def test_eq_tricky_chempot_cond():
"""
Chemical potential condition with difficult convergence for chemical potentials.
"""
eq = equilibrium(ISSUE43_DBF, ['AL', 'NI', 'CR', 'VA'], ['FCC_A1', 'GAMMA_PRIME'],
{v.MU('AL'): -135620.9960449, v.MU('CR'): -47269.29002414, v.T: 1273, v.P: 101325},
verbose=True)
chempots = np.array([-135620.9960449, -47269.29002414, -92304.23688281])
print(np.nansum(np.squeeze(eq.NP * eq.X), axis=-2))
assert_allclose(eq.GM.values, -83242.872102)
assert_allclose(np.nansum(np.squeeze(eq.NP * eq.X), axis=-2), [0.19624727, 0.38996739, 0.41378534])
assert_allclose(np.squeeze(eq.MU.values), chempots)
def test_eq_charge_halite():
"""Halite (with charged species) is correctly calculated and charged balanced in equilibrium"""
alfeo = ALFEO_DBF
comps = ['FE', 'AL', 'O', 'VA']
phases = ['HALITE']
conds = {v.P: 101325, v.N: 1, v.T: 500, v.X('FE'): 0.2, v.MU('O'): -200000}
res = equilibrium(alfeo, comps, phases, conds, verbose=True)
assert_allclose(res.GM.values, -248202.28)
def test_eq_issue259():
"""
Chemical potential condition for phase with internal degrees of freedom.
"""
dbf = ALFE_DBF
comps = ['AL', 'FE', 'VA']
eq = equilibrium(dbf, comps, ['B2_BCC'], {v.P: 101325, v.N: 1, v.T: 1013, v.MU('AL'): -95906})
assert_allclose(eq.GM.values, -65786.260)
assert_allclose(eq.MU.values.flatten(), [-95906., -52877.592122])
def test_eq_magnetic_chempot_cond():
"""
Chemical potential condition with an ill-conditioned Hessian due to magnetism (Tc->0).
This is difficult to reproduce so we only include some known examples here.
"""
# This set of conditions is known to trigger the issue
eq = equilibrium(ALFE_DBF, ['AL', 'FE', 'VA'], ['FCC_A1', 'AL13FE4'],
{v.MU('FE'): -123110, v.T: 300, v.P: 1e5}, verbose=True)
assert_allclose(np.squeeze(eq.GM.values), -35427.1, atol=0.1)
assert_allclose(np.squeeze(eq.MU.values), [-8490.7, -123110], atol=0.1)
def test_eq_ideal_chempot_cond():
TDB = """
ELEMENT A GRAPHITE 12.011 1054.0 5.7423 !
ELEMENT B BCC_A2 55.847 4489.0 27.2797 !
ELEMENT C BCC_A2 55.847 4489.0 27.2797 !
TYPE_DEFINITION % SEQ * !
PHASE TEST % 1 1 !
CONSTITUENT TEST : A,B,C: !
"""
my_phases = ['TEST']
comps = ['A', 'B', 'C']
comps = sorted(comps)
conds = dict({v.T: 1000, v.P: 101325, v.N: 1})
conds[v.MU('C')] = -1000
conds[v.X('A')] = 0.01
eq = equilibrium(Database(TDB), comps, my_phases, conds, verbose=True)
np.testing.assert_allclose(eq.GM.values.squeeze(), -3219.570565)
np.testing.assert_allclose(eq.MU.values.squeeze(), [-38289.687511, -18873.23674, -1000.])
np.testing.assert_allclose(eq.X.isel(vertex=0).values.squeeze(), [0.01, 0.103321, 0.886679], atol=1e-4)
def equilibrium(dbf, comps, phases, conditions, **kwargs):
"""
Calculate the equilibrium state of a system containing the specified
components and phases, under the specified conditions.
Model parameters are taken from 'dbf'.
Parameters
----------
dbf : Database
Thermodynamic database containing the relevant parameters.
comps : list
Names of components to consider in the calculation.
phases : list or dict
Names of phases to consider in the calculation.
conditions : dict or (list of dict)
StateVariables and their corresponding value.
verbose : bool, optional (Default: True)
Show progress of calculations.
grid_opts : dict, optional
Keyword arguments to pass to the initial grid routine.
Returns
-------
Structured equilibrium calculation.
Examples
--------
None yet.
