2 Simulation results by CLM2.1 Seasonal variations of 18O and water budgetsin land surface reservoirsOn the monthly time scale, the simulated precipitation, specific humidity and surface runoff show all the obvious bimodal seasonality, which characterizes the climatic regime of equator zones (Figure 2). The primary maximal and minimal precipitation appear respectively in April and in July with their amount difference of 528 mm, and the second maximal and minimal precipitation respectively in December and in January with the amount difference of 170 mm, merely 1/3 of the former amount and only one month in time difference, which is possibly related to the fast moving of the ITCZ in sum mer]7[. Correspondingly, theδ18O in reservoirs alsoshows the bimodal seasonality, in which the variations of theδ18O in precipitation and in surface runoff have similarity: their two maximums appear in January and July and two minimums in April and October, with the negative correlation of stable isotopic ratio with water budget; additionally, in vapor, two maximums of theδ18O appears in January and August and two minimums of theδ18O does in April and November. The second extremums are later than that in precipitation and in runoff. Such a result shows that, to a certain degree, the stable isotopes in reservoirs and vapor are impacted not only by large-scale climatic conditions, for example the solar radiation and atmospheric circulation, but also by the vapor origins]5[.- 5 -- 6 -Figure 2 The monthly variations of the O in precipitation (a),vapor (b), surface runoff (c), surface dew (d) and surface evaporation (e), with the corresponding water budgets at Manaus, Brazil.The magnitude of surface evaporation is related to atmospheric humidity. Compared Figure 2(e) with 2(b), the evaporation is relatively small at two maximal specific humidity in April and in December, but relatively great at the minimal specific humidity in July. Unlike the behaviors of precipitation, specific humidity and condensation, evaporation shows the weak seasonality and indistinctive correlation with theδ18O in evaporation.2.2 Seasonal variations of the 18O and waterThe surface infiltration water, originated primarily from atmospheric precipitation, shows a very good consistency with precipitation . As a result, 18the monthly meanδ18O in infiltration water is positively correlated to that in precipitation, but negatively to infiltration water in accordance with the amount effect. Compared with precipitation, the infiltration water is isotopic ally enriched markedly due to evaporation action.The variation of super-surface soil water is influenced not only by infiltration water but also by mass exchange with root region water and surface evaporation action. Impacted by the storage regulation and peak attenuation actions of soil, the seasonality of super-surface soil water is weakened. Correspondingly, theδ18O in superurface reservoir displays unclear seasonality and un-marked correlation with- 7 -super-surface water. However, the surface evaporation keeps isotopic consistency with super-surface reservoir because of drawing water from super-surface soil directly . Comparatively, the evaporated vapor is isotopic ally depleted.The root-region water and the subsurface runoff have all weak seasonality with slightly later time phase than precipitation. Usually, in the rainy season, bigger aquiclude and stronger sub-surface runoff corresponds to the higher water table; and in the dry season, smaller aquiclude and weaker subsurface runoff to the lower water table. Correspondingly,δ18O in reservoirs shows that, in the rainy season, the heavy precipitation and the produced strong infiltration have the marked impact on δ18O in root-region reservoir and in the decrease of theδ18Oin root-region reservoir and subsurface run- off is in apparent. Additionally, it can be found that theδ18Oin subsurface runoff is equal to that in root-region water because the mass complement mainly comes from root-region water.2.3Seasonal variations of the O and water budgets in canopy reservoirThe canopy storage water mainly comes from the precipitation interception by canopy, the replenishment from condensation is less. Therefore, the seasonal variation of theδ18O in canopy reservoir is consistent with that in precipitation. In accordance withδ18O in canopy reservoir is in-the amount effect, the versely proportional to the canopy storage water: in the rainy season, more canopy storage water corresponds toδ18O in reservoir, and in dry season, less canopy lowerδ18O in reservoir. Compared with precipitation, canopy reservoir is isotopically enriched due to evaporation action.Because vegetation transpiration process is considered not to generate isotopic fractionation, theδ18O variation in canopy transpiration keeps consistent with that in root-region water that furnishes the most of the canopy transpiration. By comparing ,the canopy transpiration varies with contrary to canopy evaporation. In the dry season, the water furnishing to canopy evaporation is less for lighter precipitation, but canopy transpiration is more due to drier atmosphere; in the rainy season, the water furnishing to canopy evaporation is more for heavier precipitation, but canopy transpiration is less due to moister atmosphere.3 Comparison between CLM simulated and actual resultsManaus is one of sampling stations attached to the global survey network set by the International Atomic Energy Agency (IAEA) in co-operation with the World Meteorological Organization (WMO). There have been 26-year stable isotopic survey records from 1965 to 1990 (absent from 1993 to 1995) at Manaus (http://www.programs/ri/gnip/gnipmain.htm).On the monthly timescale, there is the marked amount effect between the actual δ18O in precipitation and precipitation amount with the confidence level above0.001,and the simulated amount effect has good consistency with the actual that.The relationship betweenδD andδ18O in atmospheric precipitation is called as the meteoric water line (MWL). The actual global MWL by Craig is D = 8.0δ18O+10.0]12[. The slope item of 8.0 stands for comparative relationship of fractionation- 8 -rates between deuterium and oxygen-18, and the constant item of 10.0 does the deviation degree of the deuterium from that in equilibrium state. They are controlled by all of these phase-change processes from vapor evaporating in its origins to raindrops falling onto surface land.Compared with the global MWL, the actual MWL at Manaus has the slightly great slope and constant items, but the simulated one has the slightly small slope and constant items.4 Conclusions(1) Similar to the simulated variations of precipitation, specific humidity and surface runoff, the simulatedδ18O in these reservoirs also shows the bimodal seasonality with the marked negative correlations with corresponding water amount. The variation of theδ18O in dew hasvery good consistency with that in vapor because dew is condensed from vapor directly. The surface evaporation amount is related to atmospheric humidity. However, theδ18O in evaporation shows the indistinctive correlation with evaporation amount.(2) The seasonal variation of theδ18O in surface in-filtration water has a very good consistency with that in precipitation because of originating primarily from atmospheric precipitation. Impacted by storage regulation and peak attenuation actions of soil, the seasonal differences of theδ18O in super-surface and root-region reservoirs are weakened. Theδ18O in subsurface runoff equals to that in root-region water because the mass complement mainly comes from root-region water.(3) The seasonal variation of theδ18O in canopy reservoir is consistent with that in precipitation. Compared with precipitation, canopy reservoir is isotopically enriched due to evaporation action. Similar to that in surface dew, the seasonal variation of the δ18O in canopydew is consistent with that in vapor. Compared with vapor, the variation range of the δ18O in canopy dew is distinctly smaller although canopy dew is isotopically enriched.(4) Based on the available data from IAEA/WMO, the actual precipitation amount and theδ18O in precipitation have all distinct seasonality at Manaus. Moreover, the simulated amount effect between monthlyδ18O and monthly precipitation amount, and MWL (meteoric water line) are all close to the actual results.1 Aleinov I, Schmidt G A. Water isotopes in the GISS Model E land surface scheme. Glob Planet Change, 2006, 51(1-2): 108-1202 Yoshimura K, Miyazaki S, Kanae Sh, et al. Iso-MATSIRO, a land surface model that incorporates stable water isotopes. Glob Planet Change, 2006, 51(1-2): 90-1073 Gat J R. Atmospheric water balance in the Amazon basin: an isotopic evapotranspiration model. J Geophys Res, 1991, 96: 13179-131884 Hoffmann G, Werner M, Heimann M. Water isotope module of the ECHAM- 9 -atmospheric general circulation model: a study on time-scales from days to several years. J Geophys Res, 1998, 103: 16871-168965 Yoshimura K, Oki T, Ohte N, et al. A quantitative analysis of δ18O variability with a Rayleigh-type isotope circulation short-term model. J Geophys Res, 2003,108(D20): doi: 10.1029/ 2003JD0034776 Fischer M J. iCHASM, a flexible land-surface model that incorporates stable water isotopes. Glob Planet Change, 2006, 51(1-2): 121-1307 Henderson-Sellers A, MeGuffie K, Hang Z. Stable isotopes as validation tools for global climate model predictions of the impact of Amazonian deforestation. J Climate, 2002, 15: 2664-26778 Henderson-Sellers A, Fischer M, Aleinov I, et al. Stable water isotope simulation by current land-surface schemes: results of iPILPS Phase 1.Glob Planet Change, 2006, 51(1-2): 34-589 Dai Y J, Zeng X B, Dickinson R E, et al. The Common Land Model. Bull Amer Meteor Soc, 2003, 84(8): 1013-102310 Oleson K W, Dai Y J, Bonan G, et al. Technical description of the Community Land Model (CLM). NCAR/TN-461+STR, 200411 Sturm K, Hoffmann G, Langmann B, et al. Simulation of δ18O in precipitation by the regional circulation model REMOiso. Hydrol Process, 2005, 19: 3425-344412 Craig H. Isotopic variations with meteoric water. Science, 1961, 133: 1702-1703- 10 -Chinese Science Bulletin ，2009, 1007(10): 1765-1772.陆面过程模式CLM的稳定水同位素的季节变化仿真章新平1，王小云2，杨宗辆3，牛国月4，谢子出51湖南师范大学资源与环境科学系，长沙410081，中国;2青岛市气象局，青岛266003，中国;3三部地质科学院，美国德克萨斯大学奥斯汀分校，得克萨斯州78721-0254，美国.摘要本文模拟和分析了不同稳定水同位素组成在马瑙斯、巴西的月变化，运用区域土地模型- 11 -（CLM的）即将稳定同位素技术作为一个了解稳定水同位素诊断的工具，填补了观测水文气象数据和预测过程的空白。

3D视觉表面匹配技术在曲轴毛坯无序抓取中的应用

3D视觉表面匹配技术在曲轴毛坯无序抓取中的应用作者：陆贤辉申红森来源：《时代汽车》2022年第09期摘要：本文主要研究对无序排列曲轴毛坯的自动抓取，通过3D视觉技术实现空间位置搜索，并转换为机器人抓取坐标，在实际运行过程中，为解决3D算法在速度、可靠性、稳定性等方面的不足，视觉识别成功率低，机器人抓取发生碰撞的问题，利用全局模型局部匹配的算法优化基于PPF特征的投票原理，最终实现高效稳定的3D物体匹配识别，并实现了无序排列曲轴毛坯稳定、可靠、快速的抓取，满足了现实生产过程的需求。

关键词：3D视觉无序抓取 Point Pair Feature （PPF）投票原理Abstract：This paper researched the robot loading system for the free-form crankshaft. It was applied with 3D visual camera to search the object in a big steel case， and it would communicate with the robot and sent the coordinate of the specific character in the whole crankshaft. In the practice， the paper propound a new algorithm to improve the Voting Scheme based on PPF and more efficiently matched objects in the point clouds The method was met the production requirement and achieved the stability， reliability and high efficiency in the robot loading system for the free-form crankshaft.Key words：3D visual camera， free-form， Point Pair Feature （PPF）， Voting Scheme在工厂自动化、智能化的发展进程中，机器视觉技术扮演着一个重要的角色。

missForest 1.5 非参数缺失值填充 Random Forest 包说明说明书

Package‘missForest’October13,2022Type PackageTitle Nonparametric Missing Value Imputation using Random ForestVersion1.5Date2022-04-14Author Daniel J.Stekhoven<*******************.ethz.ch>Maintainer Daniel J.Stekhoven<*******************.ethz.ch>Imports randomForest,foreach,itertools,iterators,doRNGSuggests doParallelDescription The function'missForest'in this package is used toimpute missing values particularly in the case of mixed-typedata.It uses a random forest trained on the observed values ofa data matrix to predict the missing values.It can be used toimpute continuous and/or categorical data including complexinteractions and non-linear relations.It yields an out-of-bag(OOB)imputation error estimate without the need of a test setor elaborate cross-validation.It can be run in parallel tosave computation time.License GPL(>=2)URL https://,https:///stekhoven/missForest NeedsCompilation noRepository CRANDate/Publication2022-04-1414:52:29UTCR topics documented:missForest-package (2)missForest (2)mixError (6)nrmse (7)prodNA (8)varClass (8)Index101missForest-package Nonparametric Missing Value Imputation using Random ForestDescription’missForest’is used to impute missing values particularly in the case of mixed-type data.It can be used to impute continuous and/or categorical data including complex interactions and nonlinear relations.It yields an out-of-bag(OOB)imputation error estimate.Moreover,it can be run parallel to save computation time.DetailsPackage:missForestType:PackageVersion: 1.4Date:2013-12-31License:GPL(>=2)LazyLoad:yesThe main function of the package is missForest implementing the nonparametric missing value imputation.See missForest for more details.Author(s)Daniel J.Stekhoven,<*******************.ethz.ch>ReferencesStekhoven,D.J.and Buehlmann,P.(2012),’MissForest-nonparametric missing value imputation for mixed-type data’,Bioinformatics,28(1)2012,112-118,doi:10.1093/bioinformatics/btr597 missForest Nonparametric Missing Value Imputation using Random ForestDescription’missForest’is used to impute missing values particularly in the case of mixed-type data.It can be used to impute continuous and/or categorical data including complex interactions and nonlinear relations.It yields an out-of-bag(OOB)imputation error estimate.Moreover,it can be run parallel to save computation time.UsagemissForest(xmis,maxiter=10,ntree=100,variablewise=FALSE,decreasing=FALSE,verbose=FALSE,mtry=floor(sqrt(ncol(xmis))),replace=TRUE,classwt=NULL,cutoff=NULL,strata=NULL,sampsize=NULL,nodesize=NULL,maxnodes=NULL,xtrue=NA,parallelize=c( no , variables , forests )) Argumentsxmis a data matrix with missing values.The columns correspond to the variables and the rows to the observations.maxiter maximum number of iterations to be performed given the stopping criterion is not met beforehand.ntree number of trees to grow in each forest.variablewise logical.If’TRUE’the OOB error is returned for each variable separately.This can be useful as a reliability check for the imputed variables w.r.t.to a subse-quent data analysis.decreasing logical.If’FALSE’then the variables are sorted w.r.t.increasing amount of missing entries during computation.verbose logical.If’TRUE’the user is supplied with additional output between iterations,i.e.,estimated imputation error,runtime and if complete data matrix is suppliedthe true imputation error.See’xtrue’.mtry number of variables randomly sampled at each split.This argument is directly supplied to the’randomForest’function.Note that the default value is sqrt(p)for both categorical and continuous variables where p is the number of variablesin’xmis’.replace logical.If’TRUE’bootstrap sampling(with replacements)is performed else subsampling(without replacements).classwt list of priors of the classes in the categorical variables.This is equivalent to the randomForest argument,however,the user has to set the priors for all categoricalvariables in the data set(for continuous variables set it’NULL’).cutoff list of class cutoffs for each categorical variable.Same as with’classwt’(for continuous variables set it’1’).strata list of(factor)variables used for stratiﬁed sampling.Same as with’classwt’(for continuous variables set it’NULL’).sampsize list of size(s)of sample to draw.This is equivalent to the randomForest argu-ment,however,the user has to set the sizes for all variables.nodesize minimum size of terminal nodes.Has to be a vector of length2,with theﬁrst entry being the number for continuous variables and the second entry the num-ber for categorical variables.Default is1for continuous and5for categoricalvariables.maxnodes maximum number of terminal nodes for trees in the forest.xtrue plete data matrix.This can be supplied to test the performance.Upon providing the complete data matrix’verbose’will show the true impu-tation error after each iteration and the output will also contain theﬁnal trueimputation error.parallelize should’missForest’be run parallel.Default is’no’.If’variables’the data is split into pieces of the size equal to the number of cores registered in the parallelbackend.If’forests’the total number of trees in each random forests is split inthe same way.Whether’variables’or’forests’is more suitable,depends on thedata.See Details.DetailsAfter each iteration the difference between the previous and the new imputed data matrix is assessed for the continuous and categorical parts.The stopping criterion is deﬁned such that the imputation process is stopped as soon as both differences have become larger once.In case of only one type of variable the computation stops as soon as the corresponding difference goes up for theﬁrst time.However,the imputation last performed where both differences went up is generally less accurate than the previous one.Therefore,whenever the computation stops due to the stopping criterion(and not due to’maxiter’)the before last imputation matrix is returned.The normalized root mean squared error(NRMSE)is deﬁned as:mean((X true−X imp)2)var(X true)where X true the complete data matrix,X imp the imputed data matrix and’mean’/’var’being used as short notation for the empirical mean and variance computed over the continuous missing values only.The proportion of falsely classiﬁed(PFC)is also computed over the categorical missing values only.For feasibility reasons’ntree’,’mtry’,’nodesize’and’maxnodes’can be chosen smaller.The num-ber of trees can be chosen fairly small since growing many forests(e.g.p forests in each iteration) all observations get predicted a few times.The runtime behaves linear with’ntree’.In case of high-dimensional data we recommend using a small’mtry’(e.g.100should work)to obtain an appropriate imputation result within a feasible amount of time.Using an appropriate backend’missForest’can be run parallel.There are two possible ways to do this.One way is to create the random forest object in parallel(parallelize="forests").This is most useful if a single forest object takes long to compute and there are not many variables in the data.The second way is to compute multiple random forest classiﬁers parallel on different variables (parallelize="variables").This is most useful if the data contains many variables and computing the random forests is not taking too long.For details on how to register a parallel backend see for instance the documentation of’doParallel’).See the vignette for further examples on how to use missForest.I thank Steve Weston for his input regarding parallel computation of’missForest’.Valueximp imputed data matrix of same type as’xmis’.OOBerror estimated OOB imputation error.For the set of continuous variables in’xmis’the NRMSE and for the set of categorical variables the proportion of falselyclassiﬁed entries is returned.See Details for the exact deﬁnition of these errormeasures.If’variablewise’is set to’TRUE’then this will be a vector of length’p’where’p’is the number of variables and the entries will be the OOB errorfor each variable separately.error true imputation error.This is only available if’xtrue’was supplied.The error measures are the same as for’OOBerror’.Author(s)Daniel J.Stekhoven,<*******************.ethz.ch>ReferencesStekhoven,D.J.and Buehlmann,P.(2012),’MissForest-nonparametric missing value imputation for mixed-type data’,Bioinformatics,28(1)2012,112-118,doi:10.1093/bioinformatics/btr597See AlsomixError,prodNA,randomForestExamples##Nonparametric missing value imputation on mixed-type data:data(iris)summary(iris)##The data contains four continuous and one categorical variable.##Artificially produce missing values using the prodNA function:set.seed(81)iris.mis<-prodNA(iris,noNA=0.2)summary(iris.mis)##Impute missing values providing the complete matrix for##e verbose to see what happens between iterations:iris.imp<-missForest(iris.mis,xtrue=iris,verbose=TRUE)##The imputation is finished after five iterations having a final##true NRMSE of0.143and a PFC of0.036.The estimated final NRMSE##is0.157and the PFC is0.025(see Details for the reason taking##iteration4instead of iteration5as final value).##The final results can be accessed directly.The estimated error:iris.imp$OOBerror##The true imputation error(if available):iris.imp$error##And of course the imputed data matrix(do not run this):##iris.imp$ximp6mixError mixError Compute Imputation Error for Mixed-type DataDescription’mixError’is used to calculate the imputation error particularly in the case of mixed-type data.Given the complete data matrix and the data matrix containing the missing values the normalized root mean squared error for the continuous and the proportion of falsely classiﬁed entries for the categorical variables are computed.UsagemixError(ximp,xmis,xtrue)Argumentsximp imputed data matrix with variables in the columns and observations in the rows.Note there should not be any missing values.xmis data matrix with missing values.xtrue complete data matrix.Note there should not be any missing values.Valueimputation error.In case of continuous variables only this is the normalized root mean squared error (NRMSE,see’help(missForest)’for further details).In case of categorical variables onlty this is the proportion of falsely classiﬁed entries(PFC).In case of mixed-type variables both error measures are supplied.NoteThis function is internally used by missForest whenever a complete data matrix is supplied.Author(s)Daniel J.Stekhoven,<*******************.ethz.ch>See AlsomissForestExamples##Compute imputation error for mixed-type data:data(iris)##Artificially produce missing values using the prodNA function:set.seed(81)iris.mis<-prodNA(iris,noNA=0.2)nrmse7##Impute missing values using missForest :iris.imp<-missForest(iris.mis)##Compute the true imputation error manually:err.imp<-mixError(iris.imp$ximp,iris.mis,iris)err.impnrmse Normalized Root Mean Squared ErrorDescription’nrmse’computes the normalized root mean squared error for a given complete data matrix,imputed data matrix and the data matrix containing missing values.Usagenrmse(ximp,xmis,xtrue)Argumentsximp imputed data matrix with variables in the columns and observations in the rows.Note there should not be any missing values.xmis data matrix with missing values.xtrue complete data matrix.Note there should not be any missing values.Valuesee Title.NoteThe NRMSE can only be computed for continuous data.For categorical or mixed-type data see mixError.This function is internally used by mixError.Author(s)Daniel J.Stekhoven,<*******************.ethz.ch>ReferencesOba et al.(2003),’A Bayesian missing value estimation method for gene expression proﬁle data’, Bioinformatics,19(16),2088-2096See AlsomixErrorprodNA Introduce Missing Values Completely at RandomDescription’prodNA’artiﬁcially introduces missing values.Entries in the given dataframe are deleted com-pletely at random up to the speciﬁed amount.UsageprodNA(x,noNA=0.1)Argumentsx dataframe subjected to missing value introduction.noNA proportion of missing values w.r.t.the number of entries of’x’.Valuedataframe with missing values.Author(s)Daniel J.Stekhoven,<*******************.ethz.ch>See AlsomissForestExamplesdata(iris)##Introduce5%of missing values to the iris data setiris.mis<-prodNA(iris,0.05)summary(iris.mis)varClass Extract Variable Types from a DataframeDescription’varClass’returns the variable types of a dataframe.It is used internally in several functions of the ’missForest’-package.UsagevarClass(x)Argumentsx data frame with variables in the columns.Valuea vector of length p where p denotes the number of columns in’x’.The entries are"numeric"forcontinuous variables and"factor"for categorical variables.NoteThis function is internally used by missForest and mixError.Author(s)Daniel J.Stekhoven,<*******************.ethz.ch>See AlsomissForest,mixError,nrmseExamplesdata(iris)varClass(iris)##We have four continuous and one categorical variable.Index∗NAmissForest,2missForest-package,2mixError,6prodNA,8∗classesmissForest,2missForest-package,2mixError,6prodNA,8varClass,8∗errornrmse,7∗nonparametricmissForest,2missForest-package,2∗packagemissForest-package,2missForest,2,2,6,8,9missForest-package,2mixError,5,6,7,9nrmse,7,9prodNA,5,8randomForest,5varClass,810。

