CmdStan is the command line interface to Stan, a C++ package providing
- full Bayesian inference using the No-U-Turn sampler (NUTS), a variant of Hamiltonian Monte Carlo (HMC),
- approximate Bayesian inference using automatic differentiation variational inference (ADVI),
- penalized maximum likelihood estimation (MLE) using L-BFGS optimization,
- a full first- and higher-order automatic differentiation library based on C++ template overloads, and
- a supporting fully-templated matrix, linear algebra, and probability special function library.
Stan's home page, with links to everything you'll need to use Stan is:
There are separate repositories here on GitHub for interfaces:
- RStan (R interface)
- PyStan (Python interface)
- CmdStan (command-line/shell interface)
CmdStan's source-code repository is hosted here on GitHub.
The core Stan C++ code and CmdStan are licensed under new BSD.
Note that the Stan math library depends on the Intel TBB library which is licensed under the Apache 2.0 license. This dependency implies an additional restriction as compared to the new BSD lincense alone. The Apache 2.0 license is incompatible with GPL-2 licensed code if distributed as a unitary binary. You may refer to the Apache 2.0 evaluation page on the Stan Math wiki.
- Download the latest release tarball (use the "green" link) from: CmdStan releases
- Unpack the tarball.
- From the folder, type
makefor a quick tutorial on how to build models.
Installation using git
See Getting Started with CmdStan for instructions how to clone both CmdStan and Stan submodule.