improve the _three_body.py and test_M3GNetCalculator in test_ase.py

materialsvirtuallab · Aug 29, 2023 · 9e1a24b · 9e1a24b
1 parent 120a36e
commit 9e1a24b
Show file tree

Hide file tree

Showing 3 changed files with 9 additions and 13 deletions.
diff --git a/matgl/layers/_three_body.py b/matgl/layers/_three_body.py
@@ -53,11 +53,11 @@ def forward(
         end_atom_index = torch.gather(graph.edges()[1], 0, line_graph.edges()[1].to(torch.int64))
         atoms = self.update_network_atom(node_feat)
         end_atom_index = torch.unsqueeze(end_atom_index, 1)
-        atoms = torch.squeeze(atoms[end_atom_index.long()])
+        atoms = torch.squeeze(atoms[end_atom_index])
         basis = three_basis * atoms
         three_cutoff = torch.unsqueeze(three_cutoff, dim=1)  # type: ignore
-        weights = torch.reshape(three_cutoff[torch.stack(list(line_graph.edges()), dim=1)], (-1, 2))  # type: ignore
-        weights = torch.prod(weights, axis=-1)  # type: ignore
+        weights = three_cutoff[torch.stack(list(line_graph.edges()), dim=1)].view(-1, 2)  # type: ignore
+        weights = torch.prod(weights, dim=-1)  # type: ignore
         basis = basis * weights[:, None]
         new_bonds = scatter_sum(
             basis.to(matgl.float_th),

diff --git a/matgl/utils/maths.py b/matgl/utils/maths.py
@@ -101,13 +101,11 @@ def get_segment_indices_from_n(ns):
         ns: torch.Tensor, the number of atoms/bonds array
 
     Returns:
-        object:
-
-    Returns: segment indices tensor
+        torch.Tensor: segment indices tensor
     """
-    B = ns
-    A = torch.arange(B.size(dim=0))
-    return A.repeat_interleave(B, dim=0)
+    segments = torch.zeros(ns.sum(), dtype=matgl.int_th)
+    segments[ns.cumsum(0)[:-1]] = 1
+    return segments.cumsum(0)
 
 
 def get_range_indices_from_n(ns):

diff --git a/tests/ext/test_ase.py b/tests/ext/test_ase.py
@@ -8,21 +8,19 @@
 from pymatgen.io.ase import AseAtomsAdaptor
 
 from matgl import load_model
-from matgl.apps.pes import Potential
 from matgl.ext.ase import Atoms2Graph, M3GNetCalculator, MolecularDynamics, Relaxer
-from matgl.models import M3GNet
 
 
 def test_M3GNetCalculator(MoS):
     adaptor = AseAtomsAdaptor()
     s_ase = adaptor.get_atoms(MoS)  # type: ignore
-    model = M3GNet(element_types=["Mo", "S"], is_intensive=False)
-    ff = Potential(model=model)
+    ff = load_model("M3GNet-MP-2021.2.8-PES")
     calc = M3GNetCalculator(potential=ff)
     s_ase.set_calculator(calc)
     assert [s_ase.get_potential_energy().size] == [1]
     assert list(s_ase.get_forces().shape) == [2, 3]
     assert list(s_ase.get_stress().shape) == [6]
+    np.testing.assert_allclose(s_ase.get_potential_energy(), -10.312888)
 
 
 def test_M3GNetCalculator_mol(AcAla3NHMe):