def run(self):
Tx, Ty = np.load('terminal_point_{}_{}.npy'.format(
str(self.dim), str(self.n)))
lenT = Tx.size
data_pso = dict()
distancevector = gd.get_distancevector(np.copy(Tx), np.copy(Ty))
mst_size = pa.get_tree(distancevector)
for i in range(self.max_iter):
print('Iteration No :', i)
t1 = t.time()
res = call_methods(np.copy(Tx), np.copy(Ty), lenT, self.func)
t2 = t.time()
data_pso[res[0]] = (res[1], res[2], t2 - t1)
min_pso = min(data_pso.keys())
fp = open(self.file_name, 'w')
fp.write("Size of the MST = " + str(mst_size) + '\n')
fp.write('No. of Iterations :' + str(self.max_iter) + '\n')
fp.write('PSO Min Wt :' + str(min_pso) + '\n')
fp.write('X Coordinates :' + str(data_pso[min_pso][0]) + '\n')
fp.write('Y Coordinates :' + str(data_pso[min_pso][1]) + '\n')
fp.write('No. of Steiner points :' +
str(data_pso[min_pso][0].size - self.n) + '\n')
fp.write('Time Required :' + str(data_pso[min_pso][2]) + '\n')
fp.write('Error Ratio :' + str(min_pso / mst_size) + '\n')
fp.close()
def call_methods(Tx, Ty, lenT, func):
# Setting the method into a method variable for dynamic linking
global get_velocity_position
get_velocity_position = func
# Calling the method that controls the main optimisation algorithm
bestparticle = particle_swarm_test(np.copy(Tx), np.copy(Ty), lenT)
# Extracting the position of the Steiner Points from the best particle
for i in range(bestparticle.Sx.size):
if bestparticle.Sx[i] < min(Tx) or bestparticle.Sx[i] > max(Tx):
continue
if bestparticle.Sy[i] < min(Ty) or bestparticle.Sy[i] > max(Ty):
continue
Tx = np.append(Tx, bestparticle.Sx[i])
Ty = np.append(Ty, bestparticle.Sy[i])
# Finding the MST of the final RSMT
distancevector = gd.get_distancevector(Tx, Ty)
mst = pa.mst_prim(distancevector)
mst_size = pa.get_tree(distancevector)
# Building the final return tuple for the use in the main program
return_set = (mst_size, Tx, Ty)
# Ploting the RSMT
pa.draw_gridgraph(Tx, Ty, mst, lenT, lenT)
return return_set
def get_fitness(Tx, Ty, particle):
for i in range(len(particle.Sx)):
if particle.Sx[i] < min(Tx) or particle.Sx[i] > max(Tx):
continue
if particle.Sy[i] < min(Ty) or particle.Sy[i] > max(Ty):
continue
Tx.append(particle.Sx[i])
Ty.append(particle.Sy[i])
distancevector = gd.get_distancevector(Tx, Ty)
objectivefitness = pa.get_tree(distancevector)
return objectivefitness
def get_fitness(Tx, Ty, weed):
for i in range(len(weed.Wx)):
if weed.Wx[i] < min(Tx) or weed.Wx[i] > max(Tx):
continue
if weed.Wy[i] < min(Ty) or weed.Wy[i] > max(Ty):
continue
Tx.append(weed.Wx[i])
Ty.append(weed.Wy[i])
distancevector = gd.get_distancevector(Tx, Ty)
objectivefitness = pa.get_tree(distancevector)
return objectivefitness
def call_methods(Tx, Ty, lenTx, lenTy):
temp_len = Tx.size
Rx = Tx[0:temp_len - lenTx]
Ry = Ty[0:temp_len - lenTy]
len_Rx = Rx.size
Tx = Tx[0:temp_len - len_Rx]
