General improvemts

- All files are pyflakes and black clean.
- Improved logging for models.kernelridge, features.gaussian and
  features.coulombmatrix.

bin/ml4chem

@@ -1,5 +1,4 @@
 #!/usr/bin/env python3
-import time
 import click
 import os
 import sys
@@ -39,7 +38,7 @@ def main(**args):
         elif args["plot"].lower() in dim_visualization:
             method = args["plot"]
             backend = args["backend"]
-            plt = plot_atomic_features(_file, method=method, backend=backend)
+            plot_atomic_features(_file, method=method, backend=backend)
 
 
 if __name__ == "__main__":

ml4chem/data/parser.py

@@ -1,7 +1,7 @@
 import json
 from ase import Atom, Atoms
 from ase.calculators.calculator import Calculator
-from ase.io import Trajectory, write
+from ase.io import Trajectory
 
 
 # to get the total energy
@@ -68,7 +68,7 @@ class SinglePointCalculator(Calculator):
     """A SinglePointCalculator class
 
     This class creates a fake calculator that is used to populate
-    calc.results dictionaries in ASE objects. 
+    calc.results dictionaries in ASE objects.
 
     Parameters
     ----------
@@ -109,14 +109,14 @@ def get_forces(atoms):
         Returns
         -------
         forces
-            The atomic force of the molecule. 
+            The atomic force of the molecule.
         """
         return atoms.calc.results["forces"]
 
 
 def ani_to_ase(hdf5file, data_keys, trajfile=None):
     """ANI to ASE
-    
+
     Parameters
     ----------
     hdf5file : hdf5, list

ml4chem/features/coulombmatrix.py

@@ -4,7 +4,6 @@
 import logging
 import os
 import time
-import torch
 import pandas as pd
 from collections import OrderedDict
 from dscribe.descriptors import CoulombMatrix as CoulombMatrixDscribe
@@ -210,15 +209,10 @@ def calculate(self, images=None, purpose="training", data=None, svm=False):
             stacked_features += intermediate
             del intermediate
 
-        scheduler_time = time.time() - initial_time
+        # scheduler_time = time.time() - initial_time
 
         # dask.distributed.wait(stacked_features)
 
-        h, m, s = convert_elapsed_time(scheduler_time)
-        logger.info(
-            "... finished in {} hours {} minutes {:.2f}" " seconds.".format(h, m, s)
-        )
-
         logger.info("")
 
         if self.preprocessor is not None:

ml4chem/features/gaussian.py

@@ -249,6 +249,11 @@ def calculate(self, images=None, purpose="training", data=None, svm=False):
 
         self.print_fingerprint_params(self.GP)
 
+        symbol = data.unique_element_symbols[purpose][0]
+        sample = np.zeros(len(self.GP[symbol]))
+
+        self.dimension = len(sample)
+
         preprocessor = Preprocessing(self.preprocessor, purpose=purpose)
         preprocessor.set(purpose=purpose)
 
@@ -322,8 +327,6 @@ def calculate(self, images=None, purpose="training", data=None, svm=False):
             # To take advantage of dask_ml we need to convert our numpy array
             # into a dask array.
             logger.info("Converting features to dask array...")
-            symbol = data.unique_element_symbols[purpose][0]
-            sample = np.zeros(len(self.GP[symbol]))
             stacked_features = [
                 da.from_delayed(lazy, dtype=float, shape=sample.shape)
                 for lazy in stacked_features

ml4chem/models/autoencoders.py

@@ -1,6 +1,5 @@
 import dask
 import datetime
-import inspect
 import logging
 import time
 import torch
@@ -1214,7 +1213,7 @@ class Annealer(object):
     """Annealing class
 
     Based on on https://arxiv.org/abs/1903.10145.
-    
+
     Parameters
     ----------
     warm_up : int, optional
@@ -1236,12 +1235,12 @@ def __init__(self, warm_up=50, step=50, n_cycles=5):
 
     def update(self, epoch):
         """Update annealing value
-        
+
         Parameters
         ----------
         epoch : int
-            Epoch on the training process. 
-        
+            Epoch on the training process.
+
         Returns
         -------
         annealing

ml4chem/models/kernelridge.py

@@ -274,8 +274,10 @@ def get_kernel_matrix(self, feature_space, reference_features, purpose):
         kernel_matrix = []
 
         for c, chunk in enumerate(chunks):
-            print("Chunk", c)
+            chunk_initial_time = time.time()
+            logger.info("        Computing kernel functions for chunk {}...".format(c))
             intermediates = []
+
             if isinstance(chunk, dict) is False:
                 chunk = OrderedDict(chunk)
 
@@ -323,6 +325,13 @@ def get_kernel_matrix(self, feature_space, reference_features, purpose):
                         intermediates.append(kernel)
             kernel_matrix += dask.compute(intermediates, scheduler=self.scheduler)[0]
             del intermediates
+
+            chunk_final_time = time.time() - chunk_initial_time
+            h, m, s = convert_elapsed_time(chunk_final_time)
+            logger.info(
+                "          ...finished in {} hours {} minutes {:.2f} "
+                "seconds.".format(h, m, s)
+            )
             # dask.distributed.wait(kernel_matrix)
 
         del reference_features

ml4chem/models/loss.py

@@ -304,8 +304,6 @@ def VAELoss(
 
     loss = []
 
-    dim = 1
-
     if annealing is None:
         annealing = 1.0
 

ml4chem/models/merger.py

@@ -7,7 +7,7 @@
 import numpy as np
 from collections import OrderedDict
 from ml4chem.metrics import compute_rmse
-from ml4chem.utils import convert_elapsed_time, dynamic_import, get_chunks, lod_to_list
+from ml4chem.utils import dynamic_import, get_chunks, lod_to_list
 from ml4chem.optim.handler import get_optimizer, get_lr_scheduler
 
 # Setting precision and starting logger object