class MPIMigrator(object): selector = None """ This is the function slot for the selection method if you want to change the default selector, you must do this: :: migration_scheme.selector.set(Selectors.GRouletteWheel) """ def __init__(self): self.myself = None self.selector = FunctionSlot("Selector") self.GAEngine = None self.nMigrationRate = Consts.CDefGenMigrationRate self.nIndividuals = Consts.CDefMigrationNIndividuals self.nReplacement = Consts.CDefGenMigrationReplacement self.comm = MPI.COMM_WORLD self.pid = self.comm.rank self.best_selector = Selectors.GRankSelector #now this is fixed if self.pid == 0: self.source = self.comm.size - 1 else: self.source = self.comm.rank - 1 self.dest = (self.comm.rank +1) % (self.comm.size) self.all_stars = None def isReady(self): """ Returns true if is time to migrate """ if self.GAEngine.getCurrentGeneration() == 0: return False if self.GAEngine.getCurrentGeneration() % self.nMigrationRate == 0: return True else: return False def getNumReplacement(self): """ Return the number of individuals that will be replaced in the migration process """ return self.nReplacement def setNumReplacement(self, num_individuals): """ Return the number of individuals that will be replaced in the migration process :param num_individuals: the number of individuals to be replaced """ self.nReplacement = num_individuals def getNumIndividuals(self): """ Return the number of individuals that will migrate :rtype: the number of individuals to be replaced """ return self.nIndividuals def setNumIndividuals(self, num_individuals): """ Set the number of individuals that will migrate :param num_individuals: the number of individuals """ self.nIndividuals = num_individuals def setMigrationRate(self, generations): """ Sets the generation frequency supposed to migrate and receive individuals. :param generations: the number of generations """ self.nMigrationRate = generations def getMigrationRate(self): """ Return the the generation frequency supposed to migrate and receive individuals :rtype: the number of generations """ return self.nMigrationRate def setGAEngine(self, ga_engine): """ Sets the GA Engine handler """ self.GAEngine = ga_engine def start(self): """ Initializes the migration scheme """ pass def stop(self): """ Stops the migration engine """ pass def getGroupName(self): """ Gets the group name .. note:: all islands of evolution which are supposed to exchange individuals, must have the same group name. """ return self.groupName def setGroupName(self, name): """ Sets the group name :param name: the group name .. note:: all islands of evolution which are supposed to exchange individuals, must have the same group name. """ self.groupName = name def select(self): """ Pickes an individual from population using specific selection method :rtype: an individual object """ if self.selector.isEmpty(): return self.GAEngine.select(popID=self.GAEngine.currentGeneration) else: for it in self.selector.applyFunctions(self.GAEngine.internalPop, popID=self.GAEngine.currentGeneration): return it def selectPool(self, num_individuals): """ Select num_individuals number of individuals and return a pool :param num_individuals: the number of individuals to select :rtype: list with individuals """ pool = [self.select() for _ in xrange(num_individuals)] return pool def gather_bests(self): ''' Collect all the best individuals from the various populations. The result is stored in process 0 ''' best_guy = self.best_selector(self.GAEngine.internalPop, popID=self.GAEngine.currentGeneration) self.all_stars = self.comm.gather(sendobj = best_guy, root = 0) def exchange(self): """ This is the main method, is where the individuals are exchanged """ if not self.isReady(): return pool_to_send = self.selectPool(self.getNumIndividuals()) pool_received = self.comm.sendrecv(sendobj=pool_to_send, dest=self.dest, sendtag=0, recvobj=None, source=self.source, recvtag=0) population = self.GAEngine.getPopulation() pool = pool_received for i in xrange(self.getNumReplacement()): if len(pool) <= 0: break choice = rand_choice(pool) pool.remove(choice) # replace the worst population[len(population)-1-i] = choice self.gather_bests()
class GSimpleGA: """ GA Engine Class - The Genetic Algorithm Core Example: >>> ga = GSimpleGA.GSimpleGA(genome) >>> ga.selector.set(Selectors.GRouletteWheel) >>> ga.setGenerations(120) >>> ga.terminationCriteria.set(GSimpleGA.ConvergenceCriteria) :param genome: the :term:`Sample Genome` :param interactiveMode: this flag enables the Interactive Mode, the default is True :param seed: the random seed value .. note:: if you use the same random seed, all the runs of algorithm will be the same """ selector = None """ This is the function slot for the selection method if you want to change the default selector, you must do this: :: ga_engine.selector.set(Selectors.GRouletteWheel) """ stepCallback = None """ This is the :term:`step callback function` slot, if you want to set the function, you must do this: :: def your_func(ga_engine): # Here you have access to the GA Engine return False ga_engine.stepCallback.set(your_func) now *"your_func"* will be called every generation. When this function returns True, the GA Engine will stop the evolution and show a warning, if is False, the evolution continues. """ terminationCriteria = None """ This is the termination criteria slot, if you want to set one termination criteria, you must do this: :: ga_engine.terminationCriteria.set(GSimpleGA.ConvergenceCriteria) Now, when you run your GA, it will stop when the population converges. There are those termination criteria functions: :func:`GSimpleGA.RawScoreCriteria`, :func:`GSimpleGA.ConvergenceCriteria`, :func:`GSimpleGA.RawStatsCriteria`, :func:`GSimpleGA.FitnessStatsCriteria` But you can create your own termination function, this function receives one parameter which is the GA Engine, follows an example: :: def ConvergenceCriteria(ga_engine): pop = ga_engine.getPopulation() return pop[0] == pop[len(pop)-1] When this function returns True, the GA Engine will stop the evolution and show a warning, if is False, the evolution continues, this function is called every generation. """ def __init__(self, genome, owner, seed=None, interactiveMode=True): """ Initializator of GSimpleGA """ if seed: random.seed(seed) if type(interactiveMode) != BooleanType: Util.raiseException( "Interactive Mode option must be True or False", TypeError) if not isinstance(genome, GenomeBase): Util.raiseException("The genome must be a GenomeBase subclass", TypeError) self.internalPop = GPopulation(genome) self.nGenerations = Consts.CDefGAGenerations self.pMutation = Consts.CDefGAMutationRate self.pCrossover = Consts.CDefGACrossoverRate self.nElitismReplacement = Consts.CDefGAElitismReplacement self.setPopulationSize(Consts.CDefGAPopulationSize) self.minimax = Consts.minimaxType["maximize"] self.elitism = True self.owner = owner ## added 12/15 by Peter Graf, so GA can evaluate constraints ## and, now (4/16), so we can save let the owner save the state # Adapters self.dbAdapter = None self.migrationAdapter = None self.time_init = None self.interactiveMode = interactiveMode self.interactiveGen = -1 self.GPMode = False self.selector = FunctionSlot("Selector") self.stepCallback = FunctionSlot("Generation Step Callback") self.terminationCriteria = FunctionSlot("Termination Criteria") self.selector.set(Consts.CDefGASelector) self.allSlots = [ self.selector, self.stepCallback, self.terminationCriteria ] self.internalParams = {} self.currentGeneration = 0 # GP Testing for classes in Consts.CDefGPGenomes: if isinstance(self.internalPop.oneSelfGenome, classes): self.setGPMode(True) break logging.debug("A GA Engine was created, nGenerations=%d", self.nGenerations) def setGPMode(self, bool_value): """ Sets the Genetic Programming mode of the GA Engine :param bool_value: True or False """ self.GPMode = bool_value def getGPMode(self): """ Get the Genetic Programming mode of the GA Engine :rtype: True or False """ return self.GPMode def __call__(self, *args, **kwargs): """ A method to implement a callable object Example: >>> ga_engine(freq_stats=10) .. versionadded:: 0.6 The callable method. """ if kwargs.get("freq_stats", None): return self.evolve(kwargs.get("freq_stats")) else: return self.evolve() def setParams(self, **args): """ Set the internal params Example: >>> ga.setParams(gp_terminals=['x', 'y']) :param args: params to save ..versionaddd:: 0.6 Added the *setParams* method. """ self.internalParams.update(args) def getParam(self, key, nvl=None): """ Gets an internal parameter Example: >>> ga.getParam("gp_terminals") ['x', 'y'] :param key: the key of param :param nvl: if the key doesn't exist, the nvl will be returned ..versionaddd:: 0.6 Added the *getParam* method. """ return self.internalParams.get(key, nvl) def setInteractiveGeneration(self, generation): """ Sets the generation in which the GA must enter in the Interactive Mode :param generation: the generation number, use "-1" to disable .. versionadded::0.6 The *setInteractiveGeneration* method. """ if generation < -1: Util.raiseException("Generation must be >= -1", ValueError) self.interactiveGen = generation def getInteractiveGeneration(self): """ returns the generation in which the GA must enter in the Interactive Mode :rtype: the generation number or -1 if not set .. versionadded::0.6 The *getInteractiveGeneration* method. """ return self.interactiveGen def setElitismReplacement(self, numreplace): """ Set the number of best individuals to copy to the next generation on the elitism :param numreplace: the number of individuals .. versionadded:: 0.6 The *setElitismReplacement* method. """ if numreplace < 1: Util.raiseException("Replacement number must be >= 1", ValueError) self.nElitismReplacement = numreplace def setInteractiveMode(self, flag=True): """ Enable/disable the interactive mode :param flag: True or False .. versionadded: 0.6 The *setInteractiveMode* method. """ if type(flag) != BooleanType: Util.raiseException( "Interactive Mode option must be True or False", TypeError) self.interactiveMode = flag def __repr__(self): """ The string representation of the GA Engine """ ret = "- GSimpleGA\n" ret += "\tGP Mode:\t\t %s\n" % self.getGPMode() ret += "\tPopulation Size:\t %d\n" % (self.internalPop.popSize, ) ret += "\tGenerations:\t\t %d\n" % (self.nGenerations, ) ret += "\tCurrent Generation:\t %d\n" % (self.currentGeneration, ) ret += "\tMutation Rate:\t\t %.2f\n" % (self.pMutation, ) ret += "\tCrossover Rate:\t\t %.2f\n" % (self.pCrossover, ) ret += "\tMinimax Type:\t\t %s\n" % (Consts.minimaxType.keys()[ Consts.minimaxType.values().index(self.minimax)].capitalize(), ) ret += "\tElitism:\t\t %s\n" % (self.elitism, ) ret += "\tElitism Replacement:\t %d\n" % (self.nElitismReplacement, ) ret += "\tDB Adapter:\t\t %s\n" % (self.dbAdapter, ) for slot in self.allSlots: ret += "\t" + slot.__repr__() ret += "\n" return ret def setMultiProcessing(self, flag=True, full_copy=False): """ Sets the flag to enable/disable the use of python multiprocessing module. Use this option when you have more than one core on your CPU and when your evaluation function is very slow. Pyevolve will automaticly check if your Python version has **multiprocessing** support and if you have more than one single CPU core. If you don't have support or have just only one core, Pyevolve will not use the **multiprocessing** feature. Pyevolve uses the **multiprocessing** to execute the evaluation function over the individuals, so the use of this feature will make sense if you have a truly slow evaluation function (which is commom in GAs). The parameter "full_copy" defines where the individual data should be copied back after the evaluation or not. This parameter is useful when you change the individual in the evaluation function. :param flag: True (default) or False :param full_copy: True or False (default) .. warning:: Use this option only when your evaluation function is slow, so you'll get a good tradeoff between the process communication speed and the parallel evaluation. The use of the **multiprocessing** doesn't means always a better performance. .. note:: To enable the multiprocessing option, you **MUST** add the *__main__* check on your application, otherwise, it will result in errors. See more on the `Python Docs <http://docs.python.org/library/multiprocessing.html#multiprocessing-programming>`__ site. .. versionadded:: 0.6 The `setMultiProcessing` method. """ if type(flag) != BooleanType: Util.raiseException("Multiprocessing option must be True or False", TypeError) if type(full_copy) != BooleanType: Util.raiseException( "Multiprocessing 'full_copy' option must be True or False", TypeError) self.internalPop.setMultiProcessing(flag, full_copy) def setMigrationAdapter(self, migration_adapter=None): """ Sets the Migration Adapter .. versionadded:: 0.6 The `setMigrationAdapter` method. """ self.migrationAdapter = migration_adapter if self.migrationAdapter is not None: self.migrationAdapter.setGAEngine(self) def setDBAdapter(self, dbadapter=None): """ Sets the DB Adapter of the GA Engine :param dbadapter: one of the :mod:`DBAdapters` classes instance .. warning:: the use the of a DB Adapter can reduce the speed performance of the Genetic Algorithm. """ if (dbadapter is not None) and (not isinstance(dbadapter, DBBaseAdapter)): Util.raiseException( "The DB Adapter must be a DBBaseAdapter subclass", TypeError) self.dbAdapter = dbadapter def setPopulationSize(self, size): """ Sets the population size, calls setPopulationSize() of GPopulation :param size: the population size .. note:: the population size must be >= 2 """ if size < 2: Util.raiseException("population size must be >= 2", ValueError) self.internalPop.setPopulationSize(size) def setSortType(self, sort_type): """ Sets the sort type, Consts.sortType["raw"]/Consts.sortType["scaled"] Example: >>> ga_engine.setSortType(Consts.sortType["scaled"]) :param sort_type: the Sort Type """ if sort_type not in Consts.sortType.values(): Util.raiseException("sort type must be a Consts.sortType type", TypeError) self.internalPop.sortType = sort_type def setMutationRate(self, rate): """ Sets the mutation rate, between 0.0 and 1.0 :param rate: the rate, between 0.0 and 1.0 """ if (rate > 1.0) or (rate < 0.0): Util.raiseException("Mutation rate must be >= 0.0 and <= 1.0", ValueError) self.pMutation = rate def setCrossoverRate(self, rate): """ Sets the crossover rate, between 0.0 and 1.0 :param rate: the rate, between 0.0 and 1.0 """ if (rate > 1.0) or (rate < 0.0): Util.raiseException("Crossover rate must be >= 0.0 and <= 1.0", ValueError) self.pCrossover = rate def setGenerations(self, num_gens): """ Sets the number of generations to evolve :param num_gens: the number of generations """ if num_gens < 1: Util.raiseException("Number of generations must be >= 1", ValueError) self.nGenerations = num_gens def getGenerations(self): """ Return the number of generations to evolve :rtype: the number of generations .. versionadded:: 0.6 Added the *getGenerations* method """ return self.nGenerations def getMinimax(self): """ Gets the minimize/maximize mode :rtype: the Consts.minimaxType type """ return self.minimax def setMinimax(self, mtype): """ Sets the minimize/maximize mode, use Consts.minimaxType :param mtype: the minimax mode, from Consts.minimaxType """ if mtype not in Consts.minimaxType.values(): Util.raiseException("Minimax must be maximize or minimize", TypeError) self.minimax = mtype def getCurrentGeneration(self): """ Gets the current generation :rtype: the current generation """ return self.currentGeneration def setElitism(self, flag): """ Sets the elitism option, True or False :param flag: True or False """ if type(flag) != BooleanType: Util.raiseException("Elitism option must be True or False", TypeError) self.elitism = flag def getDBAdapter(self): """ Gets the DB Adapter of the GA Engine :rtype: a instance from one of the :mod:`DBAdapters` classes """ return self.dbAdapter def bestIndividual(self): """ Returns the population best individual :rtype: the best individual """ return self.internalPop.bestRaw() def __gp_catch_functions(self, prefix): """ Internally used to catch functions with some specific prefix as non-terminals of the GP core """ import __main__ as mod_main function_set = {} main_dict = mod_main.__dict__ for obj, addr in main_dict.items(): if obj[0:len(prefix)] == prefix: try: op_len = addr.func_code.co_argcount except: continue function_set[obj] = op_len if len(function_set) <= 0: Util.raiseException( "No function set found using function prefix '%s' !" % prefix, ValueError) self.setParams(gp_function_set=function_set) def initialize(self): """ Initializes the GA Engine. Create and initialize population """ self.internalPop.create(minimax=self.minimax) self.internalPop.initialize(ga_engine=self) logging.debug("The GA Engine was initialized !") def getPopulation(self): """ Return the internal population of GA Engine :rtype: the population (:class:`GPopulation.GPopulation`) """ return self.internalPop def getStatistics(self): """ Gets the Statistics class instance of current generation :rtype: the statistics instance (:class:`Statistics.Statistics`) """ return self.internalPop.getStatistics() def step(self): """ Just do one step in evolution, one generation """ genomeMom = None genomeDad = None newPop = GPopulation(self.internalPop) if (MPI.COMM_WORLD.Get_rank() == 0 ): ### assumes WE are on top of hierarchy! popsize = len(self.internalPop) numAdded = 0 maxTries = 1000 numTries = 0 crossover_empty = self.select( popID=self.currentGeneration).crossover.isEmpty() ###TODO: enforce constraints!### while numAdded < popsize: genomeMom = self.select(popID=self.currentGeneration) genomeDad = self.select(popID=self.currentGeneration) if not crossover_empty and self.pCrossover >= 1.0: for it in genomeMom.crossover.applyFunctions(mom=genomeMom, dad=genomeDad, count=2): (sister, brother) = it else: if not crossover_empty and Util.randomFlipCoin( self.pCrossover): for it in genomeMom.crossover.applyFunctions( mom=genomeMom, dad=genomeDad, count=2): (sister, brother) = it else: sister = genomeMom.clone() brother = genomeDad.clone() # logging.debug("done cloning") sister.mutate(pmut=self.pMutation, ga_engine=self) brother.mutate(pmut=self.pMutation, ga_engine=self) if (numTries > maxTries or self.owner.eval_constraints(sister)): newPop.internalPop.append(sister) numAdded += 1 print "successfully added sister" if (numAdded < popsize and (numTries > maxTries or self.owner.eval_constraints(brother))): newPop.internalPop.append(brother) print "successfully added brother" numAdded += 1 numTries += 1 #end rank0 onlye # print "rank %d start eval pop" % MPI.COMM_WORLD.Get_rank() logging.debug("Evaluating the newly created population.") newPop.evaluate() # print "rank %d done eval pop" % MPI.COMM_WORLD.Get_rank() # if (MPI.COMM_WORLD.Get_rank() == 0): # print "after eval, new pop's positions are:" # for p in newPop: # print p.wt_positions if (MPI.COMM_WORLD.Get_rank() == 0 ): ### assumes WE are on top of hierarchy! if self.elitism: logging.debug("Doing elitism, %d" % self.nElitismReplacement) if self.getMinimax() == Consts.minimaxType["maximize"]: for i in xrange(self.nElitismReplacement): #re-evaluate before being sure this is the best # self.internalPop.bestRaw(i).evaluate() if self.internalPop.bestRaw(i).score > newPop.bestRaw( i).score: newPop[len(newPop) - 1 - i] = self.internalPop.bestRaw(i) elif self.getMinimax() == Consts.minimaxType["minimize"]: for i in xrange(self.nElitismReplacement): #re-evaluate before being sure this is the best # self.internalPop.bestRaw(i).evaluate() if self.internalPop.bestRaw(i).score < newPop.bestRaw( i).score: newPop[len(newPop) - 1 - i] = self.internalPop.bestRaw(i) self.internalPop = newPop if (MPI.COMM_WORLD.Get_rank() == 0 ): ### assumes WE are on top of hierarchy! self.internalPop.sort() # if (MPI.COMM_WORLD.Get_rank() == 0): # print "after sort, internal pop's positions are:" # for p in self.internalPop: # print p.wt_positions logging.debug("The generation %d was finished.", self.currentGeneration) self.currentGeneration += 1 if (MPI.COMM_WORLD.Get_rank() == 0 ): ### assumes WE are on top of hierarchy! self.saveState() return (self.currentGeneration == self.nGenerations) def saveState(self): self.owner.saveState(self.internalPop.internalPop, self.currentGeneration) def restoreState(self, gen): self.owner.restoreState(self.internalPop.internalPop, gen) self.currentGeneration = gen def printStats(self): """ Print generation statistics :rtype: the printed statistics as string .. versionchanged:: 0.6 The return of *printStats* method. """ if (MPI.COMM_WORLD.Get_rank() == 0 ): ### assumes WE are on top of hierarchy! percent = self.currentGeneration * 100 / float(self.nGenerations) message = "Gen. %d (%.2f%%):" % (self.currentGeneration, percent) logging.info(message) print message, sys_stdout.flush() self.internalPop.statistics() stat_ret = self.internalPop.printStats() return message + stat_ret else: return "" def printTimeElapsed(self): """ Shows the time elapsed since the begin of evolution """ total_time = time() - self.time_init print "Total time elapsed: %.3f seconds." % total_time return total_time def dumpStatsDB(self): """ Dumps the current statistics to database adapter """ logging.debug("Dumping stats to the DB Adapter") self.internalPop.statistics() self.dbAdapter.insert(self) def evolve(self, freq_stats=0, restore=-1): """ Do all the generations until the termination criteria, accepts the freq_stats (default is 0) to dump statistics at n-generation Example: >>> ga_engine.evolve(freq_stats=10) (...) :param freq_stats: if greater than 0, the statistics will be printed every freq_stats generation. :rtype: returns the best individual of the evolution .. versionadded:: 0.6 the return of the best individual """ stopFlagCallback = False stopFlagTerminationCriteria = False self.time_init = time() logging.debug( "Starting the DB Adapter and the Migration Adapter if any") if self.dbAdapter: self.dbAdapter.open(self) if self.migrationAdapter: self.migrationAdapter.start() if self.getGPMode(): gp_function_prefix = self.getParam("gp_function_prefix") if gp_function_prefix is not None: self.__gp_catch_functions(gp_function_prefix) if (MPI.COMM_WORLD.Get_rank() == 0 ): ### assumes WE are on top of hierarchy! self.initialize() if (restore >= 0): self.restoreState(restore) self.internalPop.evaluate() if (MPI.COMM_WORLD.Get_rank() == 0 ): ### assumes WE are on top of hierarchy! self.internalPop.sort() # from mpi4py import MPI # if (MPI.COMM_WORLD.Get_rank() == 0): # print "after eval, new pop's positions are:" # for p in self.internalPop: # print p.wt_positions logging.debug("Starting loop over evolutionary algorithm.") while True: if (MPI.COMM_WORLD.Get_rank() == 0 ): ### assumes WE are on top of hierarchy! if self.migrationAdapter: logging.debug("Migration adapter: exchange") self.migrationAdapter.exchange() self.internalPop.clearFlags() self.internalPop.sort() if not self.stepCallback.isEmpty(): for it in self.stepCallback.applyFunctions(self): stopFlagCallback = it if not self.terminationCriteria.isEmpty(): for it in self.terminationCriteria.applyFunctions(self): stopFlagTerminationCriteria = it if freq_stats: if (self.currentGeneration % freq_stats == 0) or (self.getCurrentGeneration() == 0): self.printStats() if self.dbAdapter: if self.currentGeneration % self.dbAdapter.getStatsGenFreq( ) == 0: self.dumpStatsDB() if stopFlagTerminationCriteria: logging.debug( "Evolution stopped by the Termination Criteria !") if freq_stats: print "\n\tEvolution stopped by Termination Criteria function !\n" break if stopFlagCallback: logging.debug( "Evolution stopped by Step Callback function !") if freq_stats: print "\n\tEvolution stopped by Step Callback function !\n" break if self.interactiveMode: if sys_platform[:3] == "win": if msvcrt.kbhit(): if ord(msvcrt.getch()) == Consts.CDefESCKey: print "Loading modules for Interactive Mode...", logging.debug( "Windows Interactive Mode key detected ! generation=%d", self.getCurrentGeneration()) from pyevolve import Interaction print " done !" interact_banner = "## Pyevolve v.%s - Interactive Mode ##\nPress CTRL-Z to quit interactive mode." % ( pyevolve.__version__, ) session_locals = { "ga_engine": self, "population": self.getPopulation(), "pyevolve": pyevolve, "it": Interaction } print code.interact(interact_banner, local=session_locals) if (self.getInteractiveGeneration() >= 0) and (self.getInteractiveGeneration() == self.getCurrentGeneration()): print "Loading modules for Interactive Mode...", logging.debug( "Manual Interactive Mode key detected ! generation=%d", self.getCurrentGeneration()) from pyevolve import Interaction print " done !" interact_banner = "## Pyevolve v.%s - Interactive Mode ##" % ( pyevolve.__version__, ) session_locals = { "ga_engine": self, "population": self.getPopulation(), "pyevolve": pyevolve, "it": Interaction } print code.interact(interact_banner, local=session_locals) ## end, rank0 only if self.step(): break if (MPI.COMM_WORLD.Get_rank() == 0 ): ### assumes WE are on top of hierarchy! if freq_stats != 0: self.printStats() self.printTimeElapsed() if self.dbAdapter: logging.debug("Closing the DB Adapter") if not (self.currentGeneration % self.dbAdapter.getStatsGenFreq() == 0): self.dumpStatsDB() self.dbAdapter.commitAndClose() if self.migrationAdapter: logging.debug("Closing the Migration Adapter") if freq_stats: print "Stopping the migration adapter... ", self.migrationAdapter.stop() if freq_stats: print "done !" return self.bestIndividual() else: return None def select(self, **args): """ Select one individual from population :param args: this parameters will be sent to the selector """ for it in self.selector.applyFunctions(self.internalPop, **args): return it