Ty = Ty[0:temp_len - len_Rx]
#print("X coordinates =", Tx)
#print("Y coordinates =", Ty)
bestparticle = particle_swarm_test(Tx, Ty)
temp_len = Tx.size
Rx = Tx[0:temp_len - lenTx]
Ry = Ty[0:temp_len - lenTy]
len_Rx = Rx.size
Tx = Tx[0:temp_len - len_Rx]
Ty = Ty[0:temp_len - len_Rx]
#print("Updated X coordinates", Tx)
#print("Updated Y coordinates", Ty)
'''print("X coordinates of best particle", bestparticle.Sx)
print("Y coordinates of best particle", bestparticle.Sy)
print("Updated X coordinates", Tx)
print("Updated Y Coordinates", Ty)'''
count = 0
for i in range(len(bestparticle.Sx)):
if bestparticle.Sx[i] < min(Tx) or bestparticle.Sx[i] > max(Tx):
continue
if bestparticle.Sy[i] < min(Ty) or bestparticle.Sy[i] > max(Ty):
continue
#Tx.append(bestparticle.Sx[i])
#Ty.append(bestparticle.Sy[i])
Tx = np.append(Tx, bestparticle.Sx[i])
Ty = np.append(Ty, bestparticle.Sy[i])
count = count + 1
#print("Updated X Coordinates", Tx)
#print("Updated Y Coordinates", Ty)
distancevector = gd.get_distancevector(Tx, Ty)
mst = pa.mst_prim(distancevector)
mst_size = pa.get_tree(distancevector)
#print("Size of Steiner Tree for PSO", mst_size)
return_set = (mst_size, Tx, Ty)
#pa.draw_gridgraph(Tx, Ty, mst,lenTx,lenTx)
temp_len = Tx.size
Tx = Tx[0:temp_len - count]
Ty = Ty[0:temp_len - count]
'''print("Restored X Coordinates=", Tx)
print("Restored Y Coordinates=", Ty)
print("End of PSO")'''
return return_set
def iwo_test(Tx, Ty):
lenTx = len(Tx)
lenTy = len(Ty)
weedlist = []
for i in range(10):
Wx = gd.get_xdata(0, 500, 8)
Wy = gd.get_ydata(0, 500, 8)
Xp = reduce((lambda x, y: x + y), Wx) // len(Wx)
Yp = reduce((lambda x, y: x + y), Wy) // len(Wy)
weed = Weed(Wx, Wy, Xp, Yp, 0)
weedlist.append(weed)
# iwo_algorithm(Tx,Ty,weedlist,itermax,pmax,Smax,Smin,sigmafinal,sigmainit,n)
wlist = iwo_algorithm(Tx, Ty, weedlist, 5, 150, 10, 1, 0.01, 1, 3)
fmax = get_max_fitness(wlist)
for weed in wlist:
if fmax == weed.fitness:
bestweed = weed
break
Rx = Tx[0:len(Tx) - lenTx]
Ry = Ty[0:len(Ty) - lenTy]
Tx = Tx[0:len(Tx) - len(Rx)]
Ty = Ty[0:len(Ty) - len(Ry)]
print("X-Coordinates of best weed",
list(map(lambda x: math.ceil(x), bestweed.Wx)))
print("Y-Coordinates of best weed",
list(map(lambda y: math.ceil(y), bestweed.Wy)))
count = 0
bestweed.Wx = list(map(lambda x: math.ceil(x), bestweed.Wx))
bestweed.Wy = list(map(lambda y: math.ceil(y), bestweed.Wy))
for i in range(len(bestweed.Wx)):
if bestweed.Wx[i] < min(Tx) or bestweed.Wx[i] > max(Tx):
continue
if bestweed.Wy[i] < min(Ty) or bestweed.Wy[i] > max(Ty):
continue
Tx.append(bestweed.Wx[i])
Ty.append(bestweed.Wy[i])
count = count + 1
print("Updated X Coordinates", Tx)
print("Updated Y Coordinates", Ty)
distancevector = gd.get_distancevector(Tx, Ty)
mst = pa.mst_prim(distancevector)
mst_size = pa.get_tree(distancevector)
print("Size of Steiner Tree for IWO", mst_size)
