def func(x, ivc):
"""
This function calculate scores for parameters of swarms and need curve
:param x: population of swarms
:return: array of score
"""
# global c
# c += 1
global time_func_0, time_func_1
# time_func_0 = time.clock()
result = []
for i in range(len(x)):
# ivc_1 = get_ivc("3_win")
time_func_0 = time.clock()
create_cir('test', x[i])
logger = Logging.setup_logging()
circuit = spice.LoadFile('test.cir')
input_data = spice.Init_Data(1000, 0.3)
analysis = spice.CreateCVC(circuit, input_data, 100)
# spice.SaveFile(analysis, "test.csv")
# plot(i+c)
ivc_2 = [analysis.input_dummy, analysis.VCurrent]
result.append(compare_ivc(ivc, ivc_2))
time_func_1 = time.clock()
return result
def f(x):
global ivc
_x = copy.copy(x)
# for i in range(len(x)):
# _x[i] = x[i] * (1e-6 - 1e-12) + 1e-12 if i < 5 else x[i] * 10e6
create_cir('test', _x)
circuit = spice.LoadFile('test.cir')
input_data = spice.Init_Data(1000, 0.3, SNR=10**6)
analysis = spice.CreateCVC(circuit, input_data, 100)
# spice.SaveFile(analysis, "test.csv")
ivc_2 = [analysis.input_dummy, analysis.VCurrent]
return compare_ivc(ivc, ivc_2)
def func(x, ivc):
"""
This function calculate scores for parameters of swarms and need curve
:param x: population of swarms
:return: array of score
"""
global time_func_0, time_func_1
# time_func_0 = time.clock()
result = []
for i in range(len(x)):
time_func_0 = time.clock()
create_cir('test', x[i])
circuit = spice.LoadFile('test.cir')
input_data = spice.Init_Data(1000, 0.3, SNR=10**6)
analysis = spice.CreateCVC(circuit, input_data, 100)
ivc_2 = [analysis.input_dummy, analysis.VCurrent]
result.append(compare_ivc(ivc, ivc_2))
time_func_1 = time.clock()
return result
def func_x(x):
"""
This function calculate scores for parameters of swarms and need curve
:param x: population of swarms
:return: array of score
"""
global _x, _y, ivc_1
_x = x
# _y = y
result = []
for i in range(30):
z = []
z.extend(x[i])
z.extend(_y[i])
create_cir('test', z)
# logger = Logging.setup_logging()
circuit = spice.LoadFile('test.cir')
input_data = spice.Init_Data(1000, 0.3)
analysis = spice.CreateCVC(circuit, input_data, 100)
ivc_2 = [analysis.input_dummy, analysis.VCurrent]
result.append(compare_ivc(ivc_1, ivc_2))
return result
def func_x(x, y=None, ivc=None):
"""
This function calculate scores for parameters of swarms and need curve
:param x: population of swarms
:return: array of score
"""
global time_func_0, time_func_1, _x, _y
_x = x
# time_func_0 = time.clock()
result = []
for i in range(len(x)):
time_func_0 = time.clock()
create_cir('test', x[i][:]*y[i][:])
logger = Logging.setup_logging()
circuit = spice.LoadFile('test.cir')
input_data = spice.Init_Data(1000, 0.3)
analysis = spice.CreateCVC(circuit, input_data, 100)
ivc_2 = [analysis.input_dummy, analysis.VCurrent]
result.append(compare_ivc(ivc, ivc_2))
time_func_1 = time.clock()
return result
options = {'c1': 0.8, 'c2': 0.6, 'w': 0.6}
logger = Logging.setup_logging()
for i in range(215):
array_of_resistor = np.random.randint(0, 10**6, 8)
array_of_capacitor = np.random.randint(0, 6, 5) + 6
init_pos[i][:5] = 1/(10 ** array_of_capacitor[:])
init_pos[i][5:] = array_of_resistor[:] * cases[i][5:]
for j in range(5):
init_pos[i][j] = init_pos[i][j] if cases[i][j] == 1 else 10 ** -12
best_cost = 1
for i in range(5):
clamp = gen_velocity_clamps(12-i, 0)
print(clamp)
optimizer = ps.single.GlobalBestPSO(n_particles=215, dimensions=13, bounds=bounds,
options=options, velocity_clamp=clamp,
init_pos=init_pos)
