Enslaving random fluctuations in nonequlibrium systems
AdaptivebacksteppingElman-basedneuralcontrol for unknownnonlinearsystems
Department of Electrical Engineering, Tamkang University, No. 151, Yingzhuan Road, Tamsui District, New Taipei City 25137, Taiwan
art ic l e i nf o
Neurocomputing 136 (2014) 170 –179
Contents lists available at ScienceDirhomepage: /locate/neucom
Adaptive backstepping Elman-based neural control for unknown nonlinear systems
1. Introduction The uncertainty of system dynamics may have strong adverse effects upon system performance, constructing a controller which can achieve favorable control performance is an important issue. The system uncertainties including the unmodeled system dynamics and the external disturbances unavoidably exist in the practical systems. To overcome this problem, many neural-network-based adaptive controllers without any knowledge of the control plants have been proposed [1–9]. The success key element is the self-learning ability that neural networks are used for controller developments and for describing the system dynamics without requiring preliminary offline tuning. By adequately choosing neural network structures, training methods and sufficient input data, the neural-network-based adaptive controllers are capable to compensate for the effects of nonlinearities and system uncertainties, so that the stability, error convergence and robustness of the control system can be guaranteed. Because a RBF neural network has a simple structure, there has been considerable interest in exploring the applications to deal with the nonlinearity and uncertainty of control systems [10–13]. From the controller design viewpoint, since the output for a dynamic nonlinear system is a function of past output, past input, or both, control of this dynamic nonlinear system is not a static control problem. The RBF neural network is static neural network.
Randomness
a rX iv:mat h /1186v2[mat h.PR]1Oct21Randomness Paul Vit´a nyi ∗CWI and Universiteit van Amsterdam Abstract Here we present in a single essay a combination and completion of the several aspects of the problem of randomness of individual objects which of necessity occur scattered in our text [10].The reader can consult different arrangements of parts of the material in [7,20].Contents 1Introduction 21.1Occam’s Razor Revisited .......................31.2Lacuna of Classical Probability Theory ...............41.3Lacuna of Information Theory ....................42Randomness as Unpredictability 62.1Von Mises’Collectives ........................82.2Wald-Church Place Selection ....................113Randomness as Incompressibility 123.1Kolmogorov Complexity .......................143.2Complexity Oscillations .......................163.3Relation with Unpredictability ...................193.4Kolmogorov-Loveland Place Selection ...............204Randomness as Membership of All Large Majorities214.1Typicality ...............................214.2Randomness in Martin-L¨o f’s Sense .................244.3Random Finite Sequences ......................254.4Random Infinite Sequences .....................284.5Randomness of Individual Sequences Resolved (37)5Applications375.1Prediction (37)5.2G¨o del’s incompleteness result (38)5.3Lower bounds (39)5.4Statistical Properties of Finite Sequences (41)5.5Chaos and Predictability (45)1IntroductionPierre-Simon Laplace(1749—1827)has pointed out the following reason why intuitively a regular outcome of a random event is unlikely.“We arrange in our thought all possible events in various classes;andwe regard as extraordinary those classes which include a very smallnumber.In the game of heads and tails,if head comes up a hundredtimes in a row then this appears to us extraordinary,because thealmost infinite number of combinations that can arise in a hundredthrows are divided in regular sequences,or those in which we ob-serve a rule that is easy to grasp,and in irregular sequences,thatare incomparably more numerous”.[place,A PhilosophicalEssay on Probabilities,,Dover,1952.Originally published in1819.Translated from6th French edition.Pages16-17.]If by‘regularity’we mean that the complexity is significantly less than maximal, then the number of all regular events is small(because by simple counting the number of different objects of low complexity is small).Therefore,the event that anyone of them occurs has small probability(in the uniform distribution). Yet,the classical calculus of probabilities tells us that100heads are just as probable as any other sequence of heads and tails,even though our intuition tells us that it is less‘random’than some others.Listen to the redoubtable Dr. Samuel Johnson(1709—1784):“Dr.Beattie observed,as something remarkable which had hap-pened to him,that he chanced to see both the No.1and the No.1000,of the hackney-coaches,thefirst and the last;‘Why,Sir’,saidJohnson,‘there is an equal chance for one’s seeing those two num-bers as any other two.’He was clearly right;yet the seeing of twoextremes,each of which is in some degree more conspicuous than therest,could not but strike one in a stronger manner than the sightof any other two numbers.”[James Boswell(1740—1795),Life ofJohnson,Oxford University Press,Oxford,UK,1970.(Edited byR.W.Chapman,1904Oxford edition,as corrected by J.D.Fleeman,third edition.Originally published in1791.)Pages1319-1320.]Laplace distinguishes between the object itself and a cause of the object.2“The regular combinations occur more rarely only because they areless numerous.If we seek a cause wherever we perceive symmetry,itis not that we regard the symmetrical event as less possible than theothers,but,since this event ought to be the effect of a regular causeor that of chance,thefirst of these suppositions is more probablethan the second.On a table we see letters arranged in this order Co n s t a n t i n o p l e,and we judge that this arrangementis not the result of chance,not because it is less possible than others,for if this word were not employed in any language we would notsuspect it came from any particular cause,but this word being inuse among us,it is incomparably more probable that some personhas thus arranged the aforesaid letters than that this arrangementis due to chance.”[place,Ibid.]Let us try to turn Laplace’s argument into a formal one.First we introduce some notation.If x is afinite binary sequence,then l(x)denotes the length (number of occurrences of binary digits)in x.For example,l(010)=3.1.1Occam’s Razor RevisitedSuppose we observe a binary string x of length l(x)=n and want to know whether we must attribute the occurrence of x to pure chance or to a cause. To put things in a mathematical framework,we define chance to mean that the literal x is produced by independent tosses of a fair coin.More subtle is the interpretation of cause as meaning that the computer on our desk computes x from a program provided by independent tosses of a fair coin.The chance of generating x literally is about2−n.But the chance of generating x in the form of a short program x∗,the cause from which our computer computes x,is at least2−l(x∗).In other words,if x is regular,then l(x∗)≪n,and it is about 2n−l(x∗)times more likely that x arose as the result of computation from some simple cause(like a short program x∗)than literally by a random process.This approach will lead to an objective and absolute version of the classic maxim of William of Ockham(1290?–1349?),known as Occam’s razor:“if there are alternative explanations for a phenomenon,then,all other things being equal,we should select the simplest one”.One identifies‘simplicity of an object’with‘an object having a short effective description’.In other words,a priori we consider objects with short descriptions more likely than objects with only long descriptions.That is,objects with low complexity have high probability while objects with high complexity have low probability.This principle is intimately related with problems in both probability theory and information theory.These problems as outlined below can be interpreted as saying that the related disciplines are not‘tight’enough;they leave things unspecified which our intuition tells us should be dealt with.31.2Lacuna of Classical Probability TheoryAn adversary claims to have a true random coin and invites us to bet on the outcome.The coin produces a hundred heads in a row.We say that the coin cannot be fair.The adversary,however,appeals to probabity theory which says that each sequence of outcomes of a hundred coinflips is equally likely,1/2100, and one sequence had to come up.Probability theory gives us no basis to challenge an outcome after it has happened.We could only exclude unfairness in advance by putting a penalty side-bet on an outcome of100heads.But what about1010...?What about an initial segment of the binary expansion ofπ?Regular sequence1Pr(00000000000000000000000000)=226Random sequence1Pr(10010011011000111011010000)=being equally probable,this quantity is the number of bits needed to count all possibilities.This expresses the fact that each message in the ensemble can be communi-cated using this number of bits.However,it does not say anything about the number of bits needed to convey any individual message in the ensemble.To illustrate this,consider the ensemble consisting of all binary strings of length 9999999999999999.By Shannon’s measure,we require9999999999999999bits on the average to encode a string in such an ensemble.However,the string consisting of 99999999999999991’s can be encoded in about55bits by expressing9999999999 999999in binary and adding the repeated pattern‘1’.A requirement for this to work is that we have agreed on an algorithm that decodes the encoded string. We can compress the string still further when we note that9999999999999999 equals32×1111111111111111,and that1111111111111111consists of241’s.Thus,we have discovered an interesting phenomenon:the description of some strings can be compressed considerably,provided they exhibit enough regularity.This observation,of course,is the basis of all systems to express very large numbers and was exploited early on by Archimedes(287BC—212BC)in his treatise The Sand-Reckoner,in which he proposes a system to name very large numbers:“There are some,King Golon,who think that the number of sandis infinite in multitude[...or]that no number has been named whichis great enough to exceed its multitude.[...]But I will try to showyou,by geometrical proofs,which you will be able to follow,that,of the numbers named by me[...]some exceed not only the massof sand equal in magnitude to the earthfilled up in the way de-scribed,but also that of a mass equal in magnitude to the universe.”[Archimedes,The Sand-Reckoner,pp.420-429in:The World ofMathematics,Vol.1,J.R.Newman,Ed.,Simon and Schuster,NewYork,1956.Page420.]However,if regularity is lacking,it becomes more cumbersome to express large numbers.For instance,it seems easier to compress the number‘one billion,’than the number‘one billion seven hundred thirty-five million two hundred sixty-eight thousand and three hundred ninety-four,’even though they are of the same order of magnitude.The above example shows that we need too many bits to transmit regular objects.The converse problem,too little bits,arises as well since Shannon’s theory of information and communication deals with the specific technology problem of data transmission.That is,with the information that needs to be transmitted in order to select an object from a previously agreed upon set of alternatives;agreed upon by both the sender and the receiver of the message. If we have an ensemble consisting of the Odyssey and the sentence“let’s go drink a beer”then we can transmit the Odyssey using only one bit.Yet Greeks5feel that Homer’s book has more information contents.Our task is to widen the limited set of alternatives until it is universal.We aim at a notion of ‘absolute’information of individual objects,which is the information which by itself describes the object completely.Formulation of these considerations in an objective manner leads again to the notion of shortest programs and Kolmogorov complexity.2Randomness as UnpredictabilityWhat is the proper definition of a random sequence,the‘lacuna in probability theory’we have identified above?Let us consider how mathematicians test ran-domness of individual sequences.To measure randomness,criteria have been developed which certify this quality.Yet,in recognition that they do not mea-sure‘true’randomness,we call these criteria‘pseudo’randomness tests.For instance,statistical survey of initial segments of the sequence of decimal dig-its ofπhave failed to disclose any significant deviations of randomness.But clearly,this sequence is so regular that it can be described by a simple program to compute it,and this program can be expressed in a few bits.“Any one who considers arithmetical methods of producing randomdigits is,of course,in a state of sin.For,as has been pointed outseveral times,there is no such thing as a random number—there areonly methods to produce random numbers,and a strict arithmeticalprocedure is of course not such a method.(It is true that a problemwe suspect of being solvable by random methods may be solvable bysome rigorously defined sequence,but this is a deeper mathematicalquestion than we can go into now.)”[John Louis von Neumann(1903—1957),Various techniques used in connection with randomdigits,J.Res.Nat.Bur.Stand.Appl.Math.Series,3(1951),pp.36-38.Page36.Also,Collected Works,Vol.1,A.H.Taub,Ed.,Pergamon Press,Oxford,1963,pp.768-770.Page768.]This fact prompts more sophisticated definitions of randomness.In his famous address to the International Congress of Mathematicians in1900,David Hilbert (1862—1943)proposed twenty-three mathematical problems as a program to direct the mathematical efforts in the twentieth century.The6th problem asks for”To treat(in the same manner as geometry)by means of axioms,those physical sciences in which mathematics plays an important part;in thefirst rank are the theory of probability..”.Thus,Hilbert views probability theory as a physical applied theory.This raises the question about the properties one can expect from typical outcomes of physical random sources,which a priori has no relation whatsoever with an axiomatic mathematical theory of probabilities. That is,a mathematical system has no direct relation with physical reality.To6obtain a mathematical system that is an appropriate model of physical phe-nomena one needs to identify and codify essential properties of the phenomena under consideration by empirical observations.Notably Richard von Mises(1883—1953)proposed notions that approach the very essence of true randomness of physical phenomena.This is related with the construction of a formal mathematical theory of probability,to form a basis for real applications,in the early part of this century.While von Mises’objective was to justify the applications to the real phenomena,Andrei Niko-laevitch Kolmogorov’s(1903—1987)classic1933treatment constructs a purely axiomatic theory of probability on the basis of set theoretic axioms.“This theory was so successful,that the problem offinding the basisof real applications of the results of the mathematical theory of prob-ability became rather secondary to many investigators....[however]the basis for the applicability of the results of the mathematical the-ory of probability to real‘random phenomena’must depend in someform on the frequency concept of probability,the unavoidable natureof which has been established by von Mises in a spirited manner.”[A.N.Kolmogorov,On tables of random numbers,Sankhy¯a,SeriesA,25(1963),369-376.Page369.]The point made is that the axioms of probability theory are designed so that abstract probabilities can be computed,but nothing is said about what prob-ability really means,or how the concept can be applied meaningfully to the actual world.Von Mises analyzed this issue in detail,and suggested that a proper definition of probability depends on obtaining a proper definition of a random sequence.This makes him a‘frequentist’—a supporter of the frequency theory.The following interpretation and formulation of this theory is due to John Edensor Littlewood(1885—1977),The dilemma of probability theory,Little-wood’s Miscellany,Revised Edition,B.Bollob´a s,Ed.,Cambridge University Press,1986,pp.71-73.The frequency theory to interpret probability says, roughly,that if we perform an experiment many times,then the ratio of favor-able outcomes to the total number n of experiments will,with certainty,tend to a limit,p say,as n→∞.This tells us something about the meaning of probability,namely,the measure of the positive outcomes is p.But suppose we throw a coin1000times and wish to know what to expect.Is1000enough for convergence to happen?The statement above does not say.So we have to add something about the rate of convergence.But we cannot assert a certainty about a particular number of n throws,such as‘the proportion of heads will be p±ǫfor large enough n(withǫdepending on n)’.We can at best say‘the proportion will lie between p±ǫwith at least such and such probability(de-pending onǫand n0)whenever n>n0’.But now we defined probability in an obviously circular fashion.72.1Von Mises’CollectivesIn1919von Mises proposed to eliminate the problem by simply dividing all infi-nite sequences into special random sequences(called collectives),having relative frequency limits,which are the proper subject of the calculus of probabilities and other sequences.He postulates the existence of random sequences(thereby circumventing circularity)as certified by abundant empirical evidence,in the manner of physical laws and derives mathematical laws of probability as a con-sequence.In his view a naturally occurring sequence can be nonrandom or unlawful in the sense that it is not a proper collective.Von Mises views the theory of probabilities insofar as they are nu-merically representable as a physical theory of definitely observ-able phenomena,repetitive or mass events,for instance,as foundin games of chance,population statistics,Brownian motion.‘Prob-ability’is a primitive notion of the theory comparable to those of‘energy’or‘mass’in other physical theories.Whereas energy or mass exist infields or material objects,proba-bilities exist only in the similarly mathematical idealization of collec-tives(random sequences).All problems of the theory of probabilityconsist of deriving,according to certain rules,new collectives fromgiven ones and calculating the distributions of these new collectives.The exact formulation of the properties of the collectives is secondaryand must be based on empirical evidence.These properties are theexistence of a limiting relative frequency and randomness.The property of randomness is a generalization of the abundant experience in gambling houses,namely,the impossibility of a suc-cessful gambling system.Including this principle in the foundationof probability,von Mises argues,we proceed in the same way as thephysicists did in the case of the energy principle.Here too,the ex-perience of hunters of fortune is complemented by solid experienceof insurance companies and so forth.A fundamentally different approach is to justify a posteriori theapplication of a purely mathematically constructed theory of prob-ability,such as the theory resulting from the Kolmogorov axioms.Suppose,we can show that the appropriately defined random se-quences form a set of measure one,and without exception satisfyall laws of a given axiomatic theory of probability.Then it appearspractically justifiable to assume that as a result of an(infinite)ex-periment only random sequences appear.Von Mises’notion of infinite random sequence of0’s and1’s(collective)essen-tially appeals to the idea that no gambler,making afixed number of wagers of ‘heads’,atfixed odds[say p versus1−p]and infixed amounts,on theflips of a coin[with bias p versus1−p],can have profit in the long run from betting ac-8cording to a system instead of betting at random.Says Alonzo Church(1903—):“this definition[below]...while clear as to general intent,is too inexact in form to serve satisfactorily as the basis of a mathematical theory.”[A.Church, On the concept of a random sequence,Bull.Amer.Math.Soc.,46(1940),pp. 130-135.Page130.]Definition1An infinite sequence a1,a2,...of0’s and1’s is a random sequence in the special meaning of collective if the following two conditions are satisfied.1.Let f n is the number of1’s among thefirst n terms of the sequence.Thenf nlimn→∞we should distinguish between randomness proper(as absence of anyregularity)and stochastic randomness(which is the subject of prob-ability theory).There emerges the problem offinding reasons forthe applicability of the mathematical theory of probability to thereal world.”[A.N.Kolmogorov,On logical foundations of probabil-ity theory,Probability Theory and Mathematical Statistics,LectureNotes in Mathematics,Vol.1021,K.Itˆo and J.V.Prokhorov,Eds.,Springer-Verlag,Heidelberg,1983,pp.1-5.Page1.]Intuitively,we can distinguish between sequences that are irregular and do not satisfy the regularity implicit in stochastic randomness,and sequences that are irregular but do satisfy the regularities associated with stochastic randomness. Formally,we will distinguish the second type from thefirst type by whether or not a certain complexity measure of the initial segments goes to a definite limit. The complexity measure referred to is the length of the shortest description of the prefix(in the precise sense of Kolmogorov complexity)divided by its length. It will turn out that almost all infinite strings are irregular of the second type and satisfy all regularities of stochastic randomness.“In applying probability theory we do not confine ourselves to negat-ing regularity,but from the hypothesis of randomness of the ob-served phenomena we draw definite positive conclusions.”