pa.draw_gridgraph(Tx, Ty, mst, lenTx, lenTy)
Tx = Tx[0:len(Tx) - count]
Ty = Ty[0:len(Ty) - count]
print("Restored X Coordinates=", Tx)
print("Restored Y Coordinates=", Ty)
print("End of IWO")
def get_fitness(Tx, Ty, pigeon):
for i in range(pigeon.Sx.size):
if pigeon.Sx[i] < min(Tx) or pigeon.Sx[i] > max(Tx):
continue
if pigeon.Sy[i] < min(Ty) or pigeon.Sy[i] > max(Ty):
continue
Tx = np.append(Tx, pigeon.Sx[i])
#print(Tx)
Ty = np.append(Ty, pigeon.Sy[i])
#print(Ty)
distancevector = gd.get_distancevector(Tx, Ty)
objectivefitness = pa.get_tree(distancevector)
return objectivefitness
def call_methods(Tx, Ty, lenTx, lenTy):
# Tx = gd.get_xdata(0, 500, 10)
# Ty = gd.get_ydata(0, 500, 10)
'''
lenTx = len(Tx)
lenTy = len(Ty)
'''
# Calculating mST for PIO
#print("X coordinates =", Tx)
#print("Y coordinates =", Ty)
bestpigeon = pigeon_test(Tx, Ty)
temp_len = Tx.size
Rx = Tx[0:temp_len - lenTx]
len_Rx = Rx.size
Tx = Tx[0:temp_len - len_Rx]
Ty = Ty[0:temp_len - len_Rx]
# print("Updated X coordinates", Tx)
# print("Updated Y coordinates", Ty)
'''print("X coordinates of best pigeon", bestpigeon.Sx)
print("Y coordinates of best pigeon", bestpigeon.Sy)
print("Updated X coordinates", Tx)
print("Updated Y Coordinates", Ty)'''
count = 0
for i in range(len(bestpigeon.Sx)):
if bestpigeon.Sx[i] < min(Tx) or bestpigeon.Sx[i] > max(Tx):
continue
if bestpigeon.Sy[i] < min(Ty) or bestpigeon.Sy[i] > max(Ty):
continue
#Tx.append(math.floor(bestpigeon.Sx[i]))
#Ty.append(math.floor(bestpigeon.Sy[i]))
Tx = np.append(Tx, np.floor(bestpigeon.Sx[i]))
Ty = np.append(Ty, np.floor(bestpigeon.Sy[i]))
count = count + 1
#print("Updated X Coordinates", Tx)
#print("Updated Y Coordinates", Ty)
distancevector = gd.get_distancevector(Tx, Ty)
mst = pa.mst_prim(distancevector)
mst_size = pa.get_tree(distancevector)
#print("Size of Steiner Tree for PIO", mst_size)
#pa.draw_gridgraph(Tx, Ty, mst, lenTx, lenTy)
return_set = (mst_size, Tx, Ty)
Tx = Tx[0:len(Tx) - count]
Ty = Ty[0:len(Ty) - count]
'''print("Restored X Coordinates=", Tx)
print("Restored Y Coordinates=", Ty)
print("End of PIO")'''
return return_set
def call_methods(Tx, Ty, lenTx, lenTy):
'''
lenTx = len(Tx)
lenTy = len(Ty)
'''
Rx = Tx[0:len(Tx) - lenTx]
Ry = Ty[0:len(Tx) - lenTy]
Tx = Tx[0:len(Tx) - len(Rx)]
Ty = Ty[0:len(Ty) - len(Ry)]
print("X coordinates =", Tx)
print("Y coordinates =", Ty)
bestparticle = particle_swarm_test(Tx, Ty)
Rx = Tx[0:len(Tx) - lenTx]
Ry = Ty[0:len(Ty) - lenTy]
Tx = Tx[0:len(Tx) - len(Rx)]
Ty = Ty[0:len(Ty) - len(Ry)]
#print("Updated X coordinates", Tx)
#print("Updated Y coordinates", Ty)
print("X coordinates of best particle", bestparticle.Sx)
print("Y coordinates of best particle", bestparticle.Sy)
print("Updated X coordinates", Tx)
print("Updated Y Coordinates", Ty)
count = 0
for i in range(len(bestparticle.Sx)):