六年级写藏在心灵深处的什么英语作文

六年级写藏在心灵深处的什么英语作文全文共3篇示例，供读者参考篇1What's Hidden Deep in My HeartHave you ever felt like there's something stirring inside you, something you can't quite put your finger on? Like a secret tucked away in the depths of your being, waiting to be discovered? That's how I feel a lot of the time. There's this constant buzz, this persistent voice whispering things I can't always make out. It's like there's a whole other world within me, filled with thoughts and feelings I don't fully understand yet.Sometimes I get glimpses of it, you know? Like when I'm staring out the window during class, watching the clouds roll by. My mind starts to drift and I find myself contemplating the biggest questions – what's out there beyond our world? What great mysteries lie waiting to be unraveled? In those moments, I feel a profound sense of wonder and curiosity that seems to come from that hidden place within me.Other times, it's a feeling of determination that bubbles up, pushing me to work harder and do better. When I'm strugglingwith a tough math problem or writing an essay for English class, I can sense that inner drive, that refusal to give up. It spurs me on, filling me with a strange confidence that I can overcome any obstacle. Where does that grit and perseverance come from? I have to believe it's rooted in whatever lies buried in the depths of my heart.Then there are the times when I'm overwhelmed by this intense feeling of...I'm not even sure how to describe it. It's like a longing for something more, a yearning to explore the unknown, to seek out grand adventures and extraordinary experiences. It wells up unexpectedly while I'm playing basketball or riding my bike. Suddenly, the everyday seems dull and restrictive. That hidden voice is calling out for freedom, for the chance to spread my wings and soar.I think about things like exploring ancient ruins in remote jungles, or standing at the edge of the Grand Canyon staring out over that vast, rugged expanse. I imagine myself diving into tiny underwater caves or climbing massive glaciers. It's as if I'm meant for something bigger, grander, more epic than the here and now. Maybe that's why I'm so obsessed with stories of valiant heroes and daring adventures – they speak to thatunquenchable thirst for excitement and discovery that seems woven into my very being.At the same time, there's a softer side to those deep stirrings I can't quite name. It emerges when I'm lying on a grassy hill, staring up at a starred sky or listening to the chirping of crickets at dusk. In those peaceful moments, I'm filled with a profound appreciation for the simple beauties of the world around me and a powerful need to protect them. Watching a kaleidoscope of colorful leaves dancing on the wind or seeing a baby bird huddled in its nest brings me an inexplicable joy and tenderness.I feel profoundly connected to nature in a way I can't fully articulate. It's like the forests, oceans, and mountains are calling to some primal part of me, whispering secrets about the cycles of life. In those moments, I'm struck by how small yet infinitely wondrous our existence is in the grand scheme of things. It's honestly a little overwhelming to confront those huge existential thoughts as a sixth grader. Yet I can't deny the significance of those spiritual inklings, even if I can't grasp their deepest meanings yet.Maybe that's why my thoughts and creative pursuits always seem to circle back to the natural world. I'll spend hours sketching blades of grass, meticulously capturing every curveand texture. Or I'll craft poems about the first brush of winter's chill. I'm forever singing little songs of praise for a radiant sunset or a sky glittering with stars. Channeling the raw beauty and power of the earth through art feels like a way of expressing that inexplicable something inside me.Of course, not all the feelings篇2What's Hidden Deep Within My HeartHave you ever had a secret tucked away so deep inside that you were afraid to let anyone know about it? A hidden desire, dream or fear that felt too personal and vulnerable to share with others? Well, I have one of those deeply buried secrets and today I'm going to dig it up and reveal what's been hiding in the depths of my heart and soul.It's a pretty big secret, one that I've kept locked up tight for years now. Even my closest friends don't know the full truth. I've hinted at it here and there, but never outright stated the raw, honest feelings I've bottled up inside. Why? I guess I've been scared - scared of being judged, of being misunderstood, of having my hopes and dreams crushed before they even have a chance to sprout wings and take flight.You see, the secret weighing heavily on my heart is my true passion and life's ambition. From the outside, I probably seem like a pretty normal 6th grader. I go to school, hang out with my group of friends, play sports, watch TV shows, mess around on my phone and tablet, and get nagged by my parents to do my chores and homework. Nothing too out of the ordinary.But behind those everyday adolescent activities burns a fire fueled by a profound dream. A dream so huge and audacious that I've been almost too afraid to give it life by voicing it aloud. Because what if I say it and then people laugh at me? What if they tell me to "be realistic" and that I'm setting myself up for certain disappointment and failure? I'm not sure my heart could take that harsh of a blow.Okay, I can't dance around it any longer. I have to rip off the band-aid and let my deeply hidden aspiration out into the open. Here goes...my greatest passion and supreme ambition in life is to become a famous explorer, leading expeditions into the unknown reaches of our amazing planet.There, I said it! I want to journey to the most remote jungles, wildest mountain ranges, deepest ocean trenches and most isolated desert landscapes that Earth has to offer. I want to come face-to-face with ancient ruins, undiscovered tribes, exoticwildlife, and natural wonders that no human has ever witnessed before. I crave the thrill of being the first set of eyes to behold a place or species that's remained untouched and unexplored for eons.My dream is to follow in the footsteps of my historic explorer heroes like Ferdinand Magellan, Ernest Shackleton, Jacques Cousteau, Dian Fossey, and Neil Armstrong. These were brave, bold adventurers who cast aside fear and conformed thinking to embrace the uncertainty and dangers of venturing into the great unknown. Through their curiosity, perseverance and dogged determination, they expanded the boundaries of our world and knowledge.That's what sets my soul on fire - the quest for knowledge, the hunger to shed light on the unknown corners and mysteries of our planet. Sure, the life of an explorer is thrilling, but for me it's about more than just chasing thrills. It's a deep, primal need to understand, to constantly be probing the limits of our geography, science, cultures and species. There's just so much out there still waiting to be uncovered and understood.In my mind's eye, I can vividly picture myself one day leading a pioneering expedition into a dense, uncharted rainforest, botanists and biologists feverishly documenting a plethora ofbizarre new species with each footstep forward. Or guiding an underwater crew in a state-of-the-art submersible vehicle to explore the eerie, alien landscapes of oceanic trenches miles beneath the surface. Or maybe trekking across a remote, windswept desert, delicately excavating artifacts that could finally shed light on the puzzling rise and fall of a long-lost ancient civilization.Can you sense the excitement and passion I feel towards exploration and discovery? It quite literally makes my heart race just imagining all the wonders and revelations that await us in nature's hidden realms. There's a whole universe of experiences, knowledge and adventure out there dangling before us like a tantalizing carrot - we just have to seize it and go after it with grit and bold determination.Of course, pursuing my dream won't be easy. The life of a real explorer is grueling, dangerous, and far from glamorous despite how it gets romanticized in books and movies. I'll have to trek through sweltering jungles, frozen tundras and scorching deserts while hauling heavy gear for weeks or months at a time. I'll get eaten alive by bugs, go long stretches without modern conveniences like running water and electricity, and facecountless hazards from rapidly changing weather to menacing predators to potential conflicts with isolated tribes.Not only will I require immense physical and mental stamina, but I'll also have to be well-versed in a diverse array of fields like geography, climatology, botany, biology, anthropology and survival skills. And of course, any successful explorer needs to be an adept navigator, conservationist, videographer, writer and photographer to meticulously document and share their discoveries with the world.Then there's the terrifying life-or-death dangers that haunt every expedition into the unknown - freak accidents, raging flash floods, wildfires, rockslides, aggressive wildlife encounters, shortage of supplies, injuries or illness far from medical help, and so on. Real explorers have to willingly accept a high level of risk. Many have paid the ultimate price over the centuries in their quests to shine light on the dark, uncharted areas of our planet.But those risks, hardships and extreme demands are exactly what draws me to exploration like a moth to a flame. Could there be any greater thrill and sense of achievement than overcoming hazardous challenges to triumphantly reach an extraordinary destination that no fellow human has walked before? Justconquering and documenting the unknown is a ferociously difficult and rewarding mission in itself.Most kids my age are dreaming about becoming mainstream celebrities like athletes, actors or musicians. But not me - I fantasize about becoming a world-famous explorer whose name and history-making discoveries are etched into the annals of human exploration. I want my contributions to further our geographical, scientific and cultural knowledge to be my legacy, not scoring a winning touchdown or dropping a hit album.Now, I know there's still a very long road ahead before my dream could possibly materialize. I'm just a 6th grader after all! But I've already begun laying the groundwork by devouring every book, documentary and biography about legendary explorers and expeditions that I can get my hands on. My room is filled with maps, atlases, wilderness survival guides and real survival gear that I practice with as often as I can.I watch tons of adventurous shows and movies about people exploring the natural wonders and remote cultures across our planet. And I'm making sure to study hard in all the core subjects like science, math, writing and geography that will be essential to thrive as a pioneering explorer. Bit by bit, I'm turning my dream into an actionable pursuit.Of course, there will be naysayers and doubtful people who try to discourage me by saying my aspirations are too lofty and unrealistic for a kid. That exploring uncharted lands is the stuff of fantasy and fairytales these days since modern technology has mapped and surveyed every nook and cranny of the globe. To them I say - think bigger! There will always be new frontiers to cross and secrets to reveal about our planet and species, no matter how far technology advances.Scientists predict that millions of species still remain undiscovered and unnamed, mainly in underexplored rainforest and marine ecosystems. Think of the treasure trove of biological, ecological and medicinal discoveries awaiting us if we can successfully study these uncharted habitats! And while we may have mapped the broadest topographic contours of our continents and seafloors, I guarantee there are countless smaller pockets, crevices and underwater spaces that human eyes have yet to witness.Not to mention the rich cultural traditions, ancient ruins and artifacts, and knowledge about our ancestral roots that remain hidden within isolated tribal societies intentionally disconnected from the outside world. There is an absolute universe of enlightening discoveries about our species' history, languages,practices and origins waiting to be unearthed in every corner of the globe we have not thoroughly documented.So no, the need and opportunities for good old-fashioned exploration have not been even remotely exhausted in our rapidly advancing modern age. We've only just penetrated the outermost layer - the true depths of our planet's unknowns still await brave explorers to reveal them. That's what motivates me and fuels the fire inside to overcome any obstacle in the pursuit of this grand passion.Whew, I can't begin to express what an incredible relief it is to finally unmask that deeply buried secret I've been guarding for so long! To let that wildly ambitious dream see the light of day and courage篇3What is Hidden Deep in My HeartHave you ever felt like there is something stirring inside you, buried deep within your heart and mind, that you can't quite put your finger on? For me, that feeling has been growing stronger and stronger as I've gotten older and experienced more of the world around me. It's a tumultuous mixture of emotions, dreams,passions, worries, and so much more – all swirling together in a whirlpool that sometimes makes my head spin.Part of what is hidden in my heart is a burning curiosity about life and all its mysteries. Why are we here? What is our purpose? How did everything come into existence? The older I get, the more I find myself pondering these huge existential questions without any clear answers. I look up at the vast night sky, mesmerized by the billions of stars shining down, and I can't help but feel insignificant in the grand scheme of the universe. And yet, I'm driven by an intense urge to understand as much as I can about this amazing world we live in.Along with that curiosity is a deep passion for learning and growing as an individual. My mind is like a sponge, eagerly soaking up every bit of knowledge it can. I'm constantly amazed by all the incredible scientific discoveries, artistic masterpieces, and amazing feats of human ingenuity throughout history. Reading books, watching documentaries, trying new activities –it all fuels my thirst for greater understanding andself-improvement. I truly believe that learning opens up entire new worlds and perspectives that can enrich our lives immeasurably.However, hidden amongst those positive feelings is also a swirling vortex of anxieties, doubts, and fears that I grapple with daily. What if I'm not smart enough or talented enough to achieve my dreams? What if I disappoint the people I care about most? What if something terrible happens that I can't control? Sometimes those nagging worries feel utterly overwhelming, filling me with stress and making my heart race. I have to actively work on keeping them at bay and maintaining a positive mindset.Despite those fears, my heart is brimming with dreams for the future – dreams that I desperately want to turn into reality. I imagine myself accomplishing incredible things and leaving a positive lasting impact on the world somehow. One moment, I envision myself up on a big stage, delivering inspiring speeches to massive crowds. The next, I'm picturing myself making a huge scientific breakthrough that improves countless lives. Or maybe I'll write powerful stories and books that enlighten and move people. The possibilities seem endless and ever-changing in my restless mind.Alongside my own personal ambitions is a profound desire to help make the world a kinder, more peaceful, and sustainable place for all. Whenever I learn about major global issues likepoverty, disease, human rights violations, and environmental destruction, it weighs heavily on my heart. How can we solve these massive challenges to create real positive change? I want to grow up and devote myself to being part of the solution somehow, whether that's through scientific research, activism, politics, or using my creative voice to raise awareness. A big piece of my heart yearns to leave the world better than I found it.At the same time, I have a strong appreciation for the incredible beauty that exists all around us. A gorgeous sunset, a soaring mountain peak, a relationship filled with love and laughter – all of it awakens a deep sense of awe, gratitude, and appreciation within me. Those vibrantly colorful sights andsoul-nourishing moments remind me that life is something to be cherished, not taken for granted. They inspire me to live life to the fullest and soak in every wonderful experience.Of course, as with every young person, having fun and enjoying myself is a huge part of what's tucked away inside too! I'm always dreaming up new adventures to embark on or silly jokes to make my friends double over with laughter. That goofy, playful, adrenaline-junkie side of me craves excitement, spontaneity, and making amazing memories that will last forever. In those moments of pure unbridled joy and freedom, all myother worries and concerns seem to melt away into the background. I live fully in the present, feeling more alive than ever.Protecting those precious lighthearted times is the powerful love I feel for my family and friends. They are my everything – the people who have supported me, picked me up when I was down, showered me with affection, and most of all, shown me what truly unconditional love looks like. Spending quality time with my loved ones, being silly and creating inside jokes, or simply providing a listening ear during tough times – that human connection means more to me than anything else in this world. Underneath all of life's chaos is an unbreakable support system that I can always depend on.So in my heart swirls this beautifully chaotic combination of curiosities, passions, anxieties, hopes, gratitude, love, and so much more all tangled up together. Some days, it feels like my heart and mind are going to burst open from holding so much inside. Other times, I'm at complete peace just being present in the simple joys of everyday life.As I continue to grow older and my experiences expand, I'm sure new thoughts, dreams, and revelations will keep joining the spinning kaleidoscope of what's already there. No matter whatturbulence I face, at my core is a resilient spirit that refuses to give up on pursuing understanding, happiness, and making a positive difference – both for myself and the wider world around me. This vast well of emotions, ambitions, and perspectives is what makes me human and gives me the beautiful burden of an ever-evolving heart and soul.So while what's hidden deep in my heart is complex and often shifting, one thing will always remain clear: An unshakeable determination to live life to the fullest in pursuit of knowledge, love, and leaving an positive imprint on the world, no matter how small. This vibrant array of feelings and perspectives may seem chaotic from the outside, but to me, they represent the rich tapestry of thoughts and experiences that guide me forward into my unfolding journey of growth and self-discovery.。

piracy形容词

piracy形容词“piracy”的形容词是“pirated”或“piratical”。

一、“pirated”1. 词性解释- 形容词，主要表示“盗版的；非法翻印的；剽窃的”。

2. 意思- 与非法复制、盗用版权相关。

3. 用法- 通常用于修饰名词，作定语。

4. 近义词- counterfeit（伪造的）、bootleg（非法制造贩卖的）。

5. 例句- The police seized a large number of pirated DVDs.（警方查获了大量盗版DVD。

）- Pirated software is a serious problem in the IT industry.（盗版软件在信息技术产业是一个严重的问题。

