This vignette describes how to install CmdStan on your system to use as a backend for EpiSewer. If you cannot install CmdStan locally, you can alternatively use the docker backend (see the docker vignette).
Step 1: Install the EpiSewer package
Install the EpiSewer package from GitHub. This will also install the
cmdstanr package, which is used to interface with CmdStan
from R.
remotes::install_github("adrian-lison/EpiSewer", dependencies = TRUE)Step 2: Check the C++ toolchain
CmdStan requires a working C++ toolchain. You can check whether your system is ready using:
cmdstanr::check_cmdstan_toolchain()If the toolchain is not set up, follow the instructions printed by the above function. The cmdstanr vignette also contains platform-specific guidance for setting up the toolchain on Windows, macOS, and Linux.
Step 3: Install CmdStan
Once the toolchain is ready, install CmdStan using:
cmdstanr::install_cmdstan(cores = 2) # use more cores to speed up installationYou can verify the installation was successful by checking the CmdStan version:
cmdstanr::cmdstan_version()Step 4: Compile the EpiSewer models
The Stan models used by EpiSewer need to be compiled for your device. This is only necessary once — after installing or updating the package — and can be done using:
EpiSewer::sewer_compile()Great, you are now ready to run EpiSewer! See the README for an introduction on how to use the package.
Troubleshooting
If the models are not successfully compiled in step 4, first ensure
that CmdStan is properly set up by re-running step 2. You can also try
updating CmdStan to a newer version by re-running
cmdstanr::install_cmdstan().
If the problem persists, run
EpiSewer::sewer_compile(verbose = TRUE) and post the output
in a new GitHub issue,
along with your cmdstanr::cmdstan_version().