if bestparticle.Sx[i] < min(Tx) or bestparticle.Sx[i] > max(Tx):
continue
if bestparticle.Sy[i] < min(Ty) or bestparticle.Sy[i] > max(Ty):
continue
Tx.append(bestparticle.Sx[i])
Ty.append(bestparticle.Sy[i])
count = count + 1
print("Updated X Coordinates", Tx)
print("Updated Y Coordinates", Ty)
distancevector = gd.get_distancevector(Tx, Ty)
mst = pa.mst_prim(distancevector)
mst_size = pa.get_tree(distancevector)
print("Size of Steiner Tree for PSO", mst_size)
pa.draw_gridgraph(Tx, Ty, mst, lenTx, lenTx)
Tx = Tx[0:len(Tx) - count]
Ty = Ty[0:len(Ty) - count]
print("Restored X Coordinates=", Tx)
print("Restored Y Coordinates=", Ty)
print("End of PSO")
def call_methods(Tx, Ty, lenTx, lenTy):
# Tx = gd.get_xdata(0, 500, 10)
# Ty = gd.get_ydata(0, 500, 10)
'''
lenTx = len(Tx)
lenTy = len(Ty)
'''
# Calculating mST for PIO
print("X coordinates =", Tx)
print("Y coordinates =", Ty)
bestpigeon = pigeon_test(Tx, Ty)
Rx = Tx[0:len(Tx) - lenTx]
Ry = Ty[0:len(Ty) - lenTy]
Tx = Tx[0:len(Tx) - len(Rx)]
Ty = Ty[0:len(Ty) - len(Ry)]
# print("Updated X coordinates", Tx)
# print("Updated Y coordinates", Ty)
print("X coordinates of best pigeon", bestpigeon.Sx)
print("Y coordinates of best pigeon", bestpigeon.Sy)
print("Updated X coordinates", Tx)
print("Updated Y Coordinates", Ty)
count = 0
for i in range(len(bestpigeon.Sx)):
if bestpigeon.Sx[i] < min(Tx) or bestpigeon.Sx[i] > max(Tx):
continue
if bestpigeon.Sy[i] < min(Ty) or bestpigeon.Sy[i] > max(Ty):
continue
Tx.append(math.floor(bestpigeon.Sx[i]))
Ty.append(math.floor(bestpigeon.Sy[i]))
count = count + 1
print("Updated X Coordinates", Tx)
print("Updated Y Coordinates", Ty)
distancevector = gd.get_distancevector(Tx, Ty)
mst = pa.mst_prim(distancevector)
mst_size = pa.get_tree(distancevector)
print("Size of Steiner Tree for PIO", mst_size)
pa.draw_gridgraph(Tx, Ty, mst, lenTx, lenTy)
Tx = Tx[0:len(Tx) - count]
Ty = Ty[0:len(Ty) - count]
print("Restored X Coordinates=", Tx)
print("Restored Y Coordinates=", Ty)
print("End of PIO")
def call_methods(Tx, Ty, lenTx, lenTy):
# Tx = gd.get_xdata(0, 500, 100)
# Ty = gd.get_ydata(0, 500, 100)
lenTx = Tx.size
lenTy = Ty.size
# Calculating mST for CPSO
Rx = Tx[0:len(Tx) - lenTx]
Ry = Ty[0:len(Tx) - lenTy]
Tx = Tx[0:len(Tx) - len(Rx)]
Ty = Ty[0:len(Ty) - len(Ry)]
#print("X coordinates =", Tx)
#print("Y coordinates =", Ty)
bestparticle = constriction_factor_particle_swarm_test(Tx, Ty)
Rx = Tx[0:len(Tx) - lenTx]
Ry = Ty[0:len(Ty) - lenTy]
Tx = Tx[0:len(Tx) - len(Rx)]
Ty = Ty[0:len(Ty) - len(Ry)]
# print("Updated X coordinates", Tx)
# print("Updated Y coordinates", Ty)
'''print("X coordinates of best particle", bestparticle.Sx)
print("Y coordinates of best particle", bestparticle.Sy)
print("Updated X coordinates", Tx)
print("Updated Y Coordinates", Ty)'''
count = 0
for i in range(len(bestparticle.Sx)):