"""
active_phases = unpack_phases(phases) or sorted(dbf.phases.keys())
comps = sorted(comps)
indep_vars = ['T', 'P']
grid_opts = kwargs.pop('grid_opts', dict())
verbose = kwargs.pop('verbose', True)
phase_records = dict()
callable_dict = kwargs.pop('callables', dict())
grad_callable_dict = kwargs.pop('grad_callables', dict())
hess_callable_dict = kwargs.pop('hess_callables', dict())
points_dict = dict()
maximum_internal_dof = 0
conds = OrderedDict((key, unpack_condition(value))
for key, value in sorted(conditions.items(), key=str))
str_conds = OrderedDict((str(key), value) for key, value in conds.items())
indep_vals = list([float(x) for x in np.atleast_1d(val)]
for key, val in str_conds.items() if key in indep_vars)
components = [x for x in sorted(comps) if not x.startswith('VA')]
# Construct models for each phase; prioritize user models
models = unpack_kwarg(kwargs.pop('model', Model), default_arg=Model)
if verbose:
print('Components:', ' '.join(comps))
print('Phases:', end=' ')
for name in active_phases:
mod = models[name]
if isinstance(mod, type):
models[name] = mod = mod(dbf, comps, name)
variables = sorted(mod.energy.atoms(v.StateVariable).union(
{key
for key in conditions.keys() if key in [v.T, v.P]}),
key=str)
site_fracs = sorted(mod.energy.atoms(v.SiteFraction), key=str)
maximum_internal_dof = max(maximum_internal_dof, len(site_fracs))
# Extra factor '1e-100...' is to work around an annoying broadcasting bug for zero gradient entries
#models[name].models['_broadcaster'] = 1e-100 * Mul(*variables) ** 3
out = models[name].energy
undefs = list(out.atoms(Symbol) - out.atoms(v.StateVariable))
for undef in undefs:
out = out.xreplace({undef: float(0)})
callable_dict[name], grad_callable_dict[name], hess_callable_dict[name] = \
build_functions(out, [v.P, v.T] + site_fracs)
# Adjust gradient by the approximate chemical potentials
hyperplane = Add(*[
v.MU(i) * mole_fraction(dbf.phases[name], comps, i) for i in comps
if i != 'VA'
])
plane_obj, plane_grad, plane_hess = build_functions(
hyperplane, [v.MU(i) for i in comps if i != 'VA'] + site_fracs)
phase_records[name.upper()] = PhaseRecord(
variables=variables,
grad=grad_callable_dict[name],
hess=hess_callable_dict[name],
plane_grad=plane_grad,
plane_hess=plane_hess)
if verbose:
print(name, end=' ')
if verbose:
print('[done]', end='\n')
# 'calculate' accepts conditions through its keyword arguments
grid_opts.update(
{key: value
for key, value in str_conds.items() if key in indep_vars})
if 'pdens' not in grid_opts:
grid_opts['pdens'] = 100
coord_dict = str_conds.copy()
coord_dict['vertex'] = np.arange(len(components))
grid_shape = np.meshgrid(*coord_dict.values(), indexing='ij',
sparse=False)[0].shape
coord_dict['component'] = components
if verbose:
print('Computing initial grid', end=' ')
grid = calculate(dbf,
comps,
active_phases,
output='GM',
model=models,
callables=callable_dict,
fake_points=True,
**grid_opts)
if verbose:
print('[{0} points, {1}]'.format(len(grid.points),
sizeof_fmt(grid.nbytes)),
end='\n')
properties = xray.Dataset(
{
'NP': (list(str_conds.keys()) + ['vertex'], np.empty(grid_shape)),
'GM': (list(str_conds.keys()), np.empty(grid_shape[:-1])),
'MU':
(list(str_conds.keys()) + ['component'], np.empty(grid_shape)),
'points': (list(str_conds.keys()) + ['vertex'],
np.empty(grid_shape, dtype=np.int))
},
coords=coord_dict,
attrs={'iterations': 1},
)
# Store the potentials from the previous iteration
current_potentials = properties.MU.copy()
for iteration in range(MAX_ITERATIONS):
if verbose:
print('Computing convex hull [iteration {}]'.format(
properties.attrs['iterations']))
# lower_convex_hull will modify properties
lower_convex_hull(grid, properties)
progress = np.abs(current_potentials - properties.MU).values
converged = (progress < MIN_PROGRESS).all(axis=-1)
if verbose:
print('progress', progress.max(),
'[{} conditions updated]'.format(np.sum(~converged)))
if progress.max() < MIN_PROGRESS:
if verbose:
print('Convergence achieved')
break
current_potentials[...] = properties.MU.values
if verbose:
print('Refining convex hull')
# Insert extra dimensions for non-T,P conditions so GM broadcasts correctly
energy_broadcast_shape = grid.GM.values.shape[:len(indep_vals)] + \
