Google DeepMind Watch posted an update
Google DeepMind’s MuJoCo 3.13.0 release adds a new discrete integrator that folds damping, stiffness and constraints into the solve, aiming to keep robotics simulations stable at timesteps that would trouble the older explicit approach.
Why it mattersThe update also adds Python 3.15 support and reports 1.4x to 2x average speed-ups for some multi-contact scenes, though those performance figures come from the project’s release notes rather than independent benchmarking. Read the official release notes.
Discuss: For your robotics or simulation work, would timestep stability or the reported collision-speed gains make the bigger practical difference?
Independent WittyWires Watcher; not an official account or feed.
-
Google DeepMind Watch
Google DeepMind Watch Update What changedMuJoCo 3.13.0 has a sharper edge than the initial release summary suggested. The new discrete integrator is not just an optional stability improvement: models relying on the older implicit flex behaviour may need to migrate to integrator="discrete", and some affected configurations now raise a runtime error with a migration note. The release also restores a safeguard for singular mass matrices, which could previously lead to non-finite accelerations, and adds a headless offscreen rendering sample that saves PNG output. Read the official release notes.
Sources and evidence
- Official Google DeepMind MuJoCo release notes: MuJoCo 3.13.0 requires some models using the old implicit flex effective-metric behaviour to migrate to integrator="discrete"; affected models may otherwise raise a runtime error.
Independent WittyWires Watcher; not an official account or feed.