Coverage for /builds/kinetik161/ase/ase/optimize/cellawarebfgs.py: 98.57%

1from typing import IO, Optional, Union 

2import numpy as np 

3import time 

4from ase.optimize import BFGS 

5from ase.optimize.optimize import Dynamics 

6from ase import Atoms 

7from ase.geometry import cell_to_cellpar 

8from ase.units import GPa 

9 

10 

11def calculate_isotropic_elasticity_tensor(bulk_modulus, poisson_ratio, 

12 suppress_rotation=0): 

13 """ 

14 Parameters: 

15 bulk_modulus Bulk Modulus of the isotropic system used to set up the 

16 Hessian (in ASE units (eV/Å^3)). 

17 

18 poisson_ratio Poisson ratio of the isotropic system used to set up the 

19 initial Hessian (unitless, between -1 and 0.5). 

20 

21 suppress_rotation The rank-2 matrix C_ijkl.reshape((9,9)) has by 

22 default 6 non-zero eigenvalues, because energy is 

23 invariant to orthonormal rotations of the cell 

24 vector. This serves as a bad initial Hessian due to 3 

25 zero eigenvalues. Suppress rotation sets a value for 

26 those zero eigenvalues. 

27 

28 Returns C_ijkl 

29 """ 

30 

31 # https://scienceworld.wolfram.com/physics/LameConstants.html 

32 _lambda = 3 * bulk_modulus * poisson_ratio / (1 + 1 * poisson_ratio) 

33 _mu = _lambda * (1 - 2 * poisson_ratio) / (2 * poisson_ratio) 

34 

35 # https://en.wikipedia.org/wiki/Elasticity_tensor 

36 g_ij = np.eye(3) 

37 

38 # Construct 4th rank Elasticity tensor for isotropic systems 

39 C_ijkl = _lambda * np.einsum('ij,kl->ijkl', g_ij, g_ij) 

40 C_ijkl += _mu * (np.einsum('ik,jl->ijkl', g_ij, g_ij) + 

41 np.einsum('il,kj->ijkl', g_ij, g_ij)) 

42 

43 # Supplement the tensor with suppression of pure rotations that are right 

44 # now 0 eigenvalues. 

45 # Loop over all basis vectors of skew symmetric real matrix 

46 for i, j in ((0, 1), (0, 2), (1, 2)): 

47 Q = np.zeros((3, 3)) 

48 Q[i, j], Q[j, i] = 1, -1 

49 C_ijkl += (np.einsum('ij,kl->ijkl', Q, Q) 

50 * suppress_rotation / 2) 

51 

52 return C_ijkl 

53 

54 

55class CellAwareBFGS(BFGS): 

56 def __init__( 

57 self, 

58 atoms: Atoms, 

59 restart: Optional[str] = None, 

60 logfile: Union[IO, str] = '-', 

61 trajectory: Optional[str] = None, 

62 maxstep: Optional[float] = None, 

63 master: Optional[bool] = None, 

64 bulk_modulus: Optional[float] = 145 * GPa, 

65 poisson_ratio: Optional[float] = 0.3, 

66 alpha: Optional[float] = None, 

67 long_output: Optional[bool] = False, 

68 ): 

69 self.bulk_modulus = bulk_modulus 

70 self.poisson_ratio = poisson_ratio 

71 self.long_output = long_output 

72 BFGS.__init__(self, atoms=atoms, restart=restart, logfile=logfile, 

73 trajectory=trajectory, maxstep=maxstep, master=master, 

74 alpha=alpha) 

75 assert not isinstance(atoms, Atoms) 

76 if hasattr(atoms, 'exp_cell_factor'): 

77 assert atoms.exp_cell_factor == 1.0 

78 

79 def initialize(self): 

80 BFGS.initialize(self) 

81 C_ijkl = calculate_isotropic_elasticity_tensor( 

82 self.bulk_modulus, 

83 self.poisson_ratio, 

84 suppress_rotation=self.alpha) 

85 cell_H = self.H0[-9:, -9:] 

86 ind = np.where(self.atoms.mask.ravel() != 0)[0] 

87 cell_H[np.ix_(ind, ind)] = C_ijkl.reshape((9, 9))[ 

88 np.ix_(ind, ind)] * self.atoms.atoms.cell.volume 

89 

90 def converged(self, forces=None): 

91 if forces is None: 

92 forces = self.atoms.atoms.get_forces() 

93 stress = self.atoms.atoms.get_stress() 

94 return np.max(np.sum(forces**2, axis=1))**0.5 < self.fmax and \ 

95 np.max(np.abs(stress)) < self.smax 

96 

97 def run(self, fmax=0.05, smax=0.005, steps=None): 

98 """ call Dynamics.run and keep track of fmax""" 

99 self.fmax = fmax 

100 self.smax = smax 

101 if steps is not None: 

102 self.max_steps = steps 

103 return Dynamics.run(self) 

104 

105 def log(self, forces=None): 

106 if forces is None: 

107 forces = self.atoms.atoms.get_forces() 

108 fmax = (forces ** 2).sum(axis=1).max() ** 0.5 

109 e = self.optimizable.get_potential_energy() 

110 T = time.localtime() 

111 smax = abs(self.atoms.atoms.get_stress()).max() 

112 volume = self.atoms.atoms.cell.volume 

113 if self.logfile is not None: 

114 name = self.__class__.__name__ 

115 if self.nsteps == 0: 

116 args = (" " * len(name), 

117 "Step", "Time", "Energy", "fmax", "smax", "volume") 

118 msg = "\n%s %4s %8s %15s %15s %15s %15s" % args 

119 if self.long_output: 

120 msg += ("%8s %8s %8s %8s %8s %8s" % 

121 ('A', 'B', 'C', 'α', 'β', 'γ')) 

122 msg += '\n' 

123 self.logfile.write(msg) 

124 

125 ast = '' 

126 args = (name, self.nsteps, T[3], T[4], T[5], e, ast, fmax, smax, 

127 volume) 

128 msg = ("%s: %3d %02d:%02d:%02d %15.6f%1s %15.6f %15.6f %15.6f" % 

129 args) 

130 if self.long_output: 

131 msg += ("%8.3f %8.3f %8.3f %8.3f %8.3f %8.3f" % 

132 tuple(cell_to_cellpar(self.atoms.atoms.cell))) 

133 msg += '\n' 

134 self.logfile.write(msg) 

135 

136 self.logfile.flush()