def mutation(pop,numOfBest,mnm):
bestInGen = list()
nextGen = list()
child = list()
parentOne = list()
parentTwo = list()
# Copy over the top songs
for i in range(numOfBest):
bestInGen.append(pop[i][0][:])
nextGen.append(bestInGen[i])
neededSongs = len(pop) - numOfBest
for i in range(neededSongs):
# Randomly select two of the best songs
parentOne = bestInGen[r.randint(0,len(bestInGen)-1)]
parentTwo = bestInGen[r.randint(0,len(bestInGen)-1)]
subLen = r.randint(1,len(parentOne)-1)
pt = r.randint(0,len(parentOne) - subLen)
subList = parentOne[pt:pt+subLen]
pt = r.randint(0,len(parentOne) - subLen)
child = parentTwo[:]
child[pt:pt+subLen] = subList
# Have variable size songs (DOESN"T WORK NICELY)
# # Take a random sublist from parent one
# ptOne = r.randint(0,len(parentOne)-subLen)
# ptTwo = r.randint(0,len(parentOne))
# if ptOne > ptTwo: ptOne, ptTwo = ptTwo, ptOne
# subList = parentOne[ptOne: ptTwo]
# # splice it into a random section from parent two
# ptOne = r.randint(0,len(parentTwo))
# ptTwo = r.randint(0,len(parentTwo))
# if ptOne > ptTwo: ptOne, ptTwo = ptTwo, ptOne
# child = parentTwo[:]
# child[ptOne:ptTwo] = subList
# # Make sure child isn't empty
# if len(child) == 0: child.append(0)
# Single point mutations
for i in range(0,mnm):
note = r.randint(0,len(child)-1)
child[note] = m.setPitch(child[note],r.randint(30,110))
child[note] = m.setDur(child[note],r.randint(30,110))
# Add to next generation
nextGen.append(child)
return nextGen
def initPop(numTracks, numNotes, popSize):
song = list()
population = list()
note = 0
note = midiNote.setInst(note, 0)
for k in range(popSize):
del song[:]
for i in range(numNotes):
for j in range(numTracks):
note = midiNote.setChan(note, j+1)
note = midiNote.setTrack(note, j+1)
note = midiNote.setPitch(note, random.randint(40,90))
note = midiNote.setDur(note, random.randint(16,128))
song.append(note)
population.append(song[:]) # song[:] for splicing a copy of song
return population