(1,) * (len(str_conds) - len(indep_vals)) + (grid.GM.values.shape[-1],)
driving_forces = np.einsum('...i,...i',
properties.MU.values[..., np.newaxis, :].astype(np.float),
grid.X.values[np.index_exp[...] +
(np.newaxis,) * (len(str_conds) - len(indep_vals)) +
np.index_exp[:, :]].astype(np.float)) - \
grid.GM.values.view().reshape(energy_broadcast_shape)
for name in active_phases:
dof = len(models[name].energy.atoms(v.SiteFraction))
current_phase_indices = (grid.Phase.values == name
).reshape(energy_broadcast_shape[:-1] +
(-1, ))
# Broadcast to capture all conditions
current_phase_indices = np.broadcast_arrays(
current_phase_indices, np.empty(driving_forces.shape))[0]
# This reshape is safe as long as phases have the same number of points at all indep. conditions
current_phase_driving_forces = driving_forces[
current_phase_indices].reshape(
current_phase_indices.shape[:-1] + (-1, ))
# Note: This works as long as all points are in the same phase order for all T, P
current_site_fractions = grid.Y.values[..., current_phase_indices[
(0, ) * len(str_conds)], :]
if np.sum(
current_site_fractions[(0, ) *
len(indep_vals)][..., :dof]) == dof:
# All site fractions are 1, aka zero internal degrees of freedom
# Impossible to refine these points, so skip this phase
points_dict[name] = current_site_fractions[
(0, ) * len(indep_vals)][..., :dof]
continue
# Find the N points with largest driving force for a given set of conditions
# Remember that driving force has a sign, so we want the "most positive" values
# N is the number of components, in this context
# N points define a 'best simplex' for every set of conditions
# We also need to restrict ourselves to one phase at a time
trial_indices = np.argpartition(current_phase_driving_forces,
-len(components),
axis=-1)[..., -len(components):]
trial_indices = trial_indices.ravel()
statevar_indices = np.unravel_index(
np.arange(
np.multiply.reduce(properties.GM.values.shape +
(len(components), ))),
properties.GM.values.shape +
(len(components), ))[:len(indep_vals)]
points = current_site_fractions[np.index_exp[statevar_indices +
(trial_indices, )]]
points.shape = properties.points.shape[:-1] + (
-1, maximum_internal_dof)
# The Y arrays have been padded, so we should slice off the padding
points = points[..., :dof]
#print('Starting points shape: ', points.shape)
#print(points)
if len(points) == 0:
if name in points_dict:
del points_dict[name]
# No nearly stable points: skip this phase
continue
num_vars = len(phase_records[name].variables)
plane_grad = phase_records[name].plane_grad
plane_hess = phase_records[name].plane_hess
statevar_grid = np.meshgrid(*itertools.chain(indep_vals),
sparse=True,
indexing='ij')
# TODO: A more sophisticated treatment of constraints
num_constraints = len(dbf.phases[name].sublattices)
constraint_jac = np.zeros(
(num_constraints, num_vars - len(indep_vars)))
# Independent variables are always fixed (in this limited implementation)
#for idx in range(len(indep_vals)):
# constraint_jac[idx, idx] = 1
# This is for site fraction balance constraints
var_idx = 0 #len(indep_vals)
for idx in range(len(dbf.phases[name].sublattices)):
active_in_subl = set(
dbf.phases[name].constituents[idx]).intersection(comps)
constraint_jac[idx, var_idx:var_idx + len(active_in_subl)] = 1
var_idx += len(active_in_subl)
newton_iteration = 0
while newton_iteration < MAX_NEWTON_ITERATIONS:
flattened_points = points.reshape(
points.shape[:len(indep_vals)] + (-1, points.shape[-1]))
grad_args = itertools.chain(
[i[..., None] for i in statevar_grid], [
flattened_points[..., i]
for i in range(flattened_points.shape[-1])
])
grad = np.array(phase_records[name].grad(*grad_args),
dtype=np.float)
# Remove derivatives wrt T,P
grad = grad[..., len(indep_vars):]
grad.shape = points.shape
grad[np.isnan(grad).any(
axis=-1
)] = 0 # This is necessary for gradients on the edge of space
hess_args = itertools.chain(
[i[..., None] for i in statevar_grid], [
flattened_points[..., i]
for i in range(flattened_points.shape[-1])
])
hess = np.array(phase_records[name].hess(*hess_args),
dtype=np.float)
