def propagate(self, segments):
# Instantiate the integrator here since cython extension classes do not
# automatically implement the pickling protocol
self.model['seed'] = genrandint()
integrator = ElasticNetwork(**self.model)
for segment in segments:
starttime = time.time()
new_pcoords = np.empty((self.nsteps, self.ndim),
dtype=pcoord_dtype)
new_pcoords[0, :] = segment.pcoord[0, :]
cpos = self.ndim // 2
x = new_pcoords[0, :cpos].copy().reshape(
(-1, 3)).astype(np.float64)
v = new_pcoords[0, cpos:].copy().reshape(
(-1, 3)).astype(np.float64)
for istep in xrange(1, self.nsteps):
integrator.step(x, v, self.nsubsteps)
new_pcoords[istep, :] = np.hstack((x.ravel(), v.ravel()))
segment.pcoord = new_pcoords[...].astype(pcoord_dtype)
segment.status = Segment.SEG_STATUS_COMPLETE
segment.walltime = time.time() - starttime
del integrator
return segments
def propagate(self, segments):
# Instantiate the integrator here since cython extension classes do not
# automatically implement the pickling protocol
self.model['seed'] = genrandint()
integrator = ElasticNetwork(**self.model)
for segment in segments:
starttime = time.time()
new_pcoords = np.empty((self.nsteps,self.ndim), dtype=pcoord_dtype)
new_pcoords[0,:] = segment.pcoord[0,:]
cpos = self.ndim // 2
x = new_pcoords[0,:cpos].copy().reshape((-1,3)).astype(np.float64)
v = new_pcoords[0,cpos:].copy().reshape((-1,3)).astype(np.float64)
for istep in xrange(1, self.nsteps):
integrator.step(x, v, self.nsubsteps)
new_pcoords[istep,:] = np.hstack((x.ravel(), v.ravel()))
segment.pcoord = new_pcoords[...].astype(pcoord_dtype)
segment.status = Segment.SEG_STATUS_COMPLETE
segment.walltime = time.time() - starttime
del integrator
return segments
def run(config):
# Set up logging
logname = config.get('outputs','log')
print('Setting up logging: {}'.format(logname))
logging.basicConfig(filename=logname,level=logging.DEBUG,filemode='w')
NUM_BLOCKS = config.getint('parameters','num_blocks')
STEPS_PER_BLOCK = config.getint('parameters','steps_per_block')
model = {}
model['mass'] = config.getfloat('model','mass')
model['gamma'] = config.getfloat('model','gamma')
model['temp'] = config.getfloat('model','temp')
model['dt'] = config.getfloat('model','dt')
model['sigma'] = config.getfloat('model','sigma')
model['eps'] = config.getfloat('model','eps')
model['betamix'] = config.getfloat('model','betamix')
model.update(np.load(config.get('model','ff_data')))
model['seed'] = genrandint()
init_pos = {}
init_pos['coordsA'] = model['coordsA']
init_pos['coordsB'] = model['coordsB']
del model['coordsA']
del model['coordsB']
system = ElasticNetwork(**model)
# Setup storage
print('Setting up netcdf4 trajectory storage')
nc = trajstore(natoms=system.natoms)
nc.initialize_netcdf(config.get('outputs','trajname'))
# Initial coords and velocities
pos = init_pos[config.get('model','init_pos')]
# Assign velocities drawn from Maxwell-Boltzmann distribution
sigma = np.sqrt(model['temp']*0.001987191/model['mass'])
vel = sigma*np.random.normal(size=pos.shape)
print('Starting Simulation')
for dk in xrange(NUM_BLOCKS):
t1 = time.time()
system.step(pos,vel,STEPS_PER_BLOCK)
nc.write_frame(pos,vel)
EKin = 0.5*model['mass']*np.sum(vel**2)
T = EKin*(2./3)/(0.001987191*system.natoms)
logging.info('Completed {} of {} steps: {} s Ekin: {} Temp: {}'.format(dk,NUM_BLOCKS-1,time.time() - t1,EKin,T))
nc.ncfile.close()