diff options
author | David Luevano Alvarado <55825613+luevano@users.noreply.github.com> | 2020-03-10 14:35:36 -0700 |
---|---|---|
committer | David Luevano Alvarado <55825613+luevano@users.noreply.github.com> | 2020-03-10 14:35:36 -0700 |
commit | 1b21beea876dc99fbba3f2bf548080bc10805cc5 (patch) | |
tree | b6cdd29e52e74e35e69b137c1cca3aa841cf7ae2 | |
parent | 14412daeef35e9a01d8827567bfdd34c88de303f (diff) |
Move pos bonds to data
-rw-r--r-- | ml_exp/__init__.py | 8 | ||||
-rw-r--r-- | ml_exp/data.py | 21 | ||||
-rw-r--r-- | ml_exp/representations.py | 26 |
3 files changed, 32 insertions, 23 deletions
diff --git a/ml_exp/__init__.py b/ml_exp/__init__.py index 6e901ec2c..d304f581c 100644 --- a/ml_exp/__init__.py +++ b/ml_exp/__init__.py @@ -22,8 +22,9 @@ SOFTWARE. """ from ml_exp.compound import Compound from ml_exp.representations import coulomb_matrix, lennard_jones_matrix,\ - get_helping_data, adjacency_matrix, check_bond, bag_of_bonds + get_helping_data, adjacency_matrix, epsilon_index, check_bond, bag_of_bonds from ml_exp.qm7db import qm7db +from ml_exp.data import NUCLEAR_CHARGE, POSSIBLE_BONDS from ml_exp.kernels import laplauss_kernel from ml_exp.do_ml import simple_ml, do_ml @@ -32,9 +33,12 @@ __all__ = ['Compound', 'lennard_jones_matrix', 'get_helping_data', 'adjacency_matrix', + 'epsilon_index', 'check_bond', 'bag_of_bonds', 'qm7db', 'laplauss_kernel', 'simple_ml', - 'do_ml'] + 'do_ml', + 'NUCLEAR_CHARGE', + 'POSSIBLE_BONDS'] diff --git a/ml_exp/data.py b/ml_exp/data.py index 243aff168..65d51cdcc 100644 --- a/ml_exp/data.py +++ b/ml_exp/data.py @@ -139,3 +139,24 @@ NUCLEAR_CHARGE = { 'Lv': 116, 'Ts': 117, 'Og': 118} + +""" +NOTE: Bond distance of carbon to other elements + are (for atoms present in the qm7 dataset): + C: 1.19 - 1.54 A, 1.0 + H: 1.06 - 1.12 A, 1.0 + O: 1.43 - 2.15 A, 0.8 + N: 1.47 - 2.10 A, 1.0 + S: 1.81 - 2.55 A, 0.7 +""" +cc_bond = ''.join(sorted(['C', 'C'])) +ch_bond = ''.join(sorted(['C', 'H'])) +co_bond = ''.join(sorted(['C', 'O'])) +cn_bond = ''.join(sorted(['C', 'N'])) +cs_bond = ''.join(sorted(['C', 'S'])) + +POSSIBLE_BONDS = {cc_bond: (1.19, 1.54, 1.0), + ch_bond: (1.06, 1.12, 1.0), + co_bond: (1.43, 2.15, 0.8), + cn_bond: (1.47, 2.19, 1.0), + cs_bond: (1.81, 2.55, 0.7)} diff --git a/ml_exp/representations.py b/ml_exp/representations.py index ff5741816..3e44d626c 100644 --- a/ml_exp/representations.py +++ b/ml_exp/representations.py @@ -22,6 +22,7 @@ SOFTWARE. """ import numpy as np from collections import Counter +from ml_exp.data import POSSIBLE_BONDS def coulomb_matrix(coords, @@ -155,13 +156,6 @@ def get_helping_data(coords, nc: nuclear charge data. size: compund size. bohr_ru: if radius units should be in bohr's radius units. - NOTE: Bond distance of carbon to other elements - are (for atoms present in the qm7 dataset): - C: 1.19 - 1.54 A, 1.0 - H: 1.06 - 1.12 A, 1.0 - O: 1.43 - 2.15 A, 0.8 - N: 1.47 - 2.10 A, 1.0 - S: 1.81 - 2.55 A, 0.7 """ if bohr_ru: cr = 0.52917721067 @@ -179,16 +173,6 @@ size. Arrays are not of the right shape.') 'instead of (size).') size = n - # Possible bonds. - cc_bond = ''.join(sorted(['C', 'C'])) - ch_bond = ''.join(sorted(['C', 'H'])) - co_bond = ''.join(sorted(['C', 'O'])) - cn_bond = ''.join(sorted(['C', 'N'])) - cs_bond = ''.join(sorted(['C', 'S'])) - pos_bonds = {cc_bond: (1.19, 1.54, 1.0), ch_bond: (1.06, 1.12, 1.0), - co_bond: (1.43, 2.15, 0.8), cn_bond: (1.47, 2.19, 1.0), - cs_bond: (1.81, 2.55, 0.7)} - fnm = np.zeros((n, n), dtype=bool) bonds = [] bonds_i = [] @@ -197,9 +181,9 @@ size. Arrays are not of the right shape.') for i in range(n - 1): for j in range(i + 1, n): bond = ''.join(sorted([atoms[i], atoms[j]])) - if bond in pos_bonds.keys(): - r_min = pos_bonds[bond][0] - r_max = pos_bonds[bond][1] + if bond in POSSIBLE_BONDS.keys(): + r_min = POSSIBLE_BONDS[bond][0] + r_max = POSSIBLE_BONDS[bond][1] rv = coords[i] - coords[j] r = np.linalg.norm(rv)/cr if r >= r_min and r <= r_max: @@ -207,7 +191,7 @@ size. Arrays are not of the right shape.') fnm[j, i] = True bonds.append(bond) bonds_i.append((i, j)) - bonds_k.append(pos_bonds[bond][2]) + bonds_k.append(POSSIBLE_BONDS[bond][2]) bonds_f.append(rv*nc[i]*nc[j]/r**3) fnm = np.pad(fnm, ((0, size - n), (0, size - n)), 'constant') |