[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-9805":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":21,"rankGlobal":10,"rankLanguage":10,"license":22,"archived":23,"fork":23,"defaultBranch":24,"hasWiki":23,"hasPages":23,"topics":25,"createdAt":10,"pushedAt":10,"updatedAt":33,"readmeContent":34,"aiSummary":35,"trendingCount":16,"starSnapshotCount":16,"syncStatus":36,"lastSyncTime":37,"discoverSource":38},9805,"dm_control","google-deepmind\u002Fdm_control","google-deepmind","Google DeepMind's software stack for physics-based simulation and Reinforcement Learning environments, using MuJoCo.","",null,"Python",4615,755,124,111,0,1,8,43,6,30.64,"Apache License 2.0",false,"main",[26,27,28,29,30,31,32],"artificial-intelligence","deep-learning","machine-learning","mujoco","neural-networks","physics-simulation","reinforcement-learning","2026-06-12 02:02:12","# `dm_control`: Google DeepMind Infrastructure for Physics-Based Simulation.\n\nGoogle DeepMind's software stack for physics-based simulation and Reinforcement\nLearning environments, using MuJoCo physics.\n\nAn **introductory tutorial** for this package is available as a Colaboratory\nnotebook:\n[![Open In Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002Fgoogle-deepmind\u002Fdm_control\u002Fblob\u002Fmain\u002Ftutorial.ipynb)\n\n## Overview\n\nThis package consists of the following \"core\" components:\n\n-   [`dm_control.mujoco`]: Libraries that provide Python bindings to the MuJoCo\n    physics engine.\n\n-   [`dm_control.suite`]: A set of Python Reinforcement Learning environments\n    powered by the MuJoCo physics engine.\n\n-   [`dm_control.viewer`]: An interactive environment viewer.\n\nAdditionally, the following components are available for the creation of more\ncomplex control tasks:\n\n-   [`dm_control.mjcf`]: A library for composing and modifying MuJoCo MJCF\n    models in Python.\n\n-   `dm_control.composer`: A library for defining rich RL environments from\n    reusable, self-contained components.\n\n-   [`dm_control.locomotion`]: Additional libraries for custom tasks.\n\n-   [`dm_control.locomotion.soccer`]: Multi-agent soccer tasks.\n\nIf you use this package, please cite our accompanying [publication]:\n\n```\n@article{tunyasuvunakool2020,\n         title = {dm_control: Software and tasks for continuous control},\n         journal = {Software Impacts},\n         volume = {6},\n         pages = {100022},\n         year = {2020},\n         issn = {2665-9638},\n         doi = {https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.simpa.2020.100022},\n         url = {https:\u002F\u002Fwww.sciencedirect.com\u002Fscience\u002Farticle\u002Fpii\u002FS2665963820300099},\n         author = {Saran Tunyasuvunakool and Alistair Muldal and Yotam Doron and\n                   Siqi Liu and Steven Bohez and Josh Merel and Tom Erez and\n                   Timothy Lillicrap and Nicolas Heess and Yuval Tassa},\n}\n```\n\n## Installation\n\nInstall `dm_control` from PyPI by running\n\n```sh\npip install dm_control\n```\n\n> **Note**: **`dm_control` cannot be installed in \"editable\" mode** (i.e. `pip\n> install -e`).\n>\n> While `dm_control` has been largely updated to use the pybind11-based bindings\n> provided via the `mujoco` package, at this time it still relies on some legacy\n> components that are automatically generated from MuJoCo header files in a way\n> that is incompatible with editable mode. Attempting to install `dm_control` in\n> editable mode will result in import errors like:\n>\n> ```\n> ImportError: cannot import name 'constants' from partially initialized module 'dm_control.mujoco.wrapper.mjbindings' ...\n> ```\n>\n> The solution is to `pip uninstall dm_control` and then reinstall it without\n> the `-e` flag.