diff options
author | David Luevano <55825613+luevano@users.noreply.github.com> | 2019-12-28 11:37:22 -0700 |
---|---|---|
committer | David Luevano <55825613+luevano@users.noreply.github.com> | 2019-12-28 11:37:22 -0700 |
commit | 4704314c9b4d1066383da5c3d6ca87bba9067c8d (patch) | |
tree | 47efddb979957029945a473fde6ed2cde2c2b196 /lj_matrix | |
parent | e4f9e15588ec796f73c000a683cc9152454a913c (diff) |
Refactor code
Diffstat (limited to 'lj_matrix')
-rw-r--r-- | lj_matrix/__main__.py | 1 | ||||
-rw-r--r-- | lj_matrix/do_ml.py | 5 | ||||
-rw-r--r-- | lj_matrix/lj_matrix.py | 2 | ||||
-rw-r--r-- | lj_matrix/read_qm7_data.py | 7 |
4 files changed, 10 insertions, 5 deletions
diff --git a/lj_matrix/__main__.py b/lj_matrix/__main__.py index 811024ff0..688e5adcc 100644 --- a/lj_matrix/__main__.py +++ b/lj_matrix/__main__.py @@ -31,6 +31,7 @@ if __name__ == '__main__': ljm_diag_value=None, ljm_sigma=1.0, ljm_epsilon=1.0, + r_seed=111, save_benchmarks=False, show_msgs=True) # plot_benchmarks() diff --git a/lj_matrix/do_ml.py b/lj_matrix/do_ml.py index da9386bf7..25a55e823 100644 --- a/lj_matrix/do_ml.py +++ b/lj_matrix/do_ml.py @@ -118,6 +118,7 @@ def do_ml(min_training_size, ljm_diag_value=None, ljm_sigma=1.0, ljm_epsilon=1.0, + r_seed=111, save_benchmarks=False, max_len=25, as_eig=True, @@ -134,6 +135,7 @@ def do_ml(min_training_size, ljm_diag_value: if a special diagonal value should be used in lj matrix. ljm_sigma: sigma value for lj matrix. ljm_epsilon: epsilon value for lj matrix. + r_seed: random seed to use for the shuffling. save_benchmarks: if benchmarks should be saved. max_len: maximum amount of atoms in molecule. as_eig: if data should be returned as matrix or array of eigenvalues. @@ -147,7 +149,8 @@ def do_ml(min_training_size, max_training_size = min_training_size + training_increment_size # Data reading. - molecules, nuclear_charge, energy_pbe0, energy_delta = read_qm7_data() + molecules, nuclear_charge, energy_pbe0, energy_delta =\ + read_qm7_data(r_seed) # Matrices calculation. cm_data, ljm_data = parallel_create_matrices(molecules, diff --git a/lj_matrix/lj_matrix.py b/lj_matrix/lj_matrix.py index c3b61becb..6739ae283 100644 --- a/lj_matrix/lj_matrix.py +++ b/lj_matrix/lj_matrix.py @@ -88,7 +88,7 @@ def lj_matrix(mol_data, z = (z_i-z_j)**2 if i == j: - if not diag_value: + if diag_value is None: lj[i, j] = (0.5*Z_i**2.4) else: lj[i, j] = diag_value diff --git a/lj_matrix/read_qm7_data.py b/lj_matrix/read_qm7_data.py index 9bb7629ca..4401ca1c0 100644 --- a/lj_matrix/read_qm7_data.py +++ b/lj_matrix/read_qm7_data.py @@ -59,7 +59,7 @@ def read_db_data(zi_data, its contents as usable variables. zi_data: dictionary containing nuclear charge data. data_path: path to the data directory. - r_seed: random seed. + r_seed: random seed to use for the shuffling. """ os.chdir(data_path) @@ -122,9 +122,10 @@ def read_db_data(zi_data, return molecules, nuclear_charge, energy_pbe0, energy_delta -def read_qm7_data(): +def read_qm7_data(r_seed=111): """ Reads all the qm7 data. + r_seed: random seed to use for the shuffling. """ tic = time.perf_counter() printc('Data reading started.', 'CYAN') @@ -135,7 +136,7 @@ def read_qm7_data(): zi_data = read_nc_data(data_path) molecules, nuclear_charge, energy_pbe0, energy_delta = \ - read_db_data(zi_data, data_path) + read_db_data(zi_data, data_path, r_seed) os.chdir(init_path) toc = time.perf_counter() |