if bestparticle.Sx[i] < min(Tx) or bestparticle.Sx[i] > max(Tx):
continue
if bestparticle.Sy[i] < min(Ty) or bestparticle.Sy[i] > max(Ty):
continue
Tx = np.append(Tx, math.ceil(bestparticle.Sx[i]))
Ty = np.append(Ty, math.ceil(bestparticle.Sy[i]))
count = count + 1
#print("Updated X Coordinates", Tx)
#print("Updated Y Coordinates", Ty)
distancevector = gd.get_distancevector(Tx, Ty)
mst = pa.mst_prim(distancevector)
mst_size = pa.get_tree(distancevector)
#print("Size of Steiner Tree for Constricted-PSO", mst_size)
return_set = (mst_size, Tx, Ty)
#pa.draw_gridgraph(Tx, Ty, mst,lenTx,lenTy)
return return_set
def run(self):
Tx, Ty = np.load('terminal_point_{}_{}.npy'.format(
str(self.dim), str(self.n)))
lenT = Tx.size
data_pso = dict()
len_S = lenT - 2 # Total number of Steiner points is total Terminal points - 2
distancevector = gd.get_distancevector(np.copy(Tx), np.copy(Ty))
mst_size = pa.get_tree(distancevector)
for i in range(self.max_iter):
print('Iteration No :', i)
# Creating and initiating the Particles
create_particles(Tx, Ty, len_S)
t1 = t.time()
res = call_methods(np.copy(Tx), np.copy(Ty), lenT, self.func)
t2 = t.time()
data_pso[res[0]] = (res[1], res[2], t2 - t1, res[3])
min_pso = min(data_pso.keys())
fp = open(self.file_name, 'w')
fp.write("Size of the MST = " + str(mst_size) + '\n')
fp.write('No. of Iterations :' + str(self.max_iter) + '\n')
fp.write('PSO Min Wt :' + str(min_pso) + '\n')
fp.write('X Coordinates :' + str(data_pso[min_pso][0]) + '\n')
fp.write('Y Coordinates :' + str(data_pso[min_pso][1]) + '\n')
fp.write('No. of Steiner points :' +
str(data_pso[min_pso][0].size - self.n) + '\n')
fp.write('Time Required :' + str(data_pso[min_pso][2]) + '\n')
fp.write('Error Ratio :' + str(min_pso / mst_size) + '\n')
fp.close()
# Ploting the RSMT
pa.draw_gridgraph(Tx, Ty, data_pso[min_pso][3], lenT, lenT)
def get_fitness(Tx, Ty, particle):
# Method to find the Objective Fitness (MST Weight)
# Setting the checking boundaries
maxTx = np.amax(Tx)
minTx = np.amin(Tx)
maxTy = np.amax(Ty)
minTy = np.amin(Ty)
# Constructing the coordinate set of the Tree with Steiner Points
for i in range(particle.Sx.size):
if particle.Sx[i] < minTx or particle.Sx[i] > maxTx:
continue
if particle.Sy[i] < minTy or particle.Sy[i] > maxTy:
continue
Tx = np.append(Tx, particle.Sx[i])
Ty = np.append(Ty, particle.Sy[i])
# Finding the Objective Fitness
distancevector = gd.get_distancevector(Tx, Ty)
objectivefitness = pa.get_tree(distancevector)
return objectivefitness
def call_methods(Tx, Ty, lenT, func):
# Performance controler of the PSO
global cost_vect
global px_vect
global py_vect
# Setting the method into a method variable for dynamic linking
global get_velocity_position
get_velocity_position = func
bestparticle = Particle(fitness=math.inf)
# Running the GPSO Iterations
for i in range(gpso_max_iter):