class MigrationScheme: """ This is the base class for all migration schemes :param host: the source hostname :param port: the source host port :param group_name: the group name """ selector = None """ This is the function slot for the selection method if you want to change the default selector, you must do this: :: migration_scheme.selector.set(Selectors.GRouletteWheel) """ def __init__(self, host, port, group_name): self.myself = None self.groupName = group_name self.selector = FunctionSlot("Selector") self.setMyself(host, port) self.GAEngine = None self.nMigrationRate = Consts.CDefGenMigrationRate self.nIndividuals = Consts.CDefMigrationNIndividuals self.nReplacement = Consts.CDefGenMigrationReplacement self.networkCompression = 9 def isReady(self): """ Returns true if is time to migrate """ return True if self.GAEngine.getCurrentGeneration() % self.nMigrationRate == 0 else False def getCompressionLevel(self): """ Get the zlib compression level of network data The values are in the interval described on the :func:`Network.pickleAndCompress` """ return self.networkCompression def setCompressionLevel(self, level): """ Set the zlib compression level of network data The values are in the interval described on the :func:`Network.pickleAndCompress` :param level: the zlib compression level """ self.networkCompression = level def getNumReplacement(self): """ Return the number of individuals that will be replaced in the migration process """ return self.nReplacement def setNumReplacement(self, num_individuals): """ Return the number of individuals that will be replaced in the migration process :param num_individuals: the number of individuals to be replaced """ self.nReplacement = num_individuals def getNumIndividuals(self): """ Return the number of individuals that will migrate :rtype: the number of individuals to be replaced """ return self.nIndividuals def setNumIndividuals(self, num_individuals): """ Set the number of individuals that will migrate :param num_individuals: the number of individuals """ self.nIndividuals = num_individuals def setMigrationRate(self, generations): """ Sets the generation frequency supposed to migrate and receive individuals. :param generations: the number of generations """ self.nMigrationRate = generations def getMigrationRate(self): """ Return the the generation frequency supposed to migrate and receive individuals :rtype: the number of generations """ return self.nMigrationRate def setGAEngine(self, ga_engine): """ Sets the GA Engine handler """ self.GAEngine = ga_engine def start(self): """ Initializes the migration scheme """ pass def stop(self): """ Stops the migration engine """ pass def getGroupName(self): """ Gets the group name .. note:: all islands of evolution which are supposed to exchange individuals, must have the same group name. """ return self.groupName def setGroupName(self, name): """ Sets the group name :param name: the group name .. note:: all islands of evolution which are supposed to exchange individuals, must have the same group name. """ self.groupName = name def setMyself(self, host, port): """ Which interface you will use to send/receive data :param host: your hostname :param port: your port """ self.myself = (host, port) def select(self): """ Pickes an individual from population using specific selection method :rtype: an individual object """ if self.selector.isEmpty(): return self.GAEngine.select(popID=self.GAEngine.currentGeneration) else: for it in self.selector.applyFunctions(self.GAEngine.internalPop, popID=self.GAEngine.currentGeneration): return it def selectPool(self, num_individuals): """ Select num_individuals number of individuals and return a pool :param num_individuals: the number of individuals to select :rtype: list with individuals """ pool = [self.select() for i in xrange(num_individuals)] return pool def exchange(self): """ Exchange individuals """ pass
class GSimpleGA: """ GA Engine Class - The Genetic Algorithm Core Example: >>> ga = GSimpleGA.GSimpleGA(genome) >>> ga.selector.set(Selectors.GRouletteWheel) >>> ga.setGenerations(120) >>> ga.terminationCriteria.set(GSimpleGA.ConvergenceCriteria) :param genome: the :term:`Sample Genome` :param interactiveMode: this flag enables the Interactive Mode, the default is True :param seed: the random seed value .. note:: if you use the same random seed, all the runs of algorithm will be the same """ selector = None """ This is the function slot for the selection method if you want to change the default selector, you must do this: :: ga_engine.selector.set(Selectors.GRouletteWheel) """ stepCallback = None """ This is the :term:`step callback function` slot, if you want to set the function, you must do this: :: def your_func(ga_engine): # Here you have access to the GA Engine return False ga_engine.stepCallback.set(your_func) now *"your_func"* will be called every generation. When this function returns True, the GA Engine will stop the evolution and show a warning, if is False, the evolution continues. """ terminationCriteria = None """ This is the termination criteria slot, if you want to set one termination criteria, you must do this: :: ga_engine.terminationCriteria.set(GSimpleGA.ConvergenceCriteria) Now, when you run your GA, it will stop when the population converges. There are those termination criteria functions: :func:`GSimpleGA.RawScoreCriteria`, :func:`GSimpleGA.ConvergenceCriteria`, :func:`GSimpleGA.RawStatsCriteria`, :func:`GSimpleGA.FitnessStatsCriteria` But you can create your own termination function, this function receives one parameter which is the GA Engine, follows an example: :: def ConvergenceCriteria(ga_engine): pop = ga_engine.getPopulation() return pop[0] == pop[len(pop)-1] When this function returns True, the GA Engine will stop the evolution and show a warning, if is False, the evolution continues, this function is called every generation. """ def __init__(self, genome, seed=None, interactiveMode=True): """ Initializator of GSimpleGA """ if seed: random.seed(seed) if type(interactiveMode) != BooleanType: Util.raiseException("Interactive Mode option must be True or False", TypeError) if not isinstance(genome, GenomeBase): Util.raiseException("The genome must be a GenomeBase subclass", TypeError) self.internalPop = GPopulation(genome) self.nGenerations = Consts.CDefGAGenerations self.pMutation = Consts.CDefGAMutationRate self.pCrossover = Consts.CDefGACrossoverRate self.nElitismReplacement = Consts.CDefGAElitismReplacement self.setPopulationSize(Consts.CDefGAPopulationSize) self.minimax = Consts.minimaxType["maximize"] self.elitism = True # Adapters self.dbAdapter = None self.migrationAdapter = None self.time_init = None self.interactiveMode = interactiveMode self.interactiveGen = -1 self.GPMode = False self.selector = FunctionSlot("Selector") self.stepCallback = FunctionSlot("Generation Step Callback") self.terminationCriteria = FunctionSlot("Termination Criteria") self.selector.set(Consts.CDefGASelector) self.allSlots = [ self.selector, self.stepCallback, self.terminationCriteria ] self.internalParams = {} self.currentGeneration = 0 # GP Testing for classes in Consts.CDefGPGenomes: if isinstance(self.internalPop.oneSelfGenome, classes): self.setGPMode(True) break logging.debug("A GA Engine was created, nGenerations=%d", self.nGenerations) def setGPMode(self, bool_value): """ Sets the Genetic Programming mode of the GA Engine :param bool_value: True or False """ self.GPMode = bool_value def getGPMode(self): """ Get the Genetic Programming mode of the GA Engine :rtype: True or False """ return self.GPMode def __call__(self, *args, **kwargs): """ A method to implement a callable object Example: >>> ga_engine(freq_stats=10) .. versionadded:: 0.6 The callable method. """ if kwargs.get("freq_stats", None): return self.evolve(kwargs.get("freq_stats")) else: return self.evolve() def setParams(self, **args): """ Set the internal params Example: >>> ga.setParams(gp_terminals=['x', 'y']) :param args: params to save ..versionaddd:: 0.6 Added the *setParams* method. """ self.internalParams.update(args) def getParam(self, key, nvl=None): """ Gets an internal parameter Example: >>> ga.getParam("gp_terminals") ['x', 'y'] :param key: the key of param :param nvl: if the key doesn't exist, the nvl will be returned ..versionaddd:: 0.6 Added the *getParam* method. """ return self.internalParams.get(key, nvl) def setInteractiveGeneration(self, generation): """ Sets the generation in which the GA must enter in the Interactive Mode :param generation: the generation number, use "-1" to disable .. versionadded::0.6 The *setInteractiveGeneration* method. """ if generation < -1: Util.raiseException("Generation must be >= -1", ValueError) self.interactiveGen = generation def getInteractiveGeneration(self): """ returns the generation in which the GA must enter in the Interactive Mode :rtype: the generation number or -1 if not set .. versionadded::0.6 The *getInteractiveGeneration* method. """ return self.interactiveGen def setElitismReplacement(self, numreplace): """ Set the number of best individuals to copy to the next generation on the elitism :param numreplace: the number of individuals .. versionadded:: 0.6 The *setElitismReplacement* method. """ if numreplace < 1: Util.raiseException("Replacement number must be >= 1", ValueError) self.nElitismReplacement = numreplace def setInteractiveMode(self, flag=True): """ Enable/disable the interactive mode :param flag: True or False .. versionadded: 0.6 The *setInteractiveMode* method. """ if type(flag) != BooleanType: Util.raiseException("Interactive Mode option must be True or False", TypeError) self.interactiveMode = flag def __repr__(self): """ The string representation of the GA Engine """ ret = "- GSimpleGA\n" ret += "\tGP Mode:\t\t %s\n" % self.getGPMode() ret += "\tPopulation Size:\t %d\n" % (self.internalPop.popSize,) ret += "\tGenerations:\t\t %d\n" % (self.nGenerations,) ret += "\tCurrent Generation:\t %d\n" % (self.currentGeneration,) ret += "\tMutation Rate:\t\t %.2f\n" % (self.pMutation,) ret += "\tCrossover Rate:\t\t %.2f\n" % (self.pCrossover,) ret += "\tMinimax Type:\t\t %s\n" % (Consts.minimaxType.keys()[Consts.minimaxType.values().index(self.minimax)].capitalize(),) ret += "\tElitism:\t\t %s\n" % (self.elitism,) ret += "\tElitism Replacement:\t %d\n" % (self.nElitismReplacement,) ret += "\tDB Adapter:\t\t %s\n" % (self.dbAdapter,) for slot in self.allSlots: ret+= "\t" + slot.__repr__() ret+="\n" return ret def setMultiProcessing(self, flag=True, full_copy=False): """ Sets the flag to enable/disable the use of python multiprocessing module. Use this option when you have more than one core on your CPU and when your evaluation function is very slow. Pyevolve will automaticly check if your Python version has **multiprocessing** support and if you have more than one single CPU core. If you don't have support or have just only one core, Pyevolve will not use the **multiprocessing** feature. Pyevolve uses the **multiprocessing** to execute the evaluation function over the individuals, so the use of this feature will make sense if you have a truly slow evaluation function (which is commom in GAs). The parameter "full_copy" defines where the individual data should be copied back after the evaluation or not. This parameter is useful when you change the individual in the evaluation function. :param flag: True (default) or False :param full_copy: True or False (default) .. warning:: Use this option only when your evaluation function is slow, so you'll get a good tradeoff between the process communication speed and the parallel evaluation. The use of the **multiprocessing** doesn't means always a better performance. .. note:: To enable the multiprocessing option, you **MUST** add the *__main__* check on your application, otherwise, it will result in errors. See more on the `Python Docs <http://docs.python.org/library/multiprocessing.html#multiprocessing-programming>`__ site. .. versionadded:: 0.6 The `setMultiProcessing` method. """ if type(flag) != BooleanType: Util.raiseException("Multiprocessing option must be True or False", TypeError) if type(full_copy) != BooleanType: Util.raiseException("Multiprocessing 'full_copy' option must be True or False", TypeError) self.internalPop.setMultiProcessing(flag, full_copy) def setMigrationAdapter(self, migration_adapter=None): """ Sets the Migration Adapter .. versionadded:: 0.6 The `setMigrationAdapter` method. """ self.migrationAdapter = migration_adapter if self.migrationAdapter is not None: self.migrationAdapter.setGAEngine(self) def setDBAdapter(self, dbadapter=None): """ Sets the DB Adapter of the GA Engine :param dbadapter: one of the :mod:`DBAdapters` classes instance .. warning:: the use the of a DB Adapter can reduce the speed performance of the Genetic Algorithm. """ if (dbadapter is not None) and (not isinstance(dbadapter, DBBaseAdapter)): Util.raiseException("The DB Adapter must be a DBBaseAdapter subclass", TypeError) self.dbAdapter = dbadapter def setPopulationSize(self, size): """ Sets the population size, calls setPopulationSize() of GPopulation :param size: the population size .. note:: the population size must be >= 2 """ if size < 2: Util.raiseException("population size must be >= 2", ValueError) self.internalPop.setPopulationSize(size) def setSortType(self, sort_type): """ Sets the sort type, Consts.sortType["raw"]/Consts.sortType["scaled"] Example: >>> ga_engine.setSortType(Consts.sortType["scaled"]) :param sort_type: the Sort Type """ if sort_type not in Consts.sortType.values(): Util.raiseException("sort type must be a Consts.sortType type", TypeError) self.internalPop.sortType = sort_type def setMutationRate(self, rate): """ Sets the mutation rate, between 0.0 and 1.0 :param rate: the rate, between 0.0 and 1.0 """ if (rate>1.0) or (rate<0.0): Util.raiseException("Mutation rate must be >= 0.0 and <= 1.0", ValueError) self.pMutation = rate def setCrossoverRate(self, rate): """ Sets the crossover rate, between 0.0 and 1.0 :param rate: the rate, between 0.0 and 1.0 """ if (rate>1.0) or (rate<0.0): Util.raiseException("Crossover rate must be >= 0.0 and <= 1.0", ValueError) self.pCrossover = rate def setGenerations(self, num_gens): """ Sets the number of generations to evolve :param num_gens: the number of generations """ if num_gens < 1: Util.raiseException("Number of generations must be >= 1", ValueError) self.nGenerations = num_gens def getGenerations(self): """ Return the number of generations to evolve :rtype: the number of generations .. versionadded:: 0.6 Added the *getGenerations* method """ return self.nGenerations def getMinimax(self): """ Gets the minimize/maximize mode :rtype: the Consts.minimaxType type """ return self.minimax def setMinimax(self, mtype): """ Sets the minimize/maximize mode, use Consts.minimaxType :param mtype: the minimax mode, from Consts.minimaxType """ if mtype not in Consts.minimaxType.values(): Util.raiseException("Minimax must be maximize or minimize", TypeError) self.minimax = mtype def getCurrentGeneration(self): """ Gets the current generation :rtype: the current generation """ return self.currentGeneration def setElitism(self, flag): """ Sets the elitism option, True or False :param flag: True or False """ if type(flag) != BooleanType: Util.raiseException("Elitism option must be True or False", TypeError) self.elitism = flag def getDBAdapter(self): """ Gets the DB Adapter of the GA Engine :rtype: a instance from one of the :mod:`DBAdapters` classes """ return self.dbAdapter def bestIndividual(self): """ Returns the population best individual :rtype: the best individual """ return self.internalPop.bestRaw() def worstIndividual(self): """ Returns the population worst individual :rtype: the best individual """ return self.internalPop.worstRaw() def __gp_catch_functions(self, prefix): """ Internally used to catch functions with some specific prefix as non-terminals of the GP core """ import __main__ as mod_main function_set = {} main_dict = mod_main.