def compute_functional_and_gradients(self):
self.unpack_x()
self.u_min = tuple([max(self.u_min_min, min(self.u_min_max, v))
for v in self.u_min])
if(self.b_min > self.b_sol_max): self.b_min = self.b_sol_max
if(self.b_min < self.b_sol_min): self.b_min = self.b_sol_min
self.k_min = [max(self.k_sol_min, min(self.k_sol_max, v))
for v in self.k_min]
if(self.twin_fraction is None):
self.fmodel_core_data.update(k_sols = self.k_min, b_sol = self.b_min,
u_star = self.u_min)
tg = bulk_solvent.bulk_solvent_and_aniso_scale_target_and_grads_ls(
fm = self.fmodel_core_data.data,
fo = self.f_obs.data(),
compute_k_sol_grad = self.refine_k,
compute_b_sol_grad = self.refine_b,
compute_u_star_grad = self.refine_u)
else:
self.fmodel_core_data.update(k_sols = self.k_min, b_sol = self.b_min,
u_star = self.u_min)
self.fmodel_core_data_twin.update(k_sols = self.k_min, b_sol = self.b_min,
u_star = self.u_min)
tg = bulk_solvent.bulk_solvent_and_aniso_scale_target_and_grads_ls(
fm1 = self.fmodel_core_data.data,
fm2 = self.fmodel_core_data_twin.data,
twin_fraction = self.twin_fraction,
fo = self.f_obs.data(),
compute_k_sol_grad = self.refine_k,
compute_b_sol_grad = self.refine_b,
compute_u_star_grad = self.refine_u)
self.f = tg.target()
gk=[0.0]*self.n_shells
gb=0
gu=[0,0,0,0,0,0]
if(self.refine_k or self.refine_b):
gk = list(tg.grad_k_sols())
assert len(gk) == self.n_shells
gb = tg.grad_b_sol()
if(self.refine_u): gu = list(tg.grad_u_star())
if(self.symmetry_constraints_on_b_cart and self.refine_u):
independent_params = flex.double(
self.adp_constraints.independent_gradients(all_gradients=gu))
self.g = self.pack(
u=independent_params,
k=gk,
b=gb,
u_factor=1/self.u_factor)
else:
self.g = self.pack(u=gu, k=gk, b=gb, u_factor=1/self.u_factor)
return self.f, self.g
def compute_functional_and_gradients(self):
self.unpack_x()
self.u_min = tuple([max(self.u_min_min, min(self.u_min_max, v))
for v in self.u_min])
self.fmodel_core_data.update(k_sols = [0], b_sol = 0, u_star = self.u_min)
tg = bulk_solvent.bulk_solvent_and_aniso_scale_target_and_grads_ls(
fm = self.fmodel_core_data.data,
fo = self.f_obs.data(),
compute_k_sol_grad = False,
compute_b_sol_grad = False,
compute_u_star_grad = self.refine_u)
self.f = tg.target()
gu=[0,0,0,0,0,0]
if(self.refine_u): gu = list(tg.grad_u_star())
if(self.symmetry_constraints_on_b_cart and self.refine_u):
independent_params = flex.double(
self.adp_constraints.independent_gradients(all_gradients=gu))
self.g = self.pack(
u=independent_params,
u_factor=1/self.u_factor)
else:
self.g = self.pack(u=gu, u_factor=1/self.u_factor)
return self.f, self.g
