Merge pull request #612 from stan-dev/fix/complex-output-ordering

Fix/complex output ordering

cmdstanpy/stanfit/gq.py

@@ -523,10 +523,15 @@ def stan_variable(
             col_idxs = self._metadata.stan_vars_cols[var]
             if len(col_idxs) > 0:
                 dims.extend(self._metadata.stan_vars_dims[var])
-            # pylint: disable=redundant-keyword-arg
-            draws = self._draws[draw1:, :, col_idxs].reshape(dims, order='F')
+
+            draws = self._draws[draw1:, :, col_idxs]
+
             if self._metadata.stan_vars_types[var] == BaseType.COMPLEX:
-                draws = draws[..., 0] + 1j * draws[..., 1]
+                draws = draws[..., ::2] + 1j * draws[..., 1::2]
+                dims = dims[:-1]
+
+            draws = draws.reshape(dims, order='F')
+
             return draws
 
     def stan_variables(self, inc_warmup: bool = False) -> Dict[str, np.ndarray]:

cmdstanpy/stanfit/mcmc.py

@@ -747,9 +747,14 @@ def stan_variable(
         col_idxs = self._metadata.stan_vars_cols[var]
         if len(col_idxs) > 0:
             dims.extend(self._metadata.stan_vars_dims[var])
-        draws = self._draws[draw1:, :, col_idxs].reshape(dims, order='F')
+        draws = self._draws[draw1:, :, col_idxs]
+
         if self._metadata.stan_vars_types[var] == BaseType.COMPLEX:
-            draws = draws[..., 0] + 1j * draws[..., 1]
+            draws = draws[..., ::2] + 1j * draws[..., 1::2]
+            dims = dims[:-1]
+
+        draws = draws.reshape(dims, order='F')
+
         return draws
 
     def stan_variables(self) -> Dict[str, np.ndarray]:

cmdstanpy/stanfit/mle.py

@@ -218,12 +218,17 @@ def stan_variable(
             dims = (num_rows,) + self._metadata.stan_vars_dims[var]
             # pylint: disable=redundant-keyword-arg
             if num_rows > 1:
-                result = self._all_iters[:, col_idxs].reshape(dims, order='F')
+                result = self._all_iters[:, col_idxs]
             else:
-                result = self._mle[col_idxs].reshape(dims[1:], order="F")
+                result = self._mle[col_idxs]
+                dims = dims[1:]
 
             if self._metadata.stan_vars_types[var] == BaseType.COMPLEX:
-                result = result[..., 0] + 1j * result[..., 1]
+                result = result[..., ::2] + 1j * result[..., 1::2]
+                dims = dims[:-1]
+
+            result = result.reshape(dims, order='F')
+
             return result
 
         else:  # scalar var

cmdstanpy/stanfit/vb.py

@@ -143,12 +143,13 @@ def stan_variable(self, var: str) -> Union[np.ndarray, float]:
         shape: Tuple[int, ...] = ()
         if len(col_idxs) > 1:
             shape = self._metadata.stan_vars_dims[var]
-            result: np.ndarray = np.asarray(self._variational_mean)[
-                col_idxs
-            ].reshape(shape, order="F")
-
+            result: np.ndarray = np.asarray(self._variational_mean)[col_idxs]
             if self._metadata.stan_vars_types[var] == BaseType.COMPLEX:
-                result = result[..., 0] + 1j * result[..., 1]
+                result = result[..., ::2] + 1j * result[..., 1::2]
+                shape = shape[:-1]
+
+            result = result.reshape(shape, order="F")
+
             return result
         else:
             return float(self._variational_mean[col_idxs[0]])

cmdstanpy/utils/data_munging.py

@@ -44,12 +44,14 @@ def build_xarray_data(
     if dims:
         var_dims += tuple(f"{var_name}_dim_{i}" for i in range(len(dims)))
 
-        draws = drawset[start_row:, :, col_idxs].reshape(
-            *drawset.shape[:2], *dims, order="F"
-        )
+        draws = drawset[start_row:, :, col_idxs]
+
         if var_type == BaseType.COMPLEX:
-            draws = draws[..., 0] + 1j * draws[..., 1]
+            draws = draws[..., ::2] + 1j * draws[..., 1::2]
             var_dims = var_dims[:-1]
+            dims = dims[:-1]
+
+        draws = draws.reshape(*drawset.shape[:2], *dims, order="F")
 
         data[var_name] = (
             var_dims,

docsrc/changes.rst

@@ -7,6 +7,10 @@ What's New
 
 For full changes, see the `Releases page <https://github.com/stan-dev/cmdstanpy/releases>`__ on GitHub.
 