[A.N.Kol-mogorov,Combinatorial foundations of information theory and thecalculus of probabilities,Russian Mathematical Surveys,,38:4(1983),pp.29-40.Page34.]Considering the sequence as fair coin tosses with p=1/2,the second condition in Definition1says there is no strategyφ(principle of excluded gambling system) which assures a player betting atfixed odds and infixed amounts,on the tosses of the coin,to make infinite gain.That is,no advantage is gained in the long run by following some system,such as betting‘head’after each run of seven consecutive tails,or(more plausibly)by placing the n th bet‘head’after the appearance of n+7tails in succession.According to von Mises,the above conditions are sufficiently familiar and a uncontroverted empirical generalization to serve as the basis of an applicable calculus of probabilities.Example1It turns out that the naive mathematical approach to a concrete formulation,admitting simply all partial functions,comes to grief as follows. Let a=a1a2...be any collective.Defineφ1asφ1(a1...a i−1)=1if a i=1, and undefined otherwise.But then p=1.Definingφ0byφ0(a1...a i−1)=b i, with b i the complement of a i,for all i,we obtain by the second condition of Definition1that p=0.Consequently,if we allow functions likeφ1andφ0as strategy,then von Mises’definition cannot be satisfied at all.3102.2Wald-Church Place SelectionIn the thirties,Abraham Wald(1902—1950)proposed to restrict the a priori admissibleφto anyfixed countable set of functions.Then collectives do exist. But which countable set?In1940,Alonzo Church proposed to choose a set of functions representing‘computable’strategies.According to Church’s Thesis, this is precisely the set of recursive functions.With recursiveφ,not only is the definition completely rigorous,and random infinite sequences do exist,but moreover they are abundant since the infinite random sequences with p=1/2 form a set of measure one.From the existence of random sequences with proba-bility1/2,the existence of random sequences associated with other probabilities can be derived.Let us call sequences satisfying Definition1with recursiveφMises-Wald-Church random.That is,the involved Mises-Wald-Church place-selection rules consist of the partial recursive functions.Appeal to a theorem by Wald yields as a corollary that the set of Mises-Wald-Church random sequences associated with anyfixed probability has the cardinality of the continuum.Moreover,each Mises-Wald-Church random se-quence qualifies as a normal number.(A number is normal in the sense of´Emile F´e lix´Edouard Justin Borel(1871—1956)if each digit of the base,and each block of digits of any length,occurs with equal asymptotic frequency.)Note however, that not every normal number is Mises-Wald-Church random.This follows,for instance,from Champernowne’s sequence(or number),0.1234567891011121314151617181920...due to David G.Champernowne(1912—),which is normal in the scale of10 and where the i th digit is easily calculated from i.The definition of a Mises-Wald-Church random sequence implies that its consecutive digits cannot be effectively computed.Thus,an existence proof for Mises-Wald-Church random sequences is necessarily nonconstructive.Unfortunately,the von Mises-Wald-Church definition is not yet good enough, as was shown by Jean Ville in1939.There exist sequences that satisfy the Mises-Wald-Church definition of randomness,with limiting relative frequency of ones of1/2,but nonetheless have the property thatf nfor all n.2The probability of such a sequence of outcomes in randomflips of a fair coin is zero.Intuition:if you bet‘1’all the time against such a sequence of outcomes, then your accumulated gain is always positive!Similarly,other properties of randomness in probability theory such as the Law of the Iterated Logarithm do not follow from the Mises-Wald-Church definition.An extensive survey on these issues(and parts of the sequel)is given in[8].113Randomness as IncompressibilityAbove it turned out that describing‘randomness’in terms of‘unpredictability’is problematic and possibly unsatisfactory.Therefore,Kolmogorov tried another approach.The antithesis of‘randomness’is‘regularity’,and afinite string which is regular can be described more shortly than giving it literally.Consequently,a string which is‘incompressible’is‘random’in this sense.With respect to infinite binary sequences it is seductive to call an infinite sequence‘random’if all of its initial segments are‘random’in the above sense of being‘incompressible’.Let us see how this intuition can be made formal,and whether leads to a satisfactory solution.Intuitively,the amount of effectively usable information in afinite string is the size(number of binary digits or bits)of the shortest program that,without additional data,computes the string and terminates.A similar definition can be given for infinite strings,but in this case the program produces element after element forever.Thus,a long sequence of1’s such as10,000times11111 (1)contains little information because a program of size about log10,000bits out-puts it:for i:=1to10,000print1Likewise,the transcendental numberπ=3.1415...,an infinite sequence of seemingly‘random’decimal digits,contains but a few bits of information.(There is a short program that produces the consecutive digits ofπforever.)Such a definition would appear to make the amount of information in a string(or other object)depend on the particular programming language used.Fortunately,it can be shown that all reasonable choices of programming languages lead to quantification of the amount of‘absolute’information in indi-vidual objects that is invariant up to an additive constant.We call this quantity the‘Kolmogorov complexity’of the object.If an object is regular,then it has a shorter description than itself.We call such an object‘compressible’.More precisely,suppose we want to describe a given object by afinite binary string.We do not care whether the object has many descriptions;however,each description should describe but one object.From among all descriptions of an object we can take the length of the shortest description as a measure of the object’s complexity.It is natural to call an object‘simple’if it has at least one short description,and to call it‘complex’if all of its descriptions are long.But now we are in danger of falling in the trap so eloquently described in the Richard-Berry paradox,where we define a natural number as“the least natural number that cannot be described in less than twenty words”.If this number12does exist,we have just described it in thirteen words,contradicting its defini-tional statement.If such a number does not exist,then all natural numbers can be described in less than twenty words.(This paradox is described in[Bertrand Russell(1872—1970)and Alfred North Whitehead,Principia Mathematica,Ox-ford,1917].In a footnote they state that it“was suggested to us by Mr.G.G. Berry of the Bodleian Library”.)We need to look very carefully at the notion of‘description’.Assume that each description describes at most one object.That is,there is a specification method D which associates at most one object x with a description y.This means that D is a function from the set of descriptions,say Y,into the set of objects,say X.It seems also reasonable to require that,for each object x in X,there is a description y in Y such that D(y)=x.(Each object has a description.)To make descriptions useful we like them to befinite.This means that there are only countably many descriptions.Since there is a description for each object,there are also only countably many describable objects.How do we measure the complexity of descriptions?Taking our cue from the theory of computation,we express descriptions as finite sequences of0’s and1’s.In communication technology,if the specification method D is known to both a sender and a receiver,then a message x can be transmitted from sender to receiver by transmitting the sequence of0’s and1’s of a description y with D(y)=x.The cost of this transmission is measured by the number of occurrences of0’s and1’s in y,that is,by the length of y. The least cost of transmission of x is given by the length of a shortest y such that D(y)=x.We choose this least cost of transmission as the‘descriptional’complexity of x under specification method D.Obviously,this descriptional complexity of x depends crucially on D.The general principle involved is that the syntactic framework of the description language determines the succinctness of description.In order to objectively compare descriptional complexities of objects,to be able to say“x is more complex than z”,the descriptional complexity of x should depend on x alone.This complexity can be viewed as related to a universal description method which is a priori assumed by all senders and receivers.This complexity is optimal if no other description method assigns a lower complexity to any object.We are not really interested in optimality with respect to all description methods.For specifications to be useful at all it is necessary that the mapping from y to D(y)can be executed in an effective manner.That is,it can at least in principle be performed by humans or machines.This notion has been formalized as‘partial recursive functions’.According to generally accepted mathematical viewpoints it coincides with the intuitive notion of effective computation.The set of partial recursive functions contains an optimal function which minimizes description length of every other such function.We denote this func-tion by ly,for any other recursive function D,for all objects x,there is a description y of x under D0which is shorter than any description z of x13。
《孟德尔随机化研究指南》中英文版
《孟德尔随机化研究指南》中英文版English:"Mendelian randomization (MR) has emerged as an important tool in epidemiology and biostatistics for investigating causal relationships between risk factors and disease outcomes. In order to ensure the validity and reliability of MR studies, researchers need to follow a standardized set of guidelines. The 'Mendelian Randomization Reporting Guidelines (MR-REWG)' provide detailed recommendations for conducting, reporting, and appraising MR studies. These guidelines cover key aspects such as study design, instrument selection, data sources, statistical analysis, and result interpretation. By adhering to these guidelines, researchers can minimize bias and confounding, and produce more robust evidence for causal inference in epidemiological research."中文翻译:“孟德尔随机化(MR)已经成为流行病学和生物统计学中研究危险因素与疾病结果之间因果关系的重要工具。
《2024年随机森林算法优化研究》范文
《随机森林算法优化研究》篇一一、引言随机森林(Random Forest)是一种以决策树为基础的集成学习算法,由于其优秀的性能和稳健的表现,被广泛应用于机器学习和数据挖掘领域。
然而,随机森林算法在处理复杂问题时仍存在过拟合、效率低下等问题。
本文旨在研究随机森林算法的优化方法,提高其准确性和效率。
二、随机森林算法概述随机森林算法通过构建多个决策树来对数据进行训练和预测,每个决策树都使用随机选择的一部分特征进行训练。
最终,随机森林对各个决策树的预测结果进行集成,以得到更为准确的预测结果。
随机森林算法具有抗过拟合能力强、训练效率高、易实现等优点。
三、随机森林算法存在的问题虽然随机森林算法在很多领域取得了显著的效果,但仍然存在一些问题:1. 过拟合问题:当数据集较大或特征维度较高时,随机森林算法容易产生过拟合现象。
2. 计算效率问题:随着数据集规模的扩大,随机森林算法的计算效率会逐渐降低。
3. 特征选择问题:在构建决策树时,如何选择合适的特征是一个关键问题。
四、随机森林算法优化方法针对上述问题,本文提出以下优化方法:1. 引入集成学习技术:通过集成多个随机森林模型,可以有效提高模型的泛化能力和抗过拟合能力。
例如,可以使用Bagging、Boosting等集成学习技术来构建多个随机森林模型,并对它们的预测结果进行集成。
2. 优化决策树构建过程:在构建决策树时,可以采用特征选择方法、剪枝技术等来提高决策树的准确性和泛化能力。
此外,还可以通过调整决策树的深度、叶子节点数量等参数来优化模型性能。
3. 特征重要性评估与选择:在构建随机森林时,可以利用特征重要性评估方法来识别对模型预测结果贡献较大的特征。
然后,根据实际需求和业务背景,选择合适的特征进行建模。
这样可以减少噪声特征对模型的影响,提高模型的准确性和效率。
4. 优化模型参数:针对不同的问题和数据集,可以通过交叉验证等方法来调整随机森林算法的参数,如决策树的数量、每个决策树所使用的特征数量等。
随机森林构建方法英语作文
随机森林构建方法英语作文Random Forest Construction Method。
Random Forest is a popular machine learning algorithm that is used to solve a wide range of problems, including classification and regression. It is a type of ensemble learning method that combines multiple decision trees to produce a more accurate and robust model. In this article, we will discuss the construction method of Random Forest.Step 1: Data Preparation。
The first step in building a Random Forest model is to prepare the data. This involves cleaning the data, removing any missing values, and transforming the data into a suitable format for the algorithm. The data should be split into a training set and a testing set, with the training set used to train the model and the testing set used to evaluate its performance.Step 2: Random Sampling。
Random Forest uses a technique called bagging, which involves randomly sampling the data with replacement to create multiple subsets of the data. Each subset is used to train a decision tree, and the results are combined to produce the final model. The number of subsets is determined by the user and is typically set to a value between 100 and 1000.Step 3: Decision Tree Construction。
üèDòáò×ù für Mathematik in den Naturwissenschaften Leipzig
f¨u r Mathematikin den NaturwissenschaftenLeipzigRandom perturbations of spiking activity in apair of coupled neuronsbyBoris Gutkin,J¨u rgen Jost,and Henry TuckwellPreprint no.:492007Random perturbations of spiking activity in apair of coupled neuronsBoris Gutkin∗,J¨u rgen Jost and Henry C.Tuckwell†May14,2007AbstractWe examine the effects of stochastic input currents on thefiring be-haviour of two coupled Type1or Type2neurons.In Hodgkin-Huxleymodel neurons with standard parameters,which are Type2,in the bistableregime,synaptic transmission can initiate oscillatory joint spiking,butwhite noise can terminate it.In Type1cells(models),typified by aquadratic integrate andfire model,synaptic coupling can cause oscilla-tory behaviour in excitatory cells,but Gaussian white noise can againterminate it.We locally determine an approximate basin of attraction,A,of the periodic orbit and explain thefiring behaviour in terms of theeffects of noise on the probability of escape of trajectories from A.1IntroductionHodgkin(1948)found that various squid axon preparations responded in quali-tatively different ways to applied currents.Some preparations gave a frequency offiring which rose smoothly from zero as the current increased whereas oth-ers manifested the sudden appearance of a train of spikes at a particular input current.Cells that responded in thefirst manner were called Class1(which we refer to as Type1)whereas cells with a discontinuous frequency-current curve were called Class2(Type2).Mathematical explanations for the two types are found in the bifurcation which accompanies the transition from rest state to a periodicfiring mode.For Type1behaviour,a resting potential vanishes via a saddle-node bifurcation whereas for Type2behaviour the instability of the rest point is due to an Andronov-Hopf bifurcation,see Rinzel and Ermentrout (1989).Stochastic effects in thefiring behaviour of neurons have been widely reported, discussed and analyzed since their discovery in the1940’s.One of thefirst reports for the central nervous system was by Frank and Fuortes(1955)for catX1X3X2X4X1X2TIMEFigure1:On the left are shown the solutions of(1)-(4)for two coupled QIF model neurons with the standard parameters.X1and X2are the potential variables of neurons1and2and X3and X4are the inputs to neurons1and2, respectively.On the right is shown the periodic orbit in the(x1,x2)-plane.The square marked P was explored in detail in reference to the extent of the basin of attraction of the periodic orbit.spinal neurons.Although there have been many single neuron studies,the effect of noise on systems of coupled neurons have not been extensively investigated. Some preliminary studies are those of Gutkin,Hely and Jost(2004)and Casado and Baltan´a s(2003).2The quadratic integrate andfire modelA relatively simple neural model which exhibits Type1firing behaviour is the quadratic integrate andfire(QIF)model.We couple two model neurons in the following manner(Gutkin,Hely and Jost,2004).Let{X1(t),X2(t),t≥0}be the depolarizations of neurons1and2,where t is the time index.Then the model equations are,for subthreshold states of two identical neurons,dX1=[(X1−x R)2+β+g s X3]dt+σdW1(1)dX2=[(X2−x R)2+β+g s X4]dt+σdW2(2)dX3=−X3τ+F(X1)(4)2where X3is the synaptic input to neuron1from neuron2and X4is the synaptic input to neuron2from neuron1.The quantity x R is a resting value.g s is the coupling strength.βis the mean background input.W1and W2are independent standard Wiener processes which enter with strengthσ.This term may model variations in nonspecific inputs to the circuit as well as possibly intrinsic membrane and channel noise.By construction,we take this term to be much weaker than the mutual coupling between the cells in our circuit.The function F is given byF(x)=1+tanh(α(x−θ))whereθcharacterizes the threshold effect of synaptic activation.Since when a QIF neuron is excited and it receives no inhibition,its potential reaches an infinite value in afinite time,for numerical simulations a cutoffvalue x max is introduced so that the above model equations for the potential apply only if X1 or X2are below x max.To complete a“spike”in any neuron,taken as occurring when its potential reaches x max,its potential is instantaneously reset to some value x reset which may be taken as−x max.At the bifurcation point g s=g∗s, two heteroclinic orbits between unstable rest points turn into a periodic orbit of antiphase oscillations.3Results and theoryIn the numerical work,the following constants are employed throughout.x R= 0,x max=20,θ=10,α=1,β=−1,g s=100andτ=0.25.The initial values of the neural potentials are X1(0)=1.1,X2(0)=0and the initial values of the synaptic variables are X3(0)=X4(0)=0.When there is no noise,σ=0,the results of Figure1are obtained.The spike trains of the two coupled neurons and their synaptic inputs are shown on the left.Thefiring settles down to be quite regular and the periodic orbit,S,is shown on the right.The patch marked P is the location of the region explored in detail below.The effects of a small amount of noise are shown in Figure2.The neural excitation variables are shown on the left and the corresponding trajectories in the(x1,x2)-plane are shown on the right.In the top portion an example of the trajectory forσ=0.1is shown.Here three spikes arise in neuron1and two in neuron2,but the time between spikes increases and eventually the orbit collapses away from the periodic orbit.In the example(lower part)forσ=0.2 there are no spikes in either neuron.In10trials,the average numbers of spikes obtained for the pair of neurons were(2.5,2.2)forσ=0.1,(1.4,1.1)forσ=0.2 and(1.3,0.9)forσ=0.3;these may be compared with(5,5)for zero noise. 3.1Exit-time and orbit stabilityIf a basin of attraction for a periodic orbit can be found,then the probabil-ity that the process with noise escapes from the region of attraction gives the probability,in the present context,that spiking will cease.Since the system3TIMEX1X21 X2Figure2:On the left are shown examples of the neuronal potentials for neurons 1and2(QIF model)for two values of the noise,σ=0.1andσ=0.2.On the right are shown the trajectories corresponding to the results on the left,showing how noise pushes or keeps the trajectories out of the basin of attraction of the periodic orbit.(1)-(4)is Markovian,we may apply standardfirst-exit time theory(Tuckwell, 1989).Letting A be a set in R4and letting x=(x1,x2,x3,x4)∈A be a values of X1,X2,X3,X4)at some given time,the probability p(x1,x2,x3,x4)that the process ever escapes from A is given byL p≡σ2∂x21+σ2∂x22(5)+[(x1−x R)2+β+g s x3]∂p∂x2+ F(x2)−x3∂x3+ F(x1)−x4∂x4=0,x∈Awith boundary condition that p=1on the boundary of A(since the process is continuous).If one also adds an arbitrarily small amount of noise for X3and X4(or considers those solutions of(5)that arise from the limit of vanishing noise for X3,X4),the solution of the linear elliptic partial differential equation (5)is unique and≡1,that is,the process will eventually excape from A with probability1.Hence,the expected time f(x)of exit of the process from A satisfies L f=−1,x∈A with boundary condition f=0on the boundary of A.In fact,for small noise,the logarithm of the expected exit time from A,that4is,the time at whichfiring stops,behaves like the inverse of the square of the noise amplitude(Freidlin and Wentzell,1998).These linear partial differential equations can be solved numerically,for example by Monte-Carlo techniques.The basin of attraction A must be found in order to identify the domain of(5).We have done this approximately for the square P in Figure1.The effects of perturbations of the periodic orbit S within P on the spiking activity were found by solving(1)-(4)with various initial conditions in the absence of noise.The values of x1were from−0.43to1.57in steps of0.2and the values of x2were from-4to2also in steps of0.2.For this particular region, as expected from geometrical considerations,the system responded sensitively to to variations in x1but not x2.For example,to the left of S there tended to be no spiking activity whereas just to the right there was a full complement of spikes and further to the right(but still inside P)one spike.4Coupled Hodgkin-Huxley neuronsAs an example of a Type2neuron,we use the standard Hodgkin-Huxley(HH) model augmented with synaptic input variables as in the model for coupled QIF neurons given by equations(3)and(4),but with different parameter values. It has been long known that additive noise has a facilitative effect on single HH neurons(Yu and Lewis,1989).Coupled pairs of HH neurons have been employed with a different approach using conductance noise in order to analyze synchronization properties(e.g.Casado and Balt´a nas,2003).For the present approach,with X1and X2as the depolarizations of the two cells,we putdX1=1g K n4(V K−X1)+it was found that transient synchronization can terminate sustained activity. For Type2neurons,we have investigated coupled Hodgkin-Huxley neurons and found that in the bistable regime,noise can again terminate sustained spiking activity initiated by synaptic connections.We have investigated a minimal cir-cuit model of sustained neural activity.Such sustained activity in the prefrontal cortex has been proposed as a neural correlate of working memory(Fuster and Alexander,1973).ReferencesCasado,J.M.,Balt´a nas,J.P.(2003).Phase switching in a system of two noisy Hodgkin-Huxley neurons coupled by a diffusive interaction.Phys.Rev.E68,061917,Frank,K.,Fuortes,M.G.(1955).Potentials recorded from the spinal cord with microelectrodes,J.Physiol.130,625-654.Freidlin,M.I.,Wentzell,A.D.(1998),Random Perturbations of Dynamical Sys-tems,2nd ed.,Springer,New York Fuster,J.M.and Alexander,G.E.(1971),Neuron activity related to short-term memory.Science652-654 Gutkin,B.,Ermentrout,G.B.(1998).Dynamics of membrane excitability de-termine interval variability:a link between spike generation mechanismsand cortical spike train statistics.Neural Comp.10,1047-1065. Gutkin,B.S.et al.(2001)Turning on and offwith p.Neurosc.11:2,121-134Gutkin,B.,Hely,T.,Jost,J.(2004).Noise delays onset of sustainedfiring in a minimal model of persistent activity.Neurocomputing58-60,753-760. Hodgkin,A.L.(1948).The local changes associated with repetitive action in a non-medullated axon.J.Physiol.107,165-181.Rinzel,J.,Ermentrout,G.B.(1989).Analysis of neural excitability and oscilla-tions;in:Koch C.&Segev I.,eds.MIT Press.Tateno,T.,Harsch,A.,Robinson,H.P.C.(2004).Thresholdfiring frequency-current relationships of neurons in rat somatosensory cortex:Type1and Type2dynamics.J.Neurophysiol.92,2283-2294.Tuckwell,H.C.(1989).Stochastic Processes in the Neurosciences.SIAM,Philadel-phia.Yu,X.,Lewis,E.R.(1989).Studies with spike initiators:linearization by noise allows continuous signal modulation in neural networks.IEEE Trans.Biomed.Eng.36,36-43.6。
第四十四篇