__dict__ for obj, addr in main_dict.items(): if obj[0:len(prefix)] == prefix: try: op_len = addr.func_code.co_argcount except: continue function_set[obj] = op_len if len(function_set) <= 0: Util.raiseException("No function set found using function prefix '%s' !" % prefix, ValueError) self.setParams(gp_function_set=function_set) def initialize(self): """ Initializes the GA Engine. Create and initialize population """ self.internalPop.create(minimax=self.minimax) self.internalPop.initialize(ga_engine=self) logging.debug("The GA Engine was initialized !") def getPopulation(self): """ Return the internal population of GA Engine :rtype: the population (:class:`GPopulation.GPopulation`) """ return self.internalPop def getStatistics(self): """ Gets the Statistics class instance of current generation :rtype: the statistics instance (:class:`Statistics.Statistics`) """ return self.internalPop.getStatistics() def step(self): """ Just do one step in evolution, one generation """ genomeMom = None genomeDad = None newPop = GPopulation(self.internalPop) logging.debug("Population was cloned.") size_iterate = len(self.internalPop) # Odd population size if size_iterate % 2 != 0: size_iterate -= 1 crossover_empty = self.select(popID=self.currentGeneration).crossover.isEmpty() for i in xrange(0, size_iterate, 2): genomeMom = self.select(popID=self.currentGeneration) genomeDad = self.select(popID=self.currentGeneration) if not crossover_empty and self.pCrossover >= 1.0: for it in genomeMom.crossover.applyFunctions(mom=genomeMom, dad=genomeDad, count=2): (sister, brother) = it else: if not crossover_empty and Util.randomFlipCoin(self.pCrossover): for it in genomeMom.crossover.applyFunctions(mom=genomeMom, dad=genomeDad, count=2): (sister, brother) = it else: sister = genomeMom.clone() brother = genomeDad.clone() sister.mutate(pmut=self.pMutation, ga_engine=self) brother.mutate(pmut=self.pMutation, ga_engine=self) newPop.internalPop.append(sister) newPop.internalPop.append(brother) if len(self.internalPop) % 2 != 0: genomeMom = self.select(popID=self.currentGeneration) genomeDad = self.select(popID=self.currentGeneration) if Util.randomFlipCoin(self.pCrossover): for it in genomeMom.crossover.applyFunctions(mom=genomeMom, dad=genomeDad, count=1): (sister, brother) = it else: sister = random.choice([genomeMom, genomeDad]) sister = sister.clone() sister.mutate(pmut=self.pMutation, ga_engine=self) newPop.internalPop.append(sister) logging.debug("Evaluating the new created population.") newPop.evaluate() if self.elitism: logging.debug("Doing elitism.") if self.getMinimax() == Consts.minimaxType["maximize"]: for i in xrange(self.nElitismReplacement): #re-evaluate before being sure this is the best self.internalPop.bestRaw(i).evaluate() if self.internalPop.bestRaw(i).score > newPop.bestRaw(i).score: newPop[len(newPop)-1-i] = self.internalPop.bestRaw(i) elif self.getMinimax() == Consts.minimaxType["minimize"]: for i in xrange(self.nElitismReplacement): #re-evaluate before being sure this is the best self.internalPop.bestRaw(i).evaluate() if self.internalPop.bestRaw(i).score < newPop.bestRaw(i).score: newPop[len(newPop)-1-i] = self.internalPop.bestRaw(i) self.internalPop = newPop self.internalPop.sort() logging.debug("The generation %d was finished.", self.currentGeneration) self.currentGeneration += 1 return (self.currentGeneration == self.nGenerations) def printStats(self): """ Print generation statistics :rtype: the printed statistics as string .. versionchanged:: 0.6 The return of *printStats* method. """ percent = self.currentGeneration * 100 / float(self.nGenerations) message = "Gen. %d (%.2f%%):" % (self.currentGeneration, percent) logging.info(message) print message, sys_stdout.flush() self.internalPop.statistics() stat_ret = self.internalPop.printStats() return message + stat_ret def printTimeElapsed(self): """ Shows the time elapsed since the begin of evolution """ total_time = time()-self.time_init print "Total time elapsed: %.3f seconds." % total_time return total_time def dumpStatsDB(self): """ Dumps the current statistics to database adapter """ logging.debug("Dumping stats to the DB Adapter") self.internalPop.statistics() self.dbAdapter.insert(self) def evolve(self, freq_stats=0): """ Do all the generations until the termination criteria, accepts the freq_stats (default is 0) to dump statistics at n-generation Example: >>> ga_engine.evolve(freq_stats=10) (...) :param freq_stats: if greater than 0, the statistics will be printed every freq_stats generation. :rtype: returns the best individual of the evolution .. versionadded:: 0.6 the return of the best individual """ stopFlagCallback = False stopFlagTerminationCriteria = False self.time_init = time() logging.debug("Starting the DB Adapter and the Migration Adapter if any") if self.dbAdapter: self.dbAdapter.open(self) if self.migrationAdapter: self.migrationAdapter.start() if self.getGPMode(): gp_function_prefix = self.getParam("gp_function_prefix") if gp_function_prefix is not None: self.__gp_catch_functions(gp_function_prefix) self.initialize() self.internalPop.evaluate() self.internalPop.sort() logging.debug("Starting loop over evolutionary algorithm.") def stop_evolution(s, f): #print signal, frame if s == signal.SIGINT: if self.internalPop.multiProcessing[0]: logging.debug("CTRL-C detected, finishing evolution (stopping parallel jobs).") self.internalPop.proc_pool.terminate() self.internalPop.proc_pool.join() else: logging.debug("CTRL-C detected, finishing evolution.") if freq_stats: print "\n\tA break was detected, you have interrupted the evolution !\n" if freq_stats != 0: self.printStats() self.printTimeElapsed() if self.dbAdapter: logging.debug("Closing the DB Adapter") if not (self.currentGeneration % self.dbAdapter.getStatsGenFreq() == 0): self.dumpStatsDB() self.dbAdapter.commitAndClose() if self.migrationAdapter: logging.debug("Closing the Migration Adapter") if freq_stats: print "Stopping the migration adapter... ", self.migrationAdapter.stop() if freq_stats: print "done !" if s == signal.SIGINT: print self.bestIndividual() exit(0) else: return self.bestIndividual() signal.signal(signal.SIGINT, stop_evolution) while True: if self.migrationAdapter: logging.debug("Migration adapter: exchange") self.migrationAdapter.exchange() self.internalPop.clearFlags() self.internalPop.sort() if not self.stepCallback.isEmpty(): for it in self.stepCallback.applyFunctions(self): stopFlagCallback = it if not self.terminationCriteria.isEmpty(): for it in self.terminationCriteria.applyFunctions(self): stopFlagTerminationCriteria = it if freq_stats: if (self.currentGeneration % freq_stats == 0) or (self.getCurrentGeneration() == 0): self.printStats() #print self.bestIndividual() if self.dbAdapter: if self.currentGeneration % self.dbAdapter.getStatsGenFreq() == 0: self.dumpStatsDB() if stopFlagTerminationCriteria: logging.debug("Evolution stopped by the Termination Criteria !") if freq_stats: print "\n\tEvolution stopped by Termination Criteria function !\n" break if stopFlagCallback: logging.debug("Evolution stopped by Step Callback function !") if freq_stats: print "\n\tEvolution stopped by Step Callback function !\n" break if self.interactiveMode: if sys_platform[:3] == "win": if msvcrt.kbhit(): if ord(msvcrt.getch()) == Consts.CDefESCKey: print "Loading modules for Interactive Mode...", logging.debug("Windows Interactive Mode key detected ! generation=%d", self.getCurrentGeneration()) from pyevolve import Interaction print " done !" interact_banner = "## Pyevolve v.%s - Interactive Mode ##\nPress CTRL-Z to quit interactive mode." % (pyevolve.__version__,) session_locals = { "ga_engine" : self, "population" : self.getPopulation(), "pyevolve" : pyevolve, "it" : Interaction} print code.interact(interact_banner, local=session_locals) if (self.getInteractiveGeneration() >= 0) and (self.getInteractiveGeneration() == self.getCurrentGeneration()): print "Loading modules for Interactive Mode...", logging.debug("Manual Interactive Mode key detected ! generation=%d", self.getCurrentGeneration()) from pyevolve import Interaction print " done !" interact_banner = "## Pyevolve v.%s - Interactive Mode ##" % (pyevolve.__version__,) session_locals = { "ga_engine" : self, "population" : self.getPopulation(), "pyevolve" : pyevolve, "it" : Interaction} print code.interact(interact_banner, local=session_locals) if self.step(): break return stop_evolution(signal.SIGUSR1, None) def select(self, **args): """ Select one individual from population :param args: this parameters will be sent to the selector """ for it in self.selector.applyFunctions(self.internalPop, **args): return it
class GSimpleGA: """ GA Engine Class - The Genetic Algorithm Core Example: >>> ga = GSimpleGA.GSimpleGA(genome) >>> ga.selector.set(Selectors.GRouletteWheel) >>> ga.setGenerations(120) >>> ga.terminationCriteria.set(GSimpleGA.ConvergenceCriteria) :param genome: the :term:`Sample Genome` :param interactiveMode: this flag enables the Interactive Mode, the default is True :param seed: the random seed value .. note:: if you use the same random seed, all the runs of algorithm will be the same """ selector = None """ This is the function slot for the selection method if you want to change the default selector, you must do this: :: ga_engine.selector.set(Selectors.GRouletteWheel) """ stepCallback = None """ This is the :term:`step callback function` slot, if you want to set the function, you must do this: :: def your_func(ga_engine): # Here you have access to the GA Engine return False ga_engine.stepCallback.set(your_func) now *"your_func"* will be called every generation. When this function returns True, the GA Engine will stop the evolution and show a warning, if is False, the evolution continues. """ terminationCriteria = None """ This is the termination criteria slot, if you want to set one termination criteria, you must do this: :: ga_engine.terminationCriteria.set(GSimpleGA.ConvergenceCriteria) Now, when you run your GA, it will stop when the population converges. There are those termination criteria functions: :func:`GSimpleGA.RawScoreCriteria`, :func:`GSimpleGA.ConvergenceCriteria`, :func:`GSimpleGA.RawStatsCriteria`, :func:`GSimpleGA.FitnessStatsCriteria` But you can create your own termination function, this function receives one parameter which is the GA Engine, follows an example: :: def ConvergenceCriteria(ga_engine): pop = ga_engine.getPopulation() return pop[0] == pop[len(pop)-1] When this function returns True, the GA Engine will stop the evolution and show a warning, if is False, the evolution continues, this function is called every generation. """ def __init__(self, genome, seed=None, interactiveMode=True): """ Initializator of GSimpleGA """ if seed: random.seed(seed) if type(interactiveMode) != BooleanType: gp.Util.raiseException("Interactive Mode option must be True or False", TypeError) if not isinstance(genome, GenomeBase): gp.Util.raiseException("The genome must be a GenomeBase subclass", TypeError) self.internalPop = GPopulation(genome) self.nGenerations = Consts.CDefGAGenerations self.pMutation = Consts.CDefGAMutationRate self.pCrossover = Consts.CDefGACrossoverRate self.nElitismReplacement = Consts.CDefGAElitismReplacement self.setPopulationSize(Consts.CDefGAPopulationSize) self.minimax = Consts.minimaxType["maximize"] self.elitism = True # Adapters self.dbAdapter = None self.migrationAdapter = None self.time_init = None self.interactiveMode = interactiveMode self.interactiveGen = -1 self.GPMode = False self.selector = FunctionSlot("Selector") self.stepCallback = FunctionSlot("Generation Step Callback") self.terminationCriteria = FunctionSlot("Termination Criteria") self.selector.set(Consts.CDefGASelector) self.allSlots = [ self.selector, self.stepCallback, self.terminationCriteria ] self.internalParams = {} self.currentGeneration = 0 # gp Testing CDefGPGenomes = [GTreeGP] for classes in CDefGPGenomes: if isinstance(self.internalPop.oneSelfGenome, classes): self.setGPMode(True) break logging.debug("A GA Engine was created, nGenerations=%d", self.nGenerations) def setGPMode(self, bool_value): """ Sets the Genetic Programming mode of the GA Engine :param bool_value: True or False """ self.GPMode = bool_value def getGPMode(self): """ Get the Genetic Programming mode of the GA Engine :rtype: True or False """ return self.GPMode def __call__(self, *args, **kwargs): """ A method to implement a callable object Example: >>> ga_engine(freq_stats=10) .. versionadded:: 0.6 The callable method. """ if kwargs.get("freq_stats", None): return self.evolve(kwargs.get("freq_stats")) else: return self.evolve() def setParams(self, **args): """ Set the internal params Example: >>> ga.setParams(gp_terminals=['x', 'y']) :param args: params to save ..versionaddd:: 0.6 Added the *setParams* method. """ self.internalParams.update(args) def getParam(self, key, nvl=None): """ Gets an internal parameter Example: >>> ga.getParam("gp_terminals") ['x', 'y'] :param key: the key of param :param nvl: if the key doesn't exist, the nvl will be returned ..versionaddd:: 0.6 Added the *getParam* method. """ return self.internalParams.get(key, nvl) def setInteractiveGeneration(self, generation): """ Sets the generation in which the GA must enter in the Interactive Mode :param generation: the generation number, use "-1" to disable .. versionadded::0.6 The *setInteractiveGeneration* method. """ if generation < -1: Util.raiseException("Generation must be >= -1", ValueError) self.interactiveGen = generation def getInteractiveGeneration(self): """ returns the generation in which the GA must enter in the Interactive Mode :rtype: the generation number or -1 if not set .. versionadded::0.6 The *getInteractiveGeneration* method. """ return self.interactiveGen def setElitismReplacement(self, numreplace): """ Set the number of best individuals to copy to the next generation on the elitism :param numreplace: the number of individuals .. versionadded:: 0.6 The *setElitismReplacement* method. """ if numreplace < 1: Util.raiseException("Replacement number must be >= 1", ValueError) self.nElitismReplacement = numreplace def setInteractiveMode(self, flag=True): """ Enable/disable the interactive mode :param flag: True or False .. versionadded: 0.6 The *setInteractiveMode* method. """ if type(flag) != BooleanType: Util.raiseException("Interactive Mode option must be True or False", TypeError) self.interactiveMode = flag def __repr__(self): """ The string representation of the GA Engine """ ret = "- GSimpleGA\n" ret += "\tgp Mode:\t\t %s\n" % self.getGPMode() ret += "\tPopulation Size:\t %d\n" % (self.internalPop.popSize,) ret += "\tGenerations:\t\t %d\n" % (self.nGenerations,) ret += "\tCurrent Generation:\t %d\n" % (self.currentGeneration,) ret += "\tMutation Rate:\t\t %.2f\n" % (self.pMutation,) ret += "\tCrossover Rate:\t\t %.2f\n" % (self.pCrossover,) ret += "\tMinimax Type:\t\t %s\n" % ( Consts.minimaxType.keys()[Consts.minimaxType.values().index(self.minimax)].capitalize(),) ret += "\tElitism:\t\t %s\n" % (self.elitism,) ret += "\tElitism Replacement:\t %d\n" % (self.nElitismReplacement,) ret += "\tDB Adapter:\t\t %s\n" % (self.dbAdapter,) for slot in self.allSlots: ret+= "\t" + slot.