+CmdStanPy 1.0.6
+---------------
+
+- Fixed an issue where complex number containers in Stan program outputs were not being read in properly by CmdStanPy. The output would have the correct shape, but the values would be mixed up.
 
 CmdStanPy 1.0.6
 ---------------

test/data/complex_var.stan

@@ -16,7 +16,9 @@ generated quantities {
                            {{0, 1}, {0, 2}, {0, 3}}};
   array[2, 3] complex zs = {{3, 4i, 5}, {1i, 2i, 3i}};
   complex z = 3 + 4i;
-  
+
   array[2] int imag = {3, 4};
+
+  complex_matrix[2,3] zs_mat = to_matrix(zs);
 }
 

test/test_generate_quantities.py

@@ -440,6 +440,13 @@ def test_complex_output(self):
 
         self.assertEqual(fit.stan_variable('zs').shape, (10, 2, 3))
         self.assertEqual(fit.stan_variable('z')[0], 3 + 4j)
+
+        self.assertTrue(
+            np.allclose(
+                fit.stan_variable('zs')[0], np.array([[3, 4j, 5], [1j, 2j, 3j]])
+            )
+        )
+
         # make sure the name 'imag' isn't magic
         self.assertEqual(fit.stan_variable('imag').shape, (10, 2))
 

test/test_optimize.py

@@ -606,6 +606,13 @@ def test_complex_output(self):
 
         self.assertEqual(fit.stan_variable('zs').shape, (2, 3))
         self.assertEqual(fit.stan_variable('z'), 3 + 4j)
+
+        self.assertTrue(
+            np.allclose(
+                fit.stan_variable('zs'), np.array([[3, 4j, 5], [1j, 2j, 3j]])
+            )
+        )
+
         # make sure the name 'imag' isn't magic
         self.assertEqual(fit.stan_variable('imag').shape, (2,))
 

test/test_sample.py

@@ -1810,11 +1810,27 @@ def test_complex_output(self):
         # make sure the name 'imag' isn't magic
         self.assertEqual(fit.stan_variable('imag').shape, (10, 2))
 
+        self.assertTrue(
+            np.allclose(
+                fit.stan_variable('zs')[0], np.array([[3, 4j, 5], [1j, 2j, 3j]])
+            )
+        )
+        self.assertTrue(
+            np.allclose(
+                fit.stan_variable('zs_mat')[0],
+                np.array([[3, 4j, 5], [1j, 2j, 3j]]),
+            )
+        )
+
         self.assertNotIn("zs_dim_2", fit.draws_xr())
         # getting a raw scalar out of xarray is heavy
         self.assertEqual(
             fit.draws_xr().z.isel(chain=0, draw=1).data[()], 3 + 4j
         )
+        np.testing.assert_allclose(
+            fit.draws_xr().zs.isel(chain=0, draw=1).data,
+            np.array([[3, 4j, 5], [1j, 2j, 3j]]),
+        )
 
     def test_attrs(self):
         stan = os.path.join(DATAFILES_PATH, 'named_output.stan')

test/test_variational.py

@@ -258,6 +258,13 @@ def test_complex_output(self):
 
         self.assertEqual(fit.stan_variable('zs').shape, (2, 3))
         self.assertEqual(fit.stan_variable('z'), 3 + 4j)
+
+        self.assertTrue(
+            np.allclose(
+                fit.stan_variable('zs'), np.array([[3, 4j, 5], [1j, 2j, 3j]])
+            )
+        )
+
         # make sure the name 'imag' isn't magic
         self.assertEqual(fit.stan_variable('imag').shape, (2,))