def _interpolation_grid_force(self):
self._widths = []
self._bounds = []
self._extra_points = []
for v in self.bias_variables:
extra_points = min(self.grid_expansion, v.grid_size) if v.periodic else 0
extra_range = extra_points*v._range/(v.grid_size - 1)
self._widths.append(v.grid_size + 2*extra_points)
self._bounds += [v.min_value - extra_range, v.max_value + extra_range]
self._extra_points.append(extra_points)
self._bias = np.zeros(np.prod(self._widths))
num_bias_variables = len(self.bias_variables)
if num_bias_variables == 1:
self._table = openmm.Continuous1DFunction(self._bias, *self._bounds)
elif num_bias_variables == 2:
self._table = openmm.Continuous2DFunction(*self._widths, self._bias, *self._bounds)
else:
self._table = openmm.Continuous3DFunction(*self._widths, self._bias, *self._bounds)
expression = f'bias({",".join(v.id for v in self.bias_variables)})'
for i, v in enumerate(self.bias_variables):
expression += f';{v.id}={v._get_energy_function(i+1)}'
force = openmm.CustomCVForce(expression)
for i in range(num_bias_variables):
x = openmm.CustomExternalForce('x')
x.addParticle(0, [])
force.addCollectiveVariable(f'x{i+1}', x)
force.addTabulatedFunction('bias', self._table)
force.addGlobalParameter('Lx', 0)
return force
def get_Upair_force(Ucg, coeff, r_vals):
"""Create pairwise potentials as tabulated functions"""
N = Ucg.n_atoms
bcut = Ucg.bond_cutoff
Up_vals = Ucg.Upair_values(coeff, r_vals)
if len(Up_vals) > 1:
table_vals = np.zeros((len(Up_vals), len(r_vals)))
for i in range(len(Up_vals)):
table_vals[i,:] = Up_vals[i]
xvals = np.arange(len(Up_vals)).astype(float)
yvals = r_vals
Table_func = omm.Continuous2DFunction(len(xvals), len(yvals),
table_vals.flatten(), xvals[0], xvals[-1], yvals[0], yvals[-1])
Up_force = omm.CustomCompoundBondForce(2, "Table(p_idx, distance(p1, p2))")
Up_force.addPerBondParameter("p_idx")
Up_force.addTabulatedFunction("Table", Table_func)
else:
Table_func = omm.Continuous1DFunction(Up_vals[0], r_vals[0], r_vals[-1])
Up_force = omm.CustomCompoundBondForce(2, "Table(distance(p1, p2))")
Up_force.addTabulatedFunction("Table", Table_func)
if Ucg.pair_symmetry == "shared":
# all pairs share same interaction
for i in range(N - bcut):
for j in range(i + bcut, N):
Up_force.addBond((i, j))
elif Ucg.pair_symmetry == "seq_sep":
# pairs with same sequence sep share same interaction
for i in range(N - bcut):
for j in range(i + bcut, N):
p_idx = np.abs(j - i) - bcut
Up_force.addBond([i, j], p_idx)
elif Ucg.pair_symmetry == "unique":
# all pair have different interactions
p_idx = 0
for i in range(N - bcut):
for j in range(i + bcut, N):
Up_force.addBond((i, j), (p_idx))
p_idx += 1
else:
raise ValueError("pair_symmetry must be: shared, seq_sep, unique. Gave:" + str(Ucg.pair_symmetry))
return Up_force
def __init__(self, variables, height, frequency, grid_expansion):
self.bias_variables = [cv for cv in variables if cv.sigma is not None]
self.height = height
self.frequency = frequency
self.grid_expansion = grid_expansion
self._widths = []
self._bounds = []
self._expanded = []
self._extra_points = []
for cv in self.bias_variables:
expanded = cv.periodic # and len(self.bias_variables) > 1
extra_points = min(grid_expansion, cv.grid_size) if expanded else 0
extra_range = extra_points * cv._range / (cv.grid_size - 1)
self._widths += [cv.grid_size + 2 * extra_points]
self._bounds += [
cv.min_value - extra_range, cv.max_value + extra_range
]
self._expanded += [expanded]
self._extra_points += [extra_points]
self._bias = np.zeros(tuple(reversed(self._widths)))
if len(variables) == 1:
self._table = openmm.Continuous1DFunction(
self._bias.flatten(),
*self._bounds,
# self.bias_variables[0].periodic,
)
elif len(variables) == 2:
self._table = openmm.Continuous2DFunction(
*self._widths,
self._bias.flatten(),
*self._bounds,
)
elif len(variables) == 3:
self._table = openmm.Continuous3DFunction(
*self._widths,
self._bias.flatten(),
*self._bounds,
)
else:
raise ValueError(
'UFED requires 1, 2, or 3 biased collective variables')
parameter_list = ', '.join(f's_{cv.id}' for cv in self.bias_variables)
self.force = openmm.CustomCVForce(f'bias({parameter_list})')
for cv in self.bias_variables:
expression = f'{cv.min_value}+{cv._range}*(x/Lx-floor(x/Lx))'
parameter = openmm.CustomExternalForce(expression)
parameter.addGlobalParameter('Lx', 0.0)
parameter.addParticle(0, [])
self.force.addCollectiveVariable(f's_{cv.id}', parameter)
self.force.addTabulatedFunction('bias', self._table)
def __init__(self,
system,
variables,
temperature,
biasFactor,
height,
frequency,
saveFrequency=None,
biasDir=None):
"""Create a Metadynamics object.