）- He was caught selling pirated books on the street.（他被发现在街上贩卖盗版书籍。

）- Many consumers are attracted by the low price of pirated products, which is wrong.（许多消费者被盗版产品的低价所吸引，这是错误的。

）- The pirated version of the movie has a lot of flaws in the picture quality.（这部电影的盗版版本在画质上有很多缺陷。

） - Pirated CDs are often sold at a much lower price than the genuine ones.（盗版光盘通常以比正版低得多的价格出售。

）- Thepany is constantly fighting against pirated goods to protect its intellectual property rights.（该公司不断打击盗版商品以保护其知识产权。

一瓶有魔法的药水的英语作文

一瓶有魔法的药水的英语作文In the heart of a secluded apothecary, nestled amidst towering shelves filled with exotic herbs and arcane concoctions, there lay a captivating elixir—a bottle of Bewitching Brew. Its contents shimmered with an iridescent glow, casting an otherworldly aura upon its surroundings.Crafted by an enigmatic alchemist known only as the Enchantress, the Bewitching Brew was whispered to possess extraordinary powers, capable of granting wishes and altering the very fabric of reality. Legends spoke of its ability to heal the sick, rejuvenate the aged, and bestow upon its drinker the fortune of kings.As word of the Brew's existence spread far and wide, it became the object of desire for many. Adventurers, knights, and even royalty embarked on perilous quests to acquire its potent liquid. Some sought its ability to conquer kingdoms, others yearned for its healing properties, while a few simply craved the tantalizing allure of magic.At the center of this swirling vortex of desire stood a young woman named Anya. Driven by a thirst for knowledge and a desperate need to save her ailing father, she resolved to seek out the Enchantress and obtain the Bewitching Brew.Her journey led her through treacherous mountains, across sprawling deserts, and into the heart of an ancient forest. Along the way, she encountered mystical creatures and cunning traps, but her determination never wavered. Finally, after weeks of relentless pursuit, she stumbled upon the Enchantress's secluded cottage.The Enchantress, a wise and benevolent figure, saw the purity of Anya's intentions and agreed to grant her a single sip of the Bewitching Brew. With trembling hands, Anya raised the bottle to her lips and felt a surge of ethereal energy coursing through her veins.In that instant, she was transported to a realm of infinite possibilities. Time seemed to slow down as shecontemplated her deepest desires. With the clarity of a thousand stars, she realized that her father's health wasnot her only concern. She longed for a world where kindness prevailed, where wars ceased, and where all living beings lived in harmony.With newfound purpose, Anya poured the remaining contents of the Brew into a nearby chalice, intending touse its magic to create a better world. As she did so, a blinding light enveloped the cottage, illuminating the surrounding forest with an otherworldly glow.When the light subsided, the Bewitching Brew had vanished, and in its place stood a miraculous chalice. Legends whispered that this Chalice of Benevolence possessed the same extraordinary powers as the Brew, butits magic could only be wielded for the noblest of purposes.Anya took the Chalice in her hands and made a solemnvow to use its power wisely. She traveled throughout the land, performing countless acts of kindness and healing the sick. Her reputation as a benevolent sorceress grew, andpeople flocked to her from far and wide, seeking her aid and the blessings of the Chalice.However, the Chalice's existence also attracted the attention of those who sought to use its power for selfish gain. Warlords and wicked sorcerers schemed to seize it, believing that it would make them invincible. But the Chalice, imbued with the Enchantress's wisdom, resisted their dark machinations.In the end, the Chalice of Benevolence remained a symbol of hope and unity. It served as a reminder that even in the face of adversity, the power of kindness and compassion could prevail. And as the years passed, the legend of the Bewitching Brew faded into obscurity, overshadowed by the transformative legacy of the Chalice that it had once contained.。

我们家现在和以前的变化英语作文

我们家现在和以前的变化英语作文My Home, Then and NowMy home is my favorite place in the whole world. It's where I feel the safest and happiest. But it hasn't always been exactly the same as it is now. Over the years, many things have changed about our house and my family's life at home. Let me tell you about how things used to be and how they are different today.When I was really little, our house felt so much bigger! The ceilings seemed miles high and the rooms felt like wide open spaces that went on forever. Now that I've grown taller, the ceilings don't seem quite as towering and the rooms feel a little more cozy and compact. I used to have to walk really far to get from one end of the house to the other. My little legs would get tired making the journey! But now I can easily run from one side to the other without losing my breath.The backyard also seemed absolutely enormous when I was younger. The stretches of grass resembled a vast field and the single tree felt like a piece of a huge forest planted right there for me to explore. Nowadays the backyard doesn't feel quite as immense, but it's still my favorite place to play outside. At leastthe swing set seems normal-sized rather than miniature like it once did!Inside, my perspective on the house has changed too. I used to have to climb up on the couch since my head could barely peek over the top. Now I can flop down on it with ease. The kitchen counters that I once couldn't see over are the perfect height now. And I can actually reach the bathroom sinks instead of needed a stool!My bedroom has transformed the most over the years. When I was a baby, it had a crib and changing table. Then it became a wonderland of toys, books, and stuffed animals. These days, it's a calmer space with a bigger bed, a desk for doing homework, and some shelves for my favorite books and treasures. The walls have changed too - they've gone from pale yellow to sky blue to my current favorite color green.As I've grown up, my room has had to grow up too. We've added more furniture piece by piece as I needed it. First came a little table for coloring. Then a night stand. Then a desk. A comfy chair. Bookshelves. A TV. A vanity. Every addition makes my room feel more like the space of a capable big kid rather than a little baby's nursery. I have to admit, as much as I loved that babyroom, I'm really proud of my current big kid room. It feels just right for the person I am now.Our house has gone through other big changes too, not just the kid stuff. Five years ago, we renovated the kitchen and gave it all new appliances, countertops, and cabinets. It looked so sleek and modern when it was done! Way better than the old outdated stuff we had before. We also gave the guest bathroom a full makeover around that same time. It went from looking super old-fashioned to clean and contemporary.Some years before that, we did a big project in the backyard. First, we tore out the old cracked patio and overgrown garden beds. Then Dad worked hard building a beautiful new stone patio with a pergola overhead. He planted new garden areas too with pretty flowering bushes and edged it all with brick borders. He set up a full outdoor living space with comfy furniture and even installed a firepit! We spend so much more time outdoors now because of that project.Another major switch happened to our basement a while back. It used to be one of those dark, bare, unfinished basements that felt a little creepy. Not a place kids really wanted to hang out. But my parents hired people to renovate it into a basement guest suite. They added drywall, flooring, a bathroom, and aseparate entrance. This made a whole new living area for when my grandparents or other relatives come for extended visits. Pretty fancy!But not everything in our house has changed dramatically. Some things have stayed pretty much the same ever since I was born. Like the big comfy couch in the living room. That couch has been lived in, jumped on, napped on, and watched TV from for as long as I can remember. My dad says they've never gotten rid of it because it's still in such great shape. Worn in for sure, but not worn out. I'm glad that familiar couch is sticking around!My parents' bedroom has more or less looked the same my entire life too. They've always had the same heavy wooden bedroom furniture and the same cozy color scheme. I feel like their room is frozen in time compared to the rest of the house. That's not a bad thing though. It feels wonderfully comfortable and soothing, like the room itself is giving you a warm hug when you step inside.The home my family shares has gone through so many transformations, both small and large. We've grown and changed right along with it. Some updates have been simple, like adding new pieces to my room for my new hobbies or interests. Some have been major, like fully remodeling entire rooms andoutdoor spaces. And some areas remain blessedly the same, keeping our happy memories intact.Our house has been many things over the years - a nursery, a play space, a workspace, an outdoor haven, an oasis for visitors. Through every stage, it has adapted to our evolving needs and made us feel safe and content. It will undoubtedly keep on changing along with us as we grow older. New phases of life tend to bring new home projects! I can't wait to see how our beloved house continues to shift and mold itself around our family's future. No matter what though, this place we call home will always hold the same cozy, loving feeling it always has. Change or no change, that's one thing that will never be different.。