__repr__() ret+="\n" return ret def setPopulationSize(self, size): """ Sets the population size, calls setPopulationSize() of GPopulation :param size: the population size .. note:: the population size must be >= 2 """ if size < 2: Util.raiseException("population size must be >= 2", ValueError) self.internalPop.setPopulationSize(size) def setSortType(self, sort_type): """ Sets the sort type, Consts.sortType["raw"]/Consts.sortType["scaled"] Example: >>> ga_engine.setSortType(Consts.sortType["scaled"]) :param sort_type: the Sort Type """ if sort_type not in Consts.sortType.values(): Util.raiseException("sort type must be a Consts.sortType type", TypeError) self.internalPop.sortType = sort_type def setMutationRate(self, rate): """ Sets the mutation rate, between 0.0 and 1.0 :param rate: the rate, between 0.0 and 1.0 """ if (rate>1.0) or (rate<0.0): Util.raiseException("Mutation rate must be >= 0.0 and <= 1.0", ValueError) self.pMutation = rate def setCrossoverRate(self, rate): """ Sets the crossover rate, between 0.0 and 1.0 :param rate: the rate, between 0.0 and 1.0 """ if (rate>1.0) or (rate<0.0): Util.raiseException("Crossover rate must be >= 0.0 and <= 1.0", ValueError) self.pCrossover = rate def setGenerations(self, num_gens): """ Sets the number of generations to evolve :param num_gens: the number of generations """ if num_gens < 1: Util.raiseException("Number of generations must be >= 1", ValueError) self.nGenerations = num_gens def getGenerations(self): """ Return the number of generations to evolve :rtype: the number of generations .. versionadded:: 0.6 Added the *getGenerations* method """ return self.nGenerations def getMinimax(self): """ Gets the minimize/maximize mode :rtype: the Consts.minimaxType type """ return self.minimax def setMinimax(self, mtype): """ Sets the minimize/maximize mode, use Consts.minimaxType :param mtype: the minimax mode, from Consts.minimaxType """ if mtype not in Consts.minimaxType.values(): Util.raiseException("Minimax must be maximize or minimize", TypeError) self.minimax = mtype def getCurrentGeneration(self): """ Gets the current generation :rtype: the current generation """ return self.currentGeneration def setElitism(self, flag): """ Sets the elitism option, True or False :param flag: True or False """ if type(flag) != BooleanType: Util.raiseException("Elitism option must be True or False", TypeError) self.elitism = flag def bestIndividual(self): """ Returns the population best individual :rtype: the best individual """ return self.internalPop.bestRaw() def __gp_catch_functions(self, prefix): """ Internally used to catch functions with some specific prefix as non-terminals of the gp core """ import __main__ as mod_main function_set = {} main_dict = mod_main.__dict__ for obj, addr in main_dict.items(): if obj[0:len(prefix)] == prefix: try: op_len = addr.func_code.co_argcount except: continue function_set[obj] = op_len if len(function_set) <= 0: Util.raiseException("No function set found using function prefix '%s' !" % prefix, ValueError) self.setParams(gp_function_set=function_set) def initialize(self): """ Initializes the GA Engine. Create and initialize population """ self.internalPop.create(minimax=self.minimax) self.internalPop.initialize(ga_engine=self) logging.debug("The GA Engine was initialized !") def getPopulation(self): """ Return the internal population of GA Engine :rtype: the population (:class:`GPopulation.GPopulation`) """ return self.internalPop def getStatistics(self): """ Gets the Statistics class instance of current generation :rtype: the statistics instance (:class:`Statistics.Statistics`) """ return self.internalPop.getStatistics() def clear(self): """ Petrowski's Clearing Method """ def step(self): """ Just do one step in evolution, one generation """ genomeMom = None genomeDad = None newPop = GPopulation(self.internalPop) logging.debug("Population was cloned.") size_iterate = len(self.internalPop) # Odd population size if size_iterate % 2 != 0: size_iterate -= 1 crossover_empty = self.select(popID=self.currentGeneration).crossover.isEmpty() from utils import delog ############# delog.decache("mutate and check...") ################# for i in xrange(0, size_iterate, 2): genomeMom = self.select(popID=self.currentGeneration) genomeDad = self.select(popID=self.currentGeneration) if not crossover_empty and self.pCrossover >= 1.0: for it in genomeMom.crossover.applyFunctions(mom=genomeMom, dad=genomeDad, count=2): (sister, brother) = it else: if not crossover_empty and Util.randomFlipCoin(self.pCrossover): for it in genomeMom.crossover.applyFunctions(mom=genomeMom, dad=genomeDad, count=2): (sister, brother) = it else: sister = genomeMom.clone() brother = genomeDad.clone() sister.mutate(pmut=self.pMutation, ga_engine=self) brother.mutate(pmut=self.pMutation, ga_engine=self) newPop.internalPop.append(sister) newPop.internalPop.append(brother) delog.deprint_string("over.") if len(self.internalPop) % 2 != 0: genomeMom = self.select(popID=self.currentGeneration) genomeDad = self.select(popID=self.currentGeneration) if Util.randomFlipCoin(self.pCrossover): for it in genomeMom.crossover.applyFunctions(mom=genomeMom, dad=genomeDad, count=1): (sister, brother) = it else: sister = random.choice([genomeMom, genomeDad]) sister = sister.clone() sister.mutate(pmut=self.pMutation, ga_engine=self) newPop.internalPop.append(sister) logging.debug("Evaluating the new created population.") newPop.evaluate() newPop.sort() #Niching methods- Petrowski's clearing self.clear() if self.elitism: logging.debug("Doing elitism.") if self.getMinimax() == Consts.minimaxType["maximize"]: # in ecoc, max value is expected. for i in xrange(self.nElitismReplacement): if self.internalPop.bestRaw(i).score > newPop.bestRaw(i).score: # check duplicate to avoid repeat indivadual duplicate = False for j in xrange(len(newPop)-self.nElitismReplacement, len(newPop)): if self.internalPop.bestRaw(i).score == newPop.bestRaw(j).score: duplicate = True break if duplicate: continue newPop[len(newPop)-1-i] = newPop[i] newPop[i] = self.internalPop.bestRaw(i) elif self.getMinimax() == Consts.minimaxType["minimize"]: for i in xrange(self.nElitismReplacement): if self.internalPop.bestRaw(i).score < newPop.bestRaw(i).score: newPop[len(newPop)-1-i] = self.internalPop.bestRaw(i) self.internalPop = newPop self.internalPop.sort() logging.debug("The generation %d was finished.", self.currentGeneration) self.currentGeneration += 1 return (self.currentGeneration == self.nGenerations) def printStats(self): """ Print generation statistics :rtype: the printed statistics as string .. versionchanged:: 0.6 The return of *printStats* method. """ percent = self.currentGeneration * 100 / float(self.nGenerations) #message = "Gen. %d (%.2f%%):" % (self.currentGeneration, percent) message = "Gen. %3d: " % (self.currentGeneration) logging.info(message) print message, sys_stdout.flush() self.internalPop.statistics() stat_ret = self.internalPop.printStats() return message + stat_ret def printTimeElapsed(self): """ Shows the time elapsed since the begin of evolution """ total_time = time()-self.time_init print "Total time elapsed: %.3f seconds." % total_time return total_time def evolve(self, freq_stats=0): """ Do all the generations until the termination criteria, accepts the freq_stats (default is 0) to dump statistics at n-generation Example: >>> ga_engine.evolve(freq_stats=10) (...) :param freq_stats: if greater than 0, the statistics will be printed every freq_stats generation. :rtype: returns the best individual of the evolution .. versionadded:: 0.6 the return of the best individual """ stopFlagCallback = False stopFlagTerminationCriteria = False self.time_init = time() if self.getGPMode(): gp_function_prefix = self.getParam("gp_function_prefix") if gp_function_prefix is not None: self.__gp_catch_functions(gp_function_prefix) self.initialize() logging.debug("Starting loop over evolutionary algorithm.") try: while True: if not self.stepCallback.isEmpty(): for it in self.stepCallback.applyFunctions(self): stopFlagCallback = it if not self.terminationCriteria.isEmpty(): for it in self.terminationCriteria.applyFunctions(self): stopFlagTerminationCriteria = it if freq_stats: if (self.currentGeneration % freq_stats == 0) or (self.getCurrentGeneration() == 0): self.printStats() if stopFlagTerminationCriteria: logging.debug("Evolution stopped by the Termination Criteria !") if freq_stats: print "\n\tEvolution stopped by Termination Criteria function !\n" break if stopFlagCallback: logging.debug("Evolution stopped by Step Callback function !") if freq_stats: print "\n\tEvolution stopped by Step Callback function !\n" break if self.interactiveMode: if sys_platform[:3] == "win": if msvcrt.kbhit(): if ord(msvcrt.getch()) == Consts.CDefESCKey: print "Loading modules for Interactive Mode...", logging.debug("Windows Interactive Mode key detected ! generation=%d", self.getCurrentGeneration()) from gp import Interaction print " done !" interact_banner = "## Pyevolve v.%s - Interactive Mode ##\nPress CTRL-Z to quit interactive mode." % (__version__,) session_locals = { "ga_engine" : self, "population" : self.getPopulation(), "pyevolve" : pyevolve, "it" : Interaction} print code.interact(interact_banner, local=session_locals) if (self.getInteractiveGeneration() >= 0) and (self.getInteractiveGeneration() == self.getCurrentGeneration()): print "Loading modules for Interactive Mode...", logging.debug("Manual Interactive Mode key detected ! generation=%d", self.getCurrentGeneration()) from gp import Interaction print " done !" interact_banner = "## Pyevolve v.%s - Interactive Mode ##" % (__version__,) session_locals = { "ga_engine" : self, "population" : self.getPopulation(), "pyevolve" : pyevolve, "it" : Interaction} print code.interact(interact_banner, local=session_locals) if self.step(): break #exit if the number of generations is equal to the max. number of gens. except KeyboardInterrupt: logging.debug("CTRL-C detected, finishing evolution.") if freq_stats: print "\n\tA break was detected, you have interrupted the evolution !\n" if freq_stats != 0: self.printStats() self.printTimeElapsed() return self.bestIndividual() def select(self, **args): """ Select one individual from population :param args: this parameters will be sent to the selector """ for it in self.selector.applyFunctions(self.internalPop, **args): return it
