Showing papers on "Crossover published in 1989"
Proceedings Article•
01 Jun 1989
174 citations
TL;DR: A critical review is provided of recent articles on two-treatment crossover trials that purport to provide solutions to the problems caused by the existence of differential carryover effects, and recommends the two-period crossover design only when prior experience suggests that its assumptions are likely to be satisfied.
139 citations
Dissertation•
01 Jan 1989
TL;DR: This dissertation proposes a parallelized version of a genetic algorithm called the distributed genetic algorithm, which can achieve near-linear speedup over the traditional version of the algorithm, and discusses the issue of balancing exploration against exploitation in the distributed Genetic algorithm, by allowing different subpopulations to run with different parameters, so that someSubpopulations can emphasize exploration while others emphasize exploitation.
Abstract: The genetic algorithm is a general purpose, population-based search algorithm in which the individuals in the population represent samples from the set of all possibilities, whether they are solutions in a problem space, strategies for a game, rules in classifier systems, or arguments for problems in function optimization. The individuals evolve over time to form even better individuals by sharing and mixing their information about the space.
This dissertation proposes a parallelized version of a genetic algorithm called the distributed genetic algorithm, which can achieve near-linear speedup over the traditional version of the algorithm. This algorithm divides the large population into many equal-sized small subpopulations and runs the genetic algorithm on each subpopulation independently. Each subpopulation periodically selects some individuals and exchanges them with other subpopulations; the process known as migration.
The functions used to evaluate the performance of the distributed genetic algorithm and the traditional algorithm are called Walsh polynomials, which are based on Walsh functions. Walsh polynomials can be categorized into classes of functions with each class having a different degree of difficulty. By generating a large number of instances of the various classes, the performance difference between the distributed and traditional genetic algorithms can be analyzed statistically.
The first part of this dissertation examines the partitioned genetic algorithm, a version of the distributed genetic algorithm with no migration. The experiments on four different classes of Walsh polynomials show that the partitioned algorithm consistently outperforms the traditional algorithm as a function optimizer. This is because the physical subdivision of the population will allow each subpopulation to explore the space independently. Also, good individuals are more likely to be recognized in a smaller subpopulation than they would be in a large, diverse population. The second part of this research examines the effects of migration on the performance of the distributed genetic algorithm. The experiments show that with a moderate migration rate the distributed genetic algorithm finds better individuals than the traditional algorithm while maintaining a high overall fitness of the population.
Finally, this dissertation also discusses the issue of balancing exploration against exploitation in the distributed genetic algorithm, by allowing different subpopulations to run with different parameters, so that some subpopulations can emphasize exploration while others emphasize exploitation. The distributed algorithm is shown to be more robust than the traditional version: even when each subpopulation runs with different combinations of crossover and mutation rates, the distributed algorithm performs better than the traditional one.
136 citations
23 May 1989
TL;DR: The authors introduce genetic algorithms to solve the exact maximum-likelihood equations arising in a typical signal processing problem, and it is shown that it can outperform existing high-performance parameter estimation algorithms under difficult conditions.
Abstract: The authors introduce genetic algorithms (GAs) to solve the exact maximum-likelihood equations arising in a typical signal processing problem. GAs are an efficient and highly parallel way of maximizing complex, multimodal, multivariable functions. They are based on concepts borrowed from natural genetic evolution, involving reproduction, crossover, and mutation of a continuously evolving population of parameter estimates. An outline is presented of a GA for maximizing a Gaussian likelihood function, and it is shown that it can outperform existing high-performance parameter estimation algorithms under difficult conditions. >
69 citations
TL;DR: Hamiltonian dynamical systems with many degrees of freedom are investigated using symplectic map lattices and it is shown that anomalous diffusion exists only up to some crossover time beyond which the diffusion is normal.
Abstract: Hamiltonian dynamical systems with many degrees of freedom are investigated using symplectic map lattices. It is shown that anomalous diffusion exists only up to some crossover time beyond which the diffusion is normal. A diffusion constant, which is inversely proportional to the crossover time, exhibits a faster than any power-law dependence on the nonintegrability parameter, strongly suggesting the relevance of a bound by Nekhoroshev for Arnold diffusion. The motion in the standard mapping is also reexamined to show that the flicker noise is seen only down to some crossover frequency.
57 citations
TL;DR: In this paper, the authors provided an explanation for the observed nearly exclusive preference of the βαβ structure for forming a right-handed, rather than a left-handed crossover connection.
48 citations
TL;DR: The behavior of the largest Lyapunov exponent in a system of coupled chaotic maps shows that this crossover is remarkably sharp, and allows us to define a coherence length beyond which the system is effectively large.
