Fixed off-by-one error of inner argument to evaluate().

When selecting inner solutions to be returned by `Model.evaluate()`,
the indices were off by one. This is because Comsol uses 1-based
array indexing whereas Python's is 0-based, and the `inner` indices
were used on the Python side after having Comsol return all solutions
at once (in cases where the evaluation feature does not accept the
indices as an argument).

mph/model.py

@@ -471,7 +471,9 @@ def evaluate(self, expression, unit=None, dataset=None,
             elif inner == 'last':
                 results = results[:, -1, :]
             else:
-                results = results[:, inner, :]
+                if isinstance(inner, list):
+                    inner = array(inner)
+                results = results[:, inner-1, :]
             return results.squeeze()
         # Move on if this fails. Seems to not be a global expression then.
         except Exception:
@@ -528,7 +530,9 @@ def evaluate(self, expression, unit=None, dataset=None,
             elif inner == 'last':
                 results = results[:, -1, :]
             else:
-                results = results[:, inner, :]
+                if isinstance(inner, list):
+                    inner = array(inner)
+                results = results[:, inner-1, :]
         logger.info('Finished retrieving data.')
 
         # Remove the temporary evaluation node we added to the model.

tests/test_model.py

@@ -302,19 +302,21 @@ def test_evaluate():
     C = model.evaluate(expression, unit, dataset)
     assert C[0] == Cf
     assert C[-1] == Cl
-    # To do: This test fails because inner indices are 1-based, not 0-based.
-    # C = model.evaluate(expression, unit, dataset, inner=[1, 101])
-    # assert C[0] == Cf
-    # assert C[1] == Cl
+    C = model.evaluate(expression, unit, dataset, inner=[1, 101])
+    assert C[0] == Cf
+    assert C[1] == Cl
     # Test field evaluation of time-dependent solution.
     (dataset, expression, unit) = ('time-dependent', 'ec.normD', 'nC/m^2')
     Df = model.evaluate(expression, unit, dataset, 'first')
     assert abs(Df.max() -  7.2) < 0.1
     Dl = model.evaluate(expression, unit, dataset, 'last')
     assert abs(Dl.max() - 10.8) < 0.1
     D = model.evaluate(expression, unit, dataset)
-    assert D[0].max()  == Df.max()
-    assert D[-1].max() == Dl.max()
+    assert (D[0]  == Df).all()
+    assert (D[-1] == Dl).all()
+    D = model.evaluate(expression, unit, dataset, inner=[1, 101])
+    assert (D[0] == Df).all()
+    assert (D[1] == Dl).all()
     # Test global evaluation of parameter sweep.
     (dataset, expression, unit) = ('parametric sweep', '2*ec.intWe/U^2', 'pF')
     (indices, values) = model.outer(dataset)
@@ -331,8 +333,8 @@ def test_evaluate():
     Dl = model.evaluate(expression, unit, dataset, 'last', 2)
     assert abs(Dl.max() - 10.8) < 0.1
     D = model.evaluate(expression, unit, dataset, outer=2)
-    assert D[0].max()  == Df.max()
-    assert D[-1].max() == Dl.max()
+    assert (D[0]  == Df).all()
+    assert (D[-1] == Dl).all()
     # Test argument "dataset".
     with logging_disabled():
         assert model.evaluate('U')