From b300e365d0886695b0d0220aad1be6cb38cf3c36 Mon Sep 17 00:00:00 2001
From: David Luevano Alvarado <55825613+luevano@users.noreply.github.com>
Date: Sun, 23 Feb 2020 22:06:30 -0700
Subject: Add bos to compound, and bugfix to bos

---
 ml_exp/__init__.py        |  6 ++++--
 ml_exp/compound.py        | 15 ++++++++++++++-
 ml_exp/representations.py | 20 ++++++++++++--------
 3 files changed, 30 insertions(+), 11 deletions(-)

diff --git a/ml_exp/__init__.py b/ml_exp/__init__.py
index d81f0999a..dcb10a1df 100644
--- a/ml_exp/__init__.py
+++ b/ml_exp/__init__.py
@@ -22,7 +22,7 @@ SOFTWARE.
 """
 from ml_exp.compound import Compound
 from ml_exp.representations import coulomb_matrix, lennard_jones_matrix,\
-    first_neighbor_matrix, adjacency_matrix
+    first_neighbor_matrix, adjacency_matrix, check_bond, bag_of_stuff
 
 # If somebody does "from package import *", this is what they will
 # be able to access:
@@ -30,4 +30,6 @@ __all__ = ['Compound',
            'coulomb_matrix',
            'lennard_jones_matrix',
            'first_neighbor_matrix',
-           'adjacency_matrix']
+           'adjacency_matrix',
+           'check_bond',
+           'bag_of_stuff']
diff --git a/ml_exp/compound.py b/ml_exp/compound.py
index 595078d55..8b6af0ae9 100644
--- a/ml_exp/compound.py
+++ b/ml_exp/compound.py
@@ -23,7 +23,7 @@ SOFTWARE.
 import numpy as np
 from ml_exp.data import NUCLEAR_CHARGE
 from ml_exp.representations import coulomb_matrix, lennard_jones_matrix,\
-    first_neighbor_matrix, adjacency_matrix
+    first_neighbor_matrix, adjacency_matrix, bag_of_stuff
 
 
 class Compound:
@@ -115,6 +115,19 @@ class Compound:
                                    forces,
                                    size=size)
 
+    def gen_bos(self,
+                size=23,
+                stuff='bonds'):
+        """
+        Generate the Bag of Stuff for the compound.
+        size: compound size.
+        stuff: elements of the bag, by default the known bag of bonds.
+        """
+        self.bos = bag_of_stuff(self.cm,
+                                self.atoms,
+                                size=size,
+                                stuff=stuff)
+
     def read_xyz(self,
                  filename):
         """
diff --git a/ml_exp/representations.py b/ml_exp/representations.py
index 1fe55aa5f..ea079595e 100644
--- a/ml_exp/representations.py
+++ b/ml_exp/representations.py
@@ -310,8 +310,8 @@ def check_bond(bags,
 
 def bag_of_stuff(cm,
                  atoms,
-                 stuff='bonds',
-                 size=23):
+                 size=23,
+                 stuff='bonds'):
     """
     Creates the Bag of Bonds using the Coulomb Matrix.
     cm: coulomb matrix.
@@ -322,6 +322,10 @@ def bag_of_stuff(cm,
         raise ValueError('Coulomb Matrix hasn\'t been initialized for the \
 current compound.')
 
+    if cm.ndim == 1:
+        raise ValueError('Coulomb Matrix (CM) dimension is 1. Maybe it was \
+generated as the vector of eigenvalues, try (re-)generating the CM.')
+
     n = len(atoms)
 
     if size < n:
@@ -330,7 +334,7 @@ current compound.')
         size = n
 
     # Bond max length, calculated using only the upper triangular matrix.
-    bond_size = (size * size - size)/2 + size
+    bond_size = int((size * size - size)/2 + size)
 
     # List where each bag data is stored.
     bags = []
@@ -362,19 +366,19 @@ current compound.')
                 bonds.append(''.join(sorted([a_i, a_j])))
     bonds = atom_list + bonds
 
-    # Create the final vector for the bob.
-    bob = np.zeros(bond_size, dtype=float)
+    # Create the final vector for the bos.
+    bos = np.zeros(bond_size, dtype=float)
     c_i = 0
     for i, bond in enumerate(bonds):
         checker = check_bond(bags, bond)
         if checker[0]:
             for j, num in enumerate(sorted(bags[checker[1]][1:])[::-1]):
-                # Use c_i as the index for bob if the zero padding should
+                # Use c_i as the index for bos if the zero padding should
                 # be at the end of the vector instead of between each bond.
-                bob[i*size + j] = num
+                bos[i*size + j] = num
                 c_i += 1
         else:
             print(f'Error. Bond {bond} from bond list coudn\'t be found',
                   'in the bags list. This could be a case where the atom',
                   'is only present oncce in the molecule.')
-    return bob
+    return bos
-- 
cgit v1.2.3-54-g00ecf