newton/_src/solvers/mujoco/solver_mujoco.py (3)

1169-1171: Docstring clarity: explicitly state these apply only to limited joints

These globals are applied only when a joint has limits (limited=True). Make that explicit to avoid confusion.

-            joint_solref_limit (tuple[float, float] | None): Global solver reference parameters for all joint limits. If provided, applies these solref values to all joints created. Defaults to None (uses MuJoCo defaults).
-            joint_solimp_limit (tuple[float, float, float, float, float] | None): Global solver impedance parameters for all joint limits. If provided, applies these solimp values to all joints created. Defaults to None (uses MuJoCo defaults).
+            joint_solref_limit (tuple[float, float] | None): Global solver reference parameters for joint limits. If provided, applies these values to every joint that is limited (limited=True) during conversion. Defaults to None (uses MuJoCo defaults).
+            joint_solimp_limit (tuple[float, float, float, float, float] | None): Global solver impedance parameters for joint limits. If provided, applies these values to every joint that is limited (limited=True) during conversion. Defaults to None (uses MuJoCo defaults).

---

1175-1177: Validate input tuple lengths for new parameters

Defensive check to fail fast on malformed inputs (e.g., wrong tuple sizes) improves debuggability.

-        self.contact_stiffness_time_const = contact_stiffness_time_const
-        self.joint_solref_limit = joint_solref_limit
-        self.joint_solimp_limit = joint_solimp_limit
+        self.contact_stiffness_time_const = contact_stiffness_time_const
+        # Validate optional global joint limit params
+        if joint_solref_limit is not None and len(joint_solref_limit) != 2:
+            raise ValueError("joint_solref_limit must be a length-2 tuple (timeconst, dampratio).")
+        if joint_solimp_limit is not None and len(joint_solimp_limit) != 5:
+            raise ValueError("joint_solimp_limit must be a length-5 tuple.")
+        self.joint_solref_limit = joint_solref_limit
+        self.joint_solimp_limit = joint_solimp_limit

---

2371-2444: Multi-env parity: consider expanding jnt_solref/jnt_solimp across worlds

In multi-env mode, several arrays are tiled to all worlds. Joint limit solref/solimp are not expanded, which may lead to inconsistent behavior across worlds if MuJoCoWarp expects per-world copies. If MjWarpModel exposes these fields, add them to the expansion list.

         model_fields_to_expand = [
             # "qpos0",
             # "qpos_spring",
             # "body_pos",
             # "body_quat",
             "body_ipos",
             # "body_iquat",
             "body_mass",
             # "body_subtreemass",
             # "subtree_mass",
             "body_inertia",
             # "body_invweight0",
             # "body_gravcomp",
-            # "jnt_solref",
-            # "jnt_solimp",
+            # If present in MjWarpModel, expand joint limit params too
+            "jnt_solref",
+            "jnt_solimp",
             # "jnt_pos",
             # "jnt_axis",

If these fields are not present in MjWarpModel.dataclass_fields, this change is a no-op due to the guarded expansion loop.

Configuration used: .coderabbit.yml
Review profile: CHILL
Plan: Pro

💡 Knowledge Base configuration:

• MCP integration is disabled by default for public repositories
• Jira integration is disabled by default for public repositories
• Linear integration is disabled by default for public repositories

You can enable these sources in your CodeRabbit configuration.

Learnt from: preist-nvidia
PR: newton-physics/newton#579
File: newton/examples/example_mujoco.py:350-354
Timestamp: 2025-08-19T12:27:30.607Z
Learning: In Newton examples, Example.use_mujoco is a high-level attribute that controls whether to use MuJoCo solver vs other solvers (like XPBD), while SolverMuJoCo.use_mujoco_cpu controls the backend within MuJoCo (CPU vs Warp). These are separate concepts serving different purposes - the PR rename only applies to the solver parameter, not the Example class attributes.

Learnt from: preist-nvidia
PR: newton-physics/newton#579
File: newton/examples/example_mujoco.py:350-354
Timestamp: 2025-08-19T12:27:30.607Z
Learning: In Newton examples, there's a distinction between solver parameters and Example class attributes. The Example class can have its own use_mujoco attribute for controlling example-level behavior (like CUDA graphs, rendering logic), while the solver uses use_mujoco_cpu for backend selection. These serve different purposes and should not be conflated during API renames.