class SimplePSO(object): """ SimplePSO Engine Class - The PSO Algorithm Core Example: >>> topology = Topology.GlobalTopology(particle_rep) >>> pso = PSO.SimplePSO(topology) >>> pso.setSteps(120) :param topology: the :term:`Sample Topology`` :param interactiveMode: this flag enables the Interactive Mode :param seed: the random seed value .. note:: if you see the same random seed, all the runs of the algorithm will be the same. """ stepCallBack = None """ This is the the :term: `step callback function` slot, if you want to set the function, you must do this: :: def your_func(pso_engine): #Here you have access to the PSO Engine return False pso_engine.stepCallback.set(your_func) now *"your_func"* will be called every step. When this function returns True, the PSO Engine will stop the evolution and show a warning, if is False, the evolution continues. """ terminationCriteria = None """ This is the termination criteria slot, if you want to set one termination criteria, you mus do this: :: pso_engine.terminationCriteria.set(your_func) Now, when you run your PSO, it will stop when terminationCriteria be satisfied. To create your own termination function, you must put at least one parameter which is the PSO Engine, follows an example: :: def ConvergenceCriteria(pso_engine): swarm = pso_engine.getSwarm() return swarm[0] == swarm[len(swarm)-1] When this function returns True, the Pso Engine will stop the evolution and show a warning. If is False, the evolution continues, this function is called every step. """ def __init__(self, topology, seed=None, interactiveMode=True): """ Initializator of PSO """ #random seed random.seed(seed) #Pso type used by the particle self.psoType = Consts.CDefPsoType #Topology used self.topology = topology #Set the population size self.setSwarmSize(Consts.CDefSwarmSize) #Cognitive and Social Coefficients self.C1, self.C2 = Consts.CDefCoefficients #Time steps self.timeSteps = Consts.CDefSteps #Interactive Mode (True or False) self.interactiveMode = interactiveMode #Current step self.currentStep = 0 #Inertia Factor Minus self.inertiaFactorMinus = None #Inertia coefficient self.inertiaFactor = None #Time initial self.time_init = None #Optimization type self.minimax = Consts.minimaxType["minimize"] #Report file adapter self.reportAdapter = None #Step Callback self.stepCallback = FunctionSlot("Step Callback") #Termination Criteria self.terminationCriteria = FunctionSlot("Termination Criteria") #All slots self.allSlots = [self.stepCallback, self.terminationCriteria] print "A PSO Engine was created, timeSteps=% d" % (self.timeSteps, ) def __repr__(self): """ The String representation of the PSO Engine """ ret = "- PSO-%s-%s Execution\n" % (self.getTopologyType(), self.getPsoType()) ret += "\tSwarm Size:\t %d\n" % (self.topology.swarmSize, ) ret += "\tTime Steps:\t %d\n" % (self.timeSteps, ) ret += "\tCurrent Step:\t %d\n" % (self.currentStep, ) ret += "\tMinimax Type:\t %s\n" % (Consts.minimaxType.keys()[ Consts.minimaxType.values().index(self.minimax)].capitalize(), ) ret += "\tReport Adapter:\t %s\n" % (self.reportAdapter, ) for slot in self.allSlots: ret += "\t" + slot.__repr__() ret += "\n" return ret def setReportAdapter(self, repadapter): """ Sets the Report Adapter of the PSO Engine :param repadapter: one of the :mod:`ReportAdapters` classes instance .. warning: the use of a Report Adapter can reduce the speed performance of the PSO. """ self.reportAdapter = repadapter def setSwarmSize(self, size): """ Sets the swarm size, calls setSwarmSize() of Topology :param size: the swarm size .. note:: the swarm size must be >= 2 """ if size < 2: Util.raiseException("swarm size must be >= 2", ValueError) self.topology.setSwarmSize(size) def setPsoType(self, psoType): """ Sets the psoType, use Consts.psoType(Basic,Constricted,Inertia) Example: >>> pso_engine.setSortType(Consts.psoType["CONSTRICTED"]) :param psoType: The PSO type, from Consts.psoType """ if psoType not in Consts.psoType.values(): Util.raiseException("PsoType must be implemented !", TypeError) self.psoType = psoType def getPsoType(self): """ Return the Pso Type :rtype key: pso Type """ for key, value in Consts.psoType.items(): if value == self.psoType: return key return "" def setTimeSteps(self, num_steps): """ Sets the number of steps to converge :param num_steps: the number of steps """ if num_steps < 1: Util.raiseException("Number of steps must be >=1", ValueError) self.timeSteps = num_steps def getMinimax(self): """ Gets the minimize/maximize mode :rtype: The Consts.minimaxType type """ for key, value in Consts.minimaxType.items(): if value == self.minimax: return key return "" def getTopologyType(self): """ Returns the name of the topology :rtype name: the name of the topology """ return self.topology.__class__.__name__ def setMinimax(self, minimax): """Sets the minimize/maximize mode, use Consts.minimaxType :param minimax: the minimax mode, from Consts.minimaxType """ if minimax not in Consts.minimaxType.values(): Util.raiseException( "Optimization type must be Maximize or Minimize !", TypeError) self.minimax = minimax def getCurrentStep(self): """ Gets the current step :rtype: the current step """ return self.currentStep def getReportAdapter(self): """ Gets the Report Adapter of the PSO Engine :rtype: a instance from one of the :mod:`ReportAdapters` classes """ return self.reportAdapter def bestParticle(self): """ Returns the swarm best Particle :rtype: the best particle """ return self.topology.getBestParticle() def getTopology(self): """Return the internal topology of Pso Engine :rtype: the topology (:class: 'Topology.Topology')' """ return self.topology def getStatistics(self): """ Gets the Statistics class instance of the current step :rtype: the statistics instance (:class: `Statistics.Statistics`)` """ return self.topology.getStatistics() def dumpStatsReport(self): """ Dumps the current statistics to the report adapter """ self.topology.statistics() self.reportAdapter.insert(self.getStatistics(), self.topology, self.currentStep) def printStats(self): """ Print swarm statistics""" percent = self.currentStep * 100 / float(self.timeSteps) message = "Step: %d (%.2f%%):" % (self.currentStep, percent) print message self.topology.statistics() self.topology.printStats() def printTimeElapsed(self): """ Shows the time elapsed since the beginning of the solution construction """ print "Total time elapsed: %.3f seconds." % (time() - self.time_init) def initialize(self): """ Initializes the PSO Engine. Create and initialize the swarm """ self.topology.create(minimax=self.minimax) self.topology.initialize() print "The PSO Engine was initialized !" def constructSolution(self): """ Just do one step in execution, one step.""" for it in self.topology.position_updater.applyFunctions(self): pass for it in self.topology.information_updater.applyFunctions(self): pass if self.psoType == Consts.psoType["INERTIA"]: self.updateInertiaFactor() self.currentStep += 1 return (self.currentStep == self.timeSteps) def execute(self, freq_stats=0): """ Do all the steps until the termination criteria or time Steps achieved, accepts the freq_stats (default is 0) to dump statistics at n-step Example: >>> pso_engine.evolve(freq_stats=10) (...) :param freq_stats: if greater than 0, the statistics will be printed every freq_stats step. """ #Start time self.time_init = time() #Creates a new report if reportAdapter is not None. if self.reportAdapter: self.reportAdapter.open() #Initialize the PSO Engine self.initialize() #Already evaluates all particles print "Starting loop over evolutionary algorithm." try: while not self.constructSolution(): stopFlagCallback = False stopFlagTerminationCriteria = False if not self.stepCallback.isEmpty(): for it in self.stepCallback.applyFunctions(self): stopFlagCallback = it if not self.terminationCriteria.isEmpty(): for it in self.terminationCriteria.applyFunctions(self): stopFlagTerminationCriteria = it if freq_stats != 0: if (self.currentStep % freq_stats == 0) or (self.currentStep == 1): self.printStats() if self.reportAdapter: if self.currentStep % self.reportAdapter.statsGenFreq == 0: self.dumpStatsReport() if stopFlagTerminationCriteria: print '\n\tExecution stopped by Termination Criteria function !\n' break if stopFlagCallback: print '\n\tExecution stopped by Step Callback function!\n' break if self.interactiveMode: if sys_platform[:3] == "win": if msvcrt.kbhit(): if ord(msvcrt.getch()) == Consts.CDefESCKey: print "Loading modules for Interactive mode...", import pypso.Interaction print "done!\n" interact_banner = "## PyPSO v.%s - Interactive Mode ##\nPress CTRL-Z to quit interactive mode." % ( pypso.__version__, ) session_locals = { "pso_engine": self, "topology": self.getTopology(), "swarm_statistics": self.getTopology().swarmStats, "topology_statistics": self.getTopology().topologyStats, "pypso": pypso, "it": pypso.Interaction } print code.interact(interact_banner, local=session_locals) elif sys_platform[:5] == "linux": if Util.kbhit(): if ord(Util.getch()) == Consts.CDefESCKey: print "Loading modules for Interactive mode...", import pypso.Interaction print "done!\n" interact_banner = "## PyPSO v.%s - Interactive Mode ##\nPress CTRL-D to quit interactive mode." % ( pypso.__version__, ) session_locals = { "pso_engine": self, "topology": self.getTopology(), "swarm_statistics": self.getTopology().swarmStats, "topology_statistics": self.getTopology().topologyStats, "pypso": pypso, "it": pypso.Interaction } print code.interact(interact_banner, local=session_locals) except KeyboardInterrupt: print "\n\tA break was detected, you have interrupted the evolution !\n" if freq_stats != 0: self.printStats() self.printTimeElapsed() if self.reportAdapter: if (self.currentStep % self.reportAdapter.statsGenFreq == 0): self.dumpStatsReport() self.reportAdapter.saveAndClose()
class SimplePSO(object): """ SimplePSO Engine Class - The PSO Algorithm Core Example: >>> topology = Topology.GlobalTopology(particle_rep) >>> pso = PSO.SimplePSO(topology) >>> pso.setSteps(120) :param topology: the :term:`Sample Topology`` :param interactiveMode: this flag enables the Interactive Mode :param seed: the random seed value .. note:: if you see the same random seed, all the runs of the algorithm will be the same. """ stepCallBack = None """ This is the the :term: `step callback function` slot, if you want to set the function, you must do this: :: def your_func(pso_engine): #Here you have access to the PSO Engine return False pso_engine.stepCallback.set(your_func) now *"your_func"* will be called every step. When this function returns True, the PSO Engine will stop the evolution and show a warning, if is False, the evolution continues. """ terminationCriteria = None """ This is the termination criteria slot, if you want to set one termination criteria, you mus do this: :: pso_engine.terminationCriteria.set(your_func) Now, when you run your PSO, it will stop when terminationCriteria be satisfied. To create your own termination function, you must put at least one parameter which is the PSO Engine, follows an example: :: def ConvergenceCriteria(pso_engine): swarm = pso_engine.getSwarm() return swarm[0] == swarm[len(swarm)-1] When this function returns True, the Pso Engine will stop the evolution and show a warning. If is False, the evolution continues, this function is called every step. """ def __init__(self,topology,seed=None,interactiveMode=True): """ Initializator of PSO """ #random seed random.seed(seed) #Pso type used by the particle self.psoType = Consts.CDefPsoType #Topology used self.topology = topology #Set the population size self.setSwarmSize(Consts.CDefSwarmSize) #Cognitive and Social Coefficients self.C1,self.C2 = Consts.CDefCoefficients #Time steps self.timeSteps = Consts.CDefSteps #Interactive Mode (True or False) self.interactiveMode = interactiveMode #Current step self.currentStep = 0 #Inertia Factor Minus self.inertiaFactorMinus = None #Inertia coefficient self.inertiaFactor = None #Time initial self.time_init = None #Optimization type self.minimax = Consts.minimaxType["minimize"] #Report file adapter self.reportAdapter = None #Step Callback self.stepCallback = FunctionSlot("Step Callback") #Termination Criteria self.terminationCriteria = FunctionSlot("Termination Criteria") #All slots self.allSlots = [self.stepCallback, self.terminationCriteria] print "A PSO Engine was created, timeSteps=% d" % ( self.timeSteps, ) def __repr__(self): """ The String representation of the PSO Engine """ ret = "- PSO-%s-%s Execution\n" % (self.getTopologyType(),self.getPsoType()) ret += "\tSwarm Size:\t %d\n" % (self.topology.swarmSize,) ret += "\tTime Steps:\t %d\n" % (self.timeSteps,) ret += "\tCurrent Step:\t %d\n" % (self.currentStep,) ret += "\tMinimax Type:\t %s\n" % (Consts.minimaxType.keys()[Consts.minimaxType.values().index(self.minimax)].capitalize(),) ret += "\tReport Adapter:\t %s\n" % (self.reportAdapter,) for slot in self.allSlots: ret += "\t" + slot.__repr__() ret +="\n" return ret def setReportAdapter(self,repadapter): """ Sets the Report Adapter of the PSO Engine :param repadapter: one of the :mod:`ReportAdapters` classes instance .. warning: the use of a Report Adapter can reduce the speed performance of the PSO. """ self.reportAdapter = repadapter def setSwarmSize(self, size): """ Sets the swarm size, calls setSwarmSize() of Topology :param size: the swarm size .. note:: the swarm size must be >= 2 """ if size < 2: Util.raiseException("swarm size must be >= 2", ValueError) self.topology.setSwarmSize(size) def setPsoType(self,psoType): """ Sets the psoType, use Consts.psoType(Basic,Constricted,Inertia) Example: >>> pso_engine.setSortType(Consts.psoType["CONSTRICTED"]) :param psoType: The PSO type, from Consts.psoType """ if psoType not in Consts.psoType.values(): Util.raiseException("PsoType must be implemented !",TypeError) self.psoType = psoType def getPsoType(self): """ Return the Pso Type :rtype key: pso Type """ for key,value in Consts.psoType.items(): if value == self.psoType: return key return "" def setTimeSteps(self,num_steps): """ Sets the number of steps to converge :param num_steps: the number of steps """ if num_steps < 1: Util.raiseException("Number of steps must be >=1", ValueError) self.timeSteps = num_steps def getMinimax(self): """ Gets the minimize/maximize mode :rtype: The Consts.minimaxType type """ for key,value in Consts.minimaxType.items(): if value == self.minimax: return key return "" def getTopologyType(self): """ Returns the name of the topology :rtype name: the name of the topology """ return self.topology.__class__.