Fix CalculatorTorchSeparateSpecies zeta is [] bug

docs/source/auto_examples/example_nn_Si.ipynb

@@ -125,6 +125,13 @@
       "source": [
         "model.save(\"./final_model.pkl\")\nloss.save_optimizer_stat(\"./optimizer_stat.pkl\")\n\nmodel.write_kim_model()"
       ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "<div class=\"alert alert-info\"><h4>Note</h4><p>Now we have trained an NN for a single specie Si. If you have multiple species in\n   your system and want to use different parameters for different species,\n   take a look at the SiC_ example.</p></div>\n\n\n"
+      ]
     }
   ],
   "metadata": {

docs/source/auto_examples/example_nn_Si.py

@@ -141,3 +141,12 @@
 loss.save_optimizer_stat("./optimizer_stat.pkl")
 
 model.write_kim_model()
+
+
+##########################################################################################
+# .. note::
+#    Now we have trained an NN for a single specie Si. If you have multiple species in
+#    your system and want to use different parameters for different species,
+#    take a look at the SiC_ example.
+#
+# .. _SiC: https://github.com/mjwen/kliff/blob/master/examples/eg_nn_SiC.py

docs/source/auto_examples/example_nn_Si.py.md5

@@ -1 +1 @@
-99c760f443633dcf0ce755d5995447f7
+6baccb59d251b6e52d886f0ec66aa7ae

docs/source/auto_examples/example_nn_Si.rst

@@ -242,6 +242,8 @@ codes such as LAMMPS via the KIM API.
 
 
 
+
+
 .. rst-class:: sphx-glr-script-out
 
  Out:
@@ -253,10 +255,17 @@ codes such as LAMMPS via the KIM API.
 
 
 
+.. note::
+   Now we have trained an NN for a single specie Si. If you have multiple species in
+   your system and want to use different parameters for different species,
+   take a look at the SiC_ example.
+
+.. _SiC: https://github.com/mjwen/kliff/blob/master/examples/eg_nn_SiC.py
+
 
 .. rst-class:: sphx-glr-timing
 
-   **Total running time of the script:** ( 0 minutes  16.523 seconds)
+   **Total running time of the script:** ( 0 minutes  16.575 seconds)
 
 
 .. _sphx_glr_download_auto_examples_example_nn_Si.py:

docs/source/auto_examples/sg_execution_times.rst

@@ -5,12 +5,12 @@
 
 Computation times
 =================
-**01:57.516** total execution time for **auto_examples** files:
+**00:16.575** total execution time for **auto_examples** files:
 
 +-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_auto_examples_example_kim_SW_Si.py` (``example_kim_SW_Si.py``)                 | 01:40.195 | 0.0 MB |
+| :ref:`sphx_glr_auto_examples_example_nn_Si.py` (``example_nn_Si.py``)                         | 00:16.575 | 0.0 MB |
 +-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_auto_examples_example_nn_Si.py` (``example_nn_Si.py``)                         | 00:16.523 | 0.0 MB |
+| :ref:`sphx_glr_auto_examples_example_kim_SW_Si.py` (``example_kim_SW_Si.py``)                 | 00:00.000 | 0.0 MB |
 +-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_auto_examples_example_linear_regression.py` (``example_linear_regression.py``) | 00:00.798 | 0.0 MB |
+| :ref:`sphx_glr_auto_examples_example_linear_regression.py` (``example_linear_regression.py``) | 00:00.000 | 0.0 MB |
 +-----------------------------------------------------------------------------------------------+-----------+--------+

examples/eg_nn_SiC.py

@@ -1,10 +1,10 @@
 """
-.. _tut_nn:
+.. _tut_nn_multi_spec:
 
 Train a neural network potential
 ================================
 
-In this tutorial, we train a neural network (NN) potential for silicon
+In this tutorial, we train a neural network (NN) potential for SiC
 """
 
 

examples/example_nn_Si.py

@@ -141,3 +141,12 @@
 loss.save_optimizer_stat("./optimizer_stat.pkl")
 
 model.write_kim_model()
+
+
+##########################################################################################
+# .. note::
+#    Now we have trained an NN for a single specie Si. If you have multiple species in
+#    your system and want to use different parameters for different species,
+#    take a look at the SiC_ example.
+#
+# .. _SiC: https://github.com/mjwen/kliff/blob/master/examples/eg_nn_SiC.py

kliff/calculators/calculator_torch.py

@@ -248,7 +248,7 @@ def compute(self, batch):
         for s, zeta in zeta_by_species.items():
 
             # have no species "s" in this batch of data
-            if zeta is []:
+            if not zeta:  # zeta == []
                 continue
 
             z_tensor = torch.stack(zeta)  # convert a list of tensor to tensor

kliff/nn.py

@@ -31,12 +31,12 @@ def forward(self, input):
             shape_4D = (1, *shape, 1)
         elif dim == 3:
             if shape[0] != 1:
-                raise Exception('Shape[0] needs to be 1 for a 3D tensor.')
+                raise Exception("Shape[0] needs to be 1 for a 3D tensor.")
             shape_4D = (*shape, 1)
 
         else:
             raise Exception(
-                'Input need to be 2D or 3D tensor, but got a ' '{}D tensor.'.format(dim)
+                "Input need to be 2D or 3D tensor, but got a " "{}D tensor.".format(dim)
             )
         x = torch.reshape(input, shape_4D)
         x = torch.transpose(x, 1, 2)

kliff/wrapcalculator.py

@@ -41,7 +41,7 @@ def _update_params(self):
         """Write parameters to file KIM_MODEL_PARAMS + modelname, and also give its path to the
         enviroment variable that has the same name: KIM_MODEL_PARAMS + modelname.
         """
-        name = 'KIM_MODEL_PARAMS_' + self.params._modelname
+        name = "KIM_MODEL_PARAMS_" + self.params._modelname
         self.params.echo_params(name, print_size=True)
         path = os.getcwd() + os.path.sep + name
         os.environ[name] = path
@@ -59,13 +59,13 @@ def _parse_edn(self, fname, keys):
           Keyword in the edn format(think it as a dictionary), whose value will be returned.
         """
 
-        with open(fname, 'r') as fin:
+        with open(fname, "r") as fin:
             lines = fin.read()
         parsed = edn_format.loads(lines)
         values = []
         for k in keys:
             try:
-                v = parsed[k]['source-value']
+                v = parsed[k]["source-value"]
             except KeyError:
                 raise KeyError('Keyword "{}" not found in {}.'.format(k, self.outname))
             # make it a 1D array