def info(self, plot=True):
eu.printParamDist(self.pop, self.paramInterval, self.gIdx)
if plot:
eu.printPopFitnessStats(self.pop,
self.paramInterval,
self.gIdx,
draw_scattermatrix=True)
bestN = 20
print("--------------------------")
print(f"Best {bestN} individuals:")
eu.printIndividuals(self.toolbox.selBest(self.pop, bestN),
self.paramInterval)
def runInitial(self):
"""First round of evolution:
"""
### Evaluate the initial population
logging.info("Evaluating initial population of size %i ..." %
len(self.pop))
self.gIdx = 0 # set generation index
# evaluate
self.pop = self.evalPopulationUsingPypet(self.traj, self.toolbox,
self.pop, self.gIdx)
if self.verbose:
eu.printParamDist(self.pop, self.paramInterval, self.gIdx)
self.pop = eu.saveToPypet(self.traj, self.pop, self.gIdx)
# Only the best indviduals are selected for the population the others do not survive
self.pop[:] = self.toolbox.selBest(self.pop, k=self.traj.popsize)
self.initialPopulationSimulated = True
def info(self, plot=True, bestN=5):
validPop = [
p for p in self.pop if not np.any(np.isnan(p.fitness.values))
]
popArray = np.array(
[p[0:len(self.paramInterval._fields)] for p in validPop]).T
scores = np.array(
[validPop[i].fitness.score for i in range(len(validPop))])
# Text output
print("--- Info summary ---")
print("Valid: {}".format(len(validPop)))
print("Mean score (weighted fitness): {:.2}".format(np.mean(scores)))
eu.printParamDist(self.pop, self.paramInterval, self.gIdx)
print("--------------------")
print(f"Best {bestN} individuals:")
eu.printIndividuals(self.toolbox.selBest(self.pop, bestN),
self.paramInterval)
print("--------------------")
# Plotting
if plot:
eu.plotPopulation(self.pop,
self.paramInterval,
self.gIdx,
draw_scattermatrix=True)
def runEvolution(self):
# Start evolution
logging.info("Start of evolution")
for self.gIdx in range(self.gIdx + 1, self.gIdx + self.traj.NGEN):
# ------- Weed out the invalid individuals and replace them by random new indivuals -------- #
validpop = self.getValidPopulation(self.pop)
# add all valid individuals to the population history
self.popHist[self.gIdx] = self.getValidPopulation(self.pop)
# replace invalid individuals
nanpop = self.getInvalidPopulation(self.pop)
logging.info("Replacing {} invalid individuals.".format(
len(nanpop)))
newpop = self.toolbox.population(n=len(nanpop))
newpop = self.tagPopulation(newpop)
# self.pop = validpop + newpop
# ------- Create the next generation by crossover and mutation -------- #
### Select parents using rank selection and clone them ###
offspring = list(
map(self.toolbox.clone,
self.toolbox.selRank(self.pop, self.POP_SIZE)))
##### cross-over ####
for i in range(1, len(offspring), 2):
offspring[i - 1], offspring[i] = self.toolbox.mate(
offspring[i - 1], offspring[i], indpb=self.CXPB)
# del offspring[i - 1].fitness, offspring[i].fitness
del offspring[i -
1].fitness.values, offspring[i].fitness.values
del offspring[i -
1].fitness.wvalues, offspring[i].fitness.wvalues
# print("Deleting ID of {} and {}".format(offspring[i - 1].id, offspring[i].id))
del offspring[i - 1].id, offspring[i].id
##### Mutation ####
# Apply adaptive mutation
du.mutateUntilValid(offspring, self.paramInterval, self.toolbox)
offspring = self.tagPopulation(offspring)
# ------- Evaluate next generation -------- #
logging.info("----------- Generation %i -----------" % self.gIdx)
self.pop = offspring + newpop
self.evalPopulationUsingPypet(self.traj, self.toolbox,
offspring + newpop, self.gIdx)
# ------- Select surviving population -------- #
# select next population
self.pop = self.toolbox.selBest(validpop + offspring + newpop,
k=self.traj.popsize)
self.best_ind = self.toolbox.selBest(self.pop, 1)[0]
if self.verbose:
print("----------- Generation %i -----------" % self.gIdx)
print("Best individual is {}".format(self.best_ind))
print("Score: {}".format(self.best_ind.fitness.score))
print("Fitness: {}".format(self.best_ind.fitness.values))
print("--- Population statistics ---")
eu.printParamDist(self.pop, self.paramInterval, self.gIdx)
eu.plotPopulation(self.pop,
self.paramInterval,
self.gIdx,
draw_scattermatrix=True,
save_plots=self.trajectoryName)
# save all simulation data to pypet
try:
self.pop = eu.saveToPypet(self.traj, self.pop, self.gIdx)
except:
logging.warn("Error: Write to pypet failed!")
logging.info("--- End of evolution ---")
self.best_ind = self.toolbox.selBest(self.pop, 1)[0]
logging.info("Best individual is %s, %s" %
(self.best_ind, self.best_ind.fitness.values))
logging.info("--- End of evolution ---")
self.traj.f_store(
) # We switched off automatic storing, so we need to store manually