__name__ def setMinimax(self,minimax): """Sets the minimize/maximize mode, use Consts.minimaxType :param minimax: the minimax mode, from Consts.minimaxType """ if minimax not in Consts.minimaxType.values(): Util.raiseException("Optimization type must be Maximize or Minimize !", TypeError) self.minimax = minimax def getCurrentStep(self): """ Gets the current step :rtype: the current step """ return self.currentStep def getReportAdapter(self): """ Gets the Report Adapter of the PSO Engine :rtype: a instance from one of the :mod:`ReportAdapters` classes """ return self.reportAdapter def bestParticle(self): """ Returns the swarm best Particle :rtype: the best particle """ return self.topology.getBestParticle() def getTopology(self): """Return the internal topology of Pso Engine :rtype: the topology (:class: 'Topology.Topology')' """ return self.topology def getStatistics(self): """ Gets the Statistics class instance of the current step :rtype: the statistics instance (:class: `Statistics.Statistics`)` """ return self.topology.getStatistics() def dumpStatsReport(self): """ Dumps the current statistics to the report adapter """ self.topology.statistics() self.reportAdapter.insert(self.getStatistics(),self.topology,self.currentStep) def printStats(self): """ Print swarm statistics""" percent = self.currentStep * 100 / float(self.timeSteps) message = "Step: %d (%.2f%%):" % (self.currentStep, percent) print message self.topology.statistics() self.topology.printStats() def printTimeElapsed(self): """ Shows the time elapsed since the beginning of the solution construction """ print "Total time elapsed: %.3f seconds." % (time()-self.time_init) def initialize(self): """ Initializes the PSO Engine. Create and initialize the swarm """ self.topology.create(minimax=self.minimax) self.topology.initialize() print "The PSO Engine was initialized !" def constructSolution(self): """ Just do one step in execution, one step.""" for it in self.topology.position_updater.applyFunctions(self): pass for it in self.topology.information_updater.applyFunctions(self): pass if self.psoType == Consts.psoType["INERTIA"]: self.updateInertiaFactor() self.currentStep += 1 return (self.currentStep == self.timeSteps) def execute(self, freq_stats=0): """ Do all the steps until the termination criteria or time Steps achieved, accepts the freq_stats (default is 0) to dump statistics at n-step Example: >>> pso_engine.evolve(freq_stats=10) (...) :param freq_stats: if greater than 0, the statistics will be printed every freq_stats step. """ #Start time self.time_init = time() #Creates a new report if reportAdapter is not None. if self.reportAdapter: self.reportAdapter.open() #Initialize the PSO Engine self.initialize() #Already evaluates all particles print "Starting loop over evolutionary algorithm." try: while not self.constructSolution(): stopFlagCallback = False stopFlagTerminationCriteria = False if not self.stepCallback.isEmpty(): for it in self.stepCallback.applyFunctions(self): stopFlagCallback = it if not self.terminationCriteria.isEmpty(): for it in self.terminationCriteria.applyFunctions(self): stopFlagTerminationCriteria = it if freq_stats != 0: if (self.currentStep % freq_stats == 0) or (self.currentStep == 1): self.printStats() if self.reportAdapter: if self.currentStep % self.reportAdapter.statsGenFreq == 0: self.dumpStatsReport() if stopFlagTerminationCriteria: print '\n\tExecution stopped by Termination Criteria function !\n' break if stopFlagCallback: print '\n\tExecution stopped by Step Callback function!\n' break if self.interactiveMode: if sys_platform[:3] == "win": if msvcrt.kbhit(): if ord(msvcrt.getch()) == Consts.CDefESCKey: print "Loading modules for Interactive mode...", import pypso.Interaction print "done!\n" interact_banner = "## PyPSO v.%s - Interactive Mode ##\nPress CTRL-Z to quit interactive mode." % (pypso.__version__,) session_locals = { "pso_engine" : self, "topology" : self.getTopology(), "swarm_statistics": self.getTopology().swarmStats, "topology_statistics": self.getTopology().topologyStats, "pypso" : pypso , "it" : pypso.Interaction} print code.interact(interact_banner, local=session_locals) elif sys_platform[:5] == "linux": if Util.kbhit(): if ord(Util.getch()) == Consts.CDefESCKey: print "Loading modules for Interactive mode...", import pypso.Interaction print "done!\n" interact_banner = "## PyPSO v.%s - Interactive Mode ##\nPress CTRL-D to quit interactive mode." % (pypso.__version__,) session_locals = { "pso_engine" : self, "topology" : self.getTopology(), "swarm_statistics": self.getTopology().swarmStats, "topology_statistics": self.getTopology().topologyStats, "pypso" : pypso , "it" : pypso.Interaction} print code.interact(interact_banner, local=session_locals) except KeyboardInterrupt: print "\n\tA break was detected, you have interrupted the evolution !\n" if freq_stats != 0: self.printStats() self.printTimeElapsed() if self.reportAdapter: if (self.currentStep % self.reportAdapter.statsGenFreq == 0): self.dumpStatsReport() self.reportAdapter.saveAndClose()
class MigrationScheme(object): """ This is the base class for all migration schemes """ selector = None """ This is the function slot for the selection method if you want to change the default selector, you must do this: :: migration_scheme.selector.set(Selectors.GRouletteWheel) """ def __init__(self): self.selector = FunctionSlot("Selector") self.GAEngine = None self.nMigrationRate = Consts.CDefGenMigrationRate self.nIndividuals = Consts.CDefMigrationNIndividuals self.nReplacement = Consts.CDefGenMigrationReplacement self.networkCompression = 9 def isReady(self): """ Returns true if is time to migrate """ return True if self.GAEngine.getCurrentGeneration( ) % self.nMigrationRate == 0 else False def getCompressionLevel(self): """ Get the zlib compression level of network data The values are in the interval described on the :func:`Network.pickleAndCompress` """ return self.networkCompression def setCompressionLevel(self, level): """ Set the zlib compression level of network data The values are in the interval described on the :func:`Network.pickleAndCompress` :param level: the zlib compression level """ self.networkCompression = level def getNumReplacement(self): """ Return the number of individuals that will be replaced in the migration process """ return self.nReplacement def setNumReplacement(self, num_individuals): """ Return the number of individuals that will be replaced in the migration process :param num_individuals: the number of individuals to be replaced """ self.nReplacement = num_individuals def getNumIndividuals(self): """ Return the number of individuals that will migrate :rtype: the number of individuals to be replaced """ return self.nIndividuals def setNumIndividuals(self, num_individuals): """ Set the number of individuals that will migrate :param num_individuals: the number of individuals """ self.nIndividuals = num_individuals def setMigrationRate(self, generations): """ Sets the generation frequency supposed to migrate and receive individuals. :param generations: the number of generations """ self.nMigrationRate = generations def getMigrationRate(self): """ Return the the generation frequency supposed to migrate and receive individuals :rtype: the number of generations """ return self.nMigrationRate def setGAEngine(self, ga_engine): """ Sets the GA Engine handler """ self.GAEngine = ga_engine def start(self): """ Initializes the migration scheme """ pass def stop(self): """ Stops the migration engine """ pass def select(self): """ Pickes an individual from population using specific selection method :rtype: an individual object """ if self.selector.isEmpty(): return self.GAEngine.select(popID=self.GAEngine.currentGeneration) else: for it in self.selector.applyFunctions( self.GAEngine.internalPop, popID=self.GAEngine.currentGeneration): return it def selectPool(self, num_individuals): """ Select num_individuals number of individuals and return a pool :param num_individuals: the number of individuals to select :rtype: list with individuals """ pool = [self.select() for i in xrange(num_individuals)] return pool def exchange(self): """ Exchange individuals """ pass
class GSimpleGA: """ GA Engine Class - The Genetic Algorithm Core Example: >>> ga = GSimpleGA.GSimpleGA(genome) >>> ga.selector.set(Selectors.GRouletteWheel) >>> ga.setGenerations(120) >>> ga.terminationCriteria.set(GSimpleGA.ConvergenceCriteria) :param genome: the :term:`Sample Genome` :param interactiveMode: this flag enables the Interactive Mode, the default is True :param seed: the random seed value .. note:: if you use the same random seed, all the runs of algorithm will be the same """ selector = None """ This is the function slot for the selection method if you want to change the default selector, you must do this: :: ga_engine.selector.set(Selectors.GRouletteWheel) """ stepCallback = None """ This is the :term:`step callback function` slot, if you want to set the function, you must do this: :: def your_func(ga_engine): # Here you have access to the GA Engine return False ga_engine.stepCallback.set(your_func) now *"your_func"* will be called every generation. When this function returns True, the GA Engine will stop the evolution and show a warning, if is False, the evolution continues. """ terminationCriteria = None """ This is the termination criteria slot, if you want to set one termination criteria, you must do this: :: ga_engine.terminationCriteria.set(GSimpleGA.ConvergenceCriteria) Now, when you run your GA, it will stop when the population converges. There are those termination criteria functions: :func:`GSimpleGA.RawScoreCriteria`, :func:`GSimpleGA.ConvergenceCriteria`, :func:`GSimpleGA.RawStatsCriteria`, :func:`GSimpleGA.FitnessStatsCriteria` But you can create your own termination function, this function receives one parameter which is the GA Engine, follows an example: :: def ConvergenceCriteria(ga_engine): pop = ga_engine.getPopulation() return pop[0] == pop[len(pop)-1] When this function returns True, the GA Engine will stop the evolution and show a warning, if is False, the evolution continues, this function is called every generation. """ def __init__(self, genome, seed=None, interactiveMode=True): """ Initializator of GSimpleGA """ if seed: random.seed(seed) if type(interactiveMode) != BooleanType: Util.raiseException( "Interactive Mode option must be True or False", TypeError) if not isinstance(genome, GenomeBase): Util.raiseException("The genome must be a GenomeBase subclass", TypeError) self.internalPop = GPopulation(genome) self.nGenerations = Consts.CDefGAGenerations self.pMutation = Consts.CDefGAMutationRate self.pCrossover = Consts.CDefGACrossoverRate self.nElitismReplacement = Consts.CDefGAElitismReplacement self.setPopulationSize(Consts.CDefGAPopulationSize) self.minimax = Consts.minimaxType["maximize"] self.elitism = True # Adapters self.dbAdapter = None self.migrationAdapter = None self.time_init = None self.interactiveMode = interactiveMode self.interactiveGen = -1 self.GPMode = False self.selector = FunctionSlot("Selector") self.stepCallback = FunctionSlot("Generation Step Callback") self.terminationCriteria = FunctionSlot("Termination Criteria") self.selector.set(Consts.CDefGASelector) self.allSlots = [ self.selector, self.stepCallback, self.terminationCriteria ] self.internalParams = {} self.currentGeneration = 0 # GP Testing for classes in Consts.CDefGPGenomes: if isinstance(self.internalPop.oneSelfGenome, classes): self.setGPMode(True) break logging.debug("A GA Engine was created, nGenerations=%d", self.nGenerations) def setGPMode(self, bool_value): """ Sets the Genetic Programming mode of the GA Engine :param bool_value: True or False """ self.GPMode = bool_value def getGPMode(self): """ Get the Genetic Programming mode of the GA Engine :rtype: True or False """ return self.GPMode def __call__(self, *args, **kwargs): """ A method to implement a callable object Example: >>> ga_engine(freq_stats=10) .. versionadded:: 0.6 The callable method. """ if kwargs.get("freq_stats", None): return self.evolve(kwargs.get("freq_stats")) else: return self.evolve() def setParams(self, **args): """ Set the internal params Example: >>> ga.setParams(gp_terminals=['x', 'y']) :param args: params to save ..versionaddd:: 0.6 Added the *setParams* method. """ self.internalParams.update(args) def getParam(self, key, nvl=None): """ Gets an internal parameter Example: >>> ga.getParam("gp_terminals") ['x', 'y'] :param key: the key of param :param nvl: if the key doesn't exist, the nvl will be returned ..versionaddd:: 0.6 Added the *getParam* method. """ return self.internalParams.get(key, nvl) def setInteractiveGeneration(self, generation): """ Sets the generation in which the GA must enter in the Interactive Mode :param generation: the generation number, use "-1" to disable .. versionadded::0.6 The *setInteractiveGeneration* method. """ if generation < -1: Util.raiseException("Generation must be >= -1", ValueError) self.interactiveGen = generation def getInteractiveGeneration(self): """ returns the generation in which the GA must enter in the Interactive Mode :rtype: the generation number or -1 if not set .. versionadded::0.6 The *getInteractiveGeneration* method. """ return self.interactiveGen def setElitismReplacement(self, numreplace): """ Set the number of best individuals to copy to the next generation on the elitism :param numreplace: the number of individuals .. versionadded:: 0.6 The *setElitismReplacement* method. """ if numreplace < 1: Util.raiseException("Replacement number must be >= 1", ValueError) self.nElitismReplacement = numreplace def setInteractiveMode(self, flag=True): """ Enable/disable the interactive mode :param flag: True or False .. versionadded: 0.6 The *setInteractiveMode* method. """ if type(flag) != BooleanType: Util.raiseException( "Interactive Mode option must be True or False", TypeError) self.interactiveMode = flag def __repr__(self): """ The string representation of the GA Engine """ ret = "- GSimpleGA\n" ret += "\tGP Mode:\t\t %s\n" % self.getGPMode() ret += "\tPopulation Size:\t %d\n" % (self.internalPop.popSize, ) ret += "\tGenerations:\t\t %d\n" % (self.nGenerations, ) ret += "\tCurrent Generation:\t %d\n" % (self.currentGeneration, ) ret += "\tMutation Rate:\t\t %.2f\n" % (self.pMutation, ) ret += "\tCrossover Rate:\t\t %.2f\n" % (self.pCrossover, ) ret += "\tMinimax Type:\t\t %s\n" % (Consts.minimaxType.keys()[ Consts.minimaxType.values().index(self.minimax)].capitalize(), ) ret += "\tElitism:\t\t %s\n" % (self.elitism, ) ret += "\tElitism Replacement:\t %d\n" % (self.nElitismReplacement, ) ret += "\tDB Adapter:\t\t %s\n" % (self.dbAdapter, ) for slot in self.allSlots: ret += "\t" + slot.