Abstract: In this paper the crossover from ``small'' to ``large'' chaotic systems is studied. The behavior of the largest Lyapunov exponent in a system of coupled chaotic maps shows that this crossover is remarkably sharp, and allows us to define a coherence length beyond which the system is effectively large. Between the coherent chaos of the small system and the incoherent chaos (turbulence) of the large one there is a stable window starting at the linear instability point for the uniform chaotic state in which the lattice becomes effectively one dimensional. The scaling of the coherence length close to the onset of turbulence is investigated and compared to recent predictions.
48 citations
Proceedings Article•
20 Aug 1989TL;DR: A new crossover operator is introduced that eliminates the position dependent bias of the traditional crossover operator by shuffling the representation prior to applying crossover, and experimental results show that shuffle crossover outperforms traditional crossover on a suite of five function optimization problems.
Abstract: The traditional crossover operator used in genetic search exhibits a position-dependent bias called the dcfining-length bias. We show how this bias results in hidden biases that are difficult to anticipate and compensate for. We introduce a new crossover operator, shuffle crossover, that eliminates the position dependent bias of the traditional crossover operator by shuffling the representation prior to applying crossover. We also present experimental results that show that shuffle crossover outperforms traditional crossover on a suite of five function optimization problems.
TL;DR: It is demonstrated that crossover may have a profound effect on the statistical power of randomized studies and a method for predicting the effect of such crossover on statistical power is presented.
01 Dec 1989
TL;DR: This chapter reviews an approach to using genetic algorithms and other competition-based heuristics to learn reactive control rules given a simulation model of the environment implemented in a system called SAMUEL, which learns rules expressed in a high-level rule language.
Abstract: Publisher Summary This chapter reviews an approach to using genetic algorithms and other competition-based heuristics to learn reactive control rules given a simulation model of the environment implemented in a system called SAMUEL. SAMUEL learns rules expressed in a high-level rule language. The use of a symbolic rule language is intended to facilitate the incorporation of more traditional learning methods into the system where appropriate. SAMUEL consists of three major components: (1) a problem-specific module consisting of a World Model and its interfaces, (2) a performance module, and (3) a learning module. The Performance Module consists of CPS, a competition based production system that interacts with the World Model through the Sensor, Control, and Critic interfaces. CPS performs Matching, Conflict Resolution and Credit Assignment. The Learning Module uses a genetic algorithm to develop high performance strategies or reactive plans, expressed as a set of condition-action rules. Each strategy is evaluated by testing its performance in controlling the World Model through CPS. Genetic operators, such as crossover and mutation, produce plausible new strategies from high-performance precursors.
TL;DR: It is shown that the Gray Code can be used as the mapping function between the crossover network and a network which is equivalent to the modified data manipulator network, and the optical crossover network can be proposed as the optical interconnection fabric for many switching and computing applications.
Abstract: The crossover network is a multistage interconnection network that shows potential for use in optical computing and photonic switching applications. In this paper, a proof of topological equivalence between the crossover network and the modified data manipulator network is presented. It is shown that the Gray Code can be used as the mapping function between the crossover network and a network which is equivalent to the modified data manipulator network. Given the results of this proof, the optical crossover network can then be proposed as the optical interconnection fabric for many switching and computing applications.
01 Dec 1989
TL;DR: An incremental genetic algorithm which generates only one new member of the population and deletes only old one at a time thus equalizing the amount of computation and learning at each time interval is introduced.
Abstract: The genetic algorithm, operated in batch mode, evaluates the whole population in some environment and generates through selection, crossover and mutation a new population. In a real-time learning situation, where the population can only be evaluated sequentially, much of the computation and all of the learning is concentrated into one time interval between the evaluation of the last member of the old population and the generation of the first member of the new. This paper introduces an incremental genetic algorithm which generates only one new member of the population and deletes only old one at a time thus equalizing the amount of computation and learning at each time interval. It then compares the performance of the incremental and non-incremental genetic algorithms and of a rule based system for optimising combustion on ten simulations of multiple burner installations.
TL;DR: In this paper, a nonparametric approach to the analysis of the two-treatment, two-period, four-sequence crossover model is presented, using an approach similar to that of Koch (1972, Biometrics 28, 577-584).
Abstract: Owing to the estimability problems in the two-treatment, two-period, two-sequence crossover model, the four-sequence design (AB, BA, AA, BB) has gained popularity. This paper presents a nonparametric approach to the analysis of the two-treatment, two-period, four-sequence crossover model, using an approach similar to that of Koch (1972, Biometrics 28, 577-584). After forming differences within an experimental unit, the Benard-van Elteren test (Benard and van Elteren, 1953, Proc. Konink. Nederlandse Akad van Wetenshappen, Series A 56, 358-369) uses the transformed data to construct a test of the equality of the direct effects of the treatment. A numerical example is presented to illustrate the methodology.