Parameters
----------
system: System
the System to simulate. A CustomCVForce implementing the bias is created and
added to the System.
variables: list of BiasVariables
the collective variables to sample
temperature: temperature
the temperature at which the simulation is being run. This is used in computing
the free energy.
biasFactor: float
used in scaling the height of the Gaussians added to the bias. The collective
variables are sampled as if the effective temperature of the simulation were
temperature*biasFactor.
height: energy
the initial height of the Gaussians to add
frequency: int
the interval in time steps at which Gaussians should be added to the bias potential
saveFrequency: int (optional)
the interval in time steps at which to write out the current biases to disk. At
the same time it writes biases, it also checks for updated biases written by other
processes and loads them in. This must be a multiple of frequency.
biasDir: str (optional)
the directory to which biases should be written, and from which biases written by
other processes should be loaded
"""
if not unit.is_quantity(temperature):
temperature = temperature * unit.kelvin
if not unit.is_quantity(height):
height = height * unit.kilojoules_per_mole
if biasFactor < 1.0:
raise ValueError('biasFactor must be >= 1')
if (saveFrequency is None
and biasDir is not None) or (saveFrequency is not None
and biasDir is None):
raise ValueError('Must specify both saveFrequency and biasDir')
if saveFrequency is not None and (saveFrequency < frequency
or saveFrequency % frequency != 0):
raise ValueError('saveFrequency must be a multiple of frequency')
self.variables = variables
self.temperature = temperature
self.biasFactor = biasFactor
self.height = height
self.frequency = frequency
self.biasDir = biasDir
self.saveFrequency = saveFrequency
self._id = np.random.randint(0x7FFFFFFF)
self._saveIndex = 0
self._selfBias = np.zeros(tuple(v.gridWidth for v in variables))
self._totalBias = np.zeros(tuple(v.gridWidth for v in variables))
self._loadedBiases = {}
self._deltaT = temperature * (biasFactor - 1)
varNames = ['cv%d' % i for i in range(len(variables))]
self._force = mm.CustomCVForce('table(%s)' % ', '.join(varNames))
for name, var in zip(varNames, variables):
self._force.addCollectiveVariable(name, var.force)
widths = [v.gridWidth for v in variables]
mins = [v.minValue for v in variables]
maxs = [v.maxValue for v in variables]
if len(variables) == 1:
self._table = mm.Continuous1DFunction(self._totalBias.flatten(),
mins[0], maxs[0])
elif len(variables) == 2:
self._table = mm.Continuous2DFunction(widths[0], widths[1],
self._totalBias.flatten(),
mins[0], maxs[0], mins[1],
maxs[1])
elif len(variables) == 3:
self._table = mm.Continuous3DFunction(widths[0], widths[1],
widths[2],
self._totalBias.flatten(),
mins[0], maxs[0], mins[1],
maxs[1], mins[2], maxs[2])
else:
raise ValueError(
'Metadynamics requires 1, 2, or 3 collective variables')
self._force.addTabulatedFunction('table', self._table)
self._force.setForceGroup(31)
system.addForce(self._force)
self._syncWithDisk()