# Remove derivatives wrt T,P
hess = hess[..., len(indep_vars):, len(indep_vars):]
hess.shape = points.shape + (hess.shape[-1], )
hess[np.isnan(hess).any(axis=(-2,
-1))] = np.eye(hess.shape[-1])
plane_args = itertools.chain([
properties.MU.values[..., i][..., None]
for i in range(properties.MU.shape[-1])
], [points[..., i] for i in range(points.shape[-1])])
cast_grad = np.array(plane_grad(*plane_args), dtype=np.float)
# Remove derivatives wrt chemical potentials
cast_grad = cast_grad[..., properties.MU.shape[-1]:]
grad = grad - cast_grad
plane_args = itertools.chain([
properties.MU.values[..., i][..., None]
for i in range(properties.MU.shape[-1])
], [points[..., i] for i in range(points.shape[-1])])
cast_hess = np.array(plane_hess(*plane_args), dtype=np.float)
# Remove derivatives wrt chemical potentials
cast_hess = cast_hess[..., properties.MU.shape[-1]:,
properties.MU.shape[-1]:]
cast_hess = -cast_hess + hess
hess = cast_hess.astype(np.float, copy=False)
try:
e_matrix = np.linalg.inv(hess)
except np.linalg.LinAlgError:
print(hess)
raise
current = calculate(
dbf,
comps,
name,
output='GM',
model=models,
callables=callable_dict,
fake_points=False,
points=points.reshape(points.shape[:len(indep_vals)] +
(-1, points.shape[-1])),
**grid_opts)
current_plane = np.multiply(
current.X.values.reshape(points.shape[:-1] +
(len(components), )),
properties.MU.values[..., np.newaxis, :]).sum(axis=-1)
current_df = current.GM.values.reshape(
points.shape[:-1]) - current_plane
#print('Inv hess check: ', np.isnan(e_matrix).any())
#print('grad check: ', np.isnan(grad).any())
dy_unconstrained = -np.einsum('...ij,...j->...i', e_matrix,
grad)
#print('dy_unconstrained check: ', np.isnan(dy_unconstrained).any())
proj_matrix = np.dot(e_matrix, constraint_jac.T)
inv_matrix = np.rollaxis(np.dot(constraint_jac, proj_matrix),
0, -1)
inv_term = np.linalg.inv(inv_matrix)
#print('inv_term check: ', np.isnan(inv_term).any())
first_term = np.einsum('...ij,...jk->...ik', proj_matrix,
inv_term)
#print('first_term check: ', np.isnan(first_term).any())
# Normally a term for the residual here
# We only choose starting points which obey the constraints, so r = 0
cons_summation = np.einsum('...i,...ji->...j',
dy_unconstrained, constraint_jac)
#print('cons_summation check: ', np.isnan(cons_summation).any())
cons_correction = np.einsum('...ij,...j->...i', first_term,
cons_summation)
#print('cons_correction check: ', np.isnan(cons_correction).any())
dy_constrained = dy_unconstrained - cons_correction
#print('dy_constrained check: ', np.isnan(dy_constrained).any())
# TODO: Support for adaptive changing independent variable steps
new_direction = dy_constrained
#print('new_direction', new_direction)
#print('points', points)
# Backtracking line search
if np.isnan(new_direction).any():
print('new_direction', new_direction)
#print('Convergence angle:', -(grad*new_direction).sum(axis=-1) / (np.linalg.norm(grad, axis=-1) * np.linalg.norm(new_direction, axis=-1)))
new_points = points + INITIAL_STEP_SIZE * new_direction
alpha = np.full(new_points.shape[:-1],
INITIAL_STEP_SIZE,
dtype=np.float)
alpha[np.all(np.linalg.norm(new_direction, axis=-1) <
MIN_DIRECTION_NORM,
axis=-1)] = 0
negative_points = np.any(new_points < 0., axis=-1)
while np.any(negative_points):
alpha[negative_points] *= 0.5
new_points = points + alpha[...,
np.newaxis] * new_direction
negative_points = np.any(new_points < 0., axis=-1)
# Backtracking line search
# alpha now contains maximum possible values that keep us inside the space
# but we don't just want to take the biggest step; we want the biggest step which reduces energy
new_points = new_points.reshape(
new_points.shape[:len(indep_vals)] +
(-1, new_points.shape[-1]))
candidates = calculate(dbf,
comps,
name,
output='GM',
model=models,
callables=callable_dict,
fake_points=False,
points=new_points,
**grid_opts)
candidate_plane = np.multiply(
candidates.X.values.reshape(points.shape[:-1] +
(len(components), )),
properties.MU.values[..., np.newaxis, :]).sum(axis=-1)
energy_diff = (candidates.GM.values.reshape(
new_direction.shape[:-1]) - candidate_plane) - current_df
new_points.shape = new_direction.shape
bad_steps = energy_diff > alpha * 1e-4 * (new_direction *