__repr__() ret += "\n" return ret def setMultiProcessing(self, flag=True, full_copy=False, number_of_processes=-1): """ Sets the flag to enable/disable the use of python multiprocessing module. Use this option when you have more than one core on your CPU and when your evaluation function is very slow. Pyevolve will automaticly check if your Python version has **multiprocessing** support and if you have more than one single CPU core. If you don't have support or have just only one core, Pyevolve will not use the **multiprocessing** feature. Pyevolve uses the **multiprocessing** to execute the evaluation function over the individuals, so the use of this feature will make sense if you have a truly slow evaluation function (which is commom in GAs). The parameter "full_copy" defines where the individual data should be copied back after the evaluation or not. This parameter is useful when you change the individual in the evaluation function. :param flag: True (default) or False :param full_copy: True or False (default) :param number_of_processes: None = use the default, or specify the number .. warning:: Use this option only when your evaluation function is slow, so you'll get a good tradeoff between the process communication speed and the parallel evaluation. The use of the **multiprocessing** doesn't means always a better performance. .. note:: To enable the multiprocessing option, you **MUST** add the *__main__* check on your application, otherwise, it will result in errors. See more on the `Python Docs <http://docs.python.org/library/multiprocessing.html#multiprocessing-programming>`__ site. .. versionadded:: 0.6 The `setMultiProcessing` method. """ if type(flag) != BooleanType: Util.raiseException("Multiprocessing option must be True or False", TypeError) if type(full_copy) != BooleanType: Util.raiseException( "Multiprocessing 'full_copy' option must be True or False", TypeError) self.internalPop.setMultiProcessing(flag, full_copy, number_of_processes) def setMigrationAdapter(self, migration_adapter=None): """ Sets the Migration Adapter .. versionadded:: 0.6 The `setMigrationAdapter` method. """ if (migration_adapter is not None) and (not isinstance( migration_adapter, MigrationScheme)): Util.raiseException( "The Migration Adapter must be a MigrationScheme subclass", TypeError) self.migrationAdapter = migration_adapter if self.migrationAdapter is not None: self.migrationAdapter.setGAEngine(self) def setDBAdapter(self, dbadapter=None): """ Sets the DB Adapter of the GA Engine :param dbadapter: one of the :mod:`DBAdapters` classes instance .. warning:: the use the of a DB Adapter can reduce the speed performance of the Genetic Algorithm. """ if (dbadapter is not None) and (not isinstance(dbadapter, DBBaseAdapter)): Util.raiseException( "The DB Adapter must be a DBBaseAdapter subclass", TypeError) self.dbAdapter = dbadapter def setPopulationSize(self, size): """ Sets the population size, calls setPopulationSize() of GPopulation :param size: the population size .. note:: the population size must be >= 2 """ if size < 2: Util.raiseException("population size must be >= 2", ValueError) self.internalPop.setPopulationSize(size) def setSortType(self, sort_type): """ Sets the sort type, Consts.sortType["raw"]/Consts.sortType["scaled"] Example: >>> ga_engine.setSortType(Consts.sortType["scaled"]) :param sort_type: the Sort Type """ if sort_type not in Consts.sortType.values(): Util.raiseException("sort type must be a Consts.sortType type", TypeError) self.internalPop.sortType = sort_type def setMutationRate(self, rate): """ Sets the mutation rate, between 0.0 and 1.0 :param rate: the rate, between 0.0 and 1.0 """ if (rate > 1.0) or (rate < 0.0): Util.raiseException("Mutation rate must be >= 0.0 and <= 1.0", ValueError) self.pMutation = rate def setPreMutationRate(self, rate): """ Sets the pre-mutation rate, > 0.0. Premutation is done only when no cross-over occurs. :param rate: the rate, >= 0.0 """ if (rate < 0.0): Util.raiseException("Pre-mutation rate must be >= 0.0", ValueError) self.pPreMutation = rate def setCrossoverRate(self, rate): """ Sets the crossover rate, between 0.0 and 1.0 :param rate: the rate, between 0.0 and 1.0 """ if (rate > 1.0) or (rate < 0.0): Util.raiseException("Crossover rate must be >= 0.0 and <= 1.0", ValueError) self.pCrossover = rate def setGenerations(self, num_gens): """ Sets the number of generations to evolve :param num_gens: the number of generations """ if num_gens < 1: Util.raiseException("Number of generations must be >= 1", ValueError) self.nGenerations = num_gens def getGenerations(self): """ Return the number of generations to evolve :rtype: the number of generations .. versionadded:: 0.6 Added the *getGenerations* method """ return self.nGenerations def getMinimax(self): """ Gets the minimize/maximize mode :rtype: the Consts.minimaxType type """ return self.minimax def setMinimax(self, mtype): """ Sets the minimize/maximize mode, use Consts.minimaxType :param mtype: the minimax mode, from Consts.minimaxType """ if mtype not in Consts.minimaxType.values(): Util.raiseException("Minimax must be maximize or minimize", TypeError) self.minimax = mtype def getCurrentGeneration(self): """ Gets the current generation :rtype: the current generation """ return self.currentGeneration def setElitism(self, flag): """ Sets the elitism option, True or False :param flag: True or False """ if type(flag) != BooleanType: Util.raiseException("Elitism option must be True or False", TypeError) self.elitism = flag def getDBAdapter(self): """ Gets the DB Adapter of the GA Engine :rtype: a instance from one of the :mod:`DBAdapters` classes """ return self.dbAdapter def bestIndividual(self): """ Returns the population best individual :rtype: the best individual """ return self.internalPop.bestRaw() def __gp_catch_functions(self, prefix): """ Internally used to catch functions with some specific prefix as non-terminals of the GP core """ import __main__ as mod_main function_set = {} main_dict = mod_main.__dict__ for obj, addr in main_dict.items(): if obj[0:len(prefix)] == prefix: try: op_len = addr.func_code.co_argcount except: continue function_set[obj] = op_len if len(function_set) <= 0: Util.raiseException( "No function set found using function prefix '%s' !" % prefix, ValueError) self.setParams(gp_function_set=function_set) def initialize(self): """ Initializes the GA Engine. Create and initialize population """ self.internalPop.create(minimax=self.minimax) self.internalPop.initialize(ga_engine=self) logging.debug("The GA Engine was initialized !") def getPopulation(self): """ Return the internal population of GA Engine :rtype: the population (:class:`GPopulation.GPopulation`) """ return self.internalPop def getStatistics(self): """ Gets the Statistics class instance of current generation :rtype: the statistics instance (:class:`Statistics.Statistics`) """ return self.internalPop.getStatistics() def clear(self): """ Petrowski's Clearing Method """ def step(self): """ Just do one step in evolution, one generation """ genomeMom = None genomeDad = None newPop = GPopulation(self.internalPop) logging.debug("Population was cloned.") size_iterate = len(self.internalPop) # Odd population size if size_iterate % 2 != 0: size_iterate -= 1 #Check on the crossover function by picking a random individual - is it empty? crossover_empty = self.select( popID=self.currentGeneration).crossover.isEmpty() for i in xrange(0, size_iterate, 2): #Ok, we select 2 parents using the selector (RouletteWheel, etc.) genomeMom = self.select(popID=self.currentGeneration) genomeDad = self.select(popID=self.currentGeneration) if not crossover_empty and self.pCrossover >= 1.0: #Crossover all of them for it in genomeMom.crossover.applyFunctions(mom=genomeMom, dad=genomeDad, count=2): (sister, brother) = it else: #Filp a coin each time to determine if you should crossover if not crossover_empty and Util.randomFlipCoin( self.pCrossover): for it in genomeMom.crossover.applyFunctions(mom=genomeMom, dad=genomeDad, count=2): (sister, brother) = it else: sister = genomeMom.clone() brother = genomeDad.clone() #And "pre" mutate them sister.premutate(pmut=self.pPreMutation, ga_engine=self) brother.premutate(pmut=self.pPreMutation, ga_engine=self) #Now each offspring is mutated sister.mutate(pmut=self.pMutation, ga_engine=self) brother.mutate(pmut=self.pMutation, ga_engine=self) newPop.internalPop.append(sister) newPop.internalPop.append(brother) if len(self.internalPop) % 2 != 0: #Odd-numbered population genomeMom = self.select(popID=self.currentGeneration) genomeDad = self.select(popID=self.currentGeneration) if Util.randomFlipCoin(self.pCrossover): for it in genomeMom.crossover.applyFunctions(mom=genomeMom, dad=genomeDad, count=1): (sister, brother) = it else: sister = random.choice([genomeMom, genomeDad]) sister = sister.clone() #Do the 2 mutations sister.premutate(pmut=self.pPreMutation, ga_engine=self) sister.mutate(pmut=self.pMutation, ga_engine=self) newPop.internalPop.append(sister) #---- Evaluate fitness ------ logging.debug("Evaluating the new created population.") newPop.evaluate() #Niching methods- Petrowski's clearing self.clear() if self.elitism: #Avoid too much elitism if self.nElitismReplacement >= len(self.internalPop): self.nElitismReplacement = len(self.internalPop) - 1 logging.debug("Doing elitism.") if self.getMinimax() == Consts.minimaxType["maximize"]: #Replace the n-th worst new ones with the nth best old ones for i in xrange(self.nElitismReplacement): if self.internalPop.bestRaw(i).score > newPop.bestRaw( i).score: newPop[len(newPop) - 1 - i] = self.internalPop.bestRaw(i) elif self.getMinimax() == Consts.minimaxType["minimize"]: for i in xrange(self.nElitismReplacement): if self.internalPop.bestRaw(i).score < newPop.bestRaw( i).score: newPop[len(newPop) - 1 - i] = self.internalPop.bestRaw(i) self.internalPop = newPop self.internalPop.sort() logging.debug("The generation %d was finished.", self.currentGeneration) self.currentGeneration += 1 return (self.currentGeneration >= self.nGenerations) def printStats(self): """ Print generation statistics :rtype: the printed statistics as string .. versionchanged:: 0.6 The return of *printStats* method. """ percent = self.currentGeneration * 100 / float(self.nGenerations) message = "Gen. %d (%.2f%%):" % (self.currentGeneration, percent) logging.info(message) print message, sys_stdout.flush() self.internalPop.statistics() stat_ret = self.internalPop.printStats() return message + stat_ret def printTimeElapsed(self): """ Shows the time elapsed since the begin of evolution """ total_time = time() - self.time_init print "Total time elapsed: %.3f seconds." % total_time return total_time def dumpStatsDB(self): """ Dumps the current statistics to database adapter """ logging.debug("Dumping stats to the DB Adapter") self.internalPop.statistics() self.dbAdapter.insert(self) #---------------------------------------------------------- def evolve(self, freq_stats=0, skip_initialize=False): """ Do all the generations until the termination criteria, accepts the freq_stats (default is 0) to dump statistics at n-generation Example: >>> ga_engine.evolve(freq_stats=10) (...) :param freq_stats: if greater than 0, the statistics will be printed every freq_stats generation. :rtype: returns a tuple: best: the best individual of the evolution stopFlagCallback: stopped by the step_callback stopFlagTerminationCriteria: stopped by reching the criterion .. versionadded:: 0.6 the return of the best individual """ stopFlagCallback = False stopFlagTerminationCriteria = False self.time_init = time() logging.debug( "Starting the DB Adapter and the Migration Adapter if any") if self.dbAdapter: self.dbAdapter.open(self) if self.migrationAdapter: self.migrationAdapter.start() if self.getGPMode(): gp_function_prefix = self.getParam("gp_function_prefix") if gp_function_prefix is not None: self.__gp_catch_functions(gp_function_prefix) if not skip_initialize: #Create the population self.initialize() #Initial fitness evaluation self.internalPop.evaluate() self.internalPop.sort() logging.debug("Starting loop over evolutionary algorithm.") try: while True: if self.migrationAdapter: logging.debug("Migration adapter: exchange") self.migrationAdapter.exchange() self.internalPop.clearFlags() self.internalPop.sort() #The step callback is called before each step if not self.stepCallback.isEmpty(): for it in self.stepCallback.applyFunctions(self): stopFlagCallback = it if not self.terminationCriteria.isEmpty(): for it in self.terminationCriteria.applyFunctions(self): stopFlagTerminationCriteria = it if freq_stats: if (self.currentGeneration % freq_stats == 0) or (self.getCurrentGeneration() == 0): self.printStats() if self.dbAdapter: if self.currentGeneration % self.dbAdapter.getStatsGenFreq( ) == 0: self.dumpStatsDB() if stopFlagTerminationCriteria: logging.debug( "Evolution stopped by the Termination Criteria !") if freq_stats: print "\n\tEvolution stopped by Termination Criteria function !\n" break if stopFlagCallback: logging.debug( "Evolution stopped by Step Callback function !") if freq_stats: print "\n\tEvolution stopped by Step Callback function !\n" break #(Here was interactive mode code, removed) if self.step(): break #exit if the number of generations is equal to the max. number of gens. #(end While True) except KeyboardInterrupt: logging.debug("CTRL-C detected, finishing evolution.") if freq_stats: print "\n\tA break was detected, you have interrupted the evolution !\n" #Finished. Clean up the multiprocessing pool self.getPopulation().cleanupMultiProcessing() if freq_stats != 0: self.printStats() self.printTimeElapsed() if self.dbAdapter: logging.debug("Closing the DB Adapter") if not (self.currentGeneration % self.dbAdapter.getStatsGenFreq() == 0): self.dumpStatsDB() self.dbAdapter.commitAndClose() if self.migrationAdapter: logging.debug("Closing the Migration Adapter") if freq_stats: print "Stopping the migration adapter... ", self.migrationAdapter.stop() if freq_stats: print "done !" return (self.bestIndividual(), stopFlagCallback, stopFlagTerminationCriteria) def select(self, **args): """ Select one individual from population :param args: this parameters will be sent to the selector """ for it in self.selector.applyFunctions(self.internalPop, **args): return it