An Efficient Approach to Nondominated Sorting for Evolutionary Multiobjective OptimizationXingyi Zhang,Ye Tian,Ran Cheng,and Yaochu Jin,Senior Member,IEEEAbstract—Evolutionary algorithms have been shown to be powerful for solving multiobjective optimization problems,in which nondominated sorting is a widely adopted technique in selection.This technique,however,can be computationally expen-sive,especially when the number of individuals in the population becomes large.This is mainly because in most existing nondom-inated sorting algorithms,a solution needs to be compared with all other solutions before it can be assigned to a front.In this paper we propose a novel,computationally efficient approach to nondominated sorting,termed efficient nondominated sort(ENS). In ENS,a solution to be assigned to a front needs to be compared only with those that have already been assigned to a front,thereby avoiding many unnecessary dominance comparisons.Based on this new approach,two nondominated sorting algorithms have been suggested.Both theoretical analysis and empirical results show that the ENS-based sorting algorithms are computationally more efficient than the state-of-the-art nondominated sorting methods.Index Terms—Computational complexity,evolutionary multi-objective optimization,nondominated sorting,Pareto-optimality.I.I NTRODUCTIONM OST REAL-WORLD optimization problems are char-acterized by multiple objectives that often conflict with each other.For solving such multiobjective optimiza-tion problems(MOPs),a set of optimal solutions,known as Pareto-optimal solutions,instead of a single optimal solution, are to be achieved.Most classical optimization methods are inefficient in solving MOPs,since they can typicallyfind only one Pareto-optimal solution in one run,which means that this kind of method has to be applied multiple times to achieve a Pareto-optimal solution set.Manuscript received July25,2013;revised November10,2013;accepted February6,2014.Date of publication March13,2014;date of current version March27,2015.This work was supported in part by the National Natural Science Foundation of China under Projects61272152,61033003,91130034, 61373066,61073116,61003131,and61202011;in part by the Ph.D.Programs Foundation,Ministry of Education of China under Project20100142110072; in part by the Fundamental Research Funds for the Central Universities under Project2010ZD001;in part by the Natural Science Foundation of Anhui Higher Education Institutions of China under Projects KJ2012A010and KJ2013A007;and in part by the Scientific Research Foundation for Doctor of Anhui University under Project02203104.X.Zhang and Y.Tian are with the Key Laboratory of Intelligent Comput-ing and Signal Processing of Ministry of Education,School of Computer Science and Technology,Anhui University,Hefei230039,China(e-mail: xyzhanghust@;field910921@).R.Cheng and Y.Jin are with the Department of Computing,University of Surrey,Guildford,Surrey GU27XH,U.K.(e-mail:r.cheng@; yaochu.jin@).Digital Object Identifier10.1109/TEVC.2014.2308305Over the past20years,a variety of evolutionary algorithms have been developed to tackle MOPs,for example, Pareto envelop-based selection algorithm II(PESA-II)[1], nondominated sorting genetic algorithm-II(NSGA-II)[2], strength Pareto evolutionary algorithm2(SPEA2)[3],and memetic Pareto archived evolution strategy(M-PAES)[4],to name just a few.These multiobjective evolutionary algorithms (MOEAs)are able tofind a set of Pareto-optimal solutions in one single run.Although various approaches have been adopted for selec-tion[5],most MOEAs adopt the Pareto-based approach,that is,the qualities of the candidate solutions are compared using Pareto dominance.Among various dominance comparison mechanisms,nondominated sorting[2]has been shown to be very effective forfinding Pareto-optimal solutions.Also much work has been done to efficiently store nondominated solutions found during search in an archive[6],[7].Nondominated sorting is a procedure where solutions in the population are assigned to different fronts based on their dominance relationships.Without loss of generality,we assume that the individuals in population P can be categorized into K Pareto fronts,denoted as F i,i=1,...,K.According to nondom-inated sorting,all nondominated solutions in population P are assigned to front F1;then the nondominated solutions in P−F1,which is the set of solutions by removing the solutions assigned to front F1,are assigned to front F2.This procedure repeats until all solutions in P are assigned to a front F i,i=1,...,K.Note that the solutions belonging to front F j are dominated by at least one solution belonging to front F i,if i<j,i,j=1,2,...,K.Fig.1provides an illustrative example of a population of13solutions composed of four fronts,where both objectives are to be minimized. Nondominated sorting is computationally intensive,in par-ticular,when the population size increases.To address this problem,much research work has been dedicated to the improvement of the computational efficiency of this proce-dure.The idea of nondominated sorting wasfirst suggested in[8]as a selection strategy for evolutionary multiobjective optimization,which was implemented in a multiobjective GA, termed NSGA[9].Nondominated sorting in NSGA has a time complexity of O(MN3)and a space complexity of O(N), where M is the number of objectives and N is the number of solutions in the population.A faster version of nondomi-nated sorting,termed fast nondominated sort,was proposed in[2],where the time complexity is reduced to O(MN2). The fast nondominated sort,however,requires a larger storage1089-778X c 2014IEEE.Personal use is permitted,but republication/redistribution requires IEEE permission.See /publications_standards/publications/rights/index.html for more information.Fig.1.Population with 13solutions of a biobjective minimization problem.The individuals can be divided into four fronts.space than the nondominated sorting in NSGA,which is increased to O (N 2).Jensen [10]adopted a divide-and-conquer strategy for nondominated sorting,the time complexity of which is O (N log M −1N ).Tang et al.[11]proposed a novel nondominated sorting approach based on arena’s principle,that is,each winner will be the next arena host to be chal-lenged.This approach has been proved to have the same time complexity as the fast nondominated sort,while empirical results show that it outperforms the fast nondominated sort in terms of computational efficiency,since it can achieve a time complexity O (MN √N )in some best cases.Clymont and Keedwell [12]proposed two improved approaches to nondominated sorting,called climbing sort and deductive sort,where some dominance relationships between solutions can be inferred based on recorded comparison results.In this paper,we propose a new,computationally efficient approach to nondominated sorting,called efficient nondomi-nated sort (ENS).ENS adopts an idea different from those used in the above-mentioned methods.The main difference lies in the fact that existing nondominated sorting approaches usually compare a solution with all other solutions in the population before assigning it to a front,while ENS compares it only with those that have already been assigned to a front.This is made possible by the fact that in ENS,the population is sorted in one objective before the ENS is applied.Thus,a solution added to the fronts cannot dominate any solutions that are added before.As a result,ENS can avoid a large num-ber of redundant dominance comparisons,which significantly improves the computational efficiency.Theoretical analysis shows that the ENS approach has a space complexity of O (1),which is smaller than all existing nondominated sorting methods.Meanwhile,the time complexity of ENS will be O (MN log N )in good cases,which is much lower than that of all existing algorithms.Even in the worst case,ENS has a complexity of O (MN 2),which is the same as the fast nondominated sort.Experimental results confirm that ENS has better computational efficiency than the state of the art.The remaining of this paper is organized as follows.In Section II ,we briefly review a few widely used nondom-inated sorting approaches and analyze their computational complexity.In Section III ,we propose a new approach to non-dominated sorting,ENS,based on which two nondominatedsorting algorithms are developed.The computational com-plexities of the two algorithms are then analyzed.Simulation results are presented in Section IV to empirically compare the two ENS-based nondominated sorting algorithms with three state-of-the-art methods.Finally,conclusions and remarks are given in Section V .II.R ELATED W ORKIn this section,we review a few popular nondominated sorting approaches together with an analysis of their compu-tational complexities.A.Nondominated Sorting MethodsSince Goldberg [8]suggested the use of nondominated sorting for selection in MOEAs,a number of nondominated sorting methods have been reported in the literature over the past years.Furthermore,we review a few nondominated sorting approaches widely used in MOEAs.The nondominated sorting strategy was first adopted for selecting parents from offspring in NSGA for multiobjective optimization [9].The nondominated sorting in NSGA is carried out as follows.Each solution is compared with all other solutions in the population,and solutions that are not dominated by any other solutions are assigned to front F 1.All solutions assigned to F 1are temporarily removed from the population.Then each solution in the remaining population is compared with others and all nondominated solutions are assigned to front F 2.This operation is repeated until all solutions have been assigned to a front.This approach contains many redundant comparisons in the sense that the comparison between two solutions may be performed more than once.The time complexity of this approach is O (MN 3),which makes NSGA highly time-consuming and computationally inefficient for large populations.As an improved version of NSGA,Deb et al.[2]proposed a computationally more efficient non-dominated sorting approach,called fast nondominated sort,where the comparison between any two solutions is performed only once.Fast nondominated sort has a time complexity of O (MN 2),albeit at the cost of an increased space complexity from O (N )to O (N 2).A recursive nondominated sorting approach [10],usu-ally called Jensen’s sort,was suggested based on the divide-and-conquer mechanism,which reduces the time com-plexity to O (MN log N )for biobjective MOPs and to O (N log M −1N )for MOPs having more than two objectives.The space complexity of this approach is O (1)and O (N )for MOPs with two objectives and more than two objectives,respectively.Just as shown in O (N log M −1N ),the time com-plexity of Jensen’s sort will grow exponentially with the incre-ment of number of objectives.This means that Jensen’s sorting method will not work efficiently for MOPs with a large number of objectives.Actually,this sorting method will likely consume more runtime in simulation due to its recursive nature.In addition,as Clymont and Keedwell [12]and Fang et al.[13]pointed out,Jensen’s sorting algorithm is not applicable in many cases,for instance,when strong-dominance [14]orZHANG et al.:EFFICIENT APPROACH TO NONDOMINATED SORTING FOR EVOLUTIONARY MULTIOBJECTIVE OPTIMIZATION203Fig.2.Illustration of the commonly used strategy for nondominated sorting in most existing nondominated sorting approaches,which determines the front number of all solutions on the same front all at once,and solutions on different fronts sequentially.-dominance[15]is used in comparison or when the popu-lation contains duplicate solutions.Tang et al.[11]used arena’s principle to assign solutions to a front,which has been shown to have a better computational efficiency than the fast nondominated sort and Jensen’s sort in empirical evaluations.This approach randomly selects one solution from the population,regarded as an arena host,and all the remaining solutions in the population are compared with the arena host.The solution that dominates the arena host becomes the new arena host to replace the current one. The time complexity and space complexity of this approach are O(MN2)and O(N),respectively.Clymont and Keedwell[12]proposed two nondominated sorting approaches:climbing sort and deductive sort.As shown in[12],deductive sort often performs better than climbing sort.Deductive sort infers the dominance relationship between solutions by recording the results of comparisons,thereby avoiding some unnecessary comparisons.Deductive sort holds a time complexity of O(MN2)and a space complexity of O(N),which outperforms other approaches,for instance,the fast nondominated sort.There are a few other nondominated sorting approaches inspired by different ideas,such as the nondominated rank sort of the omni-optimizer[16],better nondominated sort[17], immune recognition-based algorithm[18],quick sort[19], sorting-based algorithm[20],and divide-and-conquer-based nondominated sorting algorithm[13].Most of these ap-proaches are effective in dealing with MOPs that have a small number of objectives,however,their efficiency often seriously degrades as the number of objectives increases.B.Analysis of Existing MethodsAlthough existing nondominated sorting approaches per-form front assignments based on various ideas,most of them can be described in a generic framework as shown in Fig.2. In this framework,solutions in different fronts are assigned front by front.For example,for a population P containing K fronts F i,1≤i≤K,all nondominated solutions in P are first assigned to front F1.Once this is done,the nondominated solutions in P−F1(the remaining population with all solutions assigned to F1being removed)can then be assigned to F2. In other words,solutions belonging to front F i+1cannot be assigned until all solutions belonging to F i have been assigned. Dominance comparisons between the solutions are the main operation in nondominated sorting,that is,thenum-Fig.3.Categorization of dominance comparison results between two solu-tions in nondominated sorting.ber of needed comparisons determines the efficiency of a nondominated sorting approach.Most existing nondominated sorting methods focus on the reduction of the number of comparisons to improve their computational efficiencies.The reason is that some dominance comparisons between solutions are unnecessary and can be spared.Taking a closer look, wefind that the result of one dominance comparison can be categorized into the following four cases,assuming that solution p m is compared with solution p n.1)Case1:p m is dominated by p n,or p n is dominatedby p m.2)Case2:p m and p n are nondominated,and they belong tothe same front F i,where F i is the current front(i.e.,the front that the solutions are being assigned to).3)Case3:p m and p n are nondominated,and they belongto the same front F i,where F i is not the current front.4)Case4:p m and p n are nondominated,but they belongto different fronts.Recall that a solution is assigned to the current front if it is not dominated by any other solutions in the current population. In Case1,if solution p m dominates solution p n,then p n does not belong to the current front and we no longer need to perform additional comparisons between p n and all other solutions in the current population,which means that such comparisons,if performed,are redundant.In Case2,both p m and p n belong to the current front,so there does not exist any solution that can dominate p m or p n,and the comparison between p m and p n should be done to verify whether one dominates the other.In fact,all solutions belonging to the current front should be compared with each other to ensure that they are all nondominated with each other.In Case3, neither p m nor p n belongs to the current front,which means that there exists at least one solution dominating p m and a solution dominating p n,and the comparison between p m and p n can be skipped.In Case4,since there exists at least one solution dominating p m or p n,a comparison between p m and p n is unnecessary.The four cases of possible comparisons are illustrated in Fig.3.As shown in Fig.3,for nondominated sorting,comparisons in Case1and comparisons in Case2cannot be avoided,which are termed necessary comparisons.The necessary comparisons204IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION,VOL.19,NO.2,APRIL2015Fig.4.Population containing six solutions for a biobjective minimization problem.in Case1refer to the comparisons between solutions in differ-ent fronts,while the necessary comparisons in Case2refer to comparisons between solutions in the same front.The number of necessary comparisons in Cases1and2is the theoretical minimum number of needed dominance comparisons for any nondominated sorting algorithm.If a nondominated sorting approach determines the front of a solution by starting to check from thefirst front to the last one,for example,from F1to F4in Fig.1,the number of necessary comparisons in Case1 can be calculated in the following way.Given a population consisting of N solutions that can be divided into K fronts, assume front F i contains N i solutions,where1≤i≤K.So, we have N1+N2+...+N K=N.If a solution p n belongs to front F i,then at least one solution dominating p n will be found in each of the preceding i−1fronts.This means that i−1comparisons in Case1are needed for solution p n.Since there are N i solutions belonging to F i,a total of(i−1)N i comparisons in Case1are needed for front F i.Therefore,the total number of necessary dominance comparisons between solutions in different fronts isNum_Comp1=Ki=1(i−1)N i.(1)Regarding the minimum number of needed comparisons in Case2,since each of the N i solutions in front F i should be compared with the other solutions in front F i,which needs a total of N i(N i−1)/2comparisons,the total number of comparisons between solutions in the same front isNum_Comp2=Ki=1N i(N i−1)2.(2)Unfortunately,most popular nondominated sorting approaches have a total number of dominance comparisons much higher than the sum of Num_Comp1and Num_Comp2.From the above discussions,we canfind that further improvements of the computational efficiencies of nondominated sorting approaches should concentrate on the reduction of unnecessary comparisons in Cases1, 3,and 4.In fact,most existing improved nondominated sorting approaches have successfully removed unnecessary dominance comparisons in Case1,however,still perform many unnecessary comparisons belonging to Cases3and4. As an example,we consider the comparisons performed in deductive sort over a population shown in Fig.4.Thispopula-Fig.5.Illustration of the proposed dominance comparison strategy,where solutions in the population can be assigned to the fronts one by one.tion contains six candidate solutions of a biobjective minimiza-tion problem,each being denoted by p i(f1,f2),i=1,2, (6)where f1and f2are the values of the two objectives of solution p i,respectively.In this example,the six candidate solutions are p1(5,4),p2(6,3),p3(7,2),p4(1,6),p5(2,5),and p6(3,1). As shown in Fig.4,there are two fronts in this population, where solutions p4,p5,and p6belong to thefirst front F1,and solutions p1,p2,and p3belong to F2.Deductive sort performs the following comparisons to assign the solutions to one of the two fronts.It begins with comparing solution p1with all other solutions in the population one by one.Solution p1isfirst compared with solution p2.Since p1is not dominated by p2, deductive sort continues to perform the comparison between p1and p3.The comparison result indicates that solution p1 is not dominated by p3either.Similarly,p1will be further compared with p4and p5,and it is concluded that p1is dominated neither by p4nor by p5.Then,p1is compared with p6,and it will be found that p1is dominated by p6, which means that p1does not belong to the current front F1. By then,solution p1will no longer be involved in further comparisons of front F1in deductive sort.In the above procedure,the following dominance compar-isons have been made:two comparisons of Case3(compar-isons between p1and p2,or p3),two comparisons of Case4 (comparisons between p1and p4,or p5),and one comparison of Case1(a comparison between p1and p6).After p1is compared with all the other solutions,deductive sort starts to consider solution p2.Dominance comparison will continue until all solutions are assigned to a front.Table I lists all comparisons performed by deductive sort for the population shown in Fig.4.From Table I,it is not difficult to see that there exist many unnecessary comparisons performed by deductive sort, which belong to Cases3and4.Among these unnecessary comparisons,several of them are duplicate comparisons,such as the comparisons between p1and p2.In fact,all duplicate comparisons in deductive sort belong to Case3.In this paper, we propose a new nondominated sorting algorithm using a strategy different from the one illustrated in Fig.2,which aims to avoid duplicate comparisons,thereby considerably reducing the number of unnecessary comparisons.III.E FFICIENT N ONDOMINATED S ORT F RAMEWORK Here,we present a new efficient nondominated sorting strategy,termed ENS,which is conceptually different fromZHANG et al.:EFFICIENT APPROACH TO NONDOMINATED SORTING FOR EVOLUTIONARY MULTIOBJECTIVE OPTIMIZATION 205TABLE IC OMPARISONS P ERFORMED BY D EDUCTIVE S ORT FORTHEP OPULATION S HOWN IN F IG .4Algorithm 1Main Steps of ENS for Nondominated SortingInput:population POutput:the set of fronts F 1:F =empty ;2:Sort P in an ascending order of the first objective value;3:for all P [n ]∈sorted P do 4:Assign solution P [n ]into F by Algorithm 2orAlgorithm 3;5:end for 6:return F ;most existing nondominated sorting methods.The main idea of the ENS approach is shown in Fig.5.By comparing Figs.2and 5,we can see that the ENS approach determines the front each solution belongs to one by one,while most existing nondominated sorting approaches determine the front of all solutions on the same front as a whole.The main merit of determining the front to which each solution belongs separately is that it can avoid duplicate comparisons,since in this approach,a solution to be assigned only needs to be compared with solutions that have already been assigned to a front.The details of ENS are given in Algorithm 1.For a minimization problem,this approach first sorts the N solutions in population P in an ascending order according to the first objective value,where N is the population size.If the first objective values of two solutions are the same,then they are sorted according to the second objective value.This procedure continues until all individuals in the population are sorted.If solutions have the same value in all objectives,their order can be arbitrary.For this sorted population P ,a solution p m will never be dominated by a solution p n ,if m <n ,since there exists at least one objective in p m whose value is smaller than that of the same objective in p n .This means that there existAlgorithm 2Sequential Search Strategy for Finding the Front of a SolutionInput:solution P [n ],the set of fronts F Output:the front number of solution P [n ]1:x =size(F );{the number of fronts having been found}2:k =1;{the front now checked}3:while true do4:compare P [n ]with the solutions in F [k ]starting fromthe last one and ending with the first one;5:if F [k ]contains no solution dominating P [n ]then 6:return k ;{move P [n ]to F [k ]}7:break ;8:else 9:k ++;10:if k >x then 11:return x +1;{move P [n ]to a new front}12:break;13:end if 14:end if 15:end whileonly two possible relationships between the two solutions:either p m dominates p n ,or p m and p n are not comparable.After finishing sorting the individuals in population P ,ENS begins to assign solutions to fronts in the sorted population P one by one,starting from the first solution p 1and ending with the last one p N .As we know,if a solution is assigned to a front,it is dominated by at least one solution in the preceding front.As pointed out earlier,a solution can never be dominated by any succeeding solution in the sorted population P .Therefore,it is sufficient to compare a solution with those that have already been assigned to a front to determine the front of this solution.The possible relationships between a solution to be assigned and those that have been assigned to a front are shown in Fig.6.Actually,if a solution p n is assigned to front F i ,F i must satisfy the following two conditions.1)There exists at least one solution in each front F j that has been assigned and dominates p n ,for 1≤j ≤i −1.2)There exists no solution in any of the assigned fronts F kthat dominates p n ,for k ≥i .In this way,the front to which a solution belongs can be determined by finding out the front that satisfies the above two conditions.In what follows,we present two strategies for searching for the front satisfying the above two conditions within the ENS framework,one using a sequential search strategy (termed ENS-SS)and the other using a binary search strategy (ENS-BS).A.Sequential Search StrategyThe pseudocode of the sequential search is presented in Algorithm 2.The idea in this search strategy is quite straight-forward.For solution p n ,the algorithm checks at first whether there exists a solution that has been assigned to the first front F 1and dominates p n .If such a solution does not exist,assign p n to front F 1.If p n is dominated by any solution in F 1,start comparing p n with the solutions assigned to F 2.If no solution in front F 2dominates p n ,assign p n to front F 2.If p n is not206IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION,VOL.19,NO.2,APRIL2015Fig.6.Relationships between p n and the solutions having been assigned to a front.assigned to any of the existing fronts,create a new front and assign p n to this new front.There is a little trick in checking whether a front has a solution dominating p n.Recall that solutions assigned to an existing front are also sorted in the same order as the population.Therefore,the comparisons between p n and the solutions assigned to the front should start with the last one in the front and end with thefirst one.This trick often leads to fewer comparisons if a solution assigned to this front dominates p n,since solutions in the end of the sorted front are more likely to dominate p n.As a result,unnecessary comparisons can be avoided.For biobjective optimization problems,the idea presented here is computationally more efficient than existing nondom-inated sorting methods.In fact,as shown in Algorithm2, only one comparison is sufficient for determining whether a solution to be assigned belongs to an existing front.The reason is as follows.In the sorting,solutions assigned to a front are sorted in the ascending order of thefirst objective,which means that the second objectives of these solutions are in the descending order,since these solutions are nondominated with each other.This means that,if p n,a solution to be assigned,is dominated by a solution in an existing front,it should be dominated by the last solution in the front,since the last solution has the smallest value in the second objective among all solutions in the front.Therefore,for biobjective optimization problems,this method can determine the front number of a solution by performing only the comparisons between this solution and the last solution in each front.In Table II,we list the comparisons performed by ENS using the sequential search strategy(ENS-SS)for nondominated sorting of the population given in Fig.4.As shown in the table,ENS-SS needs only nine comparisons in total,which is much smaller than the number of comparisons needed by deductive sort,refer to Table I.It is not difficult for the reader to check that there does not exist any comparison belonging to Case3in ENS-SS.This means that ENS-SS does not perform any duplicate comparisons,which can be attributed to the ENS strategy shown in Fig.5.It should be noted that although ENS-SS does not perform any comparison belonging to Case4 for the population shown in Fig.4,such comparisons may occur for other populations.Algorithm3Binary Search Strategy for Finding the Front of a SolutionInput:solution P[n],the set of fronts FOutput:the front number of solution P[n]1:x=size(F);{the number of fronts having been found} 2:kmin=0;{the lower bound for checking}3:kmax=x;{the upper bound for checking}4:k= (kmax+kmin)/2+1/2 ;{the front now checked} 5:while true do6:Compare P[n]with the solutions in F[k]starting from the last one and ending with thefirst one;7:if F[k]has no solution dominating P[n]then8:if k==kmin+1then9:return k;{move P[n]to F[k]}10:break;11:else12:kmax=k;13:k= (kmax+kmin)/2+1/2 ;14:end if15:else16:kmin=k;17:if(kmax==kmin+1)and(kmax<x)then18:return kmax;{move P[n]to F[kmax]}19:break;20:else if kmin==x then21:return x+1;{move P[n]to a new front}22:break;23:else24:k= (kmax+kmin)/2+1/2 ;25:end if26:end if27:end whileB.Binary Search StrategyThe pseudocode of the binary search strategy is presented in Algorithm3.Different from sequential search,the binary search strategy starts with checking the intermediate front F L/2 instead of thefirst front F1,where L is the number of fronts that have been created thus far,that is,before solution p n is assigned.If solution p n is not dominated by any solution in front F L/2 ,then solution p n will be compared with the solutions in front F L/4 .Otherwise,p n is compared with the solutions in front F 3L/4 .In this way,the binary search can determine the front to which solution p n belongs after check-ing log(L+1) fronts.If the last existing front F L has been checked and p n does not belong to this front,a new front F L+1 will be created and solution p n is assigned to this new front. The binary search strategy adopted here usually outperforms the sequential search strategy in which it requires to check fewer fronts in a population.But this does not mean that the binary search strategy can always perform fewer comparisons than the sequential search strategy in front assignment.In binary search,it can happen that more than one front that does not have any solution dominating p n needs to be checked. All solutions in these checked fronts have to be compared with p n.In sequential search,at most one front containing no solution dominating p n needs to be checked.Therefore,the。
遗传算法在模拟电路故障诊断中的应用---优秀毕业论文参考文献可复制黏贴
In order to improve the speed of fault diagnosis, the application in soft fault diagnosis of analog circuits based on sensitivity analysis combined with the genetic algorithm is presented in this paper. We have discussed the sensitivity analysis of analog circuits. Estimate the offset of the component parameters to diagnose the fault of the analog circuits. We convert the diagnosis equation, which is constituted by the incremental test node voltage and the component parameters variation, into the linear programming problem about finding the smallest independent variable based on the hard constraints of the fault diagnosis equation. And the linear programming problem with constraints is converted to the extreme solution without constraints by the penalty function. The genetic algorithm is used to solve the optimal solution. Then, the influence of the control parameters of genetic algorithm is discussed with examples. A new Self-adaptive Genetic Algorithms was proposed and the experiments show that the method has a good efficiency on the soft fault diagnosis of tolerance analog circuits and has a higher speed.
《孟德尔随机化研究指南》中英文版
《孟德尔随机化研究指南》中英文版全文共3篇示例,供读者参考篇1Randomized research is a vital component of scientific studies, allowing researchers to investigate causal relationships between variables and make accurate inferences about the effects of interventions. One of the most renowned guides for conducting randomized research is the "Mendel Randomization Research Guide," which provides detailed instructions and best practices for designing and implementing randomized controlled trials.The Mendel Randomization Research Guide offers comprehensive guidance on all aspects of randomized research, from study design and sample selection to data analysis and interpretation of results. It emphasizes the importance of randomization in reducing bias and confounding effects, thus ensuring the validity and reliability of study findings. With clear and practical recommendations, researchers can feel confident in the quality and rigor of their randomized research studies.The guide highlights the key principles of randomization, such as the use of random assignment to treatment groups, blinding of participants and researchers, and intent-to-treat analysis. It also discusses strategies for achieving balance in sample characteristics and minimizing the risk of selection bias. By following these principles and guidelines, researchers can maximize the internal validity of their studies and draw accurate conclusions about the causal effects of interventions.In addition to the technical aspects of randomized research, the Mendel Randomization Research Guide also addresses ethical considerations and practical challenges that researchers may face. It emphasizes the importance of obtaining informed consent from participants, protecting their privacy and confidentiality, and ensuring the safety and well-being of study subjects. The guide also discusses strategies for overcoming common obstacles in randomized research, such as recruitment and retention issues, data collection problems, and statistical challenges.Overall, the Mendel Randomization Research Guide is a valuable resource for researchers looking to improve the quality and validity of their randomized research studies. By following its recommendations and best practices, researchers can conductstudies that produce reliable and actionable findings, advancing scientific knowledge and contributing to evidence-based decision making in various fields.篇2Mendel Randomization Study GuideIntroductionMendel Randomization Study Guide is a comprehensive and informative resource for researchers and students interested in the field of Mendel randomization. This guide provides anin-depth overview of the principles and methods of Mendel randomization, as well as practical advice on how to design and conduct Mendel randomization studies.The guide is divided into several sections, each covering a different aspect of Mendel randomization. The first section provides a brief introduction to the history and background of Mendel randomization, tracing its origins to the work of Gregor Mendel, the father of modern genetics. It also discusses the theoretical foundations of Mendel randomization and its potential applications in causal inference.The second section of the guide focuses on the methods and techniques used in Mendel randomization studies. This includesa detailed explanation of how Mendel randomization works, as well as guidelines on how to select instrumental variables and control for potential confounders. It also discusses the strengths and limitations of Mendel randomization, and provides practical tips on how to deal with common challenges in Mendel randomization studies.The third section of the guide is dedicated to practical considerations in Mendel randomization studies. This includes advice on how to design a Mendel randomization study, collect and analyze data, and interpret the results. It also provides recommendations on how to report Mendel randomization studies and publish research findings in scientific journals.In addition, the guide includes a glossary of key terms and concepts related to Mendel randomization, as well as a list of recommended readings for further study. It also includes case studies and examples of Mendel randomization studies in practice, to illustrate the principles and techniques discussed in the guide.ConclusionIn conclusion, the Mendel Randomization Study Guide is a valuable resource for researchers and students interested in Mendel randomization. It provides a comprehensive overview ofthe principles and methods of Mendel randomization, as well as practical advice on how to design and conduct Mendel randomization studies. Whether you are new to Mendel randomization or looking to deepen your understanding of the field, this guide is an essential reference for anyone interested in causal inference and genetic epidemiology.篇3"Guide to Mendelian Randomization Studies" English VersionIntroductionMendelian randomization (MR) is a method that uses genetic variants to investigate the causal relationship between an exposure and an outcome. It is a powerful tool that can help researchers to better understand the underlying mechanisms of complex traits and diseases. The "Guide to Mendelian Randomization Studies" provides a comprehensive overview of MR studies and offers practical guidance on how to design and carry out these studies effectively.Chapter 1: Introduction to Mendelian RandomizationThis chapter provides an overview of the principles of Mendelian randomization, including the assumptions andlimitations of the method. It explains how genetic variants can be used as instrumental variables to estimate the causal effect of an exposure on an outcome, and outlines the key steps involved in conducting an MR study.Chapter 2: Choosing Genetic InstrumentsIn this chapter, the guide discusses the criteria for selecting appropriate genetic instruments for Mendelian randomization. It covers issues such as the relevance of the genetic variant to the exposure of interest, the strength of the instrument, and the potential for pleiotropy. The chapter also provides practical tips on how to search for suitable genetic variants in public databases.Chapter 3: Data Sources and ValidationThis chapter highlights the importance of using high-quality data sources for Mendelian randomization studies. It discusses the different types of data that can be used, such asgenome-wide association studies and biobanks, and offers advice on how to validate genetic instruments and ensure the reliability of the data.Chapter 4: Statistical MethodsIn this chapter, the guide explains the various statistical methods that can be used to analyze Mendelian randomization data. It covers techniques such as inverse variance weighting, MR-Egger regression, and bi-directional Mendelian randomization, and provides guidance on how to choose the most appropriate method for a given study.Chapter 5: Interpretation and ReportingThe final chapter of the guide focuses on the interpretation and reporting of Mendelian randomization results. It discusses how to assess the strength of causal inference, consider potential biases, and communicate findings effectively in research papers and presentations.ConclusionThe "Guide to Mendelian Randomization Studies" is a valuable resource for researchers who are interested in using genetic data to investigate causal relationships in epidemiological studies. By following the guidance provided in the guide, researchers can enhance the rigor and validity of their Mendelian randomization studies and contribute to a better understanding of the determinants of complex traits and diseases.。
Aspects of Gravitational Clustering
ASPECTS OF GRAVITATIONAL CLUSTERING
3
ˆk is a linear second order differmode, labeled by a wave vector k. Here L ential operator in time. Solving this set of ordinary differential equations, with given initial conditions, we can determine the evolution of each mode separately. [Similar procedure, of course, works for the case with Ω = 1. In this case, the mode functions will be more complicated than the plane waves; but, with a suitable choice of orthonormal functions, we can obtain a similar set of equations]. This solves the problem of linear gravitational clustering completely. There is, however, one major conceptual difficulty in interpreting the results of this program. In general relativity, the form (and numerical value) of the metric coefficients gαβ (or the stress-tensor components Tαβ ) can be changed by a relabeling of coordinates xα → xα′ . By such a trivial change we can make a small δTαβ large or even generate a component which was originally absent. Thus the perturbations may grow at different rates − or even decay − when we relabel coordinates. It is necessary to tackle this ambiguity before we can meaningfully talk about the growth of inhomogeneities. There are two different approaches to handling such difficulties in general relativity. The first method is to resolve the problem by force: We may choose a particular coordinate system and compute everything in that coordinate system. If the coordinate system is physically well motivated, then the quantities computed in that system can be interpreted easily; 0 to be the perturbed mass (energy) density for example, we will treat δT0 even though it is coordinate dependent. The difficulty with this method is that one cannot fix the gauge completely by simple physical arguments; the residual gauge ambiguities do create some problems. The second approach is to construct quantities − linear combinations of various perturbed physical variables − which are scalars under coordinate transformations. [see eg. the contribution by Brandenberger to this volume and references cited therein] Einstein’s equations are then rewritten as equations for these gauge invariant quantities. This approach, of course, is manifestly gauge invariant from start to finish. However, it is more complicated than the first one; besides, the gauge invariant objects do not, in general, possess any simple physical interpretation. In these lectures, we shall be mainly concerned with the first approach. Since the gauge ambiguity is a purely general relativistic effect, it is necessary to determine when such effects are significant. The effects due to the curvature of space-time will be important at length scales bigger than (or comparable to) the Hubble radius, defined as dH (t) ≡ (a/a ˙ )−1 . Writing
孟德尔随机化三大假设英文 解释
孟德尔随机化三大假设英文解释English: Mendel's three major assumptions in randomization are as follows:1. The Principle of Segregation: This assumption is based on the observation that an individual organism possesses two alleles for each gene and these alleles segregate randomly during the formation of gametes. According to Mendel, during gamete formation, the two alleles of a gene segregate from each other so that each gamete contains only one allele. This principle explains why traits can disappear in one generation and reappear in the next, as the alleles can combine in different ways in the offspring.2. The Principle of Independent Assortment: This assumption states that the segregation of alleles for one gene does not influence the segregation of alleles for other genes. According to Mendel, during gamete formation, the alleles for different genes segregate independently of each other, resulting in a random mixing of traits. This principle explains the inheritance patterns of traits that arecontrolled by multiple genes, as each gene independently contributes to the formation of traits.3. The Principle of Random Fertilization: This assumption suggests that the fusion of gametes during fertilization is a random process. According to Mendel, any sperm can fertilize any egg, regardless of the alleles they carry. This principle adds to the randomness of trait inheritance and increases the genetic variability within a population.These assumptions formed the basis of Mendel's laws of inheritance, which laid the foundation for understanding the principles of genetics. They highlighted the importance of randomness in genetic processes, emphasizing that the segregation, assortment, and fertilization of alleles are random events. Mendel's experiments with pea plants provided empirical evidence for these assumptions, enabling him to propose the laws of inheritance.中文翻译: 孟德尔在随机化方面的三个主要假设如下:1. 分离定律: 这个假设基于这样一个观察结果,即个体生物每个基因都拥有两个等位基因,这些等位基因在配子形成过程中随机分离。
一种新的部分神经进化网络的股票预测(英文)
一种新的部分神经进化网络的股票预测(英文)一种新的部分神经进化网络的股票预测自从股票市场的出现以来,人们一直在寻求能够提前预测股票走势的方法。
许多投资者和研究人员尝试使用各种技术分析工具和模型来预测股票未来的走势,但是股票市场的复杂性和难以预测性使得这变得困难重重。
因此,寻找一种能够准确预测股票走势的方法一直是金融界的热点问题。
近年来,人工智能技术在金融领域的应用日益增多。
其中,神经网络是一种被广泛使用的工具,它可以自动学习和识别模式,并根据所学的模式进行预测。
然而,传统神经网络在预测股票市场方面存在诸多问题,例如过拟合和难以处理大量数据等。
为了克服这些问题,本文提出了一种新的部分神经进化网络(Partial Neural Evolving Network, PNEN)模型来预测股票走势。
PNEN模型将神经网络和进化算法相结合,通过优化和训练来实现更准确的预测结果。
PNEN模型的核心思想是将神经网络的隐藏层拆分为多个小模块,每个小模块只负责处理一部分输入数据。
通过这种方式,模型可以更好地适应不同的市场情况和模式。
同时,采用进化算法来优化模型的参数,可以进一步提高模型的预测性能。
具体而言,PNEN模型包括以下几个步骤:1. 数据准备:从股票市场获取历史交易数据,并对数据进行预处理和归一化处理,以便更好地输入到模型中。
2. 构建模型结构:将神经网络的隐藏层拆分为多个小模块,通过进化算法来确定每个小模块的结构和参数。
进化算法通过优化模型的准确性和稳定性,以获得更好的预测结果。
3. 训练模型:使用历史数据集对模型进行训练,并通过反向传播算法来更新模型的权重和偏置。
同时,通过与进化算法的交互,不断调整模型结构和参数。
4. 预测结果:使用训练好的模型对未来的股票走势进行预测。
通过模型对市场的分析和判断,可以为投资者提供决策参考。
为了验证PNEN模型的效果,我们在实际的股票市场数据上进行了实验。
结果表明,与传统神经网络模型相比,PNEN 模型在预测股票走势方面具有更好的准确性和稳定性。
专八英语阅读