grad).sum(axis=-1)
backtracking_iterations = 0
while np.any(bad_steps):
alpha[bad_steps] *= 0.5
new_points = points + alpha[...,
np.newaxis] * new_direction
#print('new_points', new_points)
#print('bad_steps', bad_steps)
new_points = new_points.reshape(
new_points.shape[:len(indep_vals)] +
(-1, new_points.shape[-1]))
candidates = calculate(dbf,
comps,
name,
output='GM',
model=models,
callables=callable_dict,
fake_points=False,
points=new_points,
**grid_opts)
candidate_plane = np.multiply(
candidates.X.values.reshape(points.shape[:-1] +
(len(components), )),
properties.MU.values[..., np.newaxis, :]).sum(axis=-1)
energy_diff = (candidates.GM.values.reshape(
new_direction.shape[:-1]) -
candidate_plane) - current_df
#print('energy_diff', energy_diff)
new_points.shape = new_direction.shape
bad_steps = energy_diff > alpha * 1e-4 * (
new_direction * grad).sum(axis=-1)
backtracking_iterations += 1
if backtracking_iterations > MAX_BACKTRACKING:
break
biggest_step = np.max(
np.linalg.norm(new_points - points, axis=-1))
if biggest_step < 1e-2:
if verbose:
print('N-R convergence on mini-iteration',
newton_iteration, '[{}]'.format(name))
points = new_points
break
if verbose:
#print('Biggest step:', biggest_step)
#print('points', points)
#print('grad of points', grad)
#print('new_direction', new_direction)
#print('alpha', alpha)
#print('new_points', new_points)
pass
points = new_points
newton_iteration += 1
new_points = points.reshape(points.shape[:len(indep_vals)] +
(-1, points.shape[-1]))
new_points = np.concatenate(
(current_site_fractions[..., :dof], new_points), axis=-2)
points_dict[name] = new_points
if verbose:
print('Rebuilding grid', end=' ')
grid = calculate(dbf,
comps,
active_phases,
output='GM',
model=models,
callables=callable_dict,
fake_points=True,
points=points_dict,
**grid_opts)
if verbose:
print('[{0} points, {1}]'.format(len(grid.points),
sizeof_fmt(grid.nbytes)),
end='\n')
properties.attrs['iterations'] += 1
# One last call to ensure 'properties' and 'grid' are consistent with one another
lower_convex_hull(grid, properties)
ravelled_X_view = grid['X'].values.view().reshape(
-1, grid['X'].values.shape[-1])
ravelled_Y_view = grid['Y'].values.view().reshape(
-1, grid['Y'].values.shape[-1])
ravelled_Phase_view = grid['Phase'].values.view().reshape(-1)
# Copy final point values from the grid and drop the index array
# For some reason direct construction doesn't work. We have to create empty and then assign.
properties['X'] = xray.DataArray(
np.empty_like(ravelled_X_view[properties['points'].values]),
dims=properties['points'].dims + ('component', ))
properties['X'].values[...] = ravelled_X_view[properties['points'].values]
properties['Y'] = xray.DataArray(
np.empty_like(ravelled_Y_view[properties['points'].values]),
dims=properties['points'].dims + ('internal_dof', ))
properties['Y'].values[...] = ravelled_Y_view[properties['points'].values]
# TODO: What about invariant reactions? We should perform a final driving force calculation here.
# We can handle that in the same post-processing step where we identify single-phase regions.
properties['Phase'] = xray.DataArray(np.empty_like(
ravelled_Phase_view[properties['points'].values]),
dims=properties['points'].dims)
properties['Phase'].values[...] = ravelled_Phase_view[
properties['points'].values]
del properties['points']
return properties
def equilibrium(dbf, comps, phases, conditions, **kwargs):
"""
Calculate the equilibrium state of a system containing the specified
components and phases, under the specified conditions.
Model parameters are taken from 'dbf'.
Parameters
----------
dbf : Database
Thermodynamic database containing the relevant parameters.
comps : list
Names of components to consider in the calculation.
phases : list or dict
Names of phases to consider in the calculation.
conditions : dict or (list of dict)
StateVariables and their corresponding value.
verbose : bool, optional (Default: True)
Show progress of calculations.
grid_opts : dict, optional
Keyword arguments to pass to the initial grid routine.
Returns
-------
Structured equilibrium calculation.
Examples
--------
None yet.
