分布式并行遗传算法

Ref.
[55] [68] [73] [51] [23] --[59]
Year
1987 1989 1990 1991 1991 1992 1993
Main Characteristics
Generational islands on an Intel iPSC hypercube (8 CPUs). Migrate the best. Dynamic Top. Distributed populations. Good results with 20% of population migrants every 20 generations Steady-State islands with ranked selection and reduced surrogate crossover Sub-populations in a circular ladder-like 2-D topology. Migrate the best, local hill-climbing Made for nCUBE2. This is the parallel extension of the well-known simple GA of Goldberg Made for the CM-200. This places one individual in every CPU Targeted for MIMD machines and written in a very high and flexible description language
TABLE Ⅴ DETIALS OF SEVEL PARALLEL GAs
Parallel GA ASPARAGOS CoPDEB DGENESIS 1.0 ECO-GA EnGENEer GALOPPS 3.1 GAMAS GAME GAucsd 1.2 / 1.4 GDGA GENITOR II HSDGA PARAGENESIS PeGAsuS Kind of Parallelism Fine grain. Applies Hill-Climbing if no improvement Coarse grain. Every sub-pop. applies different operators Coarse grain with migrations among sub-populations Fine grain. One of the first of its class Global parallelization (parallel evaluations) Coarse grain. A very portable software Coarse grain. Uses 4 species of strings (nodes) Parallel version not available yet. Object Oriented Distributes the experiments over the network (not parallel) Coarse Grain. Admits explicit exploration/exploitation Coarse grain. Interesting crossover operator Hierarchical coarse and fine grain GA. Uses E. S. Global P. & coarse grain. Made for the CM-200 (1 ind.-1 cpu) Coarse or fine grain. High-level programming. MIMD Topology Ladder Full Connected Any Desired Grid Master / Slave Any Desired Fixed Hierarchy Any Desired <sequential> Hierarchy Ring Ring, Tree, Star, ... Local sel. (seq.) Multiple Present Applications TSP Func. Opt. and ANN’s Function Optimization Function Optimization Various Func. Opt. and Transport ANN, Func. Opt., ... TSP, Func. Opt., ... <same as GENESIS> Func. Opt. (FP-genes.) Func. Opt. and ANN’s Function Optimization Function Optimization Teaching and Func. Opt.
we now give an extensive classification of sequential and parallel Gas into three major categories according to their specific objectives. Application Oriented: These are black-box systems designed to hide the details of GAs and help the user in developing applications for specific domains. Usually they are menu-driven, and easily parameterizable. Algorithm Oriented: Based on specific algorithms. The source code is available in order to provide their easy incorporation into new applications. This class may be further sub-divided into: - Algorithm Specific: They contain one single GA. - Algorithm Libraries: They support a group of algorithms in a library format. They are highly parameterized and contain many different operators to help future applications. Tool Kits: These are flexible environments for programming a range of different GAs and applications. They can be sub-divided into: - Educational: Used for introducing GA concepts to novice users . The basic techniques to track executions and results during the evolution are easily managed. - General Purpose: Useful for modifying, developing, and supervising a wide range of operators, algorithms and applications .
5.Classification of Parallel and Sequential GAs
TABLE Ⅲ OVERVIEW OF PARALLEL DISTRIBUTED GAs BY YEAR
Par. dGA
PGA dGA GENITOR II PGA SGA-cube
PARAGENESIS
Injection island GA with hierarchical heterogeneous nodes and asynchronous migrations 128 steady-state islands on an IBM SP1 machine of 128 nodes. 2-D toroidal mesh. mr=1 Free topology, flexible migration, and policies for selection. Implemented with sockets (UDP) Very flexible. Implemented with PVM and comprising a large number of operators Synchronous. Simulated on one processor. Generational. Uses Fuzzy crossover and FP genes Every island uses its own probabilities for mutation, crossover, and specialized operators
Application Domain
Parallel training of artificial neural networks, fuzzy logic controllers, and communication protocols Synthesis of VLSI circuits Function optimization Set partitioning problem Graph partitioning problem Constraint Optimization, reordering problems, ... Traveling salesperson problem (TSP), function optimization Distributing the computing load onto a set of processing nodes The file allocation problem, XOR neural network, sine envelope sine wave function Systems modeling, protein tertiary structure prediction, and two-dimensional bin packing problems
Multiple
Master / Slave Any Desired Hypercube
Knapsack and Func. Opt.
Function Optimization Research and Business Function Optimization
5.Classification of Parallel and Sequential GAs
PeGAsuS
GAMAS
iiGA SP1-GA DGENESIS GALOPPS GDGA CoPDEB
[56]
[44] [42] [47] [30] [34] [2]
1994
1994 1994 1994 1996 1996 1996
Uses 4 very heterogeneous species (islands) and quite specialized migrations and genotypes
A Survey of Parallel Distributed Genetic Algorithms
5.Classification of Parallel and Sequential GAs 6.Technical Issues in Parallel Distributed GAs 7.Implementation Issues 8.Concluding Remarks
[68]
[72]
Walsh polynomials
Optimization of the connection weights of neuralFra Baidu biblioteknetworks (XOR, bin-adder, ...), and function optimization
5.Classification of Parallel and Sequential GAs
New genotypes and operators are being designed for dealing with constraint problems and combinatorial optimization. Besides that, the importance of cellular GAs is also growing due to recent studies in in which the search is still enhanced due to the existence of neighborhood like spatial dispositions.
5.Classification of Parallel and Sequential GAs
TABLE Ⅳ SOME APPLICATIONS OF PARALLEL DISTRIBUTED GAs
Reference
[7] [19] [31] [42] [44] [49] [51] [53] [56] [66]
PGA 2.5
PGAPack RPL2 SGA-Cube
Spatially structured selection. Allows migrations
Global parallelization (parallel evaluations) Coarse and fine grain. Very flexible and open to new models Global parallelization. Made for the nCUBE 2