cost, pos = optimizer.optimize(func, iters=12, ivc=ivc_1)
if cost < best_cost:
best_cost = cost
best_pos = pos
print("\n", best_cost, best_pos)
create_cir('test', pos)
logger = Logging.setup_logging()
circuit = spice.LoadFile('test.cir')
input_data = spice.Init_Data(1000, 0.3, SNR=10**6)
analysis = spice.CreateCVC(circuit, input_data, 100)
spice.SaveFile(analysis, "test.csv")
plot('result', '3_win')
_y = gen_init_pos(30,array_of_resistor, array_of_capacitor)
_x[:] = 1
_x[5] = 0
while True:
if iter == 0:
optimizer_disc = ps.discrete.binary.BinaryPSO(n_particles=30, dimensions=13, options=options)
else:
optimizer_disc = ps.discrete.binary.BinaryPSO(n_particles=30, dimensions=13, options=options, init_pos=_x)
cost, pos_d = optimizer_disc.optimize(func_x, iters=1, ivc=ivc_1, y=_y)
optimizer_cont = ps.single.GlobalBestPSO(n_particles=30, dimensions=13, options=options, bounds=bounds, init_pos=_y,
#velocity_clamp=clamp
)
cost, pos_c = optimizer_cont.optimize(func_y, iters=1, ivc=ivc_1, x=_x)
iter += 1
print(cost, pos_d, pos_c)
create_cir('test', pos_d * pos_c)
logger = Logging.setup_logging()
circuit = spice.LoadFile('test.cir')
input_data = spice.Init_Data(1000, 0.3)
analysis = spice.CreateCVC(circuit, input_data, 100)
spice.SaveFile(analysis, "test.csv")
plot('result'+str(iter), '3_win')
none_error += 1 if (cost-prev_cost) ** 2 < 10 ** -12 else 0
if none_error > 5 or iter > 1000:
break
# time_opt_1 = time.clock()
#
# print(time_func_1 - time_func_0)
# print("all optimization")
# print((time_opt_1 - time_opt_0)/60)
_y[i] = np.random.randint(0, 10**5, 8)
_x[i] = np.random.uniform(10**-12, 10**-6, 5)
while True:
optimizer_x = ps.single.GlobalBestPSO(n_particles=30, dimensions=5, options=options,
init_pos=_x, bounds=bounds_x)
cost, pos_x = optimizer_x.optimize(func_x, iters=1)
optimizer_y = pyswarms.discrete.binary.BinaryPSO(n_particles=30, dimensions=8, options=options,
init_pos=_y)
# bounds=bounds_y)
cost, pos_y = optimizer_y.optimize(func_y, iters=1)
iter += 1
print(cost, pos_x, pos_y)
z = []
z.extend(pos_x)
z.extend(pos_y)
create_cir('test', z)
logger = Logging.setup_logging()
circuit = spice.LoadFile('test.cir')
input_data = spice.Init_Data(1000, 0.3)
analysis = spice.CreateCVC(circuit, input_data, 100)
spice.SaveFile(analysis, "test.csv")
plot('result'+str(iter), '3_win')
none_error += 1 if (cost-prev_cost) ** 2 < 10 ** -12 else 0
if none_error > 5 or iter > 1000:
break
# time_opt_1 = time.clock()
#
# print(time_func_1 - time_func_0)
# print("all optimization")
# print((time_opt_1 - time_opt_0)/60)
# y = []
clamp = gen_velocity_clamps(cases)
logger = Logging.setup_logging()
for step in range(2):
init_pos = gen_init_pos(cases, array_of_resistor, array_of_capacitor, step)
optimizer = ps.single.GlobalBestPSO(n_particles=2 * len(cases),
dimensions=13,
bounds=bounds,
options=options,
velocity_clamp=clamp,
init_pos=init_pos)
cost, pos = optimizer.optimize(func, iters=28, ivc=ivc_1)
costs.append(cost)
poses.append(pos)
# clamps.append(clamp)
minpos = poses[np.argmin(costs)]
# minclamp = clamps[np.argmin(costs)]
time_opt_1 = time.clock()
print(pos, cost)
print(time_func_1 - time_func_0)
print("all optimization")
print((time_opt_1 - time_opt_0) / 60)
create_cir('test', minpos)
logger = Logging.setup_logging()
circuit = spice.LoadFile('test.cir')
input_data = spice.Init_Data(1000, 0.3, SNR=10**6)
analysis = spice.CreateCVC(circuit, input_data, 100)
spice.SaveFile(analysis, "test.csv")
plot('result', '3_win')