"""
active_phases = unpack_phases(phases) or sorted(dbf.phases.keys())
comps = sorted(comps)
indep_vars = ['T', 'P']
grid_opts = kwargs.pop('grid_opts', dict())
verbose = kwargs.pop('verbose', True)
phase_records = dict()
callable_dict = kwargs.pop('callables', dict())
grad_callable_dict = kwargs.pop('grad_callables', dict())
points_dict = dict()
maximum_internal_dof = 0
# Construct models for each phase; prioritize user models
models = unpack_kwarg(kwargs.pop('model', Model), default_arg=Model)
if verbose:
print('Components:', ' '.join(comps))
print('Phases:', end=' ')
for name in active_phases:
mod = models[name]
if isinstance(mod, type):
models[name] = mod = mod(dbf, comps, name)
variables = sorted(mod.energy.atoms(v.StateVariable).union(
{key
for key in conditions.keys() if key in [v.T, v.P]}),
key=str)
site_fracs = sorted(mod.energy.atoms(v.SiteFraction), key=str)
maximum_internal_dof = max(maximum_internal_dof, len(site_fracs))
# Extra factor '1e-100...' is to work around an annoying broadcasting bug for zero gradient entries
models[name].models['_broadcaster'] = 1e-100 * Mul(*variables)**3
out = models[name].energy
if name not in callable_dict:
undefs = list(out.atoms(Symbol) - out.atoms(v.StateVariable))
for undef in undefs:
out = out.xreplace({undef: float(0)})
# callable_dict takes variables in a different order due to calculate() pecularities
callable_dict[name] = make_callable(
out,
sorted((key for key in conditions.keys() if key in [v.T, v.P]),
key=str) + site_fracs)
if name not in grad_callable_dict:
grad_func = make_callable(
Matrix([out]).jacobian(variables), variables)
else:
grad_func = grad_callable_dict[name]
# Adjust gradient by the approximate chemical potentials
plane_vars = sorted(models[name].energy.atoms(v.SiteFraction), key=str)
hyperplane = Add(*[
v.MU(i) * mole_fraction(dbf.phases[name], comps, i) for i in comps
if i != 'VA'
])
# Workaround an annoying bug with zero gradient entries
# This forces numerically zero entries to broadcast correctly
hyperplane += 1e-100 * Mul(*([v.MU(i) for i in comps if i != 'VA'] +
plane_vars + [v.T, v.P]))**3
plane_grad = make_callable(
Matrix([hyperplane]).jacobian(variables),
[v.MU(i) for i in comps if i != 'VA'] + plane_vars + [v.T, v.P])
plane_hess = make_callable(hessian(hyperplane, variables),
[v.MU(i) for i in comps if i != 'VA'] +
plane_vars + [v.T, v.P])
phase_records[name.upper()] = PhaseRecord(variables=variables,
grad=grad_func,
plane_grad=plane_grad,
plane_hess=plane_hess)
if verbose:
print(name, end=' ')
if verbose:
print('[done]', end='\n')
conds = OrderedDict((key, unpack_condition(value))
for key, value in sorted(conditions.items(), key=str))
str_conds = OrderedDict((str(key), value) for key, value in conds.items())
indep_vals = list([float(x) for x in np.atleast_1d(val)]
for key, val in str_conds.items() if key in indep_vars)
components = [x for x in sorted(comps) if not x.startswith('VA')]
# 'calculate' accepts conditions through its keyword arguments
grid_opts.update(
{key: value
for key, value in str_conds.items() if key in indep_vars})
if 'pdens' not in grid_opts:
grid_opts['pdens'] = 10
coord_dict = str_conds.copy()
coord_dict['vertex'] = np.arange(len(components))
grid_shape = np.meshgrid(*coord_dict.values(), indexing='ij',
sparse=False)[0].shape
coord_dict['component'] = components
if verbose:
print('Computing initial grid', end=' ')
grid = calculate(dbf,
comps,
active_phases,
output='GM',
model=models,
callables=callable_dict,
fake_points=True,
**grid_opts)
if verbose:
print('[{0} points, {1}]'.format(len(grid.points),
sizeof_fmt(grid.nbytes)),
end='\n')
properties = xray.Dataset(
{
'NP': (list(str_conds.keys()) + ['vertex'], np.empty(grid_shape)),
'GM': (list(str_conds.keys()), np.empty(grid_shape[:-1])),
'MU':
(list(str_conds.keys()) + ['component'], np.empty(grid_shape)),
'points': (list(str_conds.keys()) + ['vertex'],
np.empty(grid_shape, dtype=np.int))
},
coords=coord_dict,
attrs={'iterations': 1},
)
# Store the potentials from the previous iteration
current_potentials = properties.MU.copy()
for iteration in range(MAX_ITERATIONS):
if verbose:
print('Computing convex hull [iteration {}]'.format(
properties.attrs['iterations']))
# lower_convex_hull will modify properties
lower_convex_hull(grid, properties)
progress = np.abs(current_potentials - properties.MU).max().values
if verbose:
print('progress', progress)
if progress < MIN_PROGRESS:
if verbose:
print('Convergence achieved')
break
current_potentials[...] = properties.MU.values
if verbose:
print('Refining convex hull')
# Insert extra dimensions for non-T,P conditions so GM broadcasts correctly
energy_broadcast_shape = grid.GM.values.shape[:len(indep_vals)] + \