应用无人机可见光影像和面向对象的随机森林模型对城市树种分类

第５２卷第３期东㊀北㊀林㊀业㊀大㊀学㊀学㊀报Ｖｏｌ．５２Ｎｏ．３２０２４年３月ＪＯＵＲＮＡＬＯＦＮＯＲＴＨＥＡＳＴＦＯＲＥＳＴＲＹＵＮＩＶＥＲＳＩＴＹＭａｒ．２０２４１）国家自然科学基金项目（３１９０１２９８），西藏自治区科学技术重点研发计划项目（ＸＺ２０２２０１ＺＹ０００３Ｇ），福建农林大学省级大学生创新创业训练项目（Ｓ２０２３１０３８９０４６），福建农林大学科技创新专项基金项目（ＫＦｂ２２０３３ＸＡ）㊂第一作者简介：陈逊龙，男，１９９８年１０月生，福建农林大学林学院，硕士研究生㊂Ｅ－ｍａｉｌ：１２２０４９６００２＠ｆａｆｕ．ｅｄｕ．ｃｎ㊂通信作者：张厚喜，福建农林大学林学院㊁南方红壤区水土保持国家林业和草原局重点实验室（福建农林大学）㊁海峡两岸红壤区水土保持协同创新中心（福建农林大学）㊁福建长汀红壤丘陵生态系统国家定位观测研究站，副教授㊂Ｅ－ｍａｉｌ：ｚｈａｎｇｈｏｕｘｉ＠１２６．ｃｏｍ㊂收稿日期：２０２３年１０月２３日㊂责任编辑：王广建㊂应用无人机可见光影像和面向对象的随机森林模型对城市树种分类１）陈逊龙㊀孙一铭㊀郭仕杰㊀段煜柯㊀唐桉琦㊀叶章熙㊀张厚喜（福建农林大学，福州，３５０００２）㊀㊀摘㊀要㊀为及时准确的了解城市树种空间分布信息，提升城市居民生活水平和推动城市生态系统可持续发展㊂以福州市仓山区城市森林为研究对象，应用无人机（ＵＡＶ）监测城市树种空间分布及其动态变化的可见光影像，根据最佳尺度对影像进行分割，并提取分割对象的光谱㊁地形㊁指数㊁纹理和几何特征㊂通过对不同类型特征的组合构建不同的分类方案，利用递归特征消除法（ＲＦＥ）筛选出优选特征子集，利用面向对象方法结合随机森林（ＲＦ）模型对城市树种进行分类㊂结果表明：在随机森林模型分类的过程中，利用光谱特征对树种分类的总体分类精度为８２．１２％；地形特征对树种分类的贡献度率为１４．９６％；指数特征和纹理特征的引入，在一定程度提高了树种的分类精度；几何特征的贡献较小，在分类过程中没有明显的贡献㊂特征优选子集的Ｓ１０方案分类精度最高，总体精度达９２．４２％，Ｋａｐｐａ系数为０．９１㊂说明特征优选能够降低高维度特征的复杂性，在大幅减少数据冗余的同时提高了分类精度㊂在最优特征子集下，随机森林（ＲＦ）算法分类的总体精度比极致梯度提升（ＸＧＢｏｏｓｔ）㊁轻量级梯度提升机（ＬｉｇｈｔＧＢＭ）和ｋ最近邻算法（ＫＮＮ）分别提高了１．１５％㊁１．８１％和１５．１５％，Ｋａｐｐａ系数分别提高了１％㊁２％和１７％㊂关键词㊀城市树种；无人机影像；面向对象；随机森林模型；地形特征分类号㊀Ｓ７７１．８ＵｒｂａｎＴｒｅｅＳｐｅｃｉｅｓＣｌａｓｓｉｆｉｃａｔｉｏｎｂｙＵＡＶＶｉｓｉｂｌｅＬｉｇｈｔＩｍａｇｅｒｙａｎｄＯＢＩＡ－ＲＦＭｏｄｅｌ／／ＣｈｅｎＸｕｎｌｏｎｇ，ＳｕｎＹｉｍ⁃ｉｎｇ，ＧｕｏＳｈｉｊｉｅ，ＤｕａｎＹｕｋｅ，ＴａｎｇＡｎｑｉ，ＹｅＺｈａｎｇｘｉ，ＺｈａｎｇＨｏｕｘｉ（ＦｕｊｉａｎＡｇｒｉｃｕｌｔｕｒｅａｎｄＦｏｒｅｓｔｒｙＵｎｉｖｅｒｓｉｔｙ，Ｆｕｚｈｏｕ３５０００２，Ｐ．Ｒ．Ｃｈｉｎａ）／／ＪｏｕｒｎａｌｏｆＮｏｒｔｈｅａｓｔＦｏｒｅｓｔｒｙＵｎｉｖｅｒｓｉｔｙ，２０２４，５２（３）：４８－５９．Ｉｎｏｒｄｅｒｔｏｏｂｔａｉｎｔｉｍｅｌｙａｎｄａｃｃｕｒａｔｅｓｐａｔｉａｌｄｉｓｔｒｉｂｕｔｉｏｎｉｎｆｏｒｍａｔｉｏｎｏｆｕｒｂａｎｔｒｅｅｓｐｅｃｉｅｓ，ｉｍｐｒｏｖｅｔｈｅｌｉｖｉｎｇｓｔａｎｄ⁃ａｒｄｓｏｆｕｒｂａｎｒｅｓｉｄｅｎｔｓ，ａｎｄｐｒｏｍｏｔｅｔｈｅｓｕｓｔａｉｎａｂｌｅｄｅｖｅｌｏｐｍｅｎｔｏｆｕｒｂａｎｅｃｏｓｙｓｔｅｍｓ，ｔｈｉｓｓｔｕｄｙｔａｋｅｓｔｈｅｕｒｂａｎｆｏｒｅｓｔｉｎＣａｎｇｓｈａｎＤｉｓｔｒｉｃｔ，ＦｕｚｈｏｕＣｉｔｙａｓｔｈｅｒｅｓｅａｒｃｈｏｂｊｅｃｔ．Ｉｔａｐｐｌｉｅｓｕｎｍａｎｎｅｄａｅｒｉａｌｖｅｈｉｃｌｅｓ（ＵＡＶｓ）ｔｏｍｏｎｉｔｏｒｔｈｅｖｉｓｉｂｌｅｌｉｇｈｔｉｍａｇｅｓｏｆｕｒｂａｎｔｒｅｅｓｐｅｃｉｅｓｓｐａｔｉａｌｄｉｓｔｒｉｂｕｔｉｏｎａｎｄｔｈｅｉｒｄｙｎａｍｉｃｃｈａｎｇｅｓ．Ｔｈｅｉｍａｇｅｓｗｅｒｅｓｅｇｍｅｎｔｅｄｂａｓｅｄｏｎｔｈｅｏｐｔｉｍａｌｓｃａｌｅ，ａｎｄｔｈｅｓｐｅｃｔｒａｌ，ｔｅｒｒａｉｎ，Ｉｎｄｅｘ，ｔｅｘｔｕｒｅ，ａｎｄｇｅｏｍｅｔｒｉｃｆｅａｔｕｒｅｓｏｆｔｈｅｓｅｇｍｅｎｔｅｄｏｂｊｅｃｔｓａｒｅｅｘｔｒａｃｔｅｄ．Ｄｉｆｆｅｒｅｎｔｃｌａｓｓｉｆｉｃａｔｉｏｎｓｃｈｅｍｅｓｗｅｒｅｃｏｎｓｔｒｕｃｔｅｄｂｙｃｏｍｂｉｎｉｎｇｄｉｆｆｅｒｅｎｔｔｙｐｅｓｏｆｆｅａｔｕｒｅｓ，ａｎｄｔｈｅｏｐｔｉｍａｌｆｅａｔｕｒｅｓｕｂｓｅｔｗａｓｓｅｌｅｃｔｅｄｕｓｉｎｇｔｈｅｒｅｃｕｒｓｉｖｅｆｅａｔｕｒｅｅｌｉｍｉｎａｔｉｏｎ（ＲＦＥ）ｍｅｔｈｏｄ．Ｔｈｅｕｒｂａｎｔｒｅｅｓｐｅｃｉｅｓｗｅｒｅｃｌａｓｓｉｆｉｅｄｕｓｉｎｇｔｈｅｏｂ⁃ｊｅｃｔ⁃ｏｒｉｅｎｔｅｄｍｅｔｈｏｄｃｏｍｂｉｎｅｄｗｉｔｈｔｈｅｒａｎｄｏｍｆｏｒｅｓｔ（ＲＦ）ｍｏｄｅｌ．ＴｈｅｒｅｓｕｌｔｓｓｈｏｗｅｄｔｈａｔｉｎｔｈｅｐｒｏｃｅｓｓｏｆＲＦｍｏｄｅｌｃｌａｓｓｉｆｉｃａｔｉｏｎ，ｔｈｅｏｖｅｒａｌｌｃｌａｓｓｉｆｉｃａｔｉｏｎａｃｃｕｒａｃｙｏｆｔｒｅｅｓｐｅｃｉｅｓｕｓｉｎｇｓｐｅｃｔｒａｌｆｅａｔｕｒｅｓｗａｓ８２．１２％．Ｔｈｅｃｏｎｔｒｉｂｕｔｉｏｎｒａｔｅｏｆｔｅｒｒａｉｎｆｅａｔｕｒｅｓｔｏｔｒｅｅｓｐｅｃｉｅｓｃｌａｓｓｉｆｉｃａｔｉｏｎｗａｓ１４．９６％．ＴｈｅｉｎｔｒｏｄｕｃｔｉｏｎｏｆＩｎｄｅｘｆｅａｔｕｒｅｓａｎｄｔｅｘｔｕｒｅｆｅａｔｕｒｅｓｉｍ⁃ｐｒｏｖｅｓｔｈｅｃｌａｓｓｉｆｉｃａｔｉｏｎａｃｃｕｒａｃｙｏｆｔｒｅｅｓｐｅｃｉｅｓｔｏａｃｅｒｔａｉｎｅｘｔｅｎｔ．Ｇｅｏｍｅｔｒｉｃｆｅａｔｕｒｅｓｈａｖｅａｓｍａｌｌｃｏｎｔｒｉｂｕｔｉｏｎａｎｄｄｏｎｏｔｈａｖｅａｓｉｇｎｉｆｉｃａｎｔｃｏｎｔｒｉｂｕｔｉｏｎｉｎｔｈｅｃｌａｓｓｉｆｉｃａｔｉｏｎｐｒｏｃｅｓｓ．ＴｈｅＳ１０ｓｃｈｅｍｅｏｆｆｅａｔｕｒｅｓｅｌｅｃｔｉｏｎｓｕｂｓｅｔｈａｄｔｈｅｈｉｇｈｅｓｔｃｌａｓ⁃ｓｉｆｉｃａｔｉｏｎａｃｃｕｒａｃｙ，ｗｉｔｈａｎｏｖｅｒａｌｌａｃｃｕｒａｃｙｏｆ９２．４２％ａｎｄａＫａｐｐａｃｏｅｆｆｉｃｉｅｎｔｏｆ０．９１．Ｔｈｉｓｉｎｄｉｃａｔｅｓｔｈａｔｆｅａｔｕｒｅｓｅｌｅｃ⁃ｔｉｏｎｃａｎｒｅｄｕｃｅｔｈｅｃｏｍｐｌｅｘｉｔｙｏｆｈｉｇｈ⁃ｄｉｍｅｎｓｉｏｎａｌｆｅａｔｕｒｅｓ，ｗｈｉｌｅｇｒｅａｔｌｙｒｅｄｕｃｉｎｇｄａｔａｒｅｄｕｎｄａｎｃｙａｎｄｉｍｐｒｏｖｉｎｇｃｌａｓｓｉｆｉ⁃ｃａｔｉｏｎａｃｃｕｒａｃｙ．Ｕｎｄｅｒｔｈｅｏｐｔｉｍａｌｆｅａｔｕｒｅｓｕｂｓｅｔ，ｔｈｅｏｖｅｒａｌｌａｃｃｕｒａｃｙｏｆｃｌａｓｓｉｆｉｃａｔｉｏｎｕｓｉｎｇｔｈｅＲＦａｌｇｏｒｉｔｈｍｗａｓｉｎ⁃ｃｒｅａｓｅｄｂｙ１．１５％，１．８１％，ａｎｄ１５．１５％ｃｏｍｐａｒｅｄｔｏｅｘｔｒｅｍｅｇｒａｄｉｅｎｔｂｏｏｓｔｉｎｇ（ＸＧＢｏｏｓｔ），ｌｉｇｈｔｇｒａｄｉｅｎｔｂｏｏｓｔｉｎｇｍａ⁃ｃｈｉｎｅ（ＬｉｇｈｔＧＢＭ），ａｎｄｋ⁃ｎｅａｒｅｓｔｎｅｉｇｈｂｏｒａｌｇｏｒｉｔｈｍ（ＫＮＮ），ｒｅｓｐｅｃｔｉｖｅｌｙ．ＴｈｅＫａｐｐａｃｏｅｆｆｉｃｉｅｎｔｗａｓｉｎｃｒｅａｓｅｄｂｙ１％，２％，ａｎｄ１７％，ｒｅｓｐｅｃｔｉｖｅｌｙ．Ｋｅｙｗｏｒｄｓ㊀Ｕｒｂａｎｔｒｅｅｓｐｅｃｉｅｓ；ＵＡＶｉｍａｇｅｒｙ；Ｏｂｊｅｃｔ－ｂａｓｅｄ；Ｒａｎｄｏｍｆｏｒｅｓｔｍｏｄｅｌ；Ｔｅｒｒａｉｎｆｅａｔｕｒｅ㊀㊀城市树木作为城市的重要组成部分是评估城市生态环境的重要指标之一，具有重要的生态㊁经济和社会效益［１］㊂随着城市化进程的不断深化，城市树木的生态效益也日渐凸显㊂然而，不同种类㊁种植结构和种植区域的城市树木会产生不同的生态环境效益［２］㊂因此，及时准确地获取城市树种的类别和空间分布信息对城市规划㊁城市树木的管理与维护具有重要意义［３］㊂传统的城市树种分类主要依靠地面调查，然而该方法存在成本高㊁耗时长且难以获取大尺度数据等不足［４］㊂近年来，遥感技术飞速发展，为城市树种的准确快速识别提供了新的途径㊂然而，传统的高分辨率卫星遥感影像易受天气和环境因素干扰㊁时效性较差且费用昂贵㊂此外，免费提供的卫星遥感影像空间分辨率低，难以适用于树种层面的识别研究［５］㊂相比传统的遥感平台，近地无人机（ＵＡＶ）能在较小空间尺度上提供高分辨率的遥感影像和地理数据，具有更高的适用性，是遥感数据获取的重要手段之一［６］㊂然而，目前有关树种信息提取的无人机遥感研究多集中于多光谱㊁高光谱影像的分类领域，但由于搭载多光谱㊁高光谱传感器的无人机普遍价格昂贵，极大地限制了其在实际生产中的推广应用㊂随着数码技术的发展，通过搭载可见光传感器的无人机获取包含树种信息的遥感影像，具有获取方便㊁成本低㊁空间分辨率高等优点，已成为遥感影像识别树种研究方向上重要的数据源之一［７］㊂根据遥感影像分类单元的不同，可将分类方法归为基于像元和面向对象两类㊂基于像元的方法主要关注局部像素的光谱信息，在处理高分辨率遥感影像时对噪声比较敏感㊁稳健性差，极易出现错分㊁漏分现象［８］㊂为弥补基于像元方法的不足，面向对象的影像分析技术（ＯＢＩＡ）逐渐被用于处理高分辨率遥感影像［９］㊂ＯＢＩＡ方法综合考虑区域相邻像素的纹理㊁形态以及空间结构等多维特征，减少了椒盐噪声的同时，通常具有更高的准确率［１０］㊂然而，随着特征维数的增加，数据处理的难度呈几何倍数增长，使得传统分类算法的应用受到一定限制㊂随机森林（ＲＦ）是一种基于集成学习思想集成多颗决策树的机器学习算法，通过对样本的决策树建模以及组合多棵决策树的预测，最终由分类树投票决定数据的分类［１１］㊂随机森林算法不仅具有模型简单㊁分类精度更高㊁校正参数更少的特点，而且鲁棒性强，不易过拟合，在遥感领域高维特征分类中得到广泛应用［１２］㊂面向对象方法可以有效减少同物异谱现象，而随机森林算法在处理高维数据时有其独特的性能优势，二者的结合在一定程度上提高了分类精度㊂宗影等［１３］将面向对象方法和随机森林算法的有机结合，有效提高了滨海湿地植被的分类精度，总体精度达８７．０７％；赵士肄等［１４］将面向对象方法和随机森林算法应用于耕地领域，并与其他机器学习分类算法进行对比验证，结果表明基于面向对象的随机森林模型取得了最高的耕地提取精度，并减弱了椒盐噪声，优化了分类结果；耿仁方等［１５］研究结果表明，基于面向对象结合随机森林算法对岩溶湿地植被具有较高的识别能力，在９５％置信区间内的总体精度为８６．７５％㊂虽然该方法的研究已经取得了一定的成功，但不同类型的特征对城市树种信息提取效果的影响尚不明确㊂因此，面向对象结合随机森林的方法对于城市树种分类的效果有待进一步探讨㊂此外，目前主流的数据源是大尺度的卫星影像和航空影像，或者是特征信息更加丰富的多光谱和激光雷达影像，而消费级无人机可见光影像在城市树种的精细分类方面还鲜有报道㊂因此，本文以福州市仓山区无人机可见光影像为研究对象，基于ＯＢＩＡ－ＲＦ模型，通过特征优选，构建最佳子集并比较不同机器学习算法的分类精度，并分析不同特征对城市树种分类的影响，构建该研究区城市行道树的最佳特征子集，比较不同分类算法对城市树种的分类效果，进一步评估ＯＢＩＡ－ＲＦ模型的分类性能和适用性，为城市生态系统保护及生态环境治理提供技术支持㊂１㊀研究区概况研究区位于福建省福州市仓山区（见图１），该区域属于南亚热带海洋性季风气候温暖湿润，冬季无严寒，夏季无酷暑㊂年日照时间１７００１９８０ｈ，年降水量９００２１００ｍｍ，气温２０２５ħ㊂福州市仓山区典型树种包括白兰（Ｍｉｃｈｅｌｉａ✕ａｌｂａ）㊁荔枝（Ｌｉ⁃ｔｃｈｉｃｈｉｎｅｎｓｉｓ）㊁芒果（Ｍａｎｇｉｆｅｒａｉｎｄｉｃａ）㊁南洋楹（Ｆａｌ⁃ｃａｔａｒｉａｆａｌｃａｔａ）㊁榕树（Ｆｉｃｕｓｍｉｃｒｏｃａｒｐａ）㊁棕榈（Ｔｒａ⁃ｃｈｙｃａｒｐｕｓｆｏｒｔｕｎｅｉ）㊁樟（Ｃｉｎｎａｍｏｍｕｍｃａｍｐｈｏｒａ）等㊂研究区地势平坦，自然环境相对复杂，具备城市的基本特征，对研究城市树种分类具有一定的代表性㊂２㊀研究方法２．１㊀无人机数据采集与预处理实验数据于２０２０年２月８日采集，采用搭载ＦＣ６３１０Ｓ可见光镜头的大疆精灵４Ｐｒｏ（ＤＪＩＰｈａｎｔｏｍ４Ｐｒｏ）无人机进行航拍获取研究区影像，为削弱阴影对分类过程的干扰，选择天气状况良好无风有云的时间段进行作业㊂飞行相关参数设置如下：航高设置为６０ｍ，航向与旁向重叠率均为７０％，镜头角度－９０ʎ，光圈值ｆ／５，曝光时间１／２００ｓ，ＩＯＳ速度为ＩＯＳ－４００㊂本次飞行共获得４５０张航拍影像，照片分辨率为５４７２ˑ３０７８㊂通过瑞士Ｐｉｘ４Ｄｍａｐｐｅｒ专业摄影测量软件对所采集的原始数据进行空中三角测量㊁点云重建㊁裁切以及镶嵌等操作，得到研究区的正射影像（ＤＯＭ）和数字地表模型（ＤＳＭ）㊂为了精确获得研究区的道路信息，采用天地图在线矢量影像作为辅助信息，并通过手绘的方式提取道路矢量数据㊂根据实际调查情况，利用缓冲分析，将缓冲距离设置为５ｍ，得到了行道树的矢量分布图，然后，将矢量布图与原始影像叠加，最终裁剪出了研究区影像㊂２．２㊀地形特征提取归一化数字表面模型（ｎＤＳＭ）是一种反映地物绝对高度的高程模型［１６］，可为地物判别提供可靠依９４第３期㊀㊀㊀㊀㊀㊀㊀陈逊龙，等：应用无人机可见光影像和面向对象的随机森林模型对城市树种分类据㊂使用ＡｒｃＭａｐ１０．２软件进行地形特征提取㊂首先，通过人工目视解译方法从ＤＳＭ中选取９５０个地面点，并批量提取栅格的高程信息，其中１００个样本点的高程数据用以验证精度㊂其次采用插值的方法生成数字高程模型（ＤＥＭ）㊂为获取更加精确的地面高程信息，比较常见的插值方法（克里金插值法㊁反距离权重法㊁样条插值法以及自然邻域法）生成的数字高程模型（ＤＥＭ），以均方根误差㊁平均绝对值误差和决定系数（Ｒ２）作为评分指标（见表１）㊂４种插值方法均可得到较高精度的ＤＥＭ数据，综合考虑决定系数（Ｒ２）㊁平均绝对值误差以及均方根误差，最终确定采用克里金插值法生成连续的ＤＥＭ数据㊂最后，根据已生成的ＤＥＭ数据，利用Ａｒｃ⁃Ｍａｐ１０．２软件中的栅格计算器，将ＤＳＭ数据与ＤＥＭ数据相减得到ｎＤＳＭ数据［１７］㊂图１㊀研究区概况图表１㊀不同插值方法精度评价方㊀法决定系数（Ｒ２）平均绝对值误差均方根误差克里金插值法０．９９０．０７０．０４反距离权重法０．９９０．０８０．０４样条插值法０．９９０．０８０．０５自然邻域法０．９９０．０７０．０４２．３㊀最佳分割尺度确定影像分割是面向对象方法中至关重要的初始环节，分割结果将直接影响分类精度［１８］㊂本研究采用尺度参数评价工具（ＥＳＰ２），结合目视解译的方法确定最佳分割尺度，所有图像分割过程均在ｅＣｏｇｎｉ⁃ｔｉｏｎ９．０Ｄｅｖｅｌｏｐｅｒ９．０软件完成㊂ＥＳＰ２是用以评价不同尺度影像整体最大差异性的工具，通过计算整体局部方差均值随尺度变化率评估不同地物所对应的最佳尺度参数［１９］㊂而ＥＳＰ２计算出的尺度参数往往是多个值，需要结合人工目视才能确定最佳分割尺度㊂形状参数和紧致度参数是准确表示不同树种轮廓，使得对象内部同质性高的关键㊂综合考虑无人机影像的特点以及影像对象形状和紧致度因子的相互关系，将形状参数设置为０．５，紧致度参数设置为０．３㊂其他必要参数为：各波段的权重值设置为１㊁起始分割尺度为４０㊁分割步长为１㊁迭代８０次㊂随着尺度的增大，局部方差均值整体呈现上升的趋势，而尺度变化率呈现下降的趋势（见图２）㊂为了获得图像的过分割和欠分割之间的临界值，选取尺度变化率峰值为５１㊁５７㊁７６㊁８０㊁８９㊁１０４㊁１０９和１１８作为相对最佳分割尺度参数，采用多尺度分割算法得到分割结果（见图３）㊂当分割尺度参数设置较大（分割尺度参数大于１０４）时，白兰㊁榕树和背景多处０５㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀东㊀北㊀林㊀业㊀大㊀学㊀学㊀报㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀第５２卷被划分为同一个对象，不同树种存在混淆现象难以被区分㊂当分割尺度参数设置较小（分割尺度小于７６）时，不同地物内部出现了过分割现象，增加了数据冗余㊂当分割尺度参数设置７６８９时，植被与背景区分相对明显，不同的树种之间能够被分割成独立的对象，整体分割效果较为理想㊂权衡分割效果与实际情况的吻合度，最终确定研究区无人机影像最佳分割尺度参数为７６，并利用该分割尺度参数进行城市行道树提取㊂图２㊀ＥＳＰ２最佳分割尺度估计图图３㊀不同尺度参数分割效果图２．４㊀对象光谱特征提取光谱特征是遥感影像的重要特征之一，地物通常具有不同的光谱特征，因此根据可见光影像中的地物光谱信息的差异可以用来区分不同的地物类型［２０］㊂植被指数利用植被在不同波段下反射和吸收的特性，增强植被信息的同时使非植被信息最小化［２１］，被广泛应用于林业病虫害防治㊁农作物生长量估计㊁生态环境监测等领域［２２］㊂在遥感图像中，不同地物通常具有复杂程度不同的边缘特征，因此，形状特征可以作为快速准确识别地物类型的有效手段［２３］㊂纹理特征是遥感影像的底层特征，不受图像亮度的影响，能够综合反映像素的灰度分布和结构信息，利用纹理特征可以有效弥补可见光影像光谱信息的不足［６］㊂在面向对象的分类过程中，结合纹理特征对于提升分类精度效果显著［２４］㊂地形特征能真实反映不同地物的高程信息，在影像分类过程中对于区分不同类型的地物具有重要意义㊂因此，本研究共选取光谱㊁指数㊁纹理㊁几何以及地形５大特征，剔除无效特征筛选出４０个子特征，具体如下：（１）光谱特征（ＳＰＥＣ）：主要包括：红色（Ｒ）波段的像元亮度的均值（ＭＲ）㊁绿色（Ｇ）波段的像元亮度的均值（ＭＧ）㊁蓝色（Ｂ）波段像元亮度的均值（ＭＢ）㊁最大差异值（Ｍｄ）㊁亮度值（Ｂｒ）㊂（２）指数特征（ＩＮＤＥ）：包括植被颜色指数（ＩＣＩＶＥ）㊁可见光波段差异植被指数（ＩＶＤＶＩ）㊁联合指数２（ＩＣＯＭ２）㊁超绿指数（ＩＥＸＧ）㊁超绿超红差分指数（ＩＥＸＧＲ）㊁植被指数（ＩＶＧＥ）㊁归一化红绿差异指数（ＩＮＧＲＤＩ）以及归一化绿蓝差异指数（ＩＮＧＢＤＩ）（见表２）㊂（３）几何特征（ＧＥＯＭ）：包括面积㊁边界长㊁宽度㊁长度㊁不对称性㊁长宽比㊁边界指数㊁圆度㊁像素个数㊁紧致度㊁体积㊁密度㊁椭圆拟合㊁主方向㊁形状指数㊁最大封闭椭圆半径㊁最小封闭椭圆半径以及矩形拟合㊂１５第３期㊀㊀㊀㊀㊀㊀㊀陈逊龙，等：应用无人机可见光影像和面向对象的随机森林模型对城市树种分类（４）纹理特征（ＧＬＣＭ）：基于灰度共生矩阵（ＧＬＣＭ）提取影像的纹理特征，包括对比度（ＴＣＯＮ）㊁相关性（ＴＣＯＲ）㊁相异性（ＴＤＩＳ）㊁熵（ＴＥＮＴ）㊁同质度（ＴＨＯＭ）㊁均值（ＴＭＥＡ）㊁角二阶矩（ＴＡＳＭ）和标准差（ＴＳＤ）等特征值［６］（见表３）㊂（５）地形特征：归一化数字表面模型（ｎＤＳＭ）㊂表２㊀植被指数及表达式指数特征公㊀式归一化红绿差异指数（ＩＮＧＲＤＩ）［２５］ＩＮＧＲＤＩ＝（ＭＧ－ＭＲ）／（ＭＧ＋ＭＲ）归一化绿蓝差异指数（ＩＮＧＢＤＩ）［２６］ＩＮＧＢＤＩ＝（ＭＧ－ＭＢ）／（ＭＧ＋ＭＲ）超绿指数（ＩＥＸＧ）［２７］ＩＥＸＧ＝２ＭＧ－ＭＢ－ＭＲ超绿超红差分指数（ＩＥＸＧＲ）［２８］ＩＥＸＧＲ＝ＭＧ－ＭＢ－２．