英语专业八级考试TEM-8阅读理解练习册(1)(英语专业2012级)UNIT 1Text AEvery minute of every day, what ecologist生态学家James Carlton calls a global ―conveyor belt‖, redistributes ocean organisms生物.It’s planetwide biological disruption生物的破坏that scientists have barely begun to understand.Dr. Carlton —an oceanographer at Williams College in Williamstown,Mass.—explains that, at any given moment, ―There are several thousand marine species traveling… in the ballast water of ships.‖ These creatures move from coastal waters where they fit into the local web of life to places where some of them could tear that web apart. This is the larger dimension of the infamous无耻的,邪恶的invasion of fish-destroying, pipe-clogging zebra mussels有斑马纹的贻贝.Such voracious贪婪的invaders at least make their presence known. What concerns Carlton and his fellow marine ecologists is the lack of knowledge about the hundreds of alien invaders that quietly enter coastal waters around the world every day. Many of them probably just die out. Some benignly亲切地,仁慈地—or even beneficially — join the local scene. But some will make trouble.In one sense, this is an old story. Organisms have ridden ships for centuries. They have clung to hulls and come along with cargo. What’s new is the scale and speed of the migrations made possible by the massive volume of ship-ballast water压载水— taken in to provide ship stability—continuously moving around the world…Ships load up with ballast water and its inhabitants in coastal waters of one port and dump the ballast in another port that may be thousands of kilometers away. A single load can run to hundreds of gallons. Some larger ships take on as much as 40 million gallons. The creatures that come along tend to be in their larva free-floating stage. When discharged排出in alien waters they can mature into crabs, jellyfish水母, slugs鼻涕虫,蛞蝓, and many other forms.Since the problem involves coastal species, simply banning ballast dumps in coastal waters would, in theory, solve it. Coastal organisms in ballast water that is flushed into midocean would not survive. Such a ban has worked for North American Inland Waterway. But it would be hard to enforce it worldwide. Heating ballast water or straining it should also halt the species spread. But before any such worldwide regulations were imposed, scientists would need a clearer view of what is going on.The continuous shuffling洗牌of marine organisms has changed the biology of the sea on a global scale. It can have devastating effects as in the case of the American comb jellyfish that recently invaded the Black Sea. It has destroyed that sea’s anchovy鳀鱼fishery by eating anchovy eggs. It may soon spread to western and northern European waters.The maritime nations that created the biological ―conveyor belt‖ should support a coordinated international effort to find out what is going on and what should be done about it. (456 words)1.According to Dr. Carlton, ocean organism‟s are_______.A.being moved to new environmentsB.destroying the planetC.succumbing to the zebra musselD.developing alien characteristics2.Oceanographers海洋学家are concerned because_________.A.their knowledge of this phenomenon is limitedB.they believe the oceans are dyingC.they fear an invasion from outer-spaceD.they have identified thousands of alien webs3.According to marine ecologists, transplanted marinespecies____________.A.may upset the ecosystems of coastal watersB.are all compatible with one anotherC.can only survive in their home watersD.sometimes disrupt shipping lanes4.The identified cause of the problem is_______.A.the rapidity with which larvae matureB. a common practice of the shipping industryC. a centuries old speciesD.the world wide movement of ocean currents5.The article suggests that a solution to the problem__________.A.is unlikely to be identifiedB.must precede further researchC.is hypothetically假设地,假想地easyD.will limit global shippingText BNew …Endangered‟ List Targets Many US RiversIt is hard to think of a major natural resource or pollution issue in North America today that does not affect rivers.Farm chemical runoff残渣, industrial waste, urban storm sewers, sewage treatment, mining, logging, grazing放牧,military bases, residential and business development, hydropower水力发电,loss of wetlands. The list goes on.Legislation like the Clean Water Act and Wild and Scenic Rivers Act have provided some protection, but threats continue.The Environmental Protection Agency (EPA) reported yesterday that an assessment of 642,000 miles of rivers and streams showed 34 percent in less than good condition. In a major study of the Clean Water Act, the Natural Resources Defense Council last fall reported that poison runoff impairs损害more than 125,000 miles of rivers.More recently, the NRDC and Izaak Walton League warned that pollution and loss of wetlands—made worse by last year’s flooding—is degrading恶化the Mississippi River ecosystem.On Tuesday, the conservation group保护组织American Rivers issued its annual list of 10 ―endangered‖ and 20 ―threatened‖ rivers in 32 states, the District of Colombia, and Canada.At the top of the list is the Clarks Fork of the Yellowstone River, whereCanadian mining firms plan to build a 74-acre英亩reservoir水库,蓄水池as part of a gold mine less than three miles from Yellowstone National Park. The reservoir would hold the runoff from the sulfuric acid 硫酸used to extract gold from crushed rock.―In the event this tailings pond failed, the impact to th e greater Yellowstone ecosystem would be cataclysmic大变动的,灾难性的and the damage irreversible不可逆转的.‖ Sen. Max Baucus of Montana, chairman of the Environment and Public Works Committee, wrote to Noranda Minerals Inc., an owner of the ― New World Mine‖.Last fall, an EPA official expressed concern about the mine and its potential impact, especially the plastic-lined storage reservoir. ― I am unaware of any studies evaluating how a tailings pond尾矿池,残渣池could be maintained to ensure its structural integrity forev er,‖ said Stephen Hoffman, chief of the EPA’s Mining Waste Section. ―It is my opinion that underwater disposal of tailings at New World may present a potentially significant threat to human health and the environment.‖The results of an environmental-impact statement, now being drafted by the Forest Service and Montana Department of State Lands, could determine the mine’s future…In its recent proposal to reauthorize the Clean Water Act, the Clinton administration noted ―dramatically improved water quality since 1972,‖ when the act was passed. But it also reported that 30 percent of riverscontinue to be degraded, mainly by silt泥沙and nutrients from farm and urban runoff, combined sewer overflows, and municipal sewage城市污水. Bottom sediments沉积物are contaminated污染in more than 1,000 waterways, the administration reported in releasing its proposal in January. Between 60 and 80 percent of riparian corridors (riverbank lands) have been degraded.As with endangered species and their habitats in forests and deserts, the complexity of ecosystems is seen in rivers and the effects of development----beyond the obvious threats of industrial pollution, municipal waste, and in-stream diversions改道to slake消除the thirst of new communities in dry regions like the Southwes t…While there are many political hurdles障碍ahead, reauthorization of the Clean Water Act this year holds promise for US rivers. Rep. Norm Mineta of California, who chairs the House Committee overseeing the bill, calls it ―probably the most important env ironmental legislation this Congress will enact.‖ (553 words)6.According to the passage, the Clean Water Act______.A.has been ineffectiveB.will definitely be renewedC.has never been evaluatedD.was enacted some 30 years ago7.“Endangered” rivers are _________.A.catalogued annuallyB.less polluted than ―threatened rivers‖C.caused by floodingD.adjacent to large cities8.The “cataclysmic” event referred to in paragraph eight would be__________.A. fortuitous偶然的,意外的B. adventitious外加的,偶然的C. catastrophicD. precarious不稳定的,危险的9. The owners of the New World Mine appear to be______.A. ecologically aware of the impact of miningB. determined to construct a safe tailings pondC. indifferent to the concerns voiced by the EPAD. willing to relocate operations10. The passage conveys the impression that_______.A. Canadians are disinterested in natural resourcesB. private and public environmental groups aboundC. river banks are erodingD. the majority of US rivers are in poor conditionText CA classic series of experiments to determine the effects ofoverpopulation on communities of rats was reported in February of 1962 in an article in Scientific American. The experiments were conducted by a psychologist, John B. Calhoun and his associates. In each of these experiments, an equal number of male and female adult rats were placed in an enclosure and given an adequate supply of food, water, and other necessities. The rat populations were allowed to increase. Calhoun knew from experience approximately how many rats could live in the enclosures without experiencing stress due to overcrowding. He allowed the population to increase to approximately twice this number. Then he stabilized the population by removing offspring that were not dependent on their mothers. He and his associates then carefully observed and recorded behavior in these overpopulated communities. At the end of their experiments, Calhoun and his associates were able to conclude that overcrowding causes a breakdown in the normal social relationships among rats, a kind of social disease. The rats in the experiments did not follow the same patterns of behavior as rats would in a community without overcrowding.The females in the rat population were the most seriously affected by the high population density: They showed deviant异常的maternal behavior; they did not behave as mother rats normally do. In fact, many of the pups幼兽,幼崽, as rat babies are called, died as a result of poor maternal care. For example, mothers sometimes abandoned their pups,and, without their mothers' care, the pups died. Under normal conditions, a mother rat would not leave her pups alone to die. However, the experiments verified that in overpopulated communities, mother rats do not behave normally. Their behavior may be considered pathologically 病理上,病理学地diseased.The dominant males in the rat population were the least affected by overpopulation. Each of these strong males claimed an area of the enclosure as his own. Therefore, these individuals did not experience the overcrowding in the same way as the other rats did. The fact that the dominant males had adequate space in which to live may explain why they were not as seriously affected by overpopulation as the other rats. However, dominant males did behave pathologically at times. Their antisocial behavior consisted of attacks on weaker male,female, and immature rats. This deviant behavior showed that even though the dominant males had enough living space, they too were affected by the general overcrowding in the enclosure.Non-dominant males in the experimental rat communities also exhibited deviant social behavior. Some withdrew completely; they moved very little and ate and drank at times when the other rats were sleeping in order to avoid contact with them. Other non-dominant males were hyperactive; they were much more active than is normal, chasing other rats and fighting each other. This segment of the rat population, likeall the other parts, was affected by the overpopulation.The behavior of the non-dominant males and of the other components of the rat population has parallels in human behavior. People in densely populated areas exhibit deviant behavior similar to that of the rats in Calhoun's experiments. In large urban areas such as New York City, London, Mexican City, and Cairo, there are abandoned children. There are cruel, powerful individuals, both men and women. There are also people who withdraw and people who become hyperactive. The quantity of other forms of social pathology such as murder, rape, and robbery also frequently occur in densely populated human communities. Is the principal cause of these disorders overpopulation? Calhoun’s experiments suggest that it might be. In any case, social scientists and city planners have been influenced by the results of this series of experiments.11. Paragraph l is organized according to__________.A. reasonsB. descriptionC. examplesD. definition12.Calhoun stabilized the rat population_________.A. when it was double the number that could live in the enclosure without stressB. by removing young ratsC. at a constant number of adult rats in the enclosureD. all of the above are correct13.W hich of the following inferences CANNOT be made from theinformation inPara. 1?A. Calhoun's experiment is still considered important today.B. Overpopulation causes pathological behavior in rat populations.C. Stress does not occur in rat communities unless there is overcrowding.D. Calhoun had experimented with rats before.14. Which of the following behavior didn‟t happen in this experiment?A. All the male rats exhibited pathological behavior.B. Mother rats abandoned their pups.C. Female rats showed deviant maternal behavior.D. Mother rats left their rat babies alone.15. The main idea of the paragraph three is that __________.A. dominant males had adequate living spaceB. dominant males were not as seriously affected by overcrowding as the otherratsC. dominant males attacked weaker ratsD. the strongest males are always able to adapt to bad conditionsText DThe first mention of slavery in the statutes法令,法规of the English colonies of North America does not occur until after 1660—some forty years after the importation of the first Black people. Lest we think that existed in fact before it did in law, Oscar and Mary Handlin assure us, that the status of B lack people down to the 1660’s was that of servants. A critique批判of the Handlins’ interpretation of why legal slavery did not appear until the 1660’s suggests that assumptions about the relation between slavery and racial prejudice should be reexamined, and that explanation for the different treatment of Black slaves in North and South America should be expanded.The Handlins explain the appearance of legal slavery by arguing that, during the 1660’s, the position of white servants was improving relative to that of black servants. Thus, the Handlins contend, Black and White servants, heretofore treated alike, each attained a different status. There are, however, important objections to this argument. First, the Handlins cannot adequately demonstrate that t he White servant’s position was improving, during and after the 1660’s; several acts of the Maryland and Virginia legislatures indicate otherwise. Another flaw in the Handlins’ interpretation is their assumption that prior to the establishment of legal slavery there was no discrimination against Black people. It is true that before the 1660’s Black people were rarely called slaves. But this shouldnot overshadow evidence from the 1630’s on that points to racial discrimination without using the term slavery. Such discrimination sometimes stopped short of lifetime servitude or inherited status—the two attributes of true slavery—yet in other cases it included both. The Handlins’ argument excludes the real possibility that Black people in the English colonies were never treated as the equals of White people.The possibility has important ramifications后果,影响.If from the outset Black people were discriminated against, then legal slavery should be viewed as a reflection and an extension of racial prejudice rather than, as many historians including the Handlins have argued, the cause of prejudice. In addition, the existence of discrimination before the advent of legal slavery offers a further explanation for the harsher treatment of Black slaves in North than in South America. Freyre and Tannenbaum have rightly argued that the lack of certain traditions in North America—such as a Roman conception of slavery and a Roman Catholic emphasis on equality— explains why the treatment of Black slaves was more severe there than in the Spanish and Portuguese colonies of South America. But this cannot be the whole explanation since it is merely negative, based only on a lack of something. A more compelling令人信服的explanation is that the early and sometimes extreme racial discrimination in the English colonies helped determine the particular nature of the slavery that followed. (462 words)16. Which of the following is the most logical inference to be drawn from the passage about the effects of “several acts of the Maryland and Virginia legislatures” (Para.2) passed during and after the 1660‟s?A. The acts negatively affected the pre-1660’s position of Black as wellas of White servants.B. The acts had the effect of impairing rather than improving theposition of White servants relative to what it had been before the 1660’s.C. The acts had a different effect on the position of white servants thandid many of the acts passed during this time by the legislatures of other colonies.D. The acts, at the very least, caused the position of White servants toremain no better than it had been before the 1660’s.17. With which of the following statements regarding the status ofBlack people in the English colonies of North America before the 1660‟s would the author be LEAST likely to agree?A. Although black people were not legally considered to be slaves,they were often called slaves.B. Although subject to some discrimination, black people had a higherlegal status than they did after the 1660’s.C. Although sometimes subject to lifetime servitude, black peoplewere not legally considered to be slaves.D. Although often not treated the same as White people, black people,like many white people, possessed the legal status of servants.18. According to the passage, the Handlins have argued which of thefollowing about the relationship between racial prejudice and the institution of legal slavery in the English colonies of North America?A. Racial prejudice and the institution of slavery arose simultaneously.B. Racial prejudice most often the form of the imposition of inheritedstatus, one of the attributes of slavery.C. The source of racial prejudice was the institution of slavery.D. Because of the influence of the Roman Catholic Church, racialprejudice sometimes did not result in slavery.19. The passage suggests that the existence of a Roman conception ofslavery in Spanish and Portuguese colonies had the effect of _________.A. extending rather than causing racial prejudice in these coloniesB. hastening the legalization of slavery in these colonies.C. mitigating some of the conditions of slavery for black people in these coloniesD. delaying the introduction of slavery into the English colonies20. The author considers the explanation put forward by Freyre andTannenbaum for the treatment accorded B lack slaves in the English colonies of North America to be _____________.A. ambitious but misguidedB. valid有根据的but limitedC. popular but suspectD. anachronistic过时的,时代错误的and controversialUNIT 2Text AThe sea lay like an unbroken mirror all around the pine-girt, lonely shores of Orr’s Island. Tall, kingly spruce s wore their regal王室的crowns of cones high in air, sparkling with diamonds of clear exuded gum流出的树胶; vast old hemlocks铁杉of primeval原始的growth stood darkling in their forest shadows, their branches hung with long hoary moss久远的青苔;while feathery larches羽毛般的落叶松,turned to brilliant gold by autumn frosts, lighted up the darker shadows of the evergreens. It was one of those hazy朦胧的, calm, dissolving days of Indian summer, when everything is so quiet that the fainest kiss of the wave on the beach can be heard, and white clouds seem to faint into the blue of the sky, and soft swathing一长条bands of violet vapor make all earth look dreamy, and give to the sharp, clear-cut outlines of the northern landscape all those mysteries of light and shade which impart such tenderness to Italian scenery.The funeral was over,--- the tread鞋底的花纹/ 踏of many feet, bearing the heavy burden of two broken lives, had been to the lonely graveyard, and had come back again,--- each footstep lighter and more unconstrained不受拘束的as each one went his way from the great old tragedy of Death to the common cheerful of Life.The solemn black clock stood swaying with its eternal ―tick-tock, tick-tock,‖ in the kitchen of the brown house on Orr’s Island. There was there that sense of a stillness that can be felt,---such as settles down on a dwelling住处when any of its inmates have passed through its doors for the last time, to go whence they shall not return. The best room was shut up and darkened, with only so much light as could fall through a little heart-shaped hole in the window-shutter,---for except on solemn visits, or prayer-meetings or weddings, or funerals, that room formed no part of the daily family scenery.The kitchen was clean and ample, hearth灶台, and oven on one side, and rows of old-fashioned splint-bottomed chairs against the wall. A table scoured to snowy whiteness, and a little work-stand whereon lay the Bible, the Missionary Herald, and the Weekly Christian Mirror, before named, formed the principal furniture. One feature, however, must not be forgotten, ---a great sea-chest水手用的储物箱,which had been the companion of Zephaniah through all the countries of the earth. Old, and battered破旧的,磨损的, and unsightly难看的it looked, yet report said that there was good store within which men for the most part respect more than anything else; and, indeed it proved often when a deed of grace was to be done--- when a woman was suddenly made a widow in a coast gale大风,狂风, or a fishing-smack小渔船was run down in the fogs off the banks, leaving in some neighboring cottage a family of orphans,---in all such cases, the opening of this sea-chest was an event of good omen 预兆to the bereaved丧亲者;for Zephaniah had a large heart and a large hand, and was apt有…的倾向to take it out full of silver dollars when once it went in. So the ark of the covenant约柜could not have been looked on with more reverence崇敬than the neighbours usually showed to Captain Pennel’s sea-chest.1. The author describes Orr‟s Island in a(n)______way.A.emotionally appealing, imaginativeB.rational, logically preciseC.factually detailed, objectiveD.vague, uncertain2.According to the passage, the “best room”_____.A.has its many windows boarded upB.has had the furniture removedC.is used only on formal and ceremonious occasionsD.is the busiest room in the house3.From the description of the kitchen we can infer that thehouse belongs to people who_____.A.never have guestsB.like modern appliancesC.are probably religiousD.dislike housework4.The passage implies that_______.A.few people attended the funeralB.fishing is a secure vocationC.the island is densely populatedD.the house belonged to the deceased5.From the description of Zephaniah we can see thathe_________.A.was physically a very big manB.preferred the lonely life of a sailorC.always stayed at homeD.was frugal and saved a lotText BBasic to any understanding of Canada in the 20 years after the Second World War is the country' s impressive population growth. For every three Canadians in 1945, there were over five in 1966. In September 1966 Canada's population passed the 20 million mark. Most of this surging growth came from natural increase. The depression of the 1930s and the war had held back marriages, and the catching-up process began after 1945. The baby boom continued through the decade of the 1950s, producing a population increase of nearly fifteen percent in the five years from 1951 to 1956. This rate of increase had been exceeded only once before in Canada's history, in the decade before 1911 when the prairies were being settled. Undoubtedly, the good economic conditions of the 1950s supported a growth in the population, but the expansion also derived from a trend toward earlier marriages and an increase in the average size of families; In 1957 the Canadian birth rate stood at 28 per thousand, one of the highest in the world. After the peak year of 1957, thebirth rate in Canada began to decline. It continued falling until in 1966 it stood at the lowest level in 25 years. Partly this decline reflected the low level of births during the depression and the war, but it was also caused by changes in Canadian society. Young people were staying at school longer, more women were working; young married couples were buying automobiles or houses before starting families; rising living standards were cutting down the size of families. It appeared that Canada was once more falling in step with the trend toward smaller families that had occurred all through theWestern world since the time of the Industrial Revolution. Although the growth in Canada’s population had slowed down by 1966 (the cent), another increase in the first half of the 1960s was only nine percent), another large population wave was coming over the horizon. It would be composed of the children of the children who were born during the period of the high birth rate prior to 1957.6. What does the passage mainly discuss?A. Educational changes in Canadian society.B. Canada during the Second World War.C. Population trends in postwar Canada.D. Standards of living in Canada.7. According to the passage, when did Canada's baby boom begin?A. In the decade after 1911.B. After 1945.C. During the depression of the 1930s.D. In 1966.8. The author suggests that in Canada during the 1950s____________.A. the urban population decreased rapidlyB. fewer people marriedC. economic conditions were poorD. the birth rate was very high9. When was the birth rate in Canada at its lowest postwar level?A. 1966.B. 1957.C. 1956.D. 1951.10. The author mentions all of the following as causes of declines inpopulation growth after 1957 EXCEPT_________________.A. people being better educatedB. people getting married earlierC. better standards of livingD. couples buying houses11.I t can be inferred from the passage that before the IndustrialRevolution_______________.A. families were largerB. population statistics were unreliableC. the population grew steadilyD. economic conditions were badText CI was just a boy when my father brought me to Harlem for the first time, almost 50 years ago. We stayed at the hotel Theresa, a grand brick structure at 125th Street and Seventh avenue. Once, in the hotel restaurant, my father pointed out Joe Louis. He even got Mr. Brown, the hotel manager, to introduce me to him, a bit punchy强力的but still champ焦急as fast as I was concerned.Much has changed since then. Business and real estate are booming. Some say a new renaissance is under way. Others decry责难what they see as outside forces running roughshod肆意践踏over the old Harlem. New York meant Harlem to me, and as a young man I visited it whenever I could. But many of my old haunts are gone. The Theresa shut down in 1966. National chains that once ignored Harlem now anticipate yuppie money and want pieces of this prime Manhattan real estate. So here I am on a hot August afternoon, sitting in a Starbucks that two years ago opened a block away from the Theresa, snatching抓取,攫取at memories between sips of high-priced coffee. I am about to open up a piece of the old Harlem---the New York Amsterdam News---when a tourist。
