Fix bug

src/metatrain/utils/data/readers/metatensor.py

@@ -54,12 +54,16 @@ def read_energy(target: DictConfig) -> Tuple[List[TensorMap], TargetInfo]:
             "Either all targets should have strain gradients or none."
         )
 
-    add_position_gradients = all(has_position_gradients)
-    add_strain_gradients = all(has_strain_gradients)
+    add_position_gradients = target["forces"]
+    add_strain_gradients = target["stress"] or target["virial"]
+    print(add_position_gradients, add_strain_gradients)
     target_info = get_energy_target_info(
         target, add_position_gradients, add_strain_gradients
     )
 
+    print(target_info.layout.block())
+    print(tensor_maps[0].block())
+
     # now check all the expected metadata (from target_info.layout) matches
     # the actual metadata in the tensor maps
     _check_tensor_maps_metadata(tensor_maps, target_info.layout)
@@ -92,7 +96,7 @@ def _check_tensor_maps_metadata(tensor_maps: List[TensorMap], layout: TensorMap)
             )
         for key in layout.keys:
             block = tensor_map.block(key)
-            block_from_layout = tensor_map.block(key)
+            block_from_layout = layout.block(key)
             if block.samples.names != block_from_layout.samples.names:
                 raise ValueError(
                     f"Unexpected samples in metatensor targets at index {i}: "

src/metatrain/utils/data/target_info.py

@@ -310,7 +310,7 @@ def _get_cartesian_target_info(target: DictConfig) -> TargetInfo:
         components = [Labels(["xyz"], torch.arange(3).reshape(-1, 1))]
     else:
         components = []
-        for component in range(target["type"][cartesian_key]["rank"]):
+        for component in range(1, target["type"][cartesian_key]["rank"] + 1):
             components.append(
                 Labels(
                     names=[f"xyz_{component}"],
@@ -357,7 +357,7 @@ def _get_spherical_target_info(target: DictConfig) -> TargetInfo:
             Labels(
                 names=["o3_mu"],
                 values=torch.arange(
-                    2 * irrep["o3_lambda"] + 1, dtype=torch.int32
+                    -irrep["o3_lambda"], irrep["o3_lambda"] + 1, dtype=torch.int32
                 ).reshape(-1, 1),
             )
         ]

tests/utils/data/test_readers_metatensor.py

@@ -68,7 +68,7 @@ def spherical_tensor_map():
                         values=torch.arange(0, 1, dtype=torch.int32).reshape(-1, 1),
                     ),
                 ],
-                properties=Labels.single(),
+                properties=Labels.range("properties", 1),
             ),
             TensorBlock(
                 values=torch.rand(1, 5, 1, dtype=torch.float64),
@@ -82,7 +82,7 @@ def spherical_tensor_map():
                         values=torch.arange(-2, 3, dtype=torch.int32).reshape(-1, 1),
                     ),
                 ],
-                properties=Labels.single(),
+                properties=Labels.range("properties", 1),
             ),
         ],
     )
@@ -109,7 +109,7 @@ def cartesian_tensor_map():
                         values=torch.arange(0, 3, dtype=torch.int32).reshape(-1, 1),
                     ),
                 ],
-                properties=Labels.single(),
+                properties=Labels.range("properties", 1),
             ),
         ],
     )
@@ -234,7 +234,7 @@ def test_read_generic_cartesian(monkeypatch, tmpdir, cartesian_tensor_map):
         assert metatensor.torch.equal(tensor_map, cartesian_tensor_map)
 
 
-def test_read_error(monkeypatch, tmpdir):
+def test_read_errors(monkeypatch, tmpdir, energy_tensor_map):
     monkeypatch.chdir(tmpdir)
 
     numpy_array = np.zeros((2, 2))
@@ -249,10 +249,19 @@ def test_read_error(monkeypatch, tmpdir):
         "type": "scalar",
         "per_atom": False,
         "num_properties": 1,
-        "forces": False,
+        "forces": True,
         "stress": False,
         "virial": False,
     }
 
     with pytest.raises(ValueError, match="Failed to read"):
         read_energy(OmegaConf.create(conf))
+
+    torch.save(
+        [energy_tensor_map, energy_tensor_map],
+        "energy.mts",
+    )
+    conf["read_from"] = "energy.mts"
+
+    with pytest.raises(ValueError, match="Unexpected gradients"):
+        read_energy(OmegaConf.create(conf))