４ＭＲ可见光波段差异植被指数（ＩＶＤＶＩ）［２１］ＩＶＤＶＩ＝（２ＭＧ－ＭＲ－ＭＢ）／（２ＭＧ＋ＭＲ＋ＭＢ）植被颜色指数（ＩＣＩＶＥ）［２９］ＩＣＩＶＥ＝０．４４ＭＲ－０．８８ＭＧ－０．３９ＭＢ＋１８．７９植被指数（ＩＶＧＥ）［３０］ＩＶＧＥ＝ＭＧ／ＭａＲＭ１－ａＢ，ａ＝０．６６７联合指数２（ＩＣＯＭ２）［３１］ＩＣＯＭ２＝０．３６ＩＥＸＧ＋０．４７ＩＣＩＶＥ＋０．１７ＩＶＧＥ㊀㊀注：ＭＲ㊁ＭＧ㊁ＭＢ分别为红㊁绿㊁蓝波段像元亮度的均值㊂表３㊀纹理特征及表达式纹理指标公㊀式角二阶矩（ＴＡＳＭ）ＴＡＳＭ＝ðＮｇｉ＝０ðＮｇｊ＝０ｐ（ｉ，ｊ）２对比度（ＴＣＯＮ）ＴＣＯＮ＝ðＮｇｉ＝０ðＮｇｊ＝０ｐ（ｉ，ｊ）ˑ（ｉ－ｊ）２相关性（ＴＣＯＲ）ＴＣＯＲ＝ðＮｇｉ＝０ðＮｇｊ＝０（（ｉ－ｕｘ）ˑ（ｊ－ｕｙ）ˑｐ（ｉ，ｊ）２）／σｘσｙ相异性（ＴＤＩＳ）ＴＤＩＳ＝ðＮｇｉ＝０ðＮｇｊ＝０ｐ（ｉ，ｊ）ˑ｜ｉ－ｊ｜熵（ＴＥＮＴ）ＴＥＮＴ＝ðＮｇｉ＝０ðＮｇｊ＝０ｐ（ｉ，ｊ）ˑｌｎｐ（ｉ，ｊ）同质度（ＴＨＯＭ）ＴＨＯＭ＝ðＮｇｉ＝０ðＮｇｊ＝０ｐ（ｉ，ｊ）ˑ（１／（１＋（ｉ＋ｊ）２））均值（ＴＭＥＡ）ＴＭＥＡ＝ðＮｇｉ＝０ðＮｇｊ＝０ｐ（ｉ，ｊ）ˑｉ标准差（ＴＳＤ）ＴＳＤ＝ðＮｇｉ＝０ðＮｇｊ＝０ｐ（ｉ，ｊ）ˑ（ｉ－ｕｘ）２㊀㊀注：其中ｉ，ｊ是像元在图像中的行列坐标，ｐ（ｉ，ｊ）为像素对的频数，Ｎｇ为灰度级数，ｕｘ㊁σｘ分别为ｐｘ的均值和标准差，ｕｙ㊁σｙ分别为ｐｙ的均值和标准差㊂２．５㊀试验样本选取本实验通过实地调查获取样本数据㊂调查者沿着研究区的主要道路记录了绿化树种，并排除了数量较少或被其他冠层遮挡的树种，最终确定了７类树种（白兰（Ｍｉｃｈｅｌｉａ✕ａｌｂａ）㊁荔枝（Ｌｉｔｃｈｉｃｈｉｎｅｎｓｉｓ）㊁芒果（Ｍａｎｇｉｆｅｒａｉｎｄｉｃａ）㊁南洋楹（Ｆａｌｃａｔａｒｉａｆａｌｃａ⁃ｔａ）㊁榕树（Ｆｉｃｕｓｍｉｃｒｏｃａｒｐａ）㊁棕榈（Ｔｒａｃｈｙｃａｒｐｕｓｆｏｒ⁃ｔｕｎｅｉ）㊁樟（Ｃｉｎｎａｍｏｍｕｍｃａｍｐｈｏｒａ））以及草地㊁灌木作为研究对象㊂根据遥感影像中不同地物类型的分布位置与大致面积比例，共选取了１１００个样本点㊂为了避免较小的样本数量影响模型分类精度，将最小样本数量设置为６０㊂采用Ｓｃｉｋｉｔ－ｌｅａｒｎ中内置的ｔｒａｉｎ＿ｔｅｓｔ＿ｓｐｌｉｔ函数进行分层抽样，按７：３的比例将数据划分为训练集和测试集（见表４），使各类别样本点数量大致与该类别的总面积成比例㊂训练集用于构建分类模型，测试集用于验证分类精度㊂表４㊀训练和验证样本地物总样本数训练样本数测试样本数白兰２００１４０６０草地５０３５１５灌木５０３５１５荔枝１４０９８４２芒果２００１４０６０南洋楹１２０８４３６榕树１４０９８４２棕榈６０４２１８樟１２０８４３６总计１１００７７０３３０２．６㊀分类模型与参数优化２．６．１㊀随机森林算法随机森林算法（ＲＦ）是一种通过集成学习的装袋思想将多棵决策树集合起来的算法，每棵决策树都充当预测目标类别的分类器㊂随机森林模型在样本数据和分类特征选择方面具有随机性，不容易过拟合，并且表现出良好的稳健性，即使在处理具有缺失值的高维数据时，仍能保持较高的分类精度㊂因此，它被认为是当今最好的算法之一［３２］㊂目前，随机森林算法已经广泛集成在各种软件包中，使用Ｓｔａｔａ数据管理统计绘图软件㊁Ｒ语言统计软件可以轻松实现㊂在模型构造的过程中，通常只需要确定每个树节点包含的特征数量（Ｍ）以及决策树数量（Ｎ），就足以保证模型的性能［３３］㊂本文采取递归特征消除法（ＲＦＥ）［３４］结合交叉验证（Ｃｒｏｓｓ－Ｖａｌｉｄａｔｉｏｎ）确定最佳特征数（见图４）㊂随着特征维数的增加，整体分类精度曲线经历几何增长㊁缓慢上升这个两个阶段后趋于平稳㊂当特征数为２０时，各分类精度曲线均处于相对最高点，因此最终将特征数量的参数设置为２０㊂在使用装袋方法生成训练集的过程中，随机森林算法会导致原始数据集中大约３７％的数据未被抽到，这部分数据被称为袋外（ＯＯＢ）数据㊂利用袋外数据对随机森林模型进行评估是一种无偏估计方法，且在一定程度上能减少计算量，提高算法的运行效率［３５］㊂因此，本文采取遍历不同数量（１１０００）决策树的方法，通过比较袋外误差的大小，确定最佳的决策树数量（见图５）㊂当决策树数量小于８５时，不同子集的袋外数据误差均随着决策树数量的增加而急剧下降，而后随着决策树数量的增加袋外数据误差的下降速度逐渐迟缓，当决策树数量为２００时，袋外数据误差处于相对最低点㊂因此，选择决策树２５㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀东㊀北㊀林㊀业㊀大㊀学㊀学㊀报㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀第５２卷的最佳数量为２００㊂图４㊀模型分类精度与特征数的关系曲线图５㊀袋外误差与决策树数量的关系曲线２．６．２㊀其他分类模型为充分探索随机森林算法对城市树种信息提取的适用性，引入当下流行的机器学习算法作为对照，包括极致梯度提升（ＸＧＢｏｏｓｔ）㊁轻量级梯度提升机（ＬｉｇｈｔＧＢＭ）以及ｋ最近邻算法（ＫＮＮ）㊂ＸＧＢｏｏｓｔ是一种基于增强学习（Ｂｏｏｓｔｉｎｇ）的集成算法，它通过在梯度下降方向上将弱分类器集成到强分类器中，并迭代生成新树以拟合先前树的残差㊂ＸＧＢｏｏｓｔ能够自动利用中央处理器（ＣＰＵ）的多线程进行分布式学习和多核计算，在保证分类准确度的前提下提高计算效率，尤其适用于处理大规模数据［３６－３７］㊂ＬｉｇｈｔＧＢＭ也属于增强学习方法，基本原理与ＸＧ⁃Ｂｏｏｓｔ相似㊂但ＬｉｇｈｔＧＢＭ使用基于直方图的决策树算法来减少存储与计算成本，并优化模型训练速度［３８］㊂ＫＮＮ算法是一种近似自变量与连续结果之间的关系的非参数方法［３９］，其基本思路是通过计算待分类样本与临近样本的距离（欧氏距离㊁曼哈顿距离）来确定所属类别，是一种简单而有效的分类算法㊂为了防止过拟合，本研究在ＪｕｐｙｔｅｒＮｏｔｅｂｏｏｋ平台上利用Ｓｃｉｋｉｔ－ｌｅａｒｎ库中的ＧｒｉｄＳｅａｒｃｈＣＶ包对这３种分类器参数进行了调优（见表５）㊂表５㊀不同分类器的超参数分类器参㊀数参数取值范围极致梯度提升（ＸＧＢｏｏｓｔ）决策树数量［５０，１００，１５０，２００］最大树深度［３，５，７，９］学习率［０．０１，０．０５，０．１０］样本抽样率［０．６，０．８，１．０］特征抽样率［０．６，０．８，１．０］轻量级梯度提升机（ＬｉｇｈｔＧＢＭ）学习率［０．０１，０．０５，０．１０］决策树数量［５０，１００，１５０，２００］叶子节点数［１０，２０，３０，４０］最大树深度［３，５，７，９］ｋ最近邻算法（ＫＮＮ）近邻数［１，２，３，４，５，６，７，８，９，１０］２．７㊀试验方案构建不同树种之间单一特征的差异有限，难以满足树种分类的要求㊂因此，本研究采取增加特征数量的方式来提高分类精度，并探究不同特征组合对分类结果的影响（见表６）㊂表６㊀研究区各地物特征值地物特征不同地物的特征值草地灌木白兰荔枝芒果南洋楹榕树棕榈樟面积６８５９．６７３６３６．７３２９２８．７９７１９４．０５７２００．１０８６８８．４８３４５７．７５２２６３．１３７３２５．０１不对称性０．５５０．４３０．４４０．４３０．４５０．４２０．４８０．５６０．４５边界指数１．７４１．４６２．０１１．９７１．９９１．８４２．０６２．１９１．７９边界长５７８．７０３４０．２０４３１．９５６５０．４１６６９．２９６７０．７３４８８．１６４０８．６３６０１．００亮度值８３．１０７８．７９１１５．６９７４．０９７７．２０８１．１５７１．６３９６．８７６３．８３植被颜色指数－２９．１４－３３．１７－４７．７８－２１．１５－１８．８９－２１．６２－２９．９６－１７．７４－２０．１８联合指数２１４．１１１６．２５２０．１９１１．７４１０．７５１１．６０１５．２７９．２５１１．８２紧致度１．８５１．６３１．８７１．９１１．８６１．８０２．００２．３６１．８５密度２．０３２．１０２．０４２．１０２．１０２．１６１．９７１．８２２．０９超绿指数７６．６０８７．７７１１７．８１５９．６２５３．９３５９．８２８０．８３４８．３１５８．５２超绿超红差分指数－２１５．５６－２０４．０６－３０４．３３－１９４．８８－２０４．０７－１９９．５４－１６８．３２－２７３．２４－１５５．９７椭圆拟合０．６８０．７５０．６３０．６７０．６７０．７１０．５９０．５００．６８角二阶矩０００００００００对比度５５６．７７７８６．２４８７７．５５５９７．２９６１４．１２７７０．７７７１４．２５７６５．５１５１４．１１相关性０．８７０．８２０．８２０．８８０．８８０．８４０．８５０．８６０．９０相异性１７．１３１９．０５２１．６６１８．１２１８．６８２０．２８１９．７２１９．６２１６．６４熵８．７９８．６６８．９０９．１４９．１８９．１９８．９４８．６９９．０７同质度０．０６０．０６０．０５０．０５０．０５０．０５０．０５０．０５０．０６均值１２７．０３１２６．０７１２５．６７１２６．８１１２６．６８１２６．７３１２６．２３１２５．９７１２６．８８３５第３期㊀㊀㊀㊀㊀㊀㊀陈逊龙，等：应用无人机可见光影像和面向对象的随机森林模型对城市树种分类续（表６）地物特征不同地物的特征值草地灌木白兰荔枝芒果南洋楹榕树棕榈樟标准差３４．１７３４．６５３６．２３３６．４０３６．２３３５．７６３６．４０３７．６６３６．３２长度１４３．４５９１．１０８５．９６１３６．３３１３６．９０１４４．３８１００．０３８９．６２１３８．３１长宽比１．８１１．７４１．５２１．５５１．５５１．５１１．６２１．７８１．６１主方向１１３．６１１３０．６３９５．３２９６．５２９４．５６９４．５５９１．４８８３．５６８１．４５最大差异值１．６４１．６９１．５３１．４７１．３６１．２０１．５０１．３４１．４６蓝色（Ｂ）波段像元亮度的均值８４．４７６６．８０９４．６２７４．２０７９．０８８１．７２６３．６４１０５．５３６４．３８绿色（Ｇ）波段像元亮度的均值１３６．２２１３３．７５１８９．３６１１６．２８１１６．９０１２０．４０１１８．７９１４１．２９１０１．７２红色（Ｒ）波段像元亮度的均值１１１．３８１１２．９２１６６．２８９８．７４１００．７９９９．２６９３．１１１２８．７５８０．５５归一化数字表面模型０．３２１．７０１２．５１７．１６１２．０４２３．２０１０．９６１１．９３８．６７归一化绿蓝差异指数０．２１０．２７０．２７０．２００．１７０．１８０．２６０．１３０．２１归一化红绿差异指数０．１００．０９０．０７０．０８０．０７０．１００．１３０．０５０．１２像素个数６８５９．６７３６３６．７３２９２８．７９７１９４．０５７２００．１０８６８８．４８３４５７．７５２２６３．１３７３２５．０１最大封闭椭圆半径０．５８０．７２０．４９０．５６０．５４０．６１０．４５０．３８０．５９最小封闭椭圆半径１．４５１．３９１．４４１．４９１．４５１．４３１．５１１．６３１．４５矩形拟合０．８２０．８６０．８００．８２０．８２０．８３０．７８０．７４０．８２圆度０．８６０．６７０．９５０．９３０．９２０．８２１．０６１．２５０．８６形状指数１．８４１．５５２．０９２．０４２．０６１．９０２．１６２．３２１．８６可见光波段差异植被指数０．１７０．２００．１９０．１５０．１３０．１４０．２１０．０９０．１７植被指数１．３６１．４３１．３８１．３０１．２６１．３０１．４６１．１８１．３７体积６８５９．６７３６３６．７３２９２８．７９７１９４．０５７２００．１０８６８８．４８３４５７．７５２２６３．１３７３２５．０１宽度８０．７８５２．７５５８．０３９０．８１９０．４９９７．８８６２．９６５１．４６８８．９４㊀㊀根据优选特征贡献率（见表７），将所选取的５大特征组合形成了１０种试验方案（Ｓ１Ｓ１０）㊂光谱特征作为每幅遥感影像的基本特征，作为基础被纳入到这１０种方案的构建中㊂其中，Ｓ１仅包含光谱特征；为了全面探究其他特征对分类结果的影响，在Ｓ１基础上引入了地形㊁指数㊁纹理等３个总体特征贡献率较高的特征，通过遍历这３个特征的各种组合得到了Ｓ２Ｓ８；Ｓ９包含了所有的特征；根据２０个优选特征组合建立Ｓ１０，具体的分类方案见表８㊂表７㊀优选特征重要性优选特征重要性／％归一化数字表面模型１４．９６最大差异值１２．４１联合指数２５．５７植被颜色指数５．４２绿色（Ｇ）波段像元亮度的均值４．８４归一化绿蓝差异指数４．６７超绿指数４．５８亮度值４．３６可见光波段差异植被指数３．４２植被指数３．２６红色（Ｒ）波段像元亮度的均值３．０５角二阶矩２．９０蓝色（Ｂ）波段像元亮度的均值２．８６超绿超红差分指数２．７８标准差２．２５归一化红绿差异指数２．２３熵２．０３相关性１．９７均值１．４１边界指数１．２８表８㊀分类方案方案特征子集特征数量Ｓ１光谱５Ｓ２光谱＋地形６Ｓ３光谱＋指数１３Ｓ４光谱＋纹理１３Ｓ５光谱＋地形＋指数１４Ｓ６光谱＋地形＋纹理１４Ｓ７光谱＋指数＋纹理２１Ｓ８光谱＋地形＋指数＋纹理２２Ｓ９光谱＋地形＋指数＋纹理＋几何４０Ｓ１０优选特征２０２．８㊀精度评价本文根据混淆矩阵对模型的分类精度进行定量评价㊂混淆矩阵也称为误差矩阵，是遥感影像二分类问题上的一种评价方法，反映了分类结果与真实地物类别之间的相关性［４０］㊂混淆矩阵的评价指标包括总体精度（ＯＡ）㊁Ｋａｐｐａ系数（Ｋｐ）㊁生产者精度（ＰＡ）以及用户精度（ＵＡ）㊂其中，总体精度指正确分类样本与总体样本的比值；生产者精度指分类结果与参考分类相符合的程度；用户精度指样本分类正确的可能性；Ｋａｐｐａ系数是用于检验遥感影像分类结果的一致性，也可以用以均衡分类效果［４１］㊂各指标计算公式如下：㊀㊀㊀㊀㊀ＯＡ＝ðｎｉ＝１ｘｉｉＮ；㊀㊀㊀㊀㊀Ｋｐ＝Ｎðｎｉ＝１ｘｉｉ－ðｎｉ＝１ｘｉ＋ｘ＋ｉＮ２－ðｎｉ＝１ｘｉ＋ｘ＋ｉ；４５㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀东㊀北㊀林㊀业㊀大㊀学㊀学㊀报㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀第５２卷㊀㊀㊀㊀㊀ＰＡ＝ｘｉｉ／ｘ＋ｉ；㊀㊀㊀㊀㊀ＵＡ＝ｘｉｉ／ｘｉ＋㊂式中：Ｎ为参与评价的样本总数；ｎ为混淆矩阵的行列数；ｘｉｉ为混淆矩阵第ｉ行㊁第ｉ列上的样本数；ｘｉ＋和ｘ＋ｉ分别为第ｉ行和第ｉ列的样本总数㊂３㊀结果与分析３．１㊀随机森林算法的不同分类方案精度由表９可知，随着不同特征类型数量的增加，总体分类精度和ｋａｐｐａ系数整体呈上升趋势㊂其中，仅利用光谱特征作为分类依据的方案Ｓ１精度最低，总体精度和ｋａｐｐａ系数分别为８２．１２％和０．７９，说明光谱特征是遥感影像最重要的特征之一，但仅利用光谱特征难以达到所需的分类精度㊂方案Ｓ２Ｓ４是在Ｓ１的基础上分别加上地形㊁指数和纹理特征，相比方案Ｓ１，这３个方案的总体分类精度分别提高了５．１５％㊁４．５５％㊁１．８２％，ｋａｐｐａ系数分别提高了０．０６㊁０．０６㊁０．０３㊂在分类过程中，地形特征相较于指数和纹理特征扮演着更重要的角色，大幅提高了分类精度㊂方案Ｓ５Ｓ７是在光谱特征的基础上加入地形㊁指数和纹理特征的两两组合，旨在研究它们之间的相互作用对分类精度的影响㊂整体而言，与Ｓ２Ｓ４相比，这３个方案的总体分类精度呈上升趋势㊂其中，Ｓ６具有最高的总体精度和ｋａｐｐａ系数，分别达到９０％和０．８８；其次是Ｓ７，和Ｓ１相比，总体精度和ｋａｐｐａ系数分别提高了７．２７％和０．０９；而Ｓ５总体精度和ｋａｐｐａ系数只增长了６．３６％和０．０８㊂表明地形与指数特征交互作用在分类过程中提供了更大的贡献度㊂方案Ｓ８是由特征重要性靠前的光谱㊁地形㊁指数以及纹理特征构成㊂与包含所有特征的方案Ｓ９相比，Ｓ８反而具有更高的总体分类精度和ｋａｐｐａ系数，分别达到９２．１２％和０．９１㊂表明几何特征对分类精度具有负向影响，它的加入降低了分类精度㊂方案Ｓ１０由优选特征组成，其获得了所有子集中最高的分类精度和ｋａｐｐａ系数，分别为９２．４２％和０．９１㊂与Ｓ９相比，分类精度提高了０．６０％㊂说明特征优选方法能消除高维复杂特征间的信息冗余，使模型仅利用较少特征数量并获得更高的运行效率和分类精度㊂表９㊀不同分类方案分类精度方案总体精度／％Ｋａｐｐａ系数方案总体精度／％Ｋａｐｐａ系数Ｓ１８２．１２０．７９Ｓ６９０．０００．８８Ｓ２８７．２７０．８５Ｓ７８９．３９０．８８Ｓ３８６．６７０．８５Ｓ８９２．１２０．９１Ｓ４８３．９４０．８２Ｓ９９１．８２０．９１Ｓ５８８．４８０．８７Ｓ１０９２．４２０．９１㊀㊀由表１０可知，虽然Ｓ１方案的用户精度与生产者精度整体上处于最低水平，但棕榈树的用户精度达到了１００％，表明棕榈与其他树种存在明显的光谱差异㊂方案Ｓ２加入地形指数后，各类地物的用户精度与生产者精度相比Ｓ１都有不同程度的提高，用户精度提升幅度１．８８％８．１８％，生产者精度提升幅度２．７８％１１．１１％，因为地形特征的加入更好的反映了不同地物之间的空间关系，从而大幅提高了分类精度㊂方案Ｓ３在Ｓ１的基础上加入了指数特征，荔枝㊁榕树以及樟的用户精度分别提升了１０．９５％㊁９．１８％和８．７２％，说明植被指数对荔枝㊁榕树以及樟分类效果显著，但对于其他树种的区分能力有限㊂方案Ｓ４加入纹理特征，芒果和樟的用户精度提升了８．８５％和９．００％，而棕榈和榕树的生产者精度分别提升了２２．２２％和１１．９％，说明这些树种的纹理结构特异性强与其他地物的差异显著，因此纹理特征的加入对分类精度有正向影响㊂方案Ｓ５与Ｓ２相比，荔枝和榕树的用户精度提升了７．０５％和５．１２％，而草地的精度下降了５．８８％；与Ｓ３相比，灌木的用户精度提升了４．４７％㊂总体而言，地形特征与指数特征的组合对分类精度的提升不显著，并且在某些树种的分类上精度出现不同程度的下降，说明这二者的组合产生了冗余信息影响了分类精度㊂方案Ｓ６与Ｓ２相比，芒果与樟的用户精度分别提升了６．４４％和７．６６％，而棕榈树和榕树的生产者精度分别提升了２７．７８％和１１．９０％，这个结果与方案Ｓ４类似，说明地形特征和纹理特征的组合与树种的分类精度呈正相关㊂方案Ｓ７与Ｓ６相比，除个别树种外，整体精度出现了不同程度的降低，波动范围为－６．２１％４．０４％㊂然而，与方案Ｓ５相比，总体分类精度有一定的提升，波动范围是－０．５８％７．５５％㊂方案Ｓ８与表现最好的方案Ｓ７相比，荔枝和榕树的总体分类精度分别提升了９．４２％和６．６７％，其他树种的总体分类精度保持稳定，这表明高维度的特征组合带来了更多的信息，在一定程度上提高了分类精度㊂综合所有特征的方案Ｓ９与Ｓ８相比，总体分类精度呈现出不升反降的现象，波动范围为－１０．２３％４．７４％，说明高纬度的特征产生了冗余信息，影响了随机森林模型的分类性能㊂优选特征子集Ｓ１０与Ｓ９相比，总体分类精度有所提升，其中灌木㊁草地以及荔枝的用户精度分别提升了１０．２３％㊁５．８８％和３．５５％㊂由此可见，特征优选通过对高维数据集的降维和优化，使模型仅利用较少的特征仍能保证良好的分类效果㊂３．２㊀应用优选特征子集对不同分类模型的精度评价由表１１可知，随机森林模型的分类精度最高，总体精度为９２．４２％，比ｋ最近邻算法（ＫＮＮ）㊁极致５５第３期㊀㊀㊀㊀㊀㊀㊀陈逊龙，等：应用无人机可见光影像和面向对象的随机森林模型对城市树种分类。