Decoherence, einselection, and the quantum origins of the classical
a r X i v :q u a n t -p h /0105127v 3 19 J u n 2003DECOHERENCE,EINSELECTION,AND THE QUANTUM ORIGINS OF THE CLASSICALWojciech Hubert ZurekTheory Division,LANL,Mail StopB288Los Alamos,New Mexico 87545Decoherence is caused by the interaction with the en-vironment which in effect monitors certain observables of the system,destroying coherence between the pointer states corresponding to their eigenvalues.This leads to environment-induced superselection or einselection ,a quantum process associated with selective loss of infor-mation.Einselected pointer states are stable.They can retain correlations with the rest of the Universe in spite of the environment.Einselection enforces classicality by imposing an effective ban on the vast majority of the Hilbert space,eliminating especially the flagrantly non-local “Schr¨o dinger cat”states.Classical structure of phase space emerges from the quantum Hilbert space in the appropriate macroscopic limit:Combination of einse-lection with dynamics leads to the idealizations of a point and of a classical trajectory.In measurements,einselec-tion replaces quantum entanglement between the appa-ratus and the measured system with the classical corre-lation.Only the preferred pointer observable of the ap-paratus can store information that has predictive power.When the measured quantum system is microscopic and isolated,this restriction on the predictive utility of its correlations with the macroscopic apparatus results in the effective “collapse of the wavepacket”.Existential in-terpretation implied by einselection regards observers as open quantum systems,distinguished only by their abil-ity to acquire,store,and process information.Spreading of the correlations with the effectively classical pointer states throughout the environment allows one to under-stand ‘classical reality’as a property based on the rela-tively objective existence of the einselected states:They can be “found out”without being re-prepared,e.g,by intercepting the information already present in the envi-ronment.The redundancy of the records of pointer states in the environment (which can be thought of as their ‘fit-ness’in the Darwinian sense)is a measure of their clas-sicality.A new symmetry appears in this setting:Envi-ronment -assisted invariance or envariance sheds a new light on the nature of ignorance of the state of the system due to quantum correlations with the environment,and leads to Born’s rules and to the reduced density matri-ces,ultimately justifying basic principles of the program of decoherence and einselection.ContentsI.INTRODUCTION2A.The problem:Hilbert space is big 21.Copenhagen Interpretation 22.Many Worlds Interpretation 3B.Decoherence and einselection 3C.The nature of the resolution and the role of envariance4D.Existential Interpretation and ‘Quantum Darwinism’4II.QUANTUM MEASUREMENTS5A.Quantum conditional dynamics51.Controlled not and a bit-by-bit measurement 62.Measurements and controlled shifts.73.Amplificationrmation transfer in measurements91.Action per bit9C.“Collapse”analogue in a classical measurement 9III.CHAOS AND LOSS OF CORRESPONDENCE11A.Loss of the quantum-classical correspondence 11B.Moyal bracket and Liouville flow 12C.Symptoms of correspondence loss131.Expectation values 132.Structure saturation13IV.ENVIRONMENT –INDUCED SUPERSELECTION 14A.Models of einselection141.Decoherence of a single qubit152.The classical domain and a quantum halo 163.Einselection and controlled shifts16B.Einselection as the selective loss of information171.Mutual information and discord18C.Decoherence,entanglement,dephasing,and noise 19D.Predictability sieve and einselection20V.EINSELECTION IN PHASE SPACE21A.Quantum Brownian motion21B.Decoherence in quantum Brownian motion231.Decoherence timescale242.Phase space view of decoherence 25C.Predictability sieve in phase space 26D.Classical limit in phase space261.Mathematical approach (¯h →0)272.Physical approach:The macroscopic limit 273.Ignorance inspires confidence in classicality 28E.Decoherence,chaos,and the Second Law281.Restoration of correspondence 282.Entropy production293.Quantum predictability horizon 30VI.EINSELECTION AND MEASUREMENTS30A.Objective existence of einselected states 30B.Measurements and memories31C.Axioms of quantum measurement theory321.Observables are Hermitean –axiom (iiia)322.Eigenvalues as outcomes –axiom (iiib)333.Immediate repeatability,axiom (iv)334.Probabilities,einselection and records342D.Probabilities from Envariance341.Envariance352.Born’s rule from envariance363.Relative frequencies from envariance384.Other approaches to probabilities39 VII.ENVIRONMENT AS A WITNESS40A.Quantum Darwinism401.Consensus and algorithmic simplicity412.Action distance413.Redundancy and mutual information424.Redundancy ratio rate43B.Observers and the Existential Interpretation43C.Events,Records,and Histories441.Relatively Objective Past45 VIII.DECOHERENCE IN THE LABORATORY46A.Decoherence due to entangling interactions46B.Simulating decoherence with classical noise471.Decoherence,Noise,and Quantum Chaos48C.Analogue of decoherence in a classical system48D.Taming decoherence491.Pointer states and noiseless subsystems492.Environment engineering493.Error correction and resilient computing50 IX.CONCLUDING REMARKS51 X.ACKNOWLEDGMENTS52 References52I.INTRODUCTIONThe issue of interpretation is as old as quantum the-ory.It dates back to the discussions of Niels Bohr, Werner Heisenberg,Erwin Schr¨o dinger,(Bohr,1928; 1949;Heisenberg,1927;Schr¨o dinger,1926;1935a,b;see also Jammer,1974;Wheeler and Zurek,1983).Perhaps the most incisive critique of the(then new)theory was due to Albert Einstein,who,searching for inconsisten-cies,distilled the essence of the conceptual difficulties of quantum mechanics through ingenious“gedankenexper-iments”.We owe him and Bohr clarification of the sig-nificance of the quantum indeterminacy in course of the Solvay congress debates(see Bohr,1949)and elucidation of the nature of quantum entanglement(Einstein,Podol-sky,and Rosen,1935;Bohr,1935,Schr¨o dinger,1935a,b). Issues identified then are still a part of the subject. Within the past two decades the focus of the re-search on the fundamental aspects of quantum theory has shifted from esoteric and philosophical to more“down to earth”as a result of three developments.To begin with, many of the old gedankenexperiments(such as the EPR “paradox”)became compelling demonstrations of quan-tum physics.More or less simultaneously the role of de-coherence begun to be appreciated and einselection was recognized as key in the emergence of st not least,various developments have led to a new view of the role of information in physics.This paper reviews progress with a focus on decoherence,einselection and the emergence of classicality,but also attempts a“pre-view”of the future of this exciting and fundamental area.A.The problem:Hilbert space is bigThe interpretation problem stems from the vastness of the Hilbert space,which,by the principle of superposi-tion,admits arbitrary linear combinations of any states as a possible quantum state.This law,thoroughly tested in the microscopic domain,bears consequences that defy classical intuition:It appears to imply that the familiar classical states should be an exceedingly rare exception. And,naively,one may guess that superposition principle should always apply literally:Everything is ultimately made out of quantum“stuff”.Therefore,there is no a priori reason for macroscopic objects to have definite position or momentum.As Einstein noted1localization with respect to macrocoordinates is not just independent, but incompatible with quantum theory.How can one then establish correspondence between the quantum and the familiar classical reality?1.Copenhagen InterpretationBohr’s solution was to draw a border between the quantum and the classical and to keep certain objects–especially measuring devices and observers–on the clas-sical side(Bohr,1928;1949).The principle of superposi-tion was suspended“by decree”in the classical domain. The exact location of this border was difficult to pinpoint, but measurements“brought to a close”quantum events. Indeed,in Bohr’s view the classical domain was more fundamental:Its laws were self-contained(they could be confirmed from within)and established the framework necessary to define the quantum.Thefirst breach in the quantum-classical border ap-peared early:In the famous Bohr–Einstein double-slit debate,quantum Heisenberg uncertainty was invoked by Bohr at the macroscopic level to preserve wave-particle duality.Indeed,as the ultimate components of classical objects are quantum,Bohr emphasized that the bound-ary must be moveable,so that even the human nervous system could be regarded as quantum provided that suit-able classical devices to detect its quantum features were available.In the words of John Archibald Wheeler(1978; 1983)who has elucidated Bohr’s position and decisively contributed to the revival of interest in these matters,“No[quantum]phenomenon is a phenomenon until it is a recorded(observed)phenomenon”.3 This is a pithy summary of a point of view–known asthe Copenhagen Interpretation(CI)–that has kept manya physicist out of despair.On the other hand,as long as acompelling reason for the quantum-classical border couldnot be found,the CI Universe would be governed by twosets of laws,with poorly defined domains of jurisdiction.This fact has kept many a student,not to mention theirteachers,in despair(Mermin1990a;b;1994).2.Many Worlds InterpretationThe approach proposed by Hugh Everett(1957a,b)and elucidated by Wheeler(1957),Bryce DeWitt(1970)and others(see DeWitt and Graham,1973;Zeh,1970;1973;Geroch,1984;Deutsch,1985,1997,2001)was toenlarge the quantum domain.Everything is now repre-sented by a unitarily evolving state vector,a gigantic su-perposition splitting to accommodate all the alternativesconsistent with the initial conditions.This is the essenceof the Many Worlds Interpretation(MWI).It does notsuffer from the dual nature of CI.However,it also doesnot explain the emergence of classical reality.The difficulty many have in accepting MWI stems fromits violation of the intuitively obvious“conservation law”–that there is just one Universe,the one we perceive.But even after this question is dealt with,,many a con-vert from CI(which claims allegiance of a majority ofphysicists)to MWI(which has steadily gained popular-ity;see Tegmark and Wheeler,2001,for an assessment)eventually realizes that the original MWI does not ad-dress the“preferred basis question”posed by Einstein1(see Wheeler,1983;Stein,1984;Bell1981,1987;Kent,1990;for critical assessments of MWI).And as long asit is unclear what singles out preferred states,perceptionof a unique outcome of a measurement and,hence,of asingle Universe cannot be explained either2.In essence,Many Worlds Interpretation does not ad-dress but only postpones the key question.The quantum-classical boundary is pushed all the way towards theobserver,right against the border between the materialUniverse and the“consciousness”,leaving it at a veryuncomfortable place to do physics.MWI is incomplete:It does not explain what is effectively classical and why.Nevertheless,it was a crucial conceptual breakthrough:4ment)can be in principle carried out without disturbing the system.Only in quantum mechanics acquisition of information inevitably alters the state of the system–the fact that becomes apparent in double-slit and related experiments(Wootters and Zurek,1979;Zurek,1983). Quantum nature of decoherence and the absence of classical analogues are a source of misconceptions.For instance,decoherence is sometimes equated with relax-ation or classical noise that can be also introduced by the environment.Indeed,all of these effects often ap-pear together and as a consequence of the“openness”. The distinction between them can be briefly summed up: Relaxation and noise are caused by the environment per-turbing the system,while decoherence and einselection are caused by the system perturbing the environment. Within the past few years decoherence and einselection became familiar to many.This does not mean that their implications are universally accepted(see comments in the April1993issue of Physics Today;d’Espagnat,1989 and1995;Bub,1997;Leggett,1998and2002;Stapp, 2001;exchange of views between Anderson,2001,and Adler,2001).In afield where controversy reigned for so long this resistance to a new paradigm is no surprise. C.The nature of the resolution and the role of envariance Our aim is to explain why does the quantum Universe appear classical.This question can be motivated only in the context of the Universe divided into systems,and must be phrased in the language of the correlations be-tween them.In the absence of systems Schr¨o dinger equa-tion dictates deterministic evolution;|Ψ(t) =exp(−iHt/¯h)|Ψ(0) ,(1.1) and the problem of interpretation seems to disappear. There is no need for“collapse”in a Universe with no systems.Yet,the division into systems is imperfect.As a consequence,the Universe is a collection of open(in-teracting)quantum systems.As the interpretation prob-lem does not arise in quantum theory unless interacting systems exist,we shall also feel free to assume that an environment exists when looking for a resolution. Decoherence and einselectionfit comfortably in the context of the Many Worlds Interpretation where they define the“branches”of the universal state vector.De-coherence makes MWI complete:It allows one to ana-lyze the Universe as it is seen by an observer,who is also subject to decoherence.Einselection justifies elements of Bohr’s CI by drawing the border between the quan-tum and the classical.This natural boundary can be sometimes shifted:Its effectiveness depends on the de-gree of isolation and on the manner in which the system is probed,but it is a very effective quantum-classical border nevertheless.Einselectionfits either MWI or CI framework:It sup-plies a statute of limitations,putting an end to the quantum jurisdiction..It delineates how much of the Universe will appear classical to observers who monitor it from within,using their limited capacity to acquire, store,and process information.It allows one to under-stand classicality as an idealization that holds in the limit of macroscopic open quantum systems.Environment imposes superselection rules by preserv-ing part of the information that resides in the correlations between the system and the measuring apparatus(Zurek, 1981,1982).The observer and the environment compete for the information about the system.Environment–because of its size and its incessant interaction with the system–wins that competition,acquiring information faster and more completely than the observer.Thus,a record useful for the purpose of prediction must be re-stricted to the observables that are already monitored by the environment.In that case,the observer and the en-vironment no longer compete and decoherence becomes unnoticeable.Indeed,typically observers use environ-ment as a“communication channel”,and monitor it to find out about the system.Spreading of the information about the system through the environment is ultimately responsible for the emer-gence of the“objective reality”.Objectivity of a state can be quantified by the redundancy with which it is recorded throughout Universe.Intercepting fragments of the environment allows observers tofind out(pointer) state of the system without perturbing it(Zurek,1993a, 1998a,and2000;see especially section VII of this pa-per for a preview of this new“environment as a witness”approach to the interpretation of quantum theory). When an effect of a transformation acting on a system can be undone by a suitable transformation acting on the environment,so that the joint state of the two remains unchanged,the transformed property of the system is said to exhibit“environment assisted invariance”or en-variance(Zurek,2002b).Observer must be obviously ignorant of the envariant properties of the system.Pure entangled states exhibit envariance.Thus,in quantum physics perfect information about the joint state of the system-environment pair can be used to prove ignorance of the state of the system.Envariance offers a new fundamental view of what is information and what is ignorance in the quantum world. It leads to Born’s rule for the probabilities and justifies the use of reduced density matrices as a description of a part of a larger combined system.Decoherence and ein-selection rely on reduced density matrices.Envariance provides a fundamental resolution of many of the inter-pretational issues.It will be discussed in section VI D.D.Existential Interpretation and‘Quantum Darwinism’What the observer knows is inseparable from what the observer is:The physical state of his memory implies his information about the Universe.Its reliability de-pends on the stability of the correlations with the exter-nal observables.In this very immediate sense decoher-5ence enforces the apparent“collapse of the wavepacket”: After a decoherence timescale,only the einselected mem-ory states will exist and retain useful correlations(Zurek, 1991;1998a,b;Tegmark,2000).The observer described by some specific einselected state(including a configu-ration of memory bits)will be able to access(“recall”) only that state.The collapse is a consequence of einse-lection and of the one-to-one correspondence between the state of his memory and of the information encoded in it. Memory is simultaneously a description of the recorded information and a part of the“identity tag”,defining observer as a physical system.It is as inconsistent to imagine observer perceiving something else than what is implied by the stable(einselected)records in his posses-sion as it is impossible to imagine the same person with a different DNA:Both cases involve information encoded in a state of a system inextricably linked with the physical identity of an individual.Distinct memory/identity states of the observer(that are also his“states of knowledge”)cannot be superposed: This censorship is strictly enforced by decoherence and the resulting einselection.Distinct memory states label and“inhabit”different branches of the Everett’s“Many Worlds”Universe.Persistence of correlations is all that is needed to recover“familiar reality”.In this manner,the distinction between epistemology and ontology is washed away:To put it succinctly(Zurek,1994)there can be no information without representation in physical states. There is usually no need to trace the collapse all the way to observer’s memory.It suffices that the states of a decohering system quickly evolve into mix-tures of the preferred(pointer)states.All that can be known in principle about a system(or about an observer, also introspectively,e.g.,by the observer himself)is its decoherence-resistant‘identity tag’–a description of its einselected state.Apart from this essentially negative function of a cen-sor the environment plays also a very different role of a“broadcasting agent”,relentlessly cloning the informa-tion about the einselected pointer states.This role of the environment as a witness in determining what exists was not appreciated until now:Throughout the past two decades,study of decoherence focused on the effect of the environment on the system.This has led to a mul-titude of technical advances we shall review,but it has also missed one crucial point of paramount conceptual importance:Observers monitor systems indirectly,by in-tercepting small fractions of their environments(e.g.,a fraction of the photons that have been reflected or emit-ted by the object of interest).Thus,if the understand-ing of why we perceive quantum Universe as classical is the principal aim,study of the nature of accessibil-ity of information spread throughout the environment should be the focus of attention.This leads one away from the models of measurement inspired by the“von Neumann chain”(1932)to studies of information trans-fer involving branching out conditional dynamics and the resulting“fan-out”of the information throughout envi-ronment(Zurek,1983,1998a,2000).This new‘quantum Darwinism’view of environment selectively amplifying einselected pointer observables of the systems of interest is complementary to the usual image of the environment as the source of perturbations that destroy quantum co-herence of the system.It suggests the redundancy of the imprint of the system in the environment may be a quantitative measure of relative objectivity and hence of classicality of quantum states.It is introduced in Sec-tions VI and VII of this review.Benefits of recognition of the role of environment in-clude not just operational definition of the objective exis-tence of the einselected states,but–as is also detailed in Section VI–a clarification of the connection between the quantum amplitudes and probabilities.Einselection con-verts arbitrary states into mixtures of well defined possi-bilities.Phases are envariant:Appreciation of envariance as a symmetry tied to the ignorance about the state of the system was the missing ingredient in the attempts of ‘no collapse’derivations of Born’s rule and in the prob-ability interpretation.While both envariance and the “environment as a witness”point of view are only begin-ning to be investigated,the extension of the program of einselection they offer allowes one to understand emer-gence of“classical reality”form the quantum substrate as a consequence of quantum laws.II.QUANTUM MEASUREMENTSThe need for a transition from quantum determinism of the global state vector to classical definiteness of states of individual systems is traditionally illustrated by the example of quantum measurements.An outcome of a “generic”measurement of the state of a quantum sys-tem is not deterministic.In the textbook discussions this random element is blamed on the“collapse of the wavepacket”,invoked whenever a quantum system comes into contact with a classical apparatus.In a fully quan-tum discussion this issue still arises,in spite(or rather because)of the overall deterministic quantum evolution of the state vector of the Universe:As pointed out by von Neumann(1932),there is no room for a‘real collapse’in the purely unitary models of measurements.A.Quantum conditional dynamicsTo illustrate the ensuing difficulties,consider a quan-tum system S initially in a state|ψ interacting with a quantum apparatus A initially in a state|A0 :|Ψ0 =|ψ |A0 = i a i|s i |A0−→ i a i|s i |A i =|Ψt .