Merge pull request #183 from google-deepmind/B7D4C144F4B8D76FD563D63AE71881C5

Split MJX tests from MuJoCo tests in Menagerie

Author: nimrod-gileadi

test/mjx_model_test.py

@@ -0,0 +1,84 @@
+# Copyright 2022 DeepMind Technologies Limited
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""Tests for all MJX models."""
+
+import pathlib
+from typing import List
+
+from absl.testing import absltest
+from absl.testing import parameterized
+import jax
+import jax.numpy as jp
+import mujoco
+from mujoco import mjx
+
+# Internal import.
+
+
+_ROOT_DIR = pathlib.Path(__file__).parent.parent
+_MODEL_DIRS = [f for f in _ROOT_DIR.iterdir() if f.is_dir()]
+_MJX_MODEL_XMLS: List[pathlib.Path] = []
+
+
+def _get_xmls(pattern: str) -> List[pathlib.Path]:
+  for d in _MODEL_DIRS:
+    # Produce tuples of test name and XML path.
+    for f in d.glob(pattern):
+      test_name = str(f).removeprefix(str(f.parent.parent))
+      yield (test_name, f)
+
+_MJX_MODEL_XMLS = list(_get_xmls('scene*mjx.xml'))
+
+# Total simulation duration, in seconds.
+_MAX_SIM_TIME = 0.1
+
+
+class MjxModelsTest(parameterized.TestCase):
+  """Tests that MJX models load and do not return NaNs."""
+
+  @parameterized.named_parameters(_MJX_MODEL_XMLS)
+  def test_compiles_and_steps(self, xml_path: pathlib.Path) -> None:
+    model = mujoco.MjModel.from_xml_path(str(xml_path))
+    model = mjx.put_model(model)
+    data = mjx.make_data(model)
+    ctrlrange = jp.where(
+        model.actuator_ctrllimited[:, None],
+        model.actuator_ctrlrange,
+        jp.array([-10.0, 10.0]),
+    )
+
+    def step(x, _):
+      data, rng = x
+      rng, key = jax.random.split(rng)
+      ctrl = jax.random.uniform(
+          key,
+          shape=(model.nu,),
+          minval=ctrlrange[:, 0],
+          maxval=ctrlrange[:, 1],
+      )
+      data = mjx.step(model, data.replace(ctrl=ctrl))
+      return (data, rng), ()
+
+    (data, _), _ = jax.lax.scan(
+        step,
+        (data, jax.random.PRNGKey(0)),
+        (),
+        length=min(_MAX_SIM_TIME // model.opt.timestep, 100),
+    )
+
+    self.assertFalse(jp.isnan(data.qpos).any())
+
+
+if __name__ == '__main__':
+  absltest.main()

test/model_test.py

@@ -18,18 +18,14 @@
 
 from absl.testing import absltest
 from absl.testing import parameterized
-import jax
-import jax.numpy as jp
 import mujoco
-from mujoco import mjx
 
 # Internal import.
 
 
 _ROOT_DIR = pathlib.Path(__file__).parent.parent
 _MODEL_DIRS = [f for f in _ROOT_DIR.iterdir() if f.is_dir()]
 _MODEL_XMLS: List[pathlib.Path] = []
-_MJX_MODEL_XMLS: List[pathlib.Path] = []
 
 
 def _get_xmls(pattern: str) -> List[pathlib.Path]:
@@ -40,7 +36,6 @@ def _get_xmls(pattern: str) -> List[pathlib.Path]:
       yield (test_name, f)
 
 _MODEL_XMLS = list(_get_xmls('scene*.xml'))
-_MJX_MODEL_XMLS = list(_get_xmls('scene*mjx.xml'))
 
 # Total simulation duration, in seconds.
 _MAX_SIM_TIME = 0.1
@@ -86,41 +81,5 @@ def test_compiles_and_steps(self, xml_path: pathlib.Path) -> None:
       self.fail(f'MuJoCo warning(s) encountered:\n{warning_info}')
 
 
-class MjxModelsTest(parameterized.TestCase):
-  """Tests that MJX models load and do not return NaNs."""
-
-  @parameterized.named_parameters(_MJX_MODEL_XMLS)
-  def test_compiles_and_steps(self, xml_path: pathlib.Path) -> None:
-    model = mujoco.MjModel.from_xml_path(str(xml_path))
-    model = mjx.put_model(model)
-    data = mjx.make_data(model)
-    ctrlrange = jp.where(
-        model.actuator_ctrllimited[:, None],
-        model.actuator_ctrlrange,
-        jp.array([-10.0, 10.0]),
-    )
-
-    def step(x, _):
-      data, rng = x
-      rng, key = jax.random.split(rng)
-      ctrl = jax.random.uniform(
-          key,
-          shape=(model.nu,),
-          minval=ctrlrange[:, 0],
-          maxval=ctrlrange[:, 1],
-      )
-      data = mjx.step(model, data.replace(ctrl=ctrl))
-      return (data, rng), ()
-
-    (data, _), _ = jax.lax.scan(
-        step,
-        (data, jax.random.PRNGKey(0)),
-        (),
-        length=min(_MAX_SIM_TIME // model.opt.timestep, 100),
-    )
-
-    self.assertFalse(jp.isnan(data.qpos).any())
-
-
 if __name__ == '__main__':
   absltest.main()