青春期的英语作文易生气就会变成红色熊猫

青春期的英语作文易生气就会变成红色熊猫When You Get Mad in Puberty, You Turn into a Red Panda!Hi there! My name is Timmy and I'm 10 years old. I'm in the 5th grade now and let me tell you, being a kid in puberty is no walk in the park! It's like this whole new weird world opened up and everything feels so different and crazy.The biggest thing is that I get mad and angry so much easier than I used to. I'll be totally fine one second, and then the next thing you know, BAM! I've turned into a raging red panda! Ok, not a literal red panda (although that would be kind of cool). But I definitely feel just as angry and irritable as one of those fluffy red creatures.It's really bizarre. Stuff that never used to bother me before can now set me off into a fiery rage fit for a furious red panda. Like if my mom tells me to clean my room or do my homework, it's like TNT just got lit inside my brain. Suddenly I'm huffing and puffing, rolling my eyes, maybe even letting out an angry roar or two.My parents are always saying "Timothy, why are you acting like such a grumpy red panda today? You need to calm down!" And I'm like "I don't know! I can't help it! Something came overme." It seriously feels like I momentarily transformed into this angry, red, uncivilized forest creature.There was this one time I was playing video games and my mom politely asked me to pause it so I could have dinner. Normal request, right? But in that moment, I just saw red (or red panda, I guess). I let out a massive irritated groan and shouted "UGGGHHHHH! Can't you see I'm in the middle of something" Then I flung the controller across the room like an angry red panda tossing bamboo. Not my finest moment, I'll admit.My friends tease me about my red panda rages too. We'll be hanging out, having a good time, and then someone will make a silly joke about my hair or my clothes or something dumb like that. Pre-puberty Timmy would've just laughed it off. But nowadays, I'll instantly puff out my cheeks, furrow my brow, and let out a hilarious (but ill-advised) red panda screech of fury. It's like some primal red panda instinct takes over my body for a few minutes.I really don't mean to get so hot-headed and transform into Red Panda Rage Monster. I think it's just all these new hormones and feelings coursing through my body during puberty. It's made me more sensitive and prone to feeling big feelings of anger or embarrassment over little things. Kinda like how redpandas can get all territorial and defensive over their tiny bamboo forest territory.The good news is, just like actual red pandas, my angry outbursts don't tend to last too long. After a few minutes of raging around and venting, I'll snap out of my red panda trance. I'll suddenly become my usual calm, easygoing self again, feeling a bit sheepish about my overreaction. Though sometimes the aftershock of my red panda temper tantrum lingers as a flushed red face or a room left in disarray.I'm trying to work on not letting my red panda rage take control so easily. Whenever I feel that rush of heat and anger start to bubble up, I try to take a few deep breaths and think happy thoughts, like basketball, pizza, or vibrant bamboo forests (red pandas love that stuff, you know). I'm realizing that a lot of the seemingly major things that set me off are really not that big a deal in the grand scheme of things. There's no need to go full crimson creature mode.My parents keep assuring me that eventually, I'll outgrow this red panda puberty phase. They say once I get through these intense childhood years, I'll be much better at keeping my cool and controlling my emotions. Part of me is looking forward to that, but another part of me will kind of miss my red pandarampages. They're kind of fun and exhilarating in a weird way! Maybe I'll just channel that feisty energy into something more positive when I'm older, like boxing or performance art.For now though, I'm just your average elementary school kid navigating the strange, unchartered waters of puberty. If you happen to spot a red panda running wild and raging through the neighborhood, clutching a video game controller or half-eaten bamboo stalk, don't be alarmed. It's probably just me. I'll be over my temper tantrum soon enough. Then you can admire my bright red panda fur and remarkable agility as I effortlessly climb back into a more mild-mannered existence. At least until the next day, when the red panda returns!。