# Running PSO algorithm
bestparticle_local = particle_swarm_test(np.copy(Tx), np.copy(Ty),
lenT)
if bestparticle_local.fitness < bestparticle.fitness:
bestparticle = bestparticle_local
# Running Gradient Cycles
px = np.mean(bestparticle_local.Sx)
py = np.mean(bestparticle_local.Sy)
for _ in range(grad_max_iter):
cost = search_function(px, py)
cost_vect.append(cost)
px_vect.append(px)
py_vect.append(py)
grad = gradient(px, py)
px, py = update_params(px, py, grad[0], grad[1])
new_Sx = bestparticle_local.Sx + round(px)
new_Sy = bestparticle_local.Sy + round(py)
new_fitness = get_fitness(np.copy(Tx), np.copy(Ty),
Particle(new_Sx, new_Sy))
if new_fitness < bestparticle.fitness:
bestparticle.fitness = new_fitness
bestparticle.Sx = new_Sx
bestparticle.Sy = new_Sy
# Extracting the position of the Steiner Points from the best particle
for i in range(bestparticle.Sx.size):
if bestparticle.Sx[i] < min(Tx) or bestparticle.Sx[i] > max(Tx):
continue
if bestparticle.Sy[i] < min(Ty) or bestparticle.Sy[i] > max(Ty):
continue
Tx = np.append(Tx, bestparticle.Sx[i])
Ty = np.append(Ty, bestparticle.Sy[i])
# Finding the MST of the final RSMT
distancevector = gd.get_distancevector(Tx, Ty)
mst = pa.mst_prim(distancevector)
mst_size = pa.get_tree(distancevector)
# Building the final return tuple for the use in the main program
return_set = (mst_size, Tx, Ty, mst)
return return_set
import prim_algorithm as pa
import gridgraphdemo as gd
import pigeon_optimization_algorithm as poa
import particle_swarm_optimization as pso
import constriction_factor_pso as cpso
Tx = gd.get_xdata(0, 500, 20) #Change values over here and test
Ty = gd.get_ydata(0, 500, 20) #Change values over here and test
lenTx = len(Tx)
lenTy = len(Ty)
print("X Coordinates", Tx)
print("Y Coordinates", Ty)
distancevector = gd.get_distancevector(Tx, Ty)
mst = pa.mst_prim(distancevector)
mst_size = pa.get_tree(distancevector)
print("Size of the MST = ", mst_size)
pa.draw_gridgraph(Tx, Ty, mst, lenTx, lenTy)
poa.call_methods(Tx, Ty, lenTx, lenTy)
pso.call_methods(Tx, Ty, lenTx, lenTy)
cpso.call_methods(Tx, Ty, lenTx, lenTy)
import weighted_particle_swarm_optimisation as wpso
from os import open
#Tx = gd.get_xdata(0, 500, 20)
#Ty = gd.get_ydata(0, 500, 20)
#lenTx = len(Tx)
#lenTy = len(Ty)
dim = 500
n = 10
Tx, Ty = np.load('terminal_point_{}_{}.npy'.format(str(dim), str(n)))
lenTx = Tx.size
lenTy = lenTx
#print("X Coordinates", Tx)
#print("Y Coordinates", Ty)
distancevector = gd.get_distancevector(Tx, Ty)
mst = pa.mst_prim(distancevector)
mst_size = pa.get_tree(distancevector)
pa.draw_gridgraph(Tx, Ty, mst, lenTx, lenTy)
data_pio = dict()
data_pso = dict()
data_cpso = dict()
data_wpso = dict()
max_iter = 5
for i in range(max_iter):
print('Iteration No :', i)
t1 = t.time()
res = poa.call_methods(Tx, Ty, lenTx, lenTy)
