Coverage for /builds/kinetik161/ase/ase/calculators/tip4p.py: 97.08%
137 statements
« prev ^ index » next coverage.py v7.2.7, created at 2023-12-10 11:04 +0000
1import numpy as np
3from ase import units
4from ase.calculators.calculator import Calculator, all_changes
5from ase.calculators.tip3p import TIP3P, angleHOH, rOH
7__all__ = ['rOH', 'angleHOH', 'TIP4P', 'sigma0', 'epsilon0']
9# Electrostatic constant and parameters:
10k_c = units.Hartree * units.Bohr
11qH = 0.52
12A = 600e3 * units.kcal / units.mol
13B = 610 * units.kcal / units.mol
14sigma0 = (A / B)**(1 / 6.)
15epsilon0 = B**2 / (4 * A)
16# https://doi.org/10.1063/1.445869
19class TIP4P(TIP3P):
20 def __init__(self, rc=7.0, width=1.0):
21 """ TIP4P potential for water.
23 :doi:`10.1063/1.445869`
25 Requires an atoms object of OHH,OHH, ... sequence
26 Correct TIP4P charges and LJ parameters set automatically.
28 Virtual interaction sites implemented in the following scheme:
29 Original atoms object has no virtual sites.
30 When energy/forces are requested:
32 * virtual sites added to temporary xatoms object
33 * energy / forces calculated
34 * forces redistributed from virtual sites to actual atoms object
36 This means you do not get into trouble when propagating your system
37 with MD while having to skip / account for massless virtual sites.
39 This also means that if using for QM/MM MD with GPAW, the EmbedTIP4P
40 class must be used.
41 """
43 TIP3P.__init__(self, rc, width)
44 self.atoms_per_mol = 3
45 self.sites_per_mol = 4
46 self.energy = None
47 self.forces = None
49 def calculate(self, atoms=None,
50 properties=['energy', 'forces'],
51 system_changes=all_changes):
52 Calculator.calculate(self, atoms, properties, system_changes)
54 assert (atoms.numbers[::3] == 8).all()
55 assert (atoms.numbers[1::3] == 1).all()
56 assert (atoms.numbers[2::3] == 1).all()
58 xpos = self.add_virtual_sites(atoms.positions)
59 xcharges = self.get_virtual_charges(atoms)
61 cell = atoms.cell
62 pbc = atoms.pbc
64 natoms = len(atoms)
65 nmol = natoms // 3
67 self.energy = 0.0
68 self.forces = np.zeros((4 * natoms // 3, 3))
70 C = cell.diagonal()
71 assert (cell == np.diag(C)).all(), 'not orthorhombic'
72 assert ((C >= 2 * self.rc) | ~pbc).all(), 'cutoff too large'
74 # Get dx,dy,dz from first atom of each mol to same atom of all other
75 # and find min. distance. Everything moves according to this analysis.
76 for a in range(nmol - 1):
77 D = xpos[(a + 1) * 4::4] - xpos[a * 4]
78 shift = np.zeros_like(D)
79 for i, periodic in enumerate(pbc):
80 if periodic:
81 shift[:, i] = np.rint(D[:, i] / C[i]) * C[i]
82 q_v = xcharges[(a + 1) * 4:]
84 # Min. img. position list as seen for molecule !a!
85 position_list = np.zeros(((nmol - 1 - a) * 4, 3))
87 for j in range(4):
88 position_list[j::4] += xpos[(a + 1) * 4 + j::4] - shift
90 # Make the smooth cutoff:
91 pbcRoo = position_list[::4] - xpos[a * 4]
92 pbcDoo = np.sum(np.abs(pbcRoo)**2, axis=-1)**(1 / 2)
93 x1 = pbcDoo > self.rc - self.width
94 x2 = pbcDoo < self.rc
95 x12 = np.logical_and(x1, x2)
96 y = (pbcDoo[x12] - self.rc + self.width) / self.width
97 t = np.zeros(len(pbcDoo))
98 t[x2] = 1.0
99 t[x12] -= y**2 * (3.0 - 2.0 * y)
100 dtdd = np.zeros(len(pbcDoo))
101 dtdd[x12] -= 6.0 / self.width * y * (1.0 - y)
102 self.energy_and_forces(a, xpos, position_list, q_v, nmol, t, dtdd)
104 if self.pcpot:
105 e, f = self.pcpot.calculate(xcharges, xpos)
106 self.energy += e
107 self.forces += f
109 f = self.redistribute_forces(self.forces)
111 self.results['energy'] = self.energy
112 self.results['forces'] = f
114 def energy_and_forces(self, a, xpos, position_list, q_v, nmol, t, dtdd):
115 """ energy and forces on molecule a from all other molecules.