TL;DR: In this paper, two methods of estimating the autocorrelation coefficient are considered, restricted maximum likelihood and conditional profile likelihood, for three-period and four-period crossover experiments.
Abstract: SUMMARY Continuous data from a crossover experiment are likely to be correlated within each subject and an efficient analysis may have to estimate the dependence in the error term. Assuming that the error term follows a first-order autoregression, and that subject effects are taken to be fixed, two methods of estimating the autocorrelation coefficient are considered. One method applies restricted maximum likelihood, the other uses conditional profile likelihood. Simulation studies for three-period and four-period designs and a numerical example are presented.
14 Nov 1989
TL;DR: The author describes how the genetic algorithm can be operated in interactive mode, generating only one new member of the population and deleting only one old one at a time, thus equalizing the amount of computation and learning at each time interval.
Abstract: The genetic algorithm, operated in batch mode, evaluates the whole population in some environment and generates a new population through selection, crossover, and mutation. In a real-time learning situation, where the population can be evaluated only sequentially, much of the computation and all of the learning is thus concentrated into one time interval between the evaluation of the last member of the old population and the generation of the first member of the new. The author describes how the genetic algorithm can be operated in interactive mode, generating only one new member of the population and deleting only one old one at a time, thus equalizing the amount of computation and learning at each time interval. He then compares the performance of the two modes of operating the algorithm and of a rule-based system for optimizing combustion on ten simulations of multiple burner installations, giving a statistical analysis of the results obtained. >
Patent•
08 May 1989TL;DR: In this article, a crossover network is implemented using two-dimensional arrays of nodes, arranged in rows and columns and there are a plurality of link stages interconnecting successive arrays.
Abstract: A crossover network implemented using two-dimensional arrays of nodes. The network is a perfect shuffle equivalent network because it is topologically equivalent to a crossover network of one-dimensional arrays of nodes. The two-dimensional arrays are arranged in rows and columns and there are a plurality of link stages interconnecting successive arrays. The network is implemented efficiently in free space optics because the network topology requires optical crossovers in some link stages that interconnect only nodes in the same column of successive arrays and optical crossovers in the other link stages that interconnect only nodes in the same row of successive arrays.
TL;DR: In this paper, the critical exponents, eigenoperators and the phase diagram of the N=2 Ashkin-Teller (AT) model were studied in two dimensions using a Monte Carlo renormalisation group.
Abstract: The authors study the critical exponents, eigenoperators and the phase diagram of the N=2 Ashkin-Teller (AT) model in two dimensions using a Monte Carlo renormalisation group. They pay special attention to the marginal operator responsible for the non-universal behaviour. The crossover exponent is measured and used to numerically calculate the thermal exponent of the eight-vertex model and of the AT model near the Potts-four-point, giving better results than a direct determination, since the crossover operator is not affected by the additional AT marginal operators.
01 Dec 1989
TL;DR: Experimental results show that genetic search using a multiple representation – Gray-coded mutation and binary-coded crossover – outperforms search using just one representation.
Abstract: Previously we demonstrated that Gray code is superior to binary code for genetic search in domains with ordered parameters. Since then we have determined that Gray code is better because it does not exhibit a counter-productive hidden bias that emerges when binary coding is used with the mutation search operator. But analysis suggests that crossover, the genetic algorithm's (GA) other search operator, should perform better with the binary representation. We present experimental results that show that genetic search using a multiple representation – Gray-coded mutation and binary-coded crossover – outperforms search using just one representation. We believe other search methods that use multiple search heuristics may also benefit from using multiple representations, one tuned for each heuristic.
TL;DR: Work is proposed to create a classifier system that can draw more effectively on the knowledge that is available in the scheduling domain and the best GA is compared with a neural network based optimizer.
TL;DR: In this article, the dielelctric breakdown model (DBM) is generalized to include lower cutoffs, which prevent growth at low fields, and multifractal theory is shown to provide quantitative predictions for the crossover from the usual DBM patterns (at small finger sizes) to a new, spiky behavior (at large sizes).
Abstract: The dielelctric breakdown model (DBM) is generalized to include lower cutoffs, which prevent growth at low fields. The new models may represent realistic situations in some DBM and some viscous fingering experiments. Multifractal theory is shown to provide quantitative predictions for the crossover from the usual DBM patterns (at small finger sizes) to a new, spiky, behavior (at large sizes). For one of the models, the theory also predicts when growth will stop. The predicted crossover scaling function is confirmed by numerical simulations.
TL;DR: The results show that by exploiting the crossover effect it is possible to obtain essentially pure benzoic acid (>99.5% purity) with a single temperature cycle.
