GA Solver en Matlab

1
GA Solver en Matlab
2
GA Solver

• X GA(FITNESSFCN,NVARS) finds the minimum of
  FITNESSFCN using
• GA. NVARS is the dimension (number of design
  variables) of the
• FITNESSFCN. FITNESSFCN accepts a vector X of
  size 1-by-NAVRS,
• and returns a scalar evaluated at X.
• X GA(FITNESSFCN,NAVRS,OPTIONS) finds the
  minimum for
• FITNESSFCN with the default optimization
  parameters replaced by values
• in the structure OPTIONS. OPTIONS can be
  created with the GAOPTIMSET
• function.
• X GA(PROBLEM) finds the minimum for
  PROBLEM. PROBLEM is a structure
• that has the following fields
• fitnessfcn ltFitness Functiongt
• nvars ltNumber of design variablesgt
• options ltOptions structure created
  with GAOPTIMSETgt
• randstate ltOptional field to reset rand
  stategt
• randnstate ltOptional field to reset
  randn stategt

3
GA Solver

• X, FVAL GA(FITNESSFCN, ...) returns FVAL,
  the value of the fitness
• function FITNESSFCN at the solution X.
• X,FVAL,REASON GA(FITNESSFCN, ...) returns
  the REASON for stopping.
• X,FVAL,REASON,OUTPUT GA(FITNESSFCN, ...)
  returns a
• structure OUTPUT with the following
  information
• randstate ltState of the function
  RAND used before GA startedgt
• randnstate ltState of the function
  RANDN used before GA startedgt
• generations ltTotal generations,
  excluding HybridFcn iterationsgt
• funccount ltTotal function
  evaluationsgt
• message ltGA termination messagegt
• X,FVAL,REASON,OUTPUT,POPULATION
  GA(FITNESSFCN, ...) returns the final
• POPULATION at termination.
• X,FVAL,REASON,OUTPUT,POPULATION,SCORES
  GA(FITNESSFCN, ...) returns the
• SCORES of the final POPULATION.

4
GA Solver

• There are several steps to the GA
• population generation
• scoring
• loop
• fitness
• scaling
• selection
• crossover
• mutation
• scoring
• migration
• output
• termination testing
• end loop
• Each of these steps can be controlled by the
  options structure created
• by GAOPTIMSET.

5
GA Solver

• Example
• Minimize 'rastriginsfcn' fitness function
  of numberOfVariables 2
• x ga(_at_rastriginsfcn,2)
• Display plotting functions while GA
  minimizes
• options gaoptimset('PlotFcns',...
• _at_gaplotbestf,_at_gaplotbestindiv,_at_gaplotexp
  ectation,_at_gaplotstopping)
• x,fval,reason,output
  ga(_at_rastriginsfcn,2,options)

6
Opciones del algoritmo I

• GAOPTIMSET Create a genetic algorithm options
  structure.
• GAOPTIMSET returns a listing of the fields in
  the options structure as
• well as valid parameters and the default
  parameter.
• OPTIONS GAOPTIMSET('PARAM',VALUE) creates a
  structure with the
• default parameters used for all PARAM not
  specified, and will use the
• passed argument VALUE for the specified
  PARAM.
• OPTIONS GAOPTIMSET('PARAM1',VALUE1,'PARAM2',
  VALUE2,....) will create a
• structure with the default parameters used
  for all fields not specified.
• Those FIELDS specified will be assigned the
  corresponding VALUE passed,
• PARAM and VALUE should be passed as pairs.
• OPTIONS GAOPTIMSET(OLDOPTS,'PARAM',VALUE)
  will create a structure named
• OPTIONS. OPTIONS is created by altering the
  PARAM specified of OLDOPTS to
• become the VALUE passed.
• OPTIONS GAOPTIMSET(OLDOPTS,'PARAM1',VALUE1,'
  PARAM2',VALUE2,...) will

7
Opciones del algoritmo I

• PopulationType - The type of Population
  being entered
• 'bitstring' 'custom'
  'doubleVector'
• PopInitRange - Initial range of values
  a population may have
• Matrix 01
• PopulationSize - Positive scalar
  indicating the number of individuals
• positive scalar 20
  
• EliteCount - Number of best
  individuals that survive to next
• generation without any
  change
• positive scalar 2
• CrossoverFraction - The fraction of genes
  swapped between individuals
• positive scalar 0.8
  
• MigrationDirection - Direction that fittest
  individuals from the various
• sub-populations may
  migrate to other sub-populations
• 'both' 'forward'
• MigrationInterval - The number of
  generations between the migration of
• the fittest individuals
  to other sub-populations
• positive scalar 20
  
• MigrationFraction - Fraction of those
  individuals scoring the best

8
Opciones del algoritmo I

• TimeLimit - The total time (in seconds)
  allowed for simulation
• positive scalar INF
  
• FitnessLimit - The lowest allowed
  score
• scalar -Inf
• StallGenLimit - If after this number of
  generations there is
• no improvement, the
  simulation will end
• positive scalar 50
  
• StallTimeLimit - If after this many
  seconds there is no improvement,
• the simulation will end
• positive scalar 20
  
• InitialPopulation - The initial population
  used in seeding the GA
• algorithm
• Matrix
• InitialScores - The initial scores used
  to determine fitness used
• in seeding the GA
  algorithm
• column vector
• positive scalar 1
• CreationFcn - Function used to
  generate initial population
• _at_gacreationuniform

9
Opciones del algoritmo I

• SelectionFcn - Function used in selecting
  parents for next generation
• _at_selectionremainder
  _at_selectionrandom
• _at_selectionroulette
  _at_selectiontournament
• _at_selectionstochunif
• CrossoverFcn - Function used to do
  crossover
• _at_crossoverheuristic
  _at_crossoverintermediate
• _at_crossoversinglepoint
  _at_crossovertwopoint
• _at_crossoverscattered
• MutationFcn - Function used in
  mutating genes
• _at_mutationuniform
  _at_mutationgaussian
• HybridFcn - Another optimization
  function to be used once GA
• has normally terminated
  (for whatever reason)
• _at_fminsearch
  _at_patternsearch _at_fminunc
• Display - Level of display
• 'off' 'iter'
  'diagnose' 'final'
• OutputFcns - Function(s) called in
  every generation. This is more
• general than PlotFcns.
• _at_gaoutputgen
• PlotFcns - Function(s) used in
  plotting various quantities