(1,) * (len(str_conds) - len(indep_vals)) + (grid.GM.values.shape[-1],)
driving_forces = np.einsum('...i,...i',
properties.MU.values[..., np.newaxis, :],
grid.X.values[np.index_exp[...] +
(np.newaxis,) * (len(str_conds) - len(indep_vals)) +
np.index_exp[:, :]]) - \
grid.GM.values.view().reshape(energy_broadcast_shape)
for name in active_phases:
dof = len(models[name].energy.atoms(v.SiteFraction))
current_phase_indices = (grid.Phase.values == name
).reshape(energy_broadcast_shape[:-1] +
(-1, ))
# Broadcast to capture all conditions
current_phase_indices = np.broadcast_arrays(
current_phase_indices, np.empty(driving_forces.shape))[0]
# This reshape is safe as long as phases have the same number of points at all indep. conditions
current_phase_driving_forces = driving_forces[
current_phase_indices].reshape(
current_phase_indices.shape[:-1] + (-1, ))
# Note: This works as long as all points are in the same phase order for all T, P
current_site_fractions = grid.Y.values[..., current_phase_indices[
(0, ) * len(str_conds)], :]
if np.sum(
current_site_fractions[(0, ) *
len(indep_vals)][..., :dof]) == dof:
# All site fractions are 1, aka zero internal degrees of freedom
# Impossible to refine these points, so skip this phase
points_dict[name] = current_site_fractions[
(0, ) * len(indep_vals)][..., :dof]
continue
# Find the N points with largest driving force for a given set of conditions
# Remember that driving force has a sign, so we want the "most positive" values
# N is the number of components, in this context
# N points define a 'best simplex' for every set of conditions
# We also need to restrict ourselves to one phase at a time
trial_indices = np.argpartition(current_phase_driving_forces,
-len(components),
axis=-1)[..., -len(components):]
trial_indices = trial_indices.ravel()
statevar_indices = np.unravel_index(
np.arange(
np.multiply.reduce(properties.GM.values.shape +
(len(components), ))),
properties.GM.values.shape +
(len(components), ))[:len(indep_vals)]
points = current_site_fractions[np.index_exp[statevar_indices +
(trial_indices, )]]
points.shape = properties.points.shape[:-1] + (
-1, maximum_internal_dof)
# The Y arrays have been padded, so we should slice off the padding
points = points[..., :dof]
# Workaround for derivative issues at endmembers
points[points == 0.] = MIN_SITE_FRACTION
if len(points) == 0:
if name in points_dict:
del points_dict[name]
# No nearly stable points: skip this phase
continue
num_vars = len(phase_records[name].variables)
plane_grad = phase_records[name].plane_grad
plane_hess = phase_records[name].plane_hess
statevar_grid = np.meshgrid(*itertools.chain(indep_vals),
sparse=True,
indexing='xy')
# TODO: A more sophisticated treatment of constraints
num_constraints = len(indep_vals) + len(
dbf.phases[name].sublattices)
constraint_jac = np.zeros((num_constraints, num_vars))
# Independent variables are always fixed (in this limited implementation)
for idx in range(len(indep_vals)):
constraint_jac[idx, idx] = 1
# This is for site fraction balance constraints
var_idx = len(indep_vals)
for idx in range(len(dbf.phases[name].sublattices)):
active_in_subl = set(
dbf.phases[name].constituents[idx]).intersection(comps)
constraint_jac[len(indep_vals) + idx,
var_idx:var_idx + len(active_in_subl)] = 1
var_idx += len(active_in_subl)
grad = phase_records[name].grad(
*itertools.chain(statevar_grid, points.T))
if grad.dtype == 'object':
# Workaround a bug in zero gradient entries
grad_zeros = np.zeros(points.T.shape[1:], dtype=np.float)
for i in np.arange(grad.shape[0]):
if isinstance(grad[i], int):
grad[i] = grad_zeros
grad = np.array(grad.tolist(), dtype=np.float)
bcasts = np.broadcast_arrays(
*itertools.chain(properties.MU.values.T, points.T))
cast_grad = -plane_grad(*itertools.chain(bcasts, [0], [0]))
cast_grad = cast_grad.T + grad.T
grad = cast_grad
grad.shape = grad.shape[:-1] # Remove extraneous dimension
# This Hessian is an approximation updated using the BFGS method
# See Nocedal and Wright, ch.3, p. 198
# Initialize as identity matrix
hess = broadcast_to(np.eye(num_vars),
grad.shape + (grad.shape[-1], )).copy()
newton_iteration = 0
while newton_iteration < MAX_NEWTON_ITERATIONS:
e_matrix = np.linalg.inv(hess)
dy_unconstrained = -np.einsum('...ij,...j->...i', e_matrix,
grad)
proj_matrix = np.dot(e_matrix, constraint_jac.T)
inv_matrix = np.rollaxis(np.dot(constraint_jac, proj_matrix),
0, -1)
inv_term = np.linalg.inv(inv_matrix)
first_term = np.einsum('...ij,...jk->...ik', proj_matrix,
inv_term)
# Normally a term for the residual here
# We only choose starting points which obey the constraints, so r = 0
cons_summation = np.einsum('...i,...ji->...j',