15篇经典英文电影赏析

2012Today I watched the movie named"2012".It is one of those "Christian-based" films with a fair amount of preaching, so be forewarned. This is the basic fault that I found with this film. Instead of telling a story and letting the viewer make his or her own decision, it’s always being shoved in peoples’ faces.This story takes place in December 2012, the family was on vacation.But unfortunately, according to Mayan prophecy, 2012 December 21, is the end of the world, but also to the day the Mayan calendar date, no next page any more.So how to stop all human beings to be destroyed is the main idea through the film.Jack Jackson (John Cusack played) to the Yellowstone National Park vacation with children, but found the lake had been dried up fond memories, while the area has also become a restricted area.He is full of doubts stolen camp near Yellowstone chance to know Charlie.Charlie told him that due to the natural environment and resources by humans and predatory destruction of the balance of the Earth system has its own collapse, humanity will soon face an unprecedented natural disaster.Charlie said that some countries have developed and built in the United secrets can escape the disaster of the ark.Jackson thought he was crazy to laugh it off and walked away.But to his surprise,the disaster happened the next day!We could see from the film that In the search for and to the Ark-based process, the Jackson family has experienced death in the face of disaster after the base finally reached the ark.However, the Ark has been fabricated can not meet the number of parts of the world heard the news coming from the affected population.At last,we can learn something from the film.To face thedisaster, from different countries make the most important human choice: "all men are equal, have equal chances of survival!"The Pursuit of Happyness,Today I watched the film named "The Pursuit of Happyness,"the first thing that flashed into my mind is that how hard it may be if you don't have money, but at end I come to understand that if you keep on your dreams ,happiness will come!With a title like The Pursuit of Happyness, you expect the characters to get to the promised land. They do, but if the journey matters more than the destination, this is a movie to skip. The Pursuit of Happyness is long, dull, and depressing. It expands into two hours a story that could have been told more effectively in one. This is not the feel-good movie of the season unless you believe that a few moments of good cheer can redeem 110 minutes of gloom.The film's most compelling scenes are those that show Chris struggling to enter the rat race. Granted, this is no Glengarry Glen Ross, but it shows the pressure these salesmen are under and how important the contact lists are. In the overall scheme of things, however, these sequences are background noise. They are neither plentiful nor lengthy. The movie spends more time following Chris on his futile sales rounds for the bone density scanner than it does accompanying him during his broker training.In the film, the hero does things actively. He always runs for his work and life. A view that shows he runs for his work that selling a machine, and we can imagine that what’s the result if the hero would not run for his machine. Maybe he will feel disappointed for a long time. Another view that shows his running for the sleep place, what a pity that he was too late, and the end is that he sleep in the toilet with his little son and keep the toilet door close stubbornly. So from the two views we should know that when we do anything, must be active, and if not, thebad result, just like the result in the film, would be coming ruthlessly.The film brings us a good deal of enlightenment. With its practical significance, we will have more spirits to participate in the future work, just like the hero never gave up, and give my families happiness, then to be a better man.AvatarAvatar is the most engaging and enthralling motion picture I have experienced this year - and "experience" is the appropriate word. There's a rush associated with coming to Pandora; this feels more like an interactive endeavor than a passive one.Avatar takes us to the planet Pandora in the year 2154. Pandora is a jungle world at which Earthmen have arrived with the intention of performing some strip-mining. Although corporations run the show, the military, led by Colonel Miles Quaritch (Stephen Lang), is on hand to provide protection and lend support.Avatar is absorbing to look at. So much time and energy was spent into creating this fictional alien species, the Na'vi, and it pays off in the respect that Cameron's visual fx team painstakingly makes it appear real and otherworldly at the same time. We might recognize the lush and green surroundings, or even some of the trippy creatures, and if it comes close to anything it's like Ferngully: The Last Rainforest squared and made semi-pre-historic. But it's the scope and grandeur, and when we see the Na'Vi in close-ups or even just far away, you can see the sweat and the detail in their faces, the human beings playing them projecting off the screen. I forgot, if only for a few moments at a time, that they were animated and done in motion-capture. If part of a filmmaker's job, in a situation like a super-mega-sci-fi epic is to make us believe in another world and place (even if it's familiar), then Cameron has done his job very admirably.Avatar has been described as a "game-changer," and perhaps it is. I'll leave that for future historians to determine. What I can say with some assuredness（确信） is this is the most technically amazing motion picture to have arrived on screens in many years - perhaps since Peter Jackson's The Return of the King.Avatar is entertainment of the highest order. It's the best movie of 2009.Forrest GumpToday I watched the film Forrest Gump,in fact, I had watched this film several times. At the very beginning, a white feather dances in the wind with the music gradually from soft to stro ng. In the blue sky the feather shining brightly seems like an angle. The film begins with the beautiful sence. It is a good start. At last, the feather appears again. It flies over the field, the house .It takes audiences to another magic world,and takes on a ture world to the audiences and stands for a miracle.The film shows kindness, thankness, honest,serious,bravry and all of beautiful things in our souls.It tells the story by Forrest Gump himself.He sits on the chair by the road and tells his experiences in his past time to the persons who are waiting for the bus.They come and go,however Forrest talking and talking.He doesn't mind w hat they are doing and how they are feeling .When I see the sences,I feel it risible.But this just accouts Forrest is a clinging man.Forrest Gump isn't a genius man.He is even thought a big fool.In all his childhood,otherboys look down upon him and ofen laugh at him.But his mother reg ards him as a normal child and tells him that he is the same as the other boys.I appriate and like the film very much.One hand the film tells several sto ries clearly in limited time.The content is affluent and the tie of charactors is complex but clear.On the other hand ,we can get revelations from it.First, if you want to be happy,you must treat youself correctly.The Day After TomorrowAfter watching 《2012》，I watch the same thee about disaster film named” The Day After Tomorrow”Some of what The Day After Tomorrow has to offer is exciting. Some is just plain stupid - like Sam's tussle with a pack of wolves (on board a ship in the middle of Manhattan), Lucy's act of self-sacrifice for a cancer patient, and the President's pep talk (that sounds like it was lifted from Independence Day). Still, a lot of the silliness is expected in this kind of motion picture, and the moments of elevated adrenaline (many of which occur during Jack's harrowing journey) and impressive visuals serve as a counterbalance. Plus, there's even a little irony thrown in for good measure when the Mexican government seals the border to keep U.S. refugees from fleeing south.The Day After Tomorrow has the good sense not to have man attempt to overcome nature's wrath (the point of such films like Armageddon and The Core). Instead, it's a given that there's nothing we can do, so the emphasis is on survival. The knowledge that victory is impossible makes for a more compelling story, since the goal becomes intensely personal: staying alive. Of course, despite the "bad science," the pro-environment message shines through. Like Super Size Me, consider it a cautionary tale. Nevertheless, Emmerich's point with The Day After Tomorrow isn't to play politics or make speeches, but to entertain. And, in the cataclysmic way he has become known for, he does so. The Day After Tomorrow is filled with bad dialogue, stock peril situations, and sketchy character development, but it's a big enough spectacle that those things don't derail the film's capacity to be enjoyed. Pass the popcorn and the cheese.The highest level of disaster film is the pursuit of visual perfection and realistic, after all, in all the disaster scene where we can not use the camera or any other things to every detail of the disaster and the process, but the disaster film provides such an opportunity, butalso asked the audience is able to accurately zone into the picture, feel it was really happened, and not a big sigh depressed,ROMAN HOLIDAYAlthough this is a black film without color,this is a classic movie.The plot describes the England princess Anne’s visit in Rome. In the evening，she went to enjoy the night view in the city, without the officials to allow. She met a kind reporter from U.S.A. two people take the trip together. And they felt i n love by time. But the officials discovered princess’s disappearance, and they all had a lot of icy perspiration. Finally the princess gave up the love between her and the reporter in order to perform her princess’s duty.The film itself is a classic of romantic wish fulfillment, exactly the sort of beautiful lie that Hollywood specialized in. Filmed in Rome, ''Holiday'' imagines that a sweet, stressed-out visiting junior royal, Princess Ann of some tiny duchy or other, would kick over her traces and slip into the night just as the sedative her doctor gave her takes effect. She awakens in the shoebox apartment of an American, Joe Bradley (Gregory Peck). Being a gentleman, he has not taken advantage of this woozy angel. Being not too much of a gentleman, he has made her sleep on the couch.However, because understood oneself to the national responsibility, reason princess finally chose left own spouse, again turned over to the imperial family. Joyful and the short 24 hours had finished, princess or princess, common people or common people, but, many growth. Meanwhile,understood the secret important to Ann,faced love and interest,Joe chose to quit the report about princess,although he would not get 5000 and would lead simply life. Unforgettable princess finally says “me most to love bravely, Rome!”She no longer is one only can nod, the smile, amenable is obedient constantly princess baby, but grows in order to has the opinion, has courage genuine princess.TitanicMaybe this is the fourth time that I watched this film.Titanic is a romance, an adventure, and a thriller all rolled into one. It contains moments of exuberance, humor, pathos, and tragedy. In their own way, the characters are all larger-than- life, but they're human enough (with all of the attendant frailties) to capture our sympathy. Perhaps the most amazing thing about Titanic is that, even though Cameron carefully recreates the death of the ship in all of its terrible grandeur, the event never eclipses the protagonists. To the end, we never cease caring about Rose and Jack.Titanic makes the "overblown movie" lists from several critics, endures extreme mockery and the ridicule of anyone who actually admits to liking it. This may have been due to people wanting to be different from the norm in 1997 and maybe this attitude never went away and too many people fell victim to it. Or maybe it was the cheesy scene's people like to imitate over and over which obviously must mean the entire movie is cheesy. However, I recently watched Titanic the other day and I'm going to say why I find it to my favorite movie of all time.The ship Titanic itself expressed a mentally during this time period in which humans felt they could overcome Mother Nature. With the sinking of this ship it made people truly eat the words "Titanic is an unsinkable ship."Without this love story you would have just come away with the basics of this story, the Titanic sinking. But this love story created sympathy in which you felt for Jack and Rose. It made you truly feel how the people on the Titanic were feeling at that exact moment. Families, loved ones and friends were being torn apart. Without the emotion created from the love story you wouldn't have felt nearly as sad or seriously towards what had actually happened to these people.Brave heartToday I watched the film named “Braveheart”, even though I have watched it before."Braveheart" received much criticism from certain History Buffs upon its release. They said that the filmmakers portrayed Scotsman William Wallace as a brave, heroic, good man. Gibson playsWilliam Wallace, a Scot who decides to revolt against the British after his wife is killed by a pack of the thievin' scoundrels! The film is LOOSELY based on his life - since no one really knows much about Wallace other than what he did: free Scotland from the English for a while, so much of the story is made up.This is a first director outing for Gibson, who not only gives us one of his best - if not the best - performance of his career, other than those great "Lethal Weapon" movies! He handles the direction very well for an actor-turned-director. He doesn't try anything memorable - no fancy camera sweeps to make us motion sick - but he directs the film like the old epics. One of many great things about the film.The cinematography is excellent. I can't think of another film that is quite so beautiful to behold. It is truly wonderful to watch the surroundings fly by the screen, purely unadulterated.As for Gibson's Scottish accent...well...he speaks surprisingly well with a Scottish accent, and doesn't sound like an American-Australian phony (isn't he an Aussie?). There is a great supporting cast in this film, as well, with a man in a rubber nose that you might not be able to place at first glance...James Horner's magnificent score is truly marvelous to behold. He mixes Scottish bagpipes and emotion into a little bundle that in and of it makes you feel emotional. It plays during the film at just the right moments and makes it easier to feel elated or depressed.All in all, I think that "Braveheart" stands as one of the best films I have ever seen. It is an epic in all sense of the word; I don't care how historically incorrect it is. If I wanted a history lesson, I wouldn't be going to see a film like this.Hatchi:A Dog's StoryThis is a simple story about a loyal dog. There is no intricate plot, and even it is only a straight still, but it was still moved me deeply with its most simple pieces of normal life which we can easily find around us such as the train station, the restaurant, the food stand and a warm home.In the movie Hatchi crossed an ocean and thousands of miles of transportation by train, and then he met Parker (the main actor) in a snowy day. Because it was snowing and the dog, Hatchi, was lost. Parker took Hatchi back to his home, and then kept it. Parker treated it very well and so did Hatchi. When Hatchi had grown up, he fetched parker to work everyday.All these years they played together, bathed together, slept together and Parker gave Hatchi full-body massage everyday. They just look like a couple. Parker looks happy because Hatchi is happy, and Hatchi’s happy due to Parker being happy.This is Hatchi’s whole life. He almost spent all his life only doing one thing—waiting for its only friend Parker, waiting to complete the uncompleted engagement.Hatchi’s story showed how great the power of loyalty and love is with all his life. He tell us what the real loyalty and true love is. Because of the real loyalty and true love Hatchi spent all his life waiting; because of the real loyalty and true love Hatchi sacrisficed all his youth just for waiting; because of the real loyalty and true love Hatchi abandon the warm house of Parker’s daughter. I think that is the real loyalty, the ture love. However,in our human life, how many of us can be as loyalty as an animal; how many person can be loyal to another one for even ten years? I think nobody can.Homeless to HarvardYesterday I watched a film, called " Homeless to Harvard”, tells the story of a gir l who overcome adverse conditions, and strive to forge ahead of the story. The heroine's parents are drug addicts and AIDS patients, her parents have no job, living in the slums, even the most basic living expenses for a child are taken by the mother to buy drugs, and her parents did not complain, there is only love, she always persisted, even if a shelter was sent terrible spent her youth, she still has not changed, she and her mother live together once again and happy.Just listen to the name we will find it a very inspirational movie. Recently stumbled always live, can not find the previous endurance, I'm not the university as a kindergarten, but the surroundings do like my childhood, do not worry about anything, had a very comfortable.I had the same poverty as the main character's childhood, but due to the diligent work of their parents. I was fortunate to have been present. I often think it is not me, my future they can not help me. Have the impulse to want to grow their own, but the inherent laziness, so I am against the original intention.My life becomes better, because my parents. Because the main character has been, unfortunately, so continue to move forward The ordinary people just be satisfied now, do not make progress, or say no change in demand, so do all the cruising life, get out of another life. This is like I am now, there is no target time-consuming, people living with me are the same decadence. Our life seems to already have a conclusion, no longer need our efforts, or just need to pay a little bit to connect through life. However, who would be willing to do so over a lifetime, but, but we found it was already old, and no courage to fight again, so life is so decadent to go.。