(1)Above,{|A i }and{|s i }are states in the Hilbert spaces of the apparatus and of the system,respectively,and a i6are complex coefficients.Conditional dynamics of such premeasurement(as the step achieved by Eq.(2.1)isoften called)can be accomplished by means of a unitary Schr¨o dinger evolution.Yet it is not enough to claim thata measurement has been achieved:Equation(2.1)leads to an uncomfortable conclusion:|Ψt is an EPR-like en-tangled state.Operationally,this EPR nature of the state emerging from the premeasurement can be made moreexplicit by re-writing the sum in a different basis:|Ψt = i a i|s i |A i = i b i|r i |B i .(2)This freedom of basis choice–basis ambiguity–is guar-anteed by the principle of superposition.Therefore,if one were to associate states of the apparatus(or the ob-server)with decompositions of|Ψt ,then even before en-quiring about the specific outcome of the measurement one would have to decide on the decomposition of|Ψt ; the change of the basis redefines the measured quantity.1.Controlled not and a bit-by-bit measurementThe interaction required to entangle the measured sys-tem and the apparatus,Eq.(2.1),is a generalization of the basic logical operation known as a“controlled not”or a c-not.Classical,c-not changes the state a t of the target when the control is1,and does nothing otherwise: 0c a t−→0c a t;1c a t−→1c¬a t(2.3) Quantum c-not is a straightforward quantum version of Eq.(2.3).It was known as a“bit by bit measurement”(Zurek,1981;1983)and used to elucidate the connection between entanglement and premeasurement already be-fore it acquired its present name and significance in the context of quantum computation(see e.g.Nielsen and Chuang,2000).Arbitrary superpositions of the control bit and of the target bit states are allowed:(α|0c +β|1c )|a t−→α|0c |a t +β|1c |¬a t (3) Above“negation”|¬a t of a state is basis dependent;¬(γ|0t +δ|1t )=γ|1t +δ|0t (2.5) With|A0 =|0t ,|A1 =|1t we have an obvious anal-ogy between the c-not and a premeasurement.In the classical controlled not the direction of informa-tion transfer is consistent with the designations of the two bits:The state of the control remains unchanged while it influences the target,Eq.(2.3).Classical measurement need not influence the system.Written in the logical ba-sis{|0 ,|1 },the truth table of the quantum c-not is essentially–that is,save for the possibility of superpo-sitions–the same as Eq.(2.3).One might have antici-pated that the direction of information transfer and the designations(“control/system”and“target/apparatus”)of the two qubits will be also unambiguous,as in the clas-sical case.This expectation is incorrect.In the conjugate basis{|+ ,|− }defined by:|± =(|0 ±|1 )/√2|1 1|S⊗(1−(|0 1|+|1 0|))A(5) Above,g is a coupling constant,and the two operators refer to the system(i.e.,to the former control),and to the apparatus pointer(the former target),respectively. It is easy to see that the states{|0 ,|1 }S of the system are unaffected by H int,since;[H int,e0|0 0|S+e1|1 1|S]=0(2.10) The measured(control)observableˆǫ=e0|0 0|+e1|1 1| is a constant of motion under H int.c-not requires inter-action time t such that gt=π/2.The states{|+ ,|− }A of the apparatus encode the information about phase between the logical states.They have exactly the same“immunity”:[H int,f+|+ +|A+f−|− −|A]=0(2.11) Hence,when the apparatus is prepared in a definite phase state(rather than in a definite pointer/logical state),it will pass on its phase onto the system,as Eqs.(2.7)-(2.8),show.Indeed,H int can be written as:H int=g|1 1|S|− −|A=g7 phases between the possible outcome states of the ap-paratus.This leads to loss of phase coherence:Phasesbecome“shared property”as we shall see in more detailin the discussion of envariance.The question“what measures what?”(decided by thedirection of the informationflow)depends on the initialstates.In“the classical practice”this ambiguity does notarise.Einselection limits the set of possible states of theapparatus to a small subset.2.Measurements and controlled shifts.The truth table of a whole class of c-not like trans-formations that includes general premeasurement,Eq.(2.1),can be written as:|s j |A k −→|s j |A k+j (2.13)Equation(2.1)follows when k=0.One can thereforemodel measurements as controlled shifts–c-shift s–generalizations of the c-not.In the bases{|s j }and{|A k },the direction of the informationflow appears tobe unambiguous–from the system S to the apparatus A.However,a complementary basis can be readily defined(Ivanovic,1981;Wootters and Fields,1989);|B k =N−1N kl)|A l .(2.14a)Above N is the dimensionality of the Hilbert space.Anal-ogous transformation can be carried out on the basis{|s i }of the system,yielding states{|r j }.Orthogonality of{|A k }implies:B l|B m =δlm.(2.15)|A k =N−1N kl)|B l (2.14b)inverts of the transformtion of Eq.(2.14a).Hence:|ψ = lαl|A l = kβk|B k ,(2.16)where the coefficientsβk are:βk=N−1Nkl)αl.(2.17)Hadamard transform of Eq.(2.6)is a special case of themore general transformation considered here.To implement the truth tables involved in premeasure-ments we define observableˆA and its conjugate:ˆA=N−1k=0k|A k A k|;ˆB=N−1 l=0l|B l B l|.(2.18a,b)The interaction Hamiltonian:H int=gˆsˆB(2.19)is an obvious generalization of Eqs.(2.9)and(2.12),withg the coupling strength andˆs:ˆs=N−1l=0l|s l s l|(2.20)In the{|A k }basisˆB is a shift operator,ˆB=iN∂ˆA.(2.21)To show how H int works,we compute:exp(−iH int t/¯h)|s j |A k =|s j N−1。
困顿感量表中文版测评医学生的效度和信度
·心理卫生评估·困顿感量表中文版测评医学生的效度和信度*龚睿婕1刘景壹1王亦晨2蔡泳3王甦平3(1上海市徐汇区疾病预防控制中心,上海2002372上海交通大学医学院附属瑞金医院,上海2000253上海交通大学医学院,上海200025通信作者:王甦平wangsuping@)【摘要】目的:引进困顿感量表(ES),评价其在医学生群体中的信效度。
方法:选取某医学院校学生1768名,将其随机分半,一半(n=855)进行探索性因子分析,另一半(n=913)进行验证性因子分析;采用病人健康问卷9条目(PHQ-9)检验效标效度。
间隔1个月后,在总样本中选取53名学生进行重测。
结果:探索性因子分析显示量表共16个条目,包含1个公因子,累计方差解释率64.66%,各条目的因子负荷值在0.23 0.77之间;验证性因子分析表明两因子模型拟合情况略优于一因子模型(χ2/df=7.00,RMSEA=0.08,GFI=0.91,CFI=0.95),各因子负荷在0.48 0.89之间。
ES得分与PHQ-9得分呈正相关(ICC=0.44)。
总量表的Cronbachα系数为0.96,2个维度的α系数分别为0.94和0.93;总量表的重测信度为0.83,2个维度的重测信度为0.80、0.83。
结论:困顿感量表中文版在医学生群体有良好的信效度,可以用于评估该群体的困顿感。
【关键词】困顿感量表;医学生;效度;信度中图分类号:B841.7文献标识码:A文章编号:1000-6729(2019)005-0393-05doi:10.3969/j.issn.1000-6729.2019.05.015(中国心理卫生杂志,2019,33(5):393-397.)Validity and reliability of the Chinese vision of theEntrapment Scale in medical studentsGONGRuijie1,LIU Jingyi1,WANG Yichen2,CAI Yong3,WANG Suping3 1Shanghai Xuhui Center for Disease Control and Prevention,Shanghai200237,China2Shanghai Jiao Tong University School of Medicine,Ruijin Hospital,Department of Hospital Infection Control,Shanghai200025,China3School of Medicine,Shanghai Jiao Tong University,Shanghai200025,ChinaCorresponding author:WANG Suping,wangsuping@【Abstract】Objective:To evaluate the validity and reliability of the Chinese vision of the Entrapment Scale (ES)in medical students.Methods:Totally1768medical students were selected.They were randomly allocated into two groups for exploratory factor analysis(n=855)and confirmatory factor analysis(n=913).All samples were assessed in criterion validity with the Patient Health Questionnaire(PHQ-9).One month later,53partici-pants were retested.Results:The exploratory factor analysis extracted1component from16items,and explained64.66%of the total variance.The factor loading of items ranged between0.23-0.77.The confirmatory factor a-nalysis verified that the fitting of the two-factor model was slightly better than that of the one-factor model(χ2/df=7.00,RMSEA=0.08,GFI=0.91,CFI=0.95).The factor loading of items ranged from0.48to0.89.TheES scores were positively correlated with the PHQ-9scores(ICC=0.44).Cronbach'sαcoefficients were0.96 for the total scale and0.83for the test-retest reliability.The internal consistency reliabilities for the2factors were*基金项目:国家自然科学基金———基于IMB模型的跨性别男男性行为者艾滋病高危行为干预策略研究0.94and0.93,and the test-retest reliabilities for the2factors were0.80and0.83.Conclusion:The Chinese vi-sion of the Entrapment Scale has good validity and reliability among Chinese medical students,and it could be used in the evaluation of entrapment.【Key words】Entrapment Scale;medical students;validity;reliability(Chin Ment Health J,2019,33(5):393-397.)困顿感(entrapment)在心理生理学中指想要摆脱威胁或者压力,但是自身没有能力而继续处于被困的一种感觉或个人心理状态[1]。
随机森林 集中度计算
随机森林集中度计算英文回答:Random Forest is a popular machine learning algorithm that combines multiple decision trees to make predictions.It is an ensemble learning method that uses the "wisdom of the crowd" principle to improve accuracy and reduce overfitting. In a Random Forest, each decision tree is trained on a random subset of the training data, and the final prediction is made by averaging the predictions ofall the trees.One important concept in Random Forest is the measureof concentration, which helps us understand how confident the model is in its predictions. The concentration can be calculated using various metrics, such as Gini impurity or entropy. These metrics quantify the purity of a node in a decision tree by measuring the distribution of class labels.A node with low impurity or entropy indicates a high concentration, meaning that the majority of the samples inthat node belong to the same class. On the other hand, a node with high impurity or entropy indicates a low concentration, meaning that the samples in that node are distributed across multiple classes.To calculate the concentration of a Random Forest, we can aggregate the concentrations of all the decision trees. For example, we can calculate the average impurity or entropy across all the trees to get an overall measure of concentration. A high average impurity or entropy suggests a low concentration, indicating that the predictions of the Random Forest are less reliable. On the other hand, a low average impurity or entropy suggests a high concentration, indicating that the predictions are more reliable.中文回答:随机森林是一种流行的机器学习算法,它将多个决策树组合起来进行预测。
Inducing Features of Random Fields (1997)
Inducing Features of Random FieldsStephen Della Pietra,Vincent Della Pietra,and John Lafferty,Member,IEEE Abstract—We present a technique for constructing randomfields from aset of training samples.The learning paradigm builds increasingly complexfields by allowing potential functions,or features,that are supported byincreasingly large subgraphs.Each feature has a weight that is trainedby minimizing the Kullback-Leibler divergence between the model and theempirical distribution of the training data.A greedy algorithm determineshow features are incrementally added to thefield and an iterative scalingalgorithm is used to estimate the optimal values of the weights.The randomfield models and techniques introduced in this paper differfrom those common to much of the computer vision literature in that theunderlying randomfields are non-Markovian and have a large number ofparameters that must be estimated.Relations to other learning approaches,including decision trees,are given.As a demonstration of the method,wedescribe its application to the problem of automatic word classification innatural language processing.Keywords—Randomfield,Kullback-Leibler divergence,iterative scal-ing,maximum entropy,EM algorithm,statistical learning,clustering,wordmorphology,natural language processing.I.I NTRODUCTIONN this paper we present a method for incrementally construct-ing randomfields.Our method builds increasingly complexfields to approximate the empirical distribution of a set of train-ing examples by allowing potential functions,or features,thatare supported by increasingly large subgraphs.Each feature isassigned a weight,and the weights are trained to minimize the Kullback-Leibler divergence between thefield and the empiri-cal distribution of the training data.Features are incrementally added to thefield using a top-down greedy algorithm,with theintent of capturing the salient properties of the empirical sam-ple while allowing generalization to new configurations.The general problem that the methods we propose address is that ofdiscovering the structure inherent in a set of sample patterns.As one of the fundamental aims of statistical inference and learn-ing,this problem is central to a wide range of tasks includingclassification,compression,and prediction.To illustrate the nature of our approach,suppose we wishto automatically characterize spellings of words according to a statistical model;this is the application we develop in Section5.Afield with no features is simply a uniform distribution on ASCII strings(where we take the distribution of string lengths as given).The most conspicuous feature of English spellings is thatthey are most commonly comprised of lower-case letters.The induction algorithm makes this observation byfirst constructing thefield1a z a z a z a z a z a z where the weight a z a z associated with adjacent lower-case letters is approximately180.Thefirst1000features that the algorithm induces include the strings s>,<re,ly>,and ing>,where the character“<”de-notes beginning-of-string and the character“>”denotes end-of-string.In addition,thefirst1000features include the regular ex-pressions[0-9][0-9](with weight915)and[a-z][A-Z] (with weight581)in addition to thefirst two features[a-z] and[a-z][a-z].A set of strings obtained by Gibbs sampling from the resultingfield is shown here:was,reaser,in,there,to,will,,,was,by, homes,thing,be,reloverated,ther,which,conists,at,fores,anditing,with,Mr.,proveral, the,,,***,on’t,prolling,prothere,,,mento,at,yaou,1,chestraing,for,have,to,intrally, of,qut,.,best,compers,***,cluseliment,uster, of,is,deveral,this,thise,of,offect,inatever, thifer,constranded,stater,vill,in,thase,in, youse,menttering,and,.,of,in,verate,of,to These examples are discussed in detail in Section5.The induction algorithm that we present has two parts:fea-ture selection and parameter estimation.The greediness of the algorithm arises in feature selection.In this step each feature in a pool of candidate features is evaluated by estimating the reduc-tion in the Kullback-Leibler divergence that would result from adding the feature to thefield.This reduction is approximated as a function of a single parameter,and the largest value of this function is called the gain of the candidate.This approximation is one of the key elements of our approach,making it practical to evaluate a large number of candidate features at each stage of the induction algorithm.The candidate with the largest gain is added to thefield.In the parameter estimation step,the parame-ters of thefield are estimated using an iterative scaling algorithm. The algorithm we use is a new statistical estimation algorithmthat we call Improved Iterative Scaling.It is an improvement of the Generalized Iterative Scaling algorithm of Darroch and Ratcliff[12]in that it does not require that the features sum to a constant.The improved algorithm is easier to implement than the Darroch and Ratcliff algorithm,and can lead to an increase in the rate of convergence by increasing the size of the step taken toward the maximum at each iteration.In Section4we give a simple,self-contained proof of the convergence of the improved algorithm that does not make use of the Kuhn-Tucker theorem or other machinery of constrained optimization.Moreover,our proof does not rely on the convergence of alternating I-projection as in Csisz´a r’s proof[10]of the Darroch-Ratcliff procedure. Both the feature selection step and the parameter estimation step require the solution of certain algebraic equations whose coefficients are determined as expectation values with respect to thefield.In many applications these expectations cannot be computed exactly because they involve a sum over an exponen-tially large number of configurations.This is true of the appli-cation that we develop in Section5.In such cases it is possible to approximate the equations that must be solved using Monte Carlo techniques to compute expectations of random variables. The application that we present uses Gibbs sampling to compute expectations,and the resulting equations are then solved using Newton’s method.Our method can be viewed in terms of the principle of max-imum entropy[19],which instructs us to assume an exponen-tial form for our distributions,with the parameters viewed as Lagrange multipliers.The techniques that we develop in this paper apply to exponential models in general.We formulate our approach in terms of randomfields because this provides a convenient framework within which to work,and because our main application is naturally cast in these terms.Our method differs from the most common applications of statistical techniques in computer vision and natural language processing.In contrast to many applications in computer vision, which involve only a few free parameters,the typical applica-tion of our method involves the estimation of thousands of free parameters.In addition,our methods apply to general exponen-tial models and randomfields–there is no underlying Markov assumption made.In contrast to the statistical techniques com-mon to natural language processing,in typical applications of our method there is no probabilisticfinite-state or push-down automaton on which the statistical model is built.In the following section we describe the form of the random field models considered in this paper and the general learning algorithm.In Section3we discuss the feature selection step of the algorithm and briefly address cases when the equations need to be estimated using Monte Carlo methods.In Section4we present the Improved Iterative Scaling algorithm for estimating the parameters,and prove the convergence of this algorithm. In Section5we present the application of inducing features of spellings,andfinally in Section6we discuss the relation between our methods and other learning approaches,as well as possible extensions of our method.II.T HE L EARNING P ARADIGMIn this section we present the basic algorithm for building up a randomfield from elementary features.The basic idea is to incrementally construct an increasingly detailedfield to approximate a reference distribution˜.Typically the distribution ˜is obtained as the empirical distribution of a set of training examples.After establishing our notation and defining the form of the randomfield models we consider,we present the training problem as a statement of two equivalent optimization problems. We then discuss the notions of a candidate feature and the gain of a candidate.Finally,we give a statement of the induction algorithm.A.Form of the randomfield modelsLet be afinite graph with vertex set and edge set,and let be afinite alphabet.The configuration space Ωis the set of all labelings of the vertices in by letters in .If andΩis a configuration,then denotesthe configuration restricted to.A randomfield on is a probability distribution onΩ.The set of all randomfields is nothing more than the simplex∆of all probability distributions onΩ.If:Ωthen the support of,written supp, is the smallest vertex subset having the property that wheneverΩwith then.We consider randomfields that are given by Gibbs distribu-tions of the form11DELLA PIETRA,DELLA PIETRA,AND LAFFERTY:INDUCING FEATURES OF RANDOM FIELDS3 associated with a non-binary feature.Having tied parameters isoften natural for a particular problem,but the presence of non-binary features generally makes the estimation of parametersmore difficult.An automorphism of a graph is a permutation of the verticesthat takes edges to edges:if and only if.A randomfield is said to have homogeneousfeatures if for each feature and automorphism of the graph,there is a feature such thatforΩ.If in addition,then thefield is said tobe homogeneous.Roughly speaking,a homogeneous featurecontributes the same weight to the distribution no matter wherein the graph it appears.Homogeneous features arise naturally inthe application of Section5.The methods that we describe in this paper apply to expo-nential models in general;that is,it is not essential that there isan underlying graph structure.However,it will be convenientto express our approach in terms of the randomfield modelsdescribed above.B.Two optimization problemsSuppose that we are given an initial model0∆,a referencedistribution˜,and a set of features01.Inpractice,it is often the case that˜is the empirical distribution ofa set of training samples12,and is thus givenby˜(1)Throughout this paper we use the notationΩfor the expectation of a function:Ωwith respect tothe probability distribution.For a function:Ωand adistribution,we use both the notation and to denote thegeneralized Gibbs distribution given by14IEEE TRANSACTIONS PATTERN ANALYSIS AND MACHINE INTELLIGENCE,VOL.19,NO.4,APRIL 1997Definition 1:Suppose that the field is given by0.The featuresare called the active features of .A feature is a candidate for if either ,or if is of the form for an atomic feature and an active feature with supp supp .The set of candidate features of is denoted .In other words,candidate features are obtained by conjoining atomic features with existing features.The condition on supports ensures that each feature is supported by a path-connected subset of .If is a candidate feature of ,then we call the 1-parameter family of random fields the induction of by .We also define˜˜(2)We think of as the improvement that feature bringsto the model when it has weight .As we show in the following section,is -convex in .