def exercise_with_disorder():
p = shelx.parse_smtbx_refinement_model(file=StringIO(disordered))
xs = p.structure
mi = xs.build_miller_set(d_min=0.6, anomalous_flag=True)
fcalc = mi.structure_factors_from_scatterers(xs,
algorithm="direct").f_calc()
xm = refinement.model(fo_sq=fcalc.norm(),
xray_structure=xs,
constraints=p.constraints,
restraints_manager=p.restraints_manager,
weighting_scheme=p.weighting_scheme,
temperature_in_celsius=p.temperature_in_celsius,
conformer_indices=p.conformer_indices)
assert xm.temperature_in_celsius == -153
ls = xm.least_squares()
expected_reparametrisation_for = {}
for i in (13, 14, 15):
for s in 'ABCDEF':
pivot_letter = 'A' if s < 'D' else 'B'
expected_reparametrisation_for["H%i%s" % (i, s)] = (
core.terminal_tetrahedral_xh3_sites,
"C%i%s" % (i, pivot_letter))
for sc, params in itertools.izip(
ls.reparametrisation.structure.scatterers(),
ls.reparametrisation.asu_scatterer_parameters):
if sc.scattering_type != 'H':
continue
(expected_type,
expected_pivot) = expected_reparametrisation_for[sc.label]
assert isinstance(params.site, expected_type)
assert ([sc1.label for sc1 in params.site.argument(0).scatterers
] == [expected_pivot])
def exercise_with_disorder():
p = shelx.parse_smtbx_refinement_model(file=StringIO(disordered))
xs = p.structure
mi = xs.build_miller_set(d_min=0.6, anomalous_flag=True)
fcalc = mi.structure_factors_from_scatterers(xs, algorithm="direct").f_calc()
xm = refinement.model(fo_sq=fcalc.norm(),
xray_structure=xs,
constraints=p.constraints,
restraints_manager=p.restraints_manager,
weighting_scheme=p.weighting_scheme,
temperature_in_celsius=p.temperature_in_celsius,
conformer_indices=p.conformer_indices)
assert xm.temperature_in_celsius == -153
ls = xm.least_squares()
expected_reparametrisation_for = {}
for i in (13, 14, 15):
for s in 'ABCDEF':
pivot_letter = 'A' if s < 'D' else 'B'
expected_reparametrisation_for["H%i%s" % (i,s)] = (
core.terminal_tetrahedral_xh3_sites, "C%i%s" % (i,pivot_letter))
for sc, params in itertools.izip(
ls.reparametrisation.structure.scatterers(),
ls.reparametrisation.asu_scatterer_parameters):
if sc.scattering_type != 'H':
continue
(expected_type, expected_pivot) = expected_reparametrisation_for[sc.label]
assert isinstance(params.site, expected_type)
assert ([sc1.label for sc1 in params.site.argument(0).scatterers] ==
[expected_pivot])
def exercise_rigu():
ins = """
CELL 0.71073 7.772741 8.721603 10.863736 90 102.9832 90
ZERR 2 0.000944 0.001056 0.001068 0 0.0107 0
LATT -1
SYMM -X,0.5+Y,-Z
SFAC C H O
UNIT 24 44 22
RIGU 0.0001 0.0001 O4 C C12
C12 1 -0.12812 0.06329 -0.17592 11.00000 0.01467 0.02689 0.02780 =
-0.00379 0.00441 -0.00377
O4 3 0.08910 0.02721 0.02186 11.00000 0.02001 0.03168 0.03125 =
-0.00504 0.00144 -0.00274
C 1 -0.05545 -0.04221 -0.06528 11.00000 0.01560 0.02699 0.02581 =
-0.00481 0.00597 -0.00068
HKLF 4
END
"""
sio = StringIO(ins)
import iotbx.shelx as shelx
model = shelx.parse_smtbx_refinement_model(file=sio)
sites_cart = model.structure.sites_cart()
u_cart = model.structure.scatterers().extract_u_cart(
model.structure.unit_cell())
arp = adp_restraint_params(sites_cart=sites_cart, u_cart=u_cart)
for rp in model._proxies['rigu']:
rr = adp_restraints.rigu(arp, rp)
U1 = matrix.sym(sym_mat3=u_cart[rp.i_seqs[0]])
U2 = matrix.sym(sym_mat3=u_cart[rp.i_seqs[1]])
vz = matrix.col(sites_cart[rp.i_seqs[1]]) - matrix.col(
sites_cart[rp.i_seqs[0]])
vy = vz.ortho()
vx = vy.cross(vz)
R = matrix.rec(
vx.normalize().elems + vy.normalize().elems + vz.normalize().elems,
(3, 3))
# with this matrix we can only test Z component as X and Y will differ
dU = ((R * U1) * R.transpose() -
(R * U2) * R.transpose()).as_sym_mat3()
assert approx_equal(dU[2], rr.delta_33())
#with the original matrix all components should match
R1 = matrix.rec(rr.RM(), (3, 3))
dU = ((R1 * U1) * R1.transpose() -
(R1 * U2) * R1.transpose()).as_sym_mat3()
assert approx_equal(dU[2], rr.delta_33())
assert approx_equal(dU[4], rr.delta_13())
assert approx_equal(dU[5], rr.delta_23())
# check the raw gradients against the reference implementation
for x, y in zip(rr.reference_gradients(R1), rr.raw_gradients()):
for idx in range(0, 6):
assert approx_equal(x[idx], y[idx])
for x, y in zip(rigu_finite_diff(R1, U1), rr.raw_gradients()):
for idx in range(0, 6):
assert approx_equal(x[idx], y[idx])