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)。

V ol.13, No.8 ©2002 Journal of Software 软件学报 1000-9825/2002/13(08)1488-06 相应于无冲突依赖的规范化对象模式森林Ã 吴永辉1,2,3, 周傲英1,21(复旦大学计算机科学与工程系,上海 200433); 2(复旦大学智能信息处理开放实验室,上海 200433); 3(中国科学院软件研究所计算机科学重点实验室,北京 100080)E-mail: slwu@摘要: 首先概括对象依赖、无冲突对象依赖集合、规范化对象模式森林和复杂对象模式规范化设计算法的基本概念和性质;然后给出并证明相应于无冲突对象依赖集合M 的规范化对象模式森林F 的性质:P (F )是惟一的、不可分解的规范化对象模式森林的路径集合;M ⇔OD (F )⇔⋈P (F );P (F )是无α环的.这对于面向对象信息系统的开发有一定的意义.关键词: 对象模式;对象依赖;无冲突;森林;路径中图法分类号: TP311 文献标识码: A随着Internet 和Web 的发展和普及,诸如数字图书馆等存储海量信息的系统应运而生.这些系统的开发存在一系列的问题.在复杂对象模式规范化设计的研究中有代表性的有Z.Tari 等人[1]和W.Y.Mok 等人[2]的工作.Z.Tari 等人的工作仅假设用户的描述不存在冲突,而没有讨论无冲突用户描述在规范化设计中具有的特性;W.Y.Mok 等人的工作则没有考虑冲突的情况.我们参考了已有的工作[1,2]和关系数据库理论[3],也在复杂对象模式规范化设计这方面进行了探索[4~7]:定义对象依赖表示对象间的语义关系;定义规范化对象模式森林作为对象范式;给出复杂对象模式的规范化设计算法——MIMI 算法,并且证明,如果F 是由MIMI 算法获得的相应于对象依赖集合M 的规范化对象模式森林,则M 蕴涵OD (F );并且OD (F )蕴涵⋈(P (F ));若以规范化覆盖M *作为MIMI 算法的输入,则获取相应于M +(M 的闭包)的路径不可分解的规范化对象模式森林;若以M −(M 所蕴涵的所有化简的对象依赖集合)作为MIMI 算法的输入,则获取相应于M +的叶结点不可分解的规范化对象模式森林.本文在已有工作的基础上,给出并证明了相应于无冲突对象依赖集合的规范化对象模式森林的性质.1 基本概念和性质的概括本节对在文献[4~7]中给出的基本概念和性质进行概括,作为全文的基础.定义1(对象依赖(OD )). 设O 是一个对象集,X ,Y ⊆O ,并且Z =O −XY .在X 和Y 之间存在聚集关系或关联关系,使得对于在X 中的每个实例x ,在Y 中一定存在一组相应的实例y 可以被x 访问,并且对于在XZ 中的任意实例xz ,在Y 中访问到的是相同的一组相应的实例y ,则在O 上X 和Y 之间存在着对象依赖(object dependency,简记为OD),表示为X →Y ,称X 决定Y ,或Y 依赖于X .根据OD 的定义,如果X →Y 在O 上成立,X ⊆V ⊂O ,则在V 上,X →Y ∩V 成立.此外,设定OD 不以∅作为左式Ã 收稿日期: 2001-06-22; 修改日期: 2002-01-08基金项目: 国家自然科学基金资助项目(60003008);中国科学院软件研究所计算机科学重点实验室资助项目(SYSKF0202) 作者简介: 吴永辉(1966－),男,浙江温岭人,博士,副教授,主要研究领域为数据库,数字图书馆,面向对象技术;周傲英(1965－),男,安徽郎溪人,博士,教授,博士生导师,主要研究领域为P2P 对象管理,WEB/XML 数据管理.吴永辉等:相应于无冲突依赖的规范化对象模式森林1489(LHS).设X,Y,Z⊆O.OD的推导公理如下:O1. 自反公理.Y⊆X⊆O蕴涵X→Y.O2. 增广规则.X→Y蕴涵XZ→Y.O3. 并规则.X→Y,X→Z蕴涵X→YZ.O4. 投影规则.X→Y,X→Z蕴涵X→Y∩Z,X→Y−Z.O5. 传递规则.X→Y,Y→Z蕴涵X→Z−Y.O6. 补规则.X→Y蕴涵X→O−XY.O7. 伪传递规则.X→Y,YW→Z蕴涵XW→Z.用⋈标记对象联接依赖,OD和OJD之间的转换规则如下:OD-OJD: X→Y蕴涵⋈(XY,X(O−XY)),其中X→Y在O上成立;OJD-OD: ⋈(X,Y)蕴涵Y→Y′,其中X⊆Y′⊆Y.以后O表示给出的全体对象的集合,M表示O上的OD集合.定义2(依赖基). 给出对象集合X,X⊆O,X的依赖基DEP(X)定义为O−X的一个划分{Y1,…,Y n},即DEP(X)={Y i|X→Y i并且对任何一个W⊂Y i,X→W在O上不成立;i=1,…,n},DEP(X)中的元素被称为X的依赖元.RDEP(X)表示相应于M的X的化简的依赖基;EDEP(X)表示相应于M的X的必要的依赖基.定义3(无冲突OD集合). 如果存在X,R⊆O,如果X有两个依赖元V1和V2,使得V1∩R和V2∩R都不为空,则称X分裂R.如果在LHS(M)中至少有一个X使得X分裂R,则称M分裂R.如果M不分裂在LHS(M)中的任意X,并且(DEP(X)∩DEP(Y))⊆DEP(X∩Y),则M是无冲突的.定义4(对象模式树). 如果用H(O)表示O上嵌套关系的层次[4,5],相应于H(O)的对象模式树T递归定义为:(1) 如果H(O)=O,则T是根为对象集O的单结点树;(2) 如果H(O)=A(O1)*...(O n)*(即A与O1,...,O n构成嵌套层次),T的根为对象集A,根的儿女是对象模式树T i的根,T i是O i相应的对象模式树,1≤i≤n.设T是O上的一棵对象模式树且e=(u,v)是T中的一条边.定义标识:F(v)表示v的父亲结点,即,结点u.A(v)为包括v在内的v的所有祖先结点的集合.D(v)为包括v在内的v的所有后继结点的集合.OS(T)是在对象模式树T中的所有对象的集合.由对象依赖的定义,OD(e)是由边e表示的OD,记为OD(A(u),D(v),OS(T)),表示在OS(T)上成立的一个OD:A(u)→D(v).OD(T)是由T中的所有的边表示的OD集合.定义5(路径). 设T是对象模式树,并且u1,...,u n是T的所有的叶结点,则T的路径集合P(T)={A(u1),...,A(u n)}.对于叶结点u,A(u)是在T中由T的根到u的路径上对象的集合.显然,在对象模式树对应的化简超图中,A(u)是一条超边.路径也表示一个语义单位.对象模式树T具有如下这些性质:P(T)是无α环的;OD(T)⇔⋈(P(T));并且OD(T)是无冲突的.在消除了对象模式树中存在的冗余和不规则之后,就可以给出能很好地表示对象间的语义关系的规范化对象模式树和规范化对象模式森的定义[4,5].复杂对象模式规范化设计的目标是产生规范化对象模式森林[4,5].引理1. 设X→W是在M+中的一个未化简的OD,那么在M+中存在一个化简的OD X′→W′,使得X′⊂X和X′W′⊆XW成立;并且如果W∈DEP(X),那么W′=HW,其中H⊆(X−X′).引理2. 如果X是非必要关键字,则X有惟一的化简依赖元;如果X是必要关键字,则X的化简依赖元多于一个.引理3. 设X,Y是对象集,V X∈DEP(X),V Y∈DEP(Y).如果V X∩Y=∅,V Y∩X=∅,并且V X∩V Y≠∅,那么V X=V Y.引理4. 设Z是M的一个关键字,则对于任意的W∈DEP(Z),Z→W是非平凡的和不可转换的.引理5. 设Z是M的一个关键字,并且X⊂Z,则存在一个V∈DEP(X)使得(1)Z⊂XV(即X不分裂Z);(2) 对于每个W∈RDEP(Z),W⊂V;(3)X→V是左部化简的,并且如果X是一个关键字,则V∈RDEP(X).引理6. 如果M是无冲突的OD集合,则如果M蕴涵OD X→V和Y→V,并且V与X和Y都没有交集,则X∩Y→V成立.1490 Journal of Software软件学报2002,13(8)引理7. 如果M是无冲突的OD集合,X,Y和V是对象集,则(1) 如果X→V,Y→V是左部化简的和右部化简的,则X=Y;(2) 如果X是关键字,V∈RDEP(X),并且Y是必要关键字使得Y∩V≠∅成立,则Y分裂XV;(3) 如果X是非必要关键字,对X的非化简依赖元V,有必要关键字X′⊂X使得X′→V;(4) 如果X分裂V,则存在一个必要的关键字X′⊆X使得X′分裂V;(5) 如果M没有分裂X,并且在M+中X→V是左部和右部化简的OD,则对于任意的分裂XV的必要关键字Y,Y∩V≠∅.引理8. 设T是相应于OD集合M的规范化对象模式树,P是T中的路径,L是P的叶结点.如果有关键字X⊂P使得X分裂P,则L⊆X.引理9. 设W=W0W1...W k是相应于M的X和Y的依赖元.那么,如果W i∈DEP(XW0),则W i∈DEP(YW0),其中1≤i≤k,W i,X和Y是O的子集.过程DECOMP(O)返回一棵不可分解的半规范化对象模式树T(叶结点不可分解).过程NEWTREE是通过删除一棵半规范化对象模式树的冗余结点来构造两个半规范化对象模式树.详细步骤参见文献[5].算法1. MIMI算法.输入:对象集合O以及O上的OD集合M.步骤:求出M的所有关键字;T1:=DECOMP(O);j:=1;设X1,...,X n是M的关键字的一个偏序的次序;For i:=1 To n DoBegin If X i⊂OS(T k),其中1≤k≤j ThenBegin V:=Ø;For 每个在T k中的满足∃Z∈DEP(X i)和D(W)=Z∩OS(T k)的结点W DoIf (W相应于X i是冗余的) or (Z∈EDEP(X i) and Z∉EDEP(A(F(W)))) Then V:=V∪{W};If V≠Ø Then Begin j:=j+1; NEWTREE(X i,V,T k,T j) EndEndEnd输出:规范化对象模式森林F={T1,...,T j}.定理1.如果T是不可分解的半规范化对象模式树,则由NEWTREE(X,V,T,T new)产生的T和T new都是不可分解的半规范化对象模式树.定理2. 如果F是由MIMI算法获得的相应于M的O的规范化对象模式森林,则M蕴涵OD(F);并且OD(F)蕴涵⋈ (P(F)).其中P(F)是规范化对象模式森林的路径集合.2 相应于无冲突OD集合的规范化对象模式森林的性质MIMI算法产生的相应于某个OD集合M的规范化对象模式森林F并不总能保持OD集合.引理10. 设F={T1,T2,…,T J}是由MIMI算法产生的规范化对象模式森林,e=〈u,v〉是T i中的一条边,1≤i≤J,并且v1是T1′中相应于v的结点,A(u)和D′(v1)分别为定义在T i和T1′中的A(u)和D(v1),则A(u)→D′(v1),并且它是左部和右部化简的.证明:设T k1,T k2,…,T kn是在F中的一个规范化对象模式树的序列,其中T k1=T1,T kn=T i,并且T kj由T k(j−1)构造,2≤j≤n.对于1≤j≤n,设T j′是T kj刚被构造时的树的情形,e j=〈u j,v j〉是T j′中的边,并且v j相应于v,即,v j=v.A(u j),D(v j)和A′(u j),D′(v j)分别在T kj和T j′中定义,则显然,A(u j)=A′(u j),D(v j)⊆D′(v j),并且D(v j)=D′(v j)∩OS(T kj).首先通过对j的归纳,证明对于1≤j≤n,A(u j)→D′(v1)是右部化简的.当j=1时,A(u1)→D′(v1)显然是右部化简的.吴永辉等:相应于无冲突依赖的规范化对象模式森林1491假设j=n−1时命题成立,即A(u n−1)→D′(v1)是右部化简的.设T″是T k(n−1)在通过执行NEWTREE(X,V,T″,T n′)构造T n′前的形式,则e n−1=〈u n−1,v n−1〉也在T″中.设D″(v n−1)是在T″中定义的D(v n−1),则D″(v n−1)=D′(v n)⊆D′(v n−1).基于定理1,T″是半规范化对象模式树,因此OD(A(u n−1),D″(v n−1),OS(T″))是右部化简的.因为T n′由T″构造,基于引理9,OD(A(u n),D″(v n−1),OS(T″))也是右部化简的.这两个OD是由M蕴涵的OD在OS(T″)上的投影,即,存在D n−1∈DEP(A(u n−1))以及D n∈DEP(A(u n))使得D″(v n−1)=D n−1∩OS(T″)以及D″(v n−1)=D n∩OS(T″).基于引理3,D n−1=D n.由归纳假设,D′(v1)∈DEP(A(u n−1))成立,并且D n−1∩D′(v n−1)≠∅,则D′(v1)=D n−1=D n,所以D′(v1)∈DEP(A(u n)),即A(u n)→D′(v1)是右部化简的.所以,对于1≤j≤n,A(u j)→D′(v1)是右部化简的.T kn是规范化对象模式树,因此OD(A(u n),D(v n),OS(T kn))是左部和右部化简的,并且显然,这一OD是A(u n)→D′(v1)在OS(T kn)上的投影.所以A(u n)→D′(v1)是左部化简和右部化简的.引理11. 设M是无冲突OD集合,Z和V是对象集合,Z→V是左部和右部化简的,M不分裂Z,V上的基本关键字集合FK(V)≠∅,并且K={Y i|Y i是必要关键字,并且Y i∩V≠∅}.在K上定义二元关系“<”如下: Y i<Y j当且仅当ZY i被Y j分裂,但没有被在K中的任何Y j′分裂,其中Y j′⊂Y j,则二元关系“<”是反自反的、反对称的和传递的.证明:(1) “<”是反自反的,即Y i<Y i不成立.因为M不分裂Z,所以Y i不分裂Z,Y i不分裂Y i Z.(2) “<”是反对称的.假设Y i<Y j,并且Y j<Y i也成立,即Y i分裂ZY j也成立.因为Y i<Y j,所以Y j分裂ZY i,又因为M 无冲突,所以存在V j∈DEP(Y j)使得Z j=V j∩Z≠∅,并且V j∩Y i=∅.同理,因为Y i分裂ZY j,所以存在V i∈DEP(Y i),使得Z i=V i∩Z≠∅,并且V i∩Y j=∅.现在证明Y j−Y i≠∅成立,假设Y j−Y i=∅成立,则Y j⊂Y i.因为M不分裂Z,所以Y i不分裂Z,Y i不分裂Y j Z,则导致矛盾.所以Y j−Y i≠∅成立,因而有Y i∩Y j⊂Y j.现在证明Z i∩Z j≠∅.因为Y i不分裂Z,Z i=V i∩Z,所以有Z−Z i⊆Y i成立,如果Z i∩Z j=∅,则Z j⊆Z−Z i⊆Y i,所以Z j⊆Y i,因为V j∩Y i=∅,所以Z j⊈Y i,则导致矛盾.所以Z i∩Z j≠∅,则V i∩V j≠∅成立.所以,由引理3和引理6,V i=V j∈DEP(Y i∩Y j),则Y i∩Y j分裂ZY i,由引理7(4),存在必要关键字Y′⊆(Y i∩Y j),使得Y′分裂ZY i;因为M无冲突,Z→V是左部化简和右部化简的,Y i∩V≠∅,所以Y i⊆ZV;亦即Y′分裂ZV;由引理7(5),Y′∩V≠∅,所以Y′∈K,则与Y i<Y j矛盾.所以Y j<Y i不成立.(3) “<”是传递的,即Y i<Y j,Y j<Y k⇒Y i<Y k.Y i<Y j蕴涵DEP(Y j)={V j1,V j2,…},其中Z⊆Y j V j1,Z∩V j1≠∅,Y i⊆Y j V j2,并且Y i∩V j2≠∅.Y j<Y k蕴涵DEP(Y k)={V k1,V k2,…},其中Z⊆Y k V k1,Z∩V k1≠∅,Y j⊆Y k V k2并且Y j∩V k2≠∅.首先证明Y k分裂ZY i.因为如果V k1∩Y i=∅并且Y i−Y k≠∅,则Y k分裂ZY i.所以要证明V k1∩Y i=∅和Y i−Y k≠∅.如果Y i−Y k=∅,则Y i⊆Y k成立,因为Y j分裂ZY i,所以Y j分裂ZY k,因为Y j<Y k,所以Y k也分裂ZY j.由上面(2)的证明可知,Y j 分裂ZY k和Y k分裂ZY j不能同时成立,导致矛盾,所以Y i−Y k≠∅成立.现在证明V k1∩Y i=∅.设Y i1=Y i∩Y j并且Y i2=Y i∩V j2,因为Y i⊆Y j V j2,所以Y i=Y i1Y i2.因为Y j∩V k1=∅,所以Y i1∩V k1=∅.因为Y k→Y k V k2,由Y j⊆Y k V k2,Y k V k2→V j2成立,则Y k→V j2−Y k V k2成立,即Y k→V j2−V k2成立.由Y k→V k1,由投影规则,Y k→(V k1−(V j2−V k2)),即Y k→(V k1−V j2)成立.因为V k1∈DEP(Y k),所以(V k1−V j2)=V k1,也就是说,V j2∩V k1=∅,因此Y i2∩V k1=∅成立.所以可以导出Y i∩V k1=∅,则Y k分裂ZY i.然后证明K中不存在Y′⊂Y k使得Y i<Y′.假设在K中存在Y′⊂Y k使得Y i<Y′.设DEP(Y′)={V1′,V2′,…},由Y i<Y′可以设Z⊆Y′V1′,Z∩V1′≠∅,Y i⊆Y′V2′,并且Y i∩V2′≠∅.因为Y j<Y k,由“<”的定义,所以Y′不分裂ZY j,即ZY j⊆Y′V1′,并且V2′∩Y j=∅.同理,设DEP(Y j)={V j1,V j2,…},其中ZY k⊆Y j V j1,Z∩V j1≠∅,Y i⊆Y j V j2,并且Y i∩V j2≠∅.由ZY k⊆Y j V j1, ZY′⊆Y j V j1,所以V j2∩Y′=∅成立.现在证明V2′∩V j2≠∅.因为V2′∩Y i≠∅,并且V j2∩Y i≠∅,又因为Y j→V j2,并且Y j不分裂Y i,所以Y i⊆Y j V j2.如果V2′∩V j2=∅,则V2′∩Y i⊆Y j成立,即Y j∩V2′≠∅,这与Y j Z⊆Y′V1′矛盾,所以V2′∩V j2≠∅.因为V2′∩V j2≠∅,则由引理3和引理6,V2′=V j2∈DEP(Y j∩Y′).因此Y j∩Y′分裂ZY i,由引理7(4),有一个必要关键字X⊆(Y j∩Y′)使得X分裂ZY i.由上述(2)的证明可知,X分裂ZV.因为Y i<Y j以及Y i<Y′,由“<”的定义蕴涵Y′⊈Y j,即Y j ∩Y′⊂Y′.所以X⊂Y′成立,并且由引理7(5),X在K中,这与Y i<Y′矛盾.所以K中不存在Y′⊂Y k使得Y i<Y′.引理12. 设M是无冲突OD集合,Z,V是对象集,并且Z→V是左部化简和右部化简的.如果M不分裂Z并且FK(V)≠∅,则在FK(V)中存在一个X0使得M不分裂ZX0.1492 Journal of Software软件学报2002,13(8)证明:设K={Y i|Y i是必要关键字并且Y i∩V≠∅}.K上定义的二元关系“<”如下:Y i<Y j当且仅当ZY i被Y j分裂,但没有被在K中的任何Y j′分裂,Y j′⊂Y j.由引理11,在K中二元关系“<”是反自反的、反对称的和传递的,则在K中至少存在这样一个Y使得对于在K中的任意其他的Y′,Y≮Y′,即Y′不分裂ZY.所以M不分裂ZY.因为Y∩V≠∅并且Y∩V∈FK(V),设X0=Y∩V,则X0∈FK(V),因为ZY⊇ZX0,所以M不分裂ZX0.引理13. 设F是由MIMI算法在O上获得的相应于M的规范化对象模式森林.如果M是无冲突的,则在P(F)中的任何路径p相应于M是不可分解的.证明:基于MIMI算法和过程DECOMP.设T1′是MIMI算法中DECOMP(O)产生的不可分解的半规范化对象模式树.引理12保证过程DECOMP的第2步[5]的选择条件总是满足的,则对于T1′中的边e=〈u1,v1〉,在T1′中存在X⊆A(u1),使得X→D(v1)是左部化简和右部化简的,并且对于任何的必要关键字Y,Y∩D(v1)≠∅,X不会被Y分裂.设e=〈u,v〉是路径p中的任意边,则在T1′中存在边e=〈u1,v1〉使得v1是v的相应结点,即v=v1.由引理10,A(u)→D(v1)是左部和右部化简的,并且由引理7(1),A(u)=X成立,并且D(v)⊆D(v1)成立,所以对于任意的必要关键字Y,Y∩D(v)≠∅,A(u)不会被Y分裂.设L是p的叶结点并且在p中A为A(F(L)),假设p相应于M是可分解的,则由引理7(4)和引理8,存在必要关键字Z⊇L,Z分裂p,即Z分裂A,导致矛盾.所以命题成立.参照文献[8]中的引理8.7及其证明可以给出并证明如下的引理14.引理14. 设M是无冲突OD集合,G(M)是它的超图表示.对于任意一个关键字偏序次序,MIMI算法产生的规范化对象模式森林F的路径集合P(F)由G(M)的最大团组成,而且M等价于对象联接依赖⋈P(F).在关系数据库理论中已经证明,如果MVD集合是无冲突的,则关系数据库模式有惟一的4NF分解[3];如果MVD集合是无冲突的,则在属性集合上有一个无分裂范式(SFNF),因为SFNF蕴涵4NF,关系数据库模式是惟一的[9].基于此,可以给出并证明如下的引理15.引理15. 设M是O上的无冲突OD集合,P(F)是O的规范化对象模式森林的路径集合,则P(F)是惟一的和不可分解的.相似于关系数据库理论中无α环数据库的特性的定理(文献[3]中的定理18.2)及其证明,可以证明复杂对象模式也有相似的无α环复杂对象模式特性,如引理16所示.引理16. 对于规范化对象模式森林F的路径集合P(F),下列命题是等价的.(1) P(F)是无α环的;(2) 对象联接依赖⋈P(F)等价于一个无冲突OD集合.相应于无冲突OD集合的规范化对象模式森林具有如下特性.定理3. 设F是由MIMI算法获得的相应于对象集合O及其OD集合M的规范化对象模式森林,而且M 是无冲突的,则(1) P(F)是O的惟一的不可分解的规范化对象模式森林的路径集合;(2) M⇔OD(F)⇔⋈P(F);并且(3) P(F)是无α环的.证明:(1) 基于定理2和引理13,可得P(F)是在O上的相应于M的规范化对象模式森林F的路径集合.由于M 是无冲突的,由引理15,P(F)是惟一的和不可分解的.(2) 基于上面的(1)和引理14,M⇔⋈P(F).并且由定理2,可以导出M⇔⋈OD(F)⇔⋈P(F);(3) 由上面的(2)和引理16,可以导出P(F)是无α环的.3 结束语本文论证了在无冲突的条件下产生的规范化对象模式森林可以更好地反映对象间的语义关系.今后,我们将把复杂对象模式规范化设计理论与XML语言、数字图书馆等相结合并加以改进.吴永辉等:相应于无冲突依赖的规范化对象模式森林1493References:[1] Tari, Z., Stokes, J., Spaccapietra, S. Object normal forms and dependency constraints for object-oriented schemata. ACMTransactions on Database Systems, 1997,22(4):513~569.[2] Mok, W.Y., Ng, Y.K., Embley, D.W. Using NNF to transform conceptual data models to object-oriented database designs. Data &Knowledge Engineering, 1998,24(3):313~336.[3] Shi, Bai-le, He, Ji-chao, Cui, Jing. The Theory and Application of Relational Database. Zhengzhou: He’nan Science andTechnology Press, 1989 (in Chinese).[4] Wu, Yong-hui, Zhou, Ao-ying. A normal form for complex object schemes. Advances in Systems Science and Applications, 2000,1(1):48~55.[5] Wu, Yong-hui, Jiang, Wen-yun, Zhou, Ao-ying. Implementation and proof for normalization design of object-oriented dataschemes. In: Chen, Jian, Chen, Ping, Bertrand, M., eds. Proceedings of the 36th International Conference on TOOLS. 2000.220~227.[6] Wu, Yong-Hui, Zhou, Ao-Ying. Research on properties of a set of object dependencies. Journal of Computer Research andDevelopment, 2001,38(12):1491~1498 (in Chinese).[7] Wu, Yong-Hui. Normalization design for complex object schemes [Ph.D. Thesis]. Shanghai: Fudan University, 2001 (in Chinese).[8] Beeri, C., Fagin, R., Maier, D., et al. On the desirability of acyclic database schemes. Journal of the ACM, 1983,30(3):479~513.[9] Beeri, C., Kifer, M. Comprehensive approach to the design of relational database schemes. In: Yuan, L., ed. Proceedings of theConference on VLDB. Singapore: ACM Press, 1984. 196~207.附中文参考文献:[3] 施伯乐,何继朝,崔靖.关系数据库的理论及应用.郑州:河南科学技术出版社,1989.[6] 吴永辉,周傲英.对象依赖集合的性质的研究.计算机研究与发展,2001,38(12):1491~1498.[7] 吴永辉.复杂对象模式的规范化设计[博士学位论文].上海:复旦大学,2001.The Normal Object Scheme Forest with Respect to Conflict-Free DependenciesÃWU Yong-hui1,2,3, ZHOU Ao-ying1,21(Department of Computer Science and Engineering, Fudan University, Shanghai 200433, China);2(Laboratory for Intelligent Information Processing, Fudan University, Shanghai 200433, China);3(Key Laboratory of Computer Science, Institute of Software, The Chinese Academy of Sciences, Beijing 100080, China)E-mail: slwu@Abstract: The properties for a normal object scheme forest with respect to a conflict-free set of ODs are shown in this paper. Firstly basic concepts and properties about object dependency, a conflict-free set of object dependencies, normal object scheme forest and the algorithm of normalization design for complex object schemes are summarized. Then the properties for a normal object scheme forest with respect to a conflict-free set of ODs are presented and proved: P(F) is a unique split-free path set for a normal object scheme forest; M⇔OD(F)⇔⋈P(F); and P(F) is α-cyclic. There is a signification in the development for the object-oriented information systems.Key words: object scheme; object dependency; conflict-free; forest; pathÃReceived June 22, 2001; accepted January 8, 2002Supported by the National Natural Science Foundation of China under Grant No.60003008; the Foundation of the Key Laboratory of Computer Science, Institute of Software, The Chinese Academy of Sciences under Grand No.SYSKF0202。