def _maximize_xyz3_distance(dist, xyz1, ang, xyz2, against_xyzs=()):
""" place another atom, maximizing its distance from another set of atoms
"""
xyz0 = vec.arbitrary_unit_perpendicular(xyz2, orig_xyz=xyz1)
against_xyzs = numpy.array(against_xyzs)
def _distance_norm(dih):
xyz3 = vec.from_internals(dist=dist,
xyz1=xyz1,
ang=ang,
xyz2=xyz2,
dih=dih,
xyz3=xyz0)
dist_norm = numpy.linalg.norm(against_xyzs - numpy.array(xyz3))
return dist_norm
def join(geo1,
geo2,
key2,
key3,
r23,
a123=85.,
a234=85.,
d1234=85.,
key1=None,
key4=None,
angstrom=True,
degree=True):
""" join two geometries based on four of their atoms, two on the first
and two on the second
Variables set the coordinates for 1-2...3-4 where 1-2 are bonded atoms in
geo1 and 3-4 are bonded atoms in geo2.
"""
key3 = key3 - count(geo1)
a123 *= phycon.DEG2RAD if degree else 1
a234 *= phycon.DEG2RAD if degree else 1
d1234 *= phycon.DEG2RAD if degree else 1
gra1, gra2 = map(connectivity_graph, (geo1, geo2))
key1 = (automol.graph.atom_neighbor_atom_key(gra1, key2)
if key1 is None else key1)
key4 = (automol.graph.atom_neighbor_atom_key(gra2, key3)
if key4 is None else key4)
syms1 = symbols(geo1)
syms2 = symbols(geo2)
xyzs1 = coordinates(geo1, angstrom=angstrom)
xyzs2 = coordinates(geo2, angstrom=angstrom)
xyz1 = xyzs1[key1]
xyz2 = xyzs1[key2]
orig_xyz3 = xyzs2[key3]
if key4 is not None:
orig_xyz4 = xyzs2[key4]
else:
orig_xyz4 = [1., 1., 1.]
r34 = vec.distance(orig_xyz3, orig_xyz4)
# Place xyz3 as far away from the atoms in geo1 as possible by optimizing
# the undetermined dihedral angle
xyz0 = vec.arbitrary_unit_perpendicular(xyz2, orig_xyz=xyz1)
def _distance_norm(dih): # objective function for minimization
dih, = dih
xyz3 = vec.from_internals(dist=r23,
xyz1=xyz2,
ang=a123,
xyz2=xyz1,
dih=dih,
xyz3=xyz0)
dist_norm = numpy.linalg.norm(numpy.subtract(xyzs1, xyz3))
# return the negative norm so that minimum value gives maximum distance
return -dist_norm
res = scipy.optimize.basinhopping(_distance_norm, 0.)
dih = res.x[0]
# Now, get the next position with the optimized dihedral angle
xyz3 = vec.from_internals(dist=r23,
xyz1=xyz2,
ang=a123,
xyz2=xyz1,
dih=dih,
xyz3=xyz0)
# Don't use the dihedral angle if 1-2-3 are linear
if numpy.abs(a123 * phycon.RAD2DEG - 180.) > 5.:
xyz4 = vec.from_internals(dist=r34,
xyz1=xyz3,
ang=a234,
xyz2=xyz2,
dih=d1234,
xyz3=xyz1)
else:
xyz4 = vec.from_internals(dist=r34, xyz1=xyz3, ang=a234, xyz2=xyz2)
align_ = vec.aligner(orig_xyz3, orig_xyz4, xyz3, xyz4)
xyzs2 = tuple(map(align_, xyzs2))
syms = syms1 + syms2
xyzs = xyzs1 + xyzs2
geo = from_data(syms, xyzs, angstrom=angstrom)
return geo
def join(geo1,
geo2,
key2,
key3,
r23,
a123=85.,
a234=85.,
d1234=85.,
key1=None,
key4=None,
angstrom=True,
degree=True):
""" join two geometries based on four of their atoms, two on the first
and two on the second
Variables set the coordinates for 1-2...3-4 where 1-2 are bonded atoms in
geo1 and 3-4 are bonded atoms in geo2.
"""
key3 = key3 - count(geo1)
a123 *= phycon.DEG2RAD if degree else 1
a234 *= phycon.DEG2RAD if degree else 1
d1234 *= phycon.DEG2RAD if degree else 1
gra1, gra2 = map(connectivity_graph, (geo1, geo2))
key1 = (automol.graph.atom_neighbor_atom_key(gra1, key2)
if key1 is None else key1)
key4 = (automol.graph.atom_neighbor_atom_key(gra2, key3)
if key4 is None else key4)
syms1 = symbols(geo1)
syms2 = symbols(geo2)
xyzs1 = coordinates(geo1, angstrom=angstrom)
xyzs2 = coordinates(geo2, angstrom=angstrom)
xyz1 = xyzs1[key1] if key1 is not None else None
xyz2 = xyzs1[key2]
orig_xyz3 = xyzs2[key3]
orig_xyz4 = xyzs2[key4] if key4 is not None else [1., 1., 1.]
if key1 is None:
# If the first fragment is monatomic, we can place the other one
# anywhere we want (direction doesn't matter)
xyz3 = numpy.add(xyz2, [0., 0., r23])
else:
# If the first fragment isn't monatomic, we need to take some care in
# where we place the second one.
# r23 and a123 are fixed, so the only degree of freedom we have to do
# this is to optimize a dihedral angle relative to an arbitrary point
# xyz0.
# This dihedral angle is optimized to maximize the distance of xyz3
# from all of the atoms in fragment 1 (xyzs1).
xyz1 = xyzs1[key1]
# Place xyz3 as far away from the atoms in geo1 as possible by
# optimizing the undetermined dihedral angle
xyz0 = vec.arbitrary_unit_perpendicular(xyz2, orig_xyz=xyz1)
def _distance_norm(dih): # objective function for minimization
dih, = dih
xyz3 = vec.from_internals(dist=r23,
xyz1=xyz2,
ang=a123,
xyz2=xyz1,
dih=dih,
xyz3=xyz0)
dist_norm = numpy.linalg.norm(numpy.subtract(xyzs1, xyz3))
# return the negative norm so that minimum value gives maximum
# distance
return -dist_norm
res = scipy.optimize.basinhopping(_distance_norm, 0.)
dih = res.x[0]
# Now, get the next position with the optimized dihedral angle
xyz3 = vec.from_internals(dist=r23,
xyz1=xyz2,
ang=a123,
xyz2=xyz1,
dih=dih,
xyz3=xyz0)
r34 = vec.distance(orig_xyz3, orig_xyz4)
# If 2 doen't have neighbors or 1-2-3 are linear, ignore the dihedral angle
if key1 is None or numpy.abs(a123 * phycon.RAD2DEG - 180.) < 5.:
xyz4 = vec.from_internals(dist=r34, xyz1=xyz3, ang=a234, xyz2=xyz2)
else:
xyz4 = vec.from_internals(dist=r34,
xyz1=xyz3,
ang=a234,
xyz2=xyz2,
dih=d1234,
xyz3=xyz1)
align_ = vec.aligner(orig_xyz3, orig_xyz4, xyz3, xyz4)
xyzs2 = tuple(map(align_, xyzs2))
syms = syms1 + syms2
xyzs = xyzs1 + xyzs2
geo = from_data(syms, xyzs, angstrom=angstrom)
return geo