(We use the suggestive notation -convex and -convex in place of the less mnemonic concave and convex terminology.)We define to be the greatest improvement that feature can give to the model while keeping all of the other features’parameters fixed:supWe refer toas the gain of the candidate .D.Incremental construction of random fieldsWe can now describe our algorithm for incrementally con-structing fields.Field Induction Algorithm.Initial Data:A reference distribution ˜and an initial model 0.Output:A field with active features 0such that arg min ¯˜.Algorithm:(0)Set 00.(1)For each candidatecompute the gain.(2)Let arg maxbe the feature with thelargest gain.(3)Computearg min¯˜,where1.(4)Set1and1,and go to step (1).This induction algorithm has two parts:feature selection and parameter estimation .Feature selection is carried out in steps (1)and (2),where the feature yielding the largest gain is incorporated into the model.Parameter estimation is carried out in step (3),where the parameters are adjusted to best represent the reference distribution.These two computations are discussed in more detail in the following two sections.III.F EA TURE S ELECTIONThe feature selection step of our induction algorithm is based upon an approximation.We approximate the improvement due to adding a single candidate feature,measured by the reduction in Kullback-Leibler divergence,by adjusting only the weight of the candidate and keeping all of the other parameters of the field fixed.In general this is only an estimate,since it may well be that adding a feature will require significant adjustments to all of the parameters in the new model.From a computational perspective,approximating the improvement in this way can enable the simultaneous evaluation of thousands of candidate features,and makes the algorithm practical.In this section we present explain the feature selection step in detail.Proposition 1:Let ,defined in (2),be the approxi-mate improvement obtained by adding feature with parameter to the field .Then if is not constant,is strictly -convex in and attains its maximum at the unique point ˆsatisfying˜ˆ(3)Proof:Using the definition (1)of the Kullback-Leibler di-vergence we can writeΩ˜log1˜222Hence,22,which is minus the varianceof with respect to ,is strictly negative,so that isstrictly convex.When is binary-valued,its gain can be expressed in a par-ticularly nice form.This is stated in the following proposition,whose proof is a simple calculation.Proposition 2:Suppose that the candidate is binary-valued.Then is maximized atˆlog˜1DELLA PIETRA,DELLA PIETRA,AND LAFFERTY:INDUCING FEATURES OF RANDOM FIELDS5where and are Bernoulli random variables given by1˜01˜101For features that are not binary-valued,but instead take values in the non-negative integers,the parameterˆthat solves(3)and thus maximizes cannot,in general,be determined in closed form.This is the case for tied binary features,and it applies to the application we describe in Section5.For these cases it is convenient to rewrite(3)slightly.Let so that.Letbe the total probability assigned to the event that the feature takes the value.Then(3)becomes0(4)This equation lends itself well to numerical solution.The gen-eral shape of the curve log is shown in Figure1.Fig.1.Derivative of the gainThe limiting value of log as is˜.The solution to equation(4)can be found using Newton’s method,which in practice converges rapidly for such functions. When the configuration spaceΩis large,so that the coeffi-cients cannot be calculated by summing over all configura-tions,Monte Carlo techniques may be used to estimate them. It is important to emphasize that the same set of random con-figurations can be used to estimate the coefficients for each candidate simultaneously.Rather than discuss the details of Monte Carlo techniques for this problem we refer to the exten-sive literature on this topic.We have obtained good results using the standard technique of Gibbs sampling[17]for the problem we describe in Section5.IV.P ARAMETER E STIMATIONIn this section we present an algorithm for selecting the pa-rameters associated with the features of a randomfield.The algorithm is a generalization of the Generalized Iterative Scal-ing algorithm of Darroch and Ratcliff[12].It reduces to the Darroch-Ratcliff algorithm when the features sum to a constant; however,the new algorithm does not make this restriction. Throughout this section we hold the set of features01,the initial model0and the reference distri-bution˜fixed,and we simplify the notation accordingly.In particular,we write instead of for.We assume that˜0whenever00.This condition is commonly written˜0,and it is equivalent to˜0.A description of the algorithm requires an additional piece of notation.Let#If the features are binary,then#is the total number of features that are“on”for the configuration.Improved Iterative Scaling.Initial Data:A reference distribution˜and an initial model0,with˜0,and non-negative features01. Output:The distribution arg min¯˜Algorithm:(0)Set00.(1)For each let be the unique solution of#˜(5)(2)Set1and1.(3)If has converged,set and terminate.Oth-erwise go to step(1).In other words,this algorithm constructs a distributionlim0whereand is determined as the solution to the equation#˜When used in the-th iteration of thefield induction algorithm, where a candidate feature is added to thefield,we choose the initial distribution0to be0ˆ,whereˆis the parameter that maximizes the gain of.In practice,this provides a good starting point from which to begin iterative scaling.In fact,we can view this distribution as the result of applying one iteration of an Iterative Proportional Fitting Procedure[5],[9] to project onto the linear family of distributions with-marginals constrained to˜.Our main result in this section isProposition3:Suppose is the sequence in∆determined by the Improved Iterative Scaling algorithm.Then˜decreases monotonically to˜and converges toarg min¯˜arg min0.In the remainder of this section we present a self-contained proof of the convergence of the algorithm.The key idea of the proof is to express the incremental step of the algorithm in terms of an auxiliary function which bounds from below the log-likelihood objective function.This technique is the standard means of analyzing the EM algorithm[13],but it has not previ-ously been applied to iterative scaling.Our analysis of iterative scaling is different and simpler than previous treatments.In particular,in contrast to Csisz´a r’s proof of the Darroch-Ratcliff6IEEE TRANSACTIONS PATTERN ANALYSIS AND MACHINE INTELLIGENCE,VOL.19,NO.4,APRIL1997procedure[10],our proof does not rely upon the convergence of alternating I-projection[9].We begin by formulating the basic duality theorem which states that the maximum likelihood problem for a Gibbs dis-tribution and the maximum entropy problem subject to linear constraints have the same solution.We then turn to the task of computing this solution.After introducingauxiliary functions in a general setting,we apply this method to prove convergence of the Improved Iterative Scaling algorithm.Wefinish the section by discussing Monte Carlo methods for estimating the equations when the size of the configuration space prevents the explicit calculation of feature expectations.A.DualityThe duality between the maximum likelihood and maximum entropy problems is expressed in the following Proposition. Proposition4:Suppose that˜0.Then there exists a unique∆satisfying(1)¯(2)for any and¯(3)arg min¯˜(4)arg min0.Moreover,any of these four properties determines uniquely. This result is well known,although perhaps not quite in this packaging.In the language of constrained optimization,it ex-presses the fact that the maximum likelihood problem for Gibbs distributions is the convex dual to the maximum entropy prob-lem for linear constraints.Property(2)is called the Pythagorean property since it resembles the Pythagorean theorem if we imag-ine that is the square of Euclidean distance andare the vertices of a right triangle.We include a proof of this result in Appendix A to make this paper self-contained and also to carefully address the technical issues arising from the fact that is not closed.The proposition would not be true if we replaced¯with;in fact,might be empty.Our proof is elementary and does not rely on the Kuhn-Tucker theorem or other machinery of constrained optimization.B.Auxiliary functionsWe now turn to the task of computing.Fix˜and let :∆be the log-likelihood objective function˜Definition2:A function:∆is an auxiliary function for if(1)For all∆and(2)is continuous in∆and1in with00andWe can use an auxiliary function to construct an iterative algorithm for maximizing.We start with0and recursively define1by1with arg maxIt is clear from property(1)of the definition that each step of this procedure increases.The following proposition implies that in fact the sequence will reach the maximum of. Proposition5:Suppose is any sequence in∆withand1where satisfiessup(6)Then increases monotonically to max¯and converges to arg max¯.Equation(6)assumes that the supremum sup is achieved atfinite.In Appendix B,under slightly stronger assumptions,we present an extension that allows some compo-nents of to take the value.To use the proposition to construct a practical algorithm we must determine an auxiliary function for which satisfying the required condition can be determined efficiently. In Section4.3we present a choice of auxiliary function which yields the Improved Iterative Scaling updates.To prove Proposition5wefirst prove three lemmas. Lemma1:If∆is a cluster point of,then0for all.Proof:Let be a sub-sequence converging to.Then for any11Thefirst inequality follows from property(6)of.The sec-ond and third inequalities are a consequence of the monotonicity of.The lemma follows by taking limits and using the fact that and are continuous.Lemma2:If∆is a cluster point of,thenDELLA PIETRA,DELLA PIETRA,AND LAFFERTY:INDUCING FEATURES OF RANDOM FIELDS7 Proof of Proposition5:Suppose that is a cluster point of.Then it follows from Lemma2that#.It is easy to check that extends to acontinuous function on∆.Lemma4:is an extended auxiliary function for. The key ingredient in the proof of the lemma is the-convexity of the logarithm and the-convexity of the exponential,as ex-pressed in the inequalitiesif0,1(8) log1for all0(9) Proof of Lemma4:Because extends to a continuous func-tion on∆,it suffices to prove that it satisfies properties(1)and(2)of Definition2.To prove property(1)note that˜log(10)˜1(11)˜1#(12)(13)Equality(10)is a simple calculation.Inequality(11)follows from inequality(9).Inequality(12)follows from the definitionof#and Jensen’s inequality(8).Property(2)of Definition2is straightforward to verify.Proposition3follows immediately from the above lemma and the extended Proposition5.Indeed,it is easy to check thatdefined in Proposition3achieves the maximum of, so that it satisfies the condition of Proposition5in Appendix B.D.Monte Carlo methodsThe Improved Iterative Scaling algorithm described in the previous section is well-suited to numerical techniques since all of the features take non-negative values.In each iteration of this algorithm it is necessary to solve a polynomial equation for each feature.That is,we can express equation5in the formwhere is the largest value of#and#˜0(14)where is thefield for the-th iteration and.This equation has no solution precisely when0for0. Otherwise,it can be efficiently solved using Newton’s method since all of the coefficients,0,are non-negative.When Monte Carlo methods are to be used because the configuration spaceΩis large,the coefficients can be simultaneously estimated for all and by generating a single set of samples from the distribution.V.A PPLICA TION:W ORD M ORPHOLOGYWord clustering algorithms are useful for many natural lan-guage processing tasks.One such algorithm[6],called mutual information clustering,is based upon the construction of simple bigram language models using the maximum likelihood crite-rion.The algorithm gives a hierarchical binary classification of words that has been used for a variety of purposes,including the construction of decision tree language and parsing models,and sense disambiguation for machine translation[7].A fundamental shortcoming of the mutual information word clustering algorithm given in[6]is that it takes as fundamental the word spellings themselves.This increases the severity of the problem of small counts that is present in virtually every statistical learning algorithm.For example,the word“Hamil-tonianism”appears only once in the365,893,263-word corpus used to collect bigrams for the clustering experiments described in[6].Clearly this is insufficient evidence on which to base a statistical clustering decision.The basic motivation behind the feature-based approach is that by querying features of spellings, a clustering algorithm could notice that such a word begins with a capital letter,ends in“ism”or contains“ian,”and profit from how these features are used for other words in similar contexts. In this section we describe how we applied the randomfield induction algorithm to discover morphological features of words, and we present sample results.This application demonstrates how our technique gradually sharpens the probability mass from the enormous set of all possible configurations,in this case ASCII strings,onto a set of configurations that is increasingly similar to those in the training sample.It achieves this by introducing both “positive”features which many of the training samples exhibit, as well as“negative”features which do not appear in the sample, or appear only rarely.A description of how the resulting features8IEEE TRANSACTIONS PATTERN ANALYSIS AND MACHINE INTELLIGENCE,VOL.19,NO.4,APRIL1997 were used to improve mutual information clustering is given in[20],and is beyond the scope of the present paper;we refer thereader to[6],[20]for a more detailed treatment of this topic.In Section5.1we formulate the problem in terms of the no-tation and results of Sections2,3,and4.In Section5.2wedescribe how thefield induction algorithm is actually carried outin this application.In Section5.3we explain the results of theinduction algorithm by presenting a series of examples.A.Problem formulationTo discover features of spellings we take as configurationspace the set of all stringsΩin the ASCII alphabet.Weconstruct a probability distribution onΩbyfirst predictingthe length,and then predicting the actual spelling;thus,where is the length distributionand is the spelling distribution.We take the length distributionas given.We model the spelling distribution over stringsof length as a randomfield.LetΩbe the configuration spaceof all ASCII strings of length.ThenΩ102since eachis an ASCII character.To reduce the number of parameters,we tie features,as de-scribed in Section2.1,so that a feature has the same weightindependent of where it appears in the string.Because of this itis natural to view the graph underlyingΩas a regular-gon.Thegroup of automorphisms of this graph is the set of all rotations,and the resultingfield is homogeneous as defined in Section2.Not only is eachfield homogeneous,but in addition,we tiefeatures acrossfields for different values of.Thus,the weightof a feature is independent of.To introduce a dependenceon the length,as well as on whether or not a feature applies atthe beginning or end of a string,we adopt the following artificialconstruction.We take as the graph ofΩan1-gon ratherthan an-gon,and label a distinguished vertex by the length,keeping this label heldfixed.To complete the description of thefields that are induced,weneed to specify the set of atomic features.The atomic featuresthat we allow fall into three types.Thefirst type is the class offeatures of the form1if0otherwise.where is any ASCII character,and denotes an arbitrary char-acter position in the string.The second type of atomic featuresinvolve the special vertex<>that carries the length of the string.These are the features1if<>0otherwise<>1if<>for some 0otherwiseThe atomic feature<>introduces a dependence on whether a string of characters lies at the beginning or end of the string,and the atomic features introduce a dependence on the length of the string.To tie together the length dependence for long strings, we also introduce an atomic feature7for strings of length7 or greater.Thefinal type of atomic feature asks whether a character lies in one of four sets,[a-z],[A-Z],[0-9],[@-&],denoting arbitrary lowercase letters,uppercase letters,digits,and punctu-ation.For example,the atomic feature[a-z]1if[a-z]0otherwisetests whether or not a character is lowercase.To illustratethe notation that we use,let us suppose that the the following features are active for afield:“ends in ism,”“a string of at least7characters beginning with a capital letter”and“con-tains ian.”Then the probability of the word“Hamiltonianism”would be given as14Hamiltonianism14141。
- 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)。
PACS No.:05.40.+j, 82.20.Mj
I. Introduction There has been much discussion recently[1-10] on how to model a physical system that could extract work out of random fluctuations without having to apply directly an obvious biased force taking, in fact, cue from —- or perhaps yearning to explain —– an experimentally observed phenomena[11] of predominantly unidirectional motion of macromolecules (biological motors) along microtubules. We reason in the present work that system inhomogeneity may provide a clear and unifying framework to approach the problem of macroscopic motion under discussion. Macroscopic unidirectional motion of a particle is not possible thermodynamically in the prsence of equilibrium fluctuations. However, such a motion can be obtained in an nonequilibrium situation where the principle of detailed balance does not hold. The existing popular models,[1-10] currently in the literature, mostly take the fluctuations to be nonequilibrium, that is, consider nonwhite or at least nonguassian-white (colored) noise together with a ratchetlike periodic system potential to aid asymmetric motion of an overdamped Brownian particle. The ratchetlike periodic system potentials, V (q ), obviously violate parity V (q ) = V (−q ). For such a ratchetlike potential one can readily calculate steady current flow J (F ) of a Brownian particle in the presence of an external field F . It turns out that J (F ) is not an odd function of F and, in general, J (F ) = −J (−F ). In other words, reversal of the external force may not lead to a reversed current of the same magnitude in sharp contrast to the case of a nonratchetlike (symmetric) periodic potential system where J (F )= −J (−F ) follows. From this general observation, in a ratchetlike potential, it can be easily concluded that on application of a zero time averaged periodic field, say F =F0 sinωt, one can obtain net unidirectional current. Of course, the direction and magnitude of the average velocity rnal field parameters, F0 and ω . A careful tuning of the relevant parameters may even result in the reversal of the macroscopic current [2]. This is the basic physics behind some of the physical models used to obtain current rectification in a periodic potential system. There are models, however, that do not use oscillating external fields. Instead, colored noise of zero average strength—-dichotomous, Ornstein-Uhlenbeck, Kangaroo processes,..., [3-5]—– is used to drive the Brownian particle to obtain macroscopic motion in a ratchetlike potential system. There are further interesting models where the potential barriers themselves are allowed to fluctuate, for instance, with finite time correlations between two states under the influence of a noise source. An example being an overdamped Brownian particle subjected to a ratchetlike periodic potential, where the ratchetlike saw-tooth potential is switched on to its full strength for time τon during which the Brownian particle slides down the potential slope to the bottom of the potential trench. At the end of τon , the system is put in the other (of f ) state during which the potential is set equal to a constant (say = 0) for an interval τof f and the particle executes force-free diffusive motion. At the end of τof f the system is put back in the on state for interval τon . This process of flipping of states is repeated ad-infinitum. If τof f is adjusted in such a way that by the end of τof f the diffusive motion just takes the particle out of the (now nonexistent) potential trench in the steeper slope direction (smaller distance) of the saw-tooth potential but fails to do so in the gentler slope direction (larger distance), the immediate next on interval will take the particle to the adjacent trench minimum in the steeper slope side of the saw-tooth potential. Repititon of such sequential flippings of states for a large number of times leads to a net unidirectional macroscopic current of the Brownian particle. It should be noted that a symmetrical nonratchetlike potential would, instead, have yielded symmetrical excursions of the Brownian particle and, hence, no net unidirectional
Enslaving random fluctuations in nonequlibrium systems
arXiv:cond-mat/9603103v1 14 Mar 1996
Mangal C. Mahato, T. P. Pareek and A. M. Jayannavar Institute of Physics, Sachivalaya Marg, Bhubaneswar-751005, INDIA.
Abstract Several physical models have recently been proposed to obtain unidirectional motion of an overdamped Brownian particle in a periodic potential system. The asymmetric ratchetlike form of the periodic potential and the presence of correlated nonequilibrium fluctuating forces are considered essential to obtain such a macroscopic motion in homogeneous systems. In the present work, instead, inhomogeneous systems are considered, wherein the friction coefficient and/or temperature could vary in space. We show that unidirectional motion can be obtained even in a symmetric nonratchetlike periodic potential system in the presence of white noise fluctuations. We consider four different cases of system inhomogeneity We argue that all these different models work under the same basic principle of alteration of relative stability of otherwise locally stable states in the presence of temperature inhomogeneity.