\n\n## Versioning\n\nStarting from version 1.0.0, we adopt semantic versioning.\n\nPrior to version 1.0.0, the `dm_control` Python package was versioned `0.0.N`,\nwhere `N` was an internal revision number that increased by an arbitrary amount\nat every single Git commit.\n\nIf you want to install an unreleased version of `dm_control` directly from our\nrepository, you can do so by running `pip install\ngit+https:\u002F\u002Fgithub.com\u002Fgoogle-deepmind\u002Fdm_control.git`.\n\n## Rendering\n\nThe MuJoCo Python bindings support three different OpenGL rendering backends:\nEGL (headless, hardware-accelerated), GLFW (windowed, hardware-accelerated), and\nOSMesa (purely software-based). At least one of these three backends must be\navailable in order render through `dm_control`.\n\n*   Hardware rendering with a windowing system is supported via GLFW and GLEW.\n    On Linux these can be installed using your distribution's package manager.\n    For example, on Debian and Ubuntu, this can be done by running `sudo apt-get\n    install libglfw3 libglew2.0`. Please note that:\n\n    -   [`dm_control.viewer`] can only be used with GLFW.\n    -   GLFW will not work on headless machines.\n\n*   \"Headless\" hardware rendering (i.e. without a windowing system such as X11)\n    requires [EXT_platform_device] support in the EGL driver. Recent Nvidia\n    drivers support this. You will also need GLEW. On Debian and Ubuntu, this\n    can be installed via `sudo apt-get install libglew2.0`.\n\n*   Software rendering requires GLX and OSMesa. On Debian and Ubuntu these can\n    be installed using `sudo apt-get install libgl1-mesa-glx libosmesa6`.\n\nBy default, `dm_control` will attempt to use GLFW first, then EGL, then OSMesa.\nYou can also specify a particular backend to use by setting the `MUJOCO_GL=`\nenvironment variable to `\"glfw\"`, `\"egl\"`, or `\"osmesa\"`, respectively. When\nrendering with EGL, you can also specify which GPU to use for rendering by\nsetting the environment variable `MUJOCO_EGL_DEVICE_ID=` to the target GPU ID.\n\n## Additional instructions for Homebrew users on macOS\n\n1.  The above instructions using `pip` should work, provided that you use a\n    Python interpreter that is installed by Homebrew (rather than the\n    system-default one).\n\n2.  Before running, the `DYLD_LIBRARY_PATH` environment variable needs to be\n    updated with the path to the GLFW library. This can be done by running\n    `export DYLD_LIBRARY_PATH=$(brew --prefix)\u002Flib:$DYLD_LIBRARY_PATH`.\n\n[EXT_platform_device]: https:\u002F\u002Fwww.khronos.org\u002Fregistry\u002FEGL\u002Fextensions\u002FEXT\u002FEGL_EXT_platform_device.txt\n[Releases page on the MuJoCo GitHub repository]: https:\u002F\u002Fgithub.com\u002Fgoogle-deepmind\u002Fmujoco\u002Freleases\n[MuJoCo website]: https:\u002F\u002Fmujoco.org\u002F\n[publication]: https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.simpa.2020.100022\n[`ctypes`]: https:\u002F\u002Fdocs.python.org\u002F3\u002Flibrary\u002Fctypes.html\n[`dm_control.mjcf`]: dm_control\u002Fmjcf\u002FREADME.md\n[`dm_control.mujoco`]: dm_control\u002Fmujoco\u002FREADME.md\n[`dm_control.suite`]: dm_control\u002Fsuite\u002FREADME.md\n[`dm_control.viewer`]: dm_control\u002Fviewer\u002FREADME.md\n[`dm_control.locomotion`]: dm_control\u002Flocomotion\u002FREADME.md\n[`dm_control.locomotion.soccer`]: dm_control\u002Flocomotion\u002Fsoccer\u002FREADME.md\n","dm_control 是 Google DeepMind 开发的一套基于 MuJoCo 物理引擎的仿真和强化学习环境软件栈。它提供了 Python 绑定来访问 MuJoCo 物理引擎，并包含一系列用于创建复杂控制任务的组件，如 dm_control.suite 提供了多种强化学习环境，dm_control.viewer 则支持交互式环境查看。此外，还有专门针对多智能体足球任务等高级应用的扩展库。该项目适合需要进行物理模拟或开发强化学习算法的研究人员及开发者使用，尤其是在机器人学、生物力学等领域中探索连续控制问题时。",2,"2026-06-11 03:24:49","top_topic"]