def setup_output(self):
self.woutp = WriteOutputMC()
def setup_output(self):
self.woutp = WriteOutputMC()
class MonteCarlo:
def __init__(self,
mol,
pot,
delta,
nstep,
temp,
restart=False,
frozen_atom_number=None):
self.mol, self.pot = mol, pot
self.nstep = nstep
self.temp, self.restart = temp, restart
self.interaction_save = np.zeros((len(self.mol), len(self.mol)))
self.which_frozen = [False] * len(self.mol)
if not frozen_atom_number == None:
for i in frozen_atom_number:
self.which_frozen[i] = True
if self.pot.get_check_pbc:
self.mol.set_positions([
self.pot.get_pbc_adjusted(pos)
for pos in self.mol.get_positions()
])
if isinstance(delta, float):
self.delta = [delta] * len(self.mol)
elif isinstance(delta, list) or isinstance(delta, np.ndarray):
if len(delta) == len(mol):
self.delta = np.array(delta)
else:
print 'check the array size of delta'
sys.exit()
else:
print 'delta type error'
sys.exit()
self.beta = 1.0 / (temp * kelvin2hartree)
self.count, self.count_accept = 0, 0
self.setup_output()
def access_writeoutput(self):
return self.woutp
def setup_output(self):
self.woutp = WriteOutputMC()
#if self.restart: self.woutp.restart(self)
#else: self.woutp.start()
def run(self):
self.prepare()
# log at the initial point
self.woutp.logging(self)
# loop of sampling
while self.count < self.nstep:
self.count += 1
self.step()
self.woutp.logging(self)
self.woutp.finalize(self)
def prepare(self):
if self.restart: self.woutp.restart(self)
for i in xrange(len(self.mol)):
self.interaction_save[i, :] = self.pot.get_interaction_energy_mc(i)
self.mol.set_potential_energy(0.5 * np.sum(self.interaction_save))
def step(self):
# select the transfered atom
ind = randint(0, len(self.mol) - 1)
if self.which_frozen[ind]: return
x_save = copy(self.mol.get_positions_index(ind))
if self.pot.get_check_pbc():
self.mol.set_positions_index(ind, self.pot.get_pbc_adjusted(x_save \
+ self.delta[ind] * random_v(self.mol.get_ndims())))
else:
self.mol.set_positions_index(ind, x_save \
+ self.delta[ind] * random_v(self.mol.get_ndims()))
int_trial = self.pot.get_interaction_energy(ind)
delta_e = np.sum(int_trial) - np.sum(self.interaction_save[ind])
judge = exp(-self.beta * delta_e) > random()
# the trial movement is judged
if judge:
self.count_accept += 1
self.interaction_save[ind, :] = int_trial
self.interaction_save[:, ind] = int_trial
self.mol.set_potential_energy(self.mol.get_potential_energy() +
delta_e)
else:
self.mol.set_positions_index(ind, x_save)
class MonteCarlo:
def __init__(self, mol, pot, delta, nstep, temp, restart = False, frozen_atom_number = None ):
self.mol, self.pot = mol, pot
self.nstep = nstep
self.temp, self.restart = temp, restart
self.interaction_save = np.zeros((len(self.mol), len(self.mol)))
self.which_frozen = [False] * len(self.mol)
if not frozen_atom_number == None:
for i in frozen_atom_number: self.which_frozen[i] = True
if self.pot.get_check_pbc:
self.mol.set_positions([self.pot.get_pbc_adjusted(pos) for pos in self.mol.get_positions()])
if isinstance(delta, float):
self.delta = [delta] * len(self.mol)
elif isinstance(delta, list) or isinstance(delta, np.ndarray):
if len(delta) == len(mol):
self.delta = np.array(delta)
else:
print 'check the array size of delta'
sys.exit()
else:
print 'delta type error'
sys.exit()
self.beta = 1.0 / (temp * kelvin2hartree)
self.count, self.count_accept = 0, 0
self.setup_output()
def access_writeoutput(self):
return self.woutp
def setup_output(self):
self.woutp = WriteOutputMC()
#if self.restart: self.woutp.restart(self)
#else: self.woutp.start()
def run(self):
self.prepare()
# log at the initial point
self.woutp.logging(self)
# loop of sampling
while self.count < self.nstep:
self.count += 1
self.step()
self.woutp.logging(self)
self.woutp.finalize(self)
def prepare(self):
if self.restart: self.woutp.restart(self)
for i in xrange(len(self.mol)):
self.interaction_save[i,:] = self.pot.get_interaction_energy_mc(i)
self.mol.set_potential_energy(0.5 * np.sum(self.interaction_save))
def step(self):
# select the transfered atom
ind = randint(0,len(self.mol)-1)
if self.which_frozen[ind]: return
x_save = copy(self.mol.get_positions_index(ind))
if self.pot.get_check_pbc():
self.mol.set_positions_index(ind, self.pot.get_pbc_adjusted(x_save \
+ self.delta[ind] * random_v(self.mol.get_ndims())))
else:
self.mol.set_positions_index(ind, x_save \
+ self.delta[ind] * random_v(self.mol.get_ndims()))
int_trial = self.pot.get_interaction_energy(ind)
delta_e = np.sum(int_trial) - np.sum(self.interaction_save[ind])
judge = exp(- self.beta * delta_e) > random()
# the trial movement is judged
if judge:
self.count_accept += 1
self.interaction_save[ind,:] = int_trial
self.interaction_save[:,ind] = int_trial
self.mol.set_potential_energy(self.mol.get_potential_energy() + delta_e)
else: self.mol.set_positions_index(ind, x_save)