116 cutoff is based on O-O Distance. """
118 # LJ part - only O-O interactions
119 epsil = np.tile([epsilon0], nmol - 1 - a)
120 sigma = np.tile([sigma0], nmol - 1 - a)
121 DOO = position_list[::4] - xpos[a * 4]
122 d2 = (DOO**2).sum(1)
123 d = np.sqrt(d2)
124 e_lj = 4 * epsil * (sigma**12 / d**12 - sigma**6 / d**6)
125 f_lj = (4 * epsil * (12 * sigma**12 / d**13 -
126 6 * sigma**6 / d**7) * t -
127 e_lj * dtdd)[:, np.newaxis] * DOO / d[:, np.newaxis]
129 self.forces[a * 4] -= f_lj.sum(0)
130 self.forces[(a + 1) * 4::4] += f_lj
132 # Electrostatics
133 e_elec = 0
134 all_cut = np.repeat(t, 4)
135 for i in range(4):
136 D = position_list - xpos[a * 4 + i]
137 d2_all = (D**2).sum(axis=1)
138 d_all = np.sqrt(d2_all)
139 e = k_c * q_v[i] * q_v / d_all
140 e_elec += np.dot(all_cut, e).sum()
141 e_f = e.reshape(nmol - a - 1, 4).sum(1)
142 F = (e / d_all * all_cut)[:, np.newaxis] * D / d_all[:, np.newaxis]
143 FOO = -(e_f * dtdd)[:, np.newaxis] * DOO / d[:, np.newaxis]
144 self.forces[(a + 1) * 4 + 0::4] += FOO
145 self.forces[a * 4] -= FOO.sum(0)
146 self.forces[(a + 1) * 4:] += F
147 self.forces[a * 4 + i] -= F.sum(0)
149 self.energy += np.dot(e_lj, t) + e_elec
151 def add_virtual_sites(self, pos):
152 # Order: OHHM,OHHM,...
153 # DOI: 10.1002/(SICI)1096-987X(199906)20:8
154 b = 0.15
155 xatomspos = np.zeros((4 * len(pos) // 3, 3))
156 for w in range(0, len(pos), 3):
157 r_i = pos[w] # O pos
158 r_j = pos[w + 1] # H1 pos
159 r_k = pos[w + 2] # H2 pos
160 n = (r_j + r_k) / 2 - r_i
161 n /= np.linalg.norm(n)
162 r_d = r_i + b * n
164 x = 4 * w // 3
165 xatomspos[x + 0] = r_i
166 xatomspos[x + 1] = r_j
167 xatomspos[x + 2] = r_k
168 xatomspos[x + 3] = r_d
170 return xatomspos
172 def get_virtual_charges(self, atoms):
173 charges = np.empty(len(atoms) * 4 // 3)
174 charges[0::4] = 0.00 # O
175 charges[1::4] = qH # H1
176 charges[2::4] = qH # H2
177 charges[3::4] = - 2 * qH # X1
178 return charges
180 def redistribute_forces(self, forces):
181 f = forces
182 b = 0.15
183 a = 0.5
184 pos = self.atoms.positions
185 for w in range(0, len(pos), 3):
186 r_i = pos[w] # O pos
187 r_j = pos[w + 1] # H1 pos
188 r_k = pos[w + 2] # H2 pos
189 r_ij = r_j - r_i
190 r_jk = r_k - r_j
191 r_d = r_i + b * (r_ij + a * r_jk) / np.linalg.norm(r_ij + a * r_jk)
192 r_id = r_d - r_i
193 gamma = b / np.linalg.norm(r_ij + a * r_jk)
195 x = w * 4 // 3
196 Fd = f[x + 3] # force on M
197 F1 = (np.dot(r_id, Fd) / np.dot(r_id, r_id)) * r_id
198 Fi = Fd - gamma * (Fd - F1) # Force from M on O
199 Fj = (1 - a) * gamma * (Fd - F1) # Force from M on H1
200 Fk = a * gamma * (Fd - F1) # Force from M on H2
202 f[x] += Fi
203 f[x + 1] += Fj
204 f[x + 2] += Fk
206 # remove virtual sites from force array
207 f = np.delete(f, list(range(3, f.shape[0], 4)), axis=0)
208 return f