def colSelection(col_list, c0, c1):
nCols = len(col_list)
fitness_list = []
for colony in col_list:
fitness_list.append(fitness(colony, c0, c1))
totFit = sum(fitness_list)
# This array holds votes for each colony in col_list
colonyVotes = np.zeros(nCols)
# Iterate over the total number of new colonies to be created
for j in range(nCols):
# Assign random number over size of total fitness
temp1 = random.uniform(0, totFit)
# Initialize an accumulator
tot = 0
# Get cumulative sum of fitnesses to decide on colony
for i in range(nCols):
# Update cumulative sum
tot = tot + fitness_list[i]
# If less than the random number, increase vote
if temp1<tot:
colonyVotes[i] = colonyVotes[i] + 1
break
return colonyVotes
# Initialize a list of ant colonies
colony_list = colCreate(colCreateParms)
for i in range(evo_cycles):
# create a colony index
j = 0
# Initialize minimum to a high number
tempMin = 1e5
# Cycle each colony through TSP 'iters' times
for colony in colony_list:
# Submit to TSP cycling
colony = tsp(colony, iters, n_mat)
# Display update and store fitness of current colony/evo
print("evo= " + str(i) + " , colony= "+str(j))
fitness_mat[i,j]=fitness(colony, c0, c1)
# Get colony minimum for this evo and store it
iterMin = colony.minL
lMin_mat[i,j]=iterMin
lAvg_mat[i,j]=colony.L/colony.N
# Check if it's best so far from all colonies
# If so, store the index of the colony
if iterMin<tempMin:
tempMin = iterMin
bestIdx = j
# Increment colony index
j = j+1
# Store the alpha dist. for the best L_min
alphaBestMin.append(colony_list[bestIdx].getAlphaDist())