dy_unconstrained, constraint_jac)
cons_correction = np.einsum('...ij,...j->...i', first_term,
cons_summation)
dy_constrained = dy_unconstrained - cons_correction
# TODO: Support for adaptive changing independent variable steps
new_direction = dy_constrained[..., len(indep_vals):]
# Backtracking line search
new_points = points + INITIAL_STEP_SIZE * new_direction
alpha = np.full(new_points.shape[:-1],
INITIAL_STEP_SIZE,
dtype=np.float)
negative_points = np.any(new_points < 0., axis=-1)
while np.any(negative_points):
alpha[negative_points] *= 0.1
new_points = points + alpha[...,
np.newaxis] * new_direction
negative_points = np.any(new_points < 0., axis=-1)
# If we made "near" zero progress on any points, don't update the Hessian until
# we've rebuilt the convex hull
# Nocedal and Wright recommend against skipping Hessian updates
# They recommend using a damped update approach, pp. 538-539 of their book
# TODO: Check the projected gradient norm, not the step length
if np.any(
np.max(np.abs(alpha[..., np.newaxis] * new_direction),
axis=-1) < MIN_STEP_LENGTH):
break
# Workaround for derivative issues at endmembers
new_points[new_points == 0.] = 1e-16
# BFGS update to Hessian
new_grad = phase_records[name].grad(
*itertools.chain(statevar_grid, new_points.T))
if new_grad.dtype == 'object':
# Workaround a bug in zero gradient entries
grad_zeros = np.zeros(new_points.T.shape[1:],
dtype=np.float)
for i in np.arange(new_grad.shape[0]):
if isinstance(new_grad[i], int):
new_grad[i] = grad_zeros
new_grad = np.array(new_grad.tolist(), dtype=np.float)
bcasts = np.broadcast_arrays(
*itertools.chain(properties.MU.values.T, new_points.T))
cast_grad = -plane_grad(*itertools.chain(bcasts, [0], [0]))
cast_grad = cast_grad.T + new_grad.T
new_grad = cast_grad
new_grad.shape = new_grad.shape[:
-1] # Remove extraneous dimension
# Notation used here consistent with Nocedal and Wright
s_k = np.empty(points.shape[:-1] +
(points.shape[-1] + len(indep_vals), ))
# Zero out independent variable changes for now
s_k[..., :len(indep_vals)] = 0
s_k[..., len(indep_vals):] = new_points - points
y_k = new_grad - grad
s_s_term = np.einsum('...j,...k->...jk', s_k, s_k)
s_b_s_term = np.einsum('...i,...ij,...j', s_k, hess, s_k)
y_y_y_s_term = np.einsum('...j,...k->...jk', y_k, y_k) / \
np.einsum('...i,...i', y_k, s_k)[..., np.newaxis, np.newaxis]
update = np.einsum('...ij,...jk,...kl->...il', hess, s_s_term, hess) / \
s_b_s_term[..., np.newaxis, np.newaxis] + y_y_y_s_term
hess = hess - update
cast_hess = -plane_hess(
*itertools.chain(bcasts, [0], [0])).T + hess
hess = -cast_hess #TODO: Why does this fix things?
# TODO: Verify that the chosen step lengths reduce the energy
points = new_points
grad = new_grad
newton_iteration += 1
new_points = new_points.reshape(
new_points.shape[:len(indep_vals)] +
(-1, new_points.shape[-1]))
new_points = np.concatenate(
(current_site_fractions[..., :dof], new_points), axis=-2)
points_dict[name] = new_points
if verbose:
print('Rebuilding grid', end=' ')
grid = calculate(dbf,
comps,
active_phases,
output='GM',
model=models,
callables=callable_dict,
fake_points=True,
points=points_dict,
**grid_opts)
if verbose:
print('[{0} points, {1}]'.format(len(grid.points),
sizeof_fmt(grid.nbytes)),
end='\n')
properties.attrs['iterations'] += 1
# One last call to ensure 'properties' and 'grid' are consistent with one another
lower_convex_hull(grid, properties)
ravelled_X_view = grid['X'].values.view().reshape(
-1, grid['X'].values.shape[-1])
ravelled_Y_view = grid['Y'].values.view().reshape(
-1, grid['Y'].values.shape[-1])
ravelled_Phase_view = grid['Phase'].values.view().reshape(-1)
# Copy final point values from the grid and drop the index array
# For some reason direct construction doesn't work. We have to create empty and then assign.
properties['X'] = xray.DataArray(
np.empty_like(ravelled_X_view[properties['points'].values]),
dims=properties['points'].dims + ('component', ))
properties['X'].values[...] = ravelled_X_view[properties['points'].values]
properties['Y'] = xray.DataArray(
np.empty_like(ravelled_Y_view[properties['points'].values]),
dims=properties['points'].dims + ('internal_dof', ))
properties['Y'].values[...] = ravelled_Y_view[properties['points'].values]
# TODO: What about invariant reactions? We should perform a final driving force calculation here.
# We can handle that in the same post-processing step where we identify single-phase regions.
properties['Phase'] = xray.DataArray(np.empty_like(
ravelled_Phase_view[properties['points'].values]),
dims=properties['points'].dims)
properties['Phase'].values[...] = ravelled_Phase_view[
properties['points'].values]
del properties['points']
return properties