Use np.array_split instead of np.split to support uneven splits in sp…

…sa._batch_evaluate (#8634)

* Use np.array_split instead of np.split to support uneven splits in spsa._batch_evaluate

* better testing

Co-authored-by: Julien Gacon <gaconju@gmail.com>

* Update test/python/algorithms/optimizers/test_spsa.py

Co-authored-by: Julien Gacon <gaconju@gmail.com>

* len of a boolean

* Apply suggestions from code review

* Fix Sphinx ref

Co-authored-by: Julien Gacon <gaconju@gmail.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

qiskit/algorithms/optimizers/spsa.py

@@ -727,7 +727,7 @@ def _batch_evaluate(function, points, max_evals_grouped):
         num_batches += 1
 
     # split the points
-    batched_points = np.split(np.asarray(points), num_batches)
+    batched_points = np.array_split(np.asarray(points), num_batches)
 
     results = []
     for batch in batched_points:

releasenotes/notes/qiskit-nature-797-8f1b0975309b8756.yaml

@@ -0,0 +1,6 @@
+---
+fixes:
+  - |
+    When the class :class:`~.SPSA` was using `np.split` (from NumPy) for splitting the jobs in even batches,
+    resulting in an exception if a perfectly even split was not possible. Now, it uses `np.array_split`, which is safer
+    for these cases.

test/python/algorithms/optimizers/test_spsa.py

@@ -181,3 +181,17 @@ def callback(nfev, point, fval, update, accepted):
         for i, (key, values) in enumerate(history.items()):
             self.assertTrue(all(isinstance(value, expected_types[i]) for value in values))
             self.assertEqual(len(history[key]), maxiter)
+
+    @data(1, 2, 3, 4)
+    def test_estimate_stddev(self, max_evals_grouped):
+        """Test the estimate_stddev
+        See https://github.com/Qiskit/qiskit-nature/issues/797"""
+
+        def objective(x):
+            if len(x.shape) == 2:
+                return np.array([sum(x_i) for x_i in x])
+            return sum(x)
+
+        point = np.ones(5)
+        result = SPSA.estimate_stddev(objective, point, avg=10, max_evals_grouped=max_evals_grouped)
+        self.assertAlmostEqual(result, 0)