Abstract: This paper presents the experimental technique and data acquired in the evaluation of a separation process in the crossover region of a model supercritical fluid system consisting of benzoic acid, 1–10 decanediol, and carbon dioxide. The results show that by exploiting the crossover effect it is possible to obtain essentially pure benzoic acid (>99.5% purity) with a single temperature cycle. A number of other issues are also discussed including the tradeoff between yield and purity and the manner in which the process variables are specified, which involves the interplay of experiment and model calculations.
01 Jan 1989
TL;DR: The authors describe the thermal conductivity, the thermal diffusivity and the viscosity of a number of fluids using the crossover model for the transport properties of fluids by Olchowy and Sengers with a special emphasis on the region in which the critical enhancement gives a sizeable contribution.
Abstract: Transport properties of fluids in the critical region can be represented by a sum of two terms: the critical enhancement, which asymptotically close to the critical point obeys a power law in the reduced temperature, and the background part, which is independent of critical fluctuations. Although the critical enhancement is dominant only in the vicinity of the critical point and vanishes far away from it, its contributions are considerable over a wide range of temperatures and densities. Using the crossover model for the transport properties of fluids by Olchowy and Sengers the authors describe the thermal conductivity, the thermal diffusivity and the viscosity of a number of fluids. The results are presented with a special emphasis on the region in which the critical enhancement gives a sizeable contribution.
Journal Article•
TL;DR: In this paper, a new method is presented for the design and evaluation of loudspeaker crossover filters, which can be derived from conventional filters or based on a measured filter from a reference (favorable) system.
Abstract: A new method is presented for the design and evaluation of loudspeaker crossover filters. The desired system characteristic can be prescribed by a (complex) acoustic transfer function rather than an electrical one only. It may be derived from conventional filters or based on a measured filter from a reference (favorable) system. Double blind listening tests are performed to verify subjectively the similarity between the reference system and its experimental counterpart. The drivers of the experimental loudspeaker are preceded by digital filters, enabling the imitation of several different favorable loudspeakers. Multidimensional scaling techniques are applied to represent the results of the listening tests. These results affirm the strength of the design method
TL;DR: An efficient algorithm has been devised to compute the crossovers in satellite altimetry with the ability to predict (by analytical means) the crossover coordinates to within 6 km and 1 sec of the true values.
Abstract: An efficient algorithm has been devised to compute the crossovers in satellite altimetry The significance of the crossovers is twofold First, they are needed to perform the crossover adjustment to remove the orbit error Secondly, they yield important insight into oceanic variability Nevertheless, there is no published algorithm to make this very time consuming task easier, which is the goal of this report The success of the algorithm is predicated on the ability to predict (by analytical means) the crossover coordinates to within 6 km and 1 sec of the true values Hence, only one interpolation/extrapolation step on the data is needed to derive the crossover coordinates in contrast to the many interpolation/extrapolation operations usually needed to arrive at the same accuracy level if deprived of this information
TL;DR: It is shown here that although the angular structure itself can be obtained from the crossover exponents, in contrast to multifractals, the asymptotic form of this angular-structure factor is described by the Legendre transform of the (analytically continued) infinite set ofrossover exponents only in the special case of two dimensions.
Abstract: For transitions described by the XY model, it has been shown that each Fourier component of the angular-structure factor is controlled by a different crossover exponent. Each crossover exponent is associated with a particular symmetry-breaking field. It is shown here that although, as for multifractals, the angular structure itself can be obtained from the crossover exponents, in contrast to multifractals, the asymptotic form of this angular-structure factor is described by the Legendre transform of the (analytically continued) infinite set of crossover exponents only in the special case of two dimensions.
08 Feb 1989
TL;DR: The crossover network holds an isomorphic relationship to the perfect shuffle and the Banyan network and using a Michelson-based optical setup the crossover network can be implemented with a higher efficiency in terms of the number of connections and conservation of light.
Abstract: The crossover network is introduced as a new interconnection network for optical digital computers. It belongs to the class of multistage interconnection networks such as the perfect shuffle and the Banyan network. For connecting N = 2k input ports to N output ports (where k is an arbitrary integer) log2(N) stages are required. The crossover network holds an isomorphic relationship to the perfect shuffle and the Banyan network. However, using a Michelson-based optical setup the crossover network can be implemented with a higher efficiency in terms of the number of connections and conservation of light. Experimental results are presented where more than 1000 interconnections are implemented.© (1989) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
TL;DR: In this paper, a theoretical procedure for the temperature and wave-number-dependent order-parameter correlation function is presented, which is based on an analysis of the renormalization-group equation.
Abstract: A theoretical procedure is presented for the temperature- and wave-number-dependent order-parameter correlation function which incorporates singular scaling behavior near the critical point and the analytic Ornstein-Zernike function away from the critical point. The theory is based on an analysis of the renormalization-group equation. Near the critical point, the crossover formalism produces a Wegner correction-to-scaling term for the correlation function.