In principle if a Python and R environment are available on your system, including facilities to run Jupyter Notebooks, everything should "just work" (haha, famous last words). The instructions below are to ensure that your setup contains all the necessary dependencies to run the various tutorials.
Make your life easy and install the anaconda distribution, this can be dropped wherever you like on your hard drive without needing root permissions. Subsequently install the following:
> conda install -c astropy corner
> pip install pygaia
> pip install daft (optional)
where the installation of daft is optional.
Follow the instructions here. In
particular, use pip install pystan. The necessary compilers can be installed using conda.
There is no equivalent to anaconda for R but installation from source should be fairly painless, or
use the package manager of your Linux distro. Subsequently make sure that your can run R notebooks
by following the instructions on
irkernel.github.io (the "binary" version was used when writing this
INSTALL file). Step 1/2 should be done as "root" for system wide R installations. Step 2/2 should be
done as user. This should result in the magrittr R package being installed (needed for some of the
tutorials). Subsequently install the following packages from within R (as root for system wide R
installations):
> install.packages("mvtnorm")
> install.packages("PolynomF")
> install.packages("fields")
> install.packages("RColorBrewer")
> install.packages("png")
> install.packages("ggplot2")
Note that the period luminosity relation tutorial requires the
installation of R packages related to graph drawing, as well as the rpy2 interface between R and
Python. This can be avoided by using the modified
version of that tutorial.
The recommended installation method for RStan (working with RStudio) can be found here.
To install from source follow the instructions here for the installation of Rstan. Where the last step is to do from within R (run as root for a system wide installation):
> Sys.setenv(MAKEFLAGS = "-j4")
> install.packages("rstan", type = "source")
Note that a Makevars file in the ~/.R folder with the following contents is useful for
optimization of the compiled Stan models and supressing of irrelevant warnings (see RStan install
instructions):
CXXFLAGS=-O3 -mtune=native -march=native -Wno-unused-variable -Wno-unused-function -Wno-builtin-macro-redefined
CXXFLAGS+=-flto -Wno-unused-local-typedefs
CXXFLAGS += -Wno-ignored-attributes -Wno-deprecated-declarations
The anaconda distribution is again recommended. After installation open the Anaconda prompt and install the following:
> conda install -c astropy corner
> pip install pygaia
> pip install daft (optional)
where the installation of daft is optional.
IT IS IMPORTANT to now add the location of the anaconda and python executables to your path: Go to "Settings" (from the windows start menu for example) and then search for "environment" in the search field at the top of the window. Select "Edit environment variables for your account" and then edit the "Path" variable, adding the following two paths:
C:\Users\....\Anaconda3
C:\Users\....\Anaconda3\Scripts
Installation instructions for pystan can be found
here. Note in particular that
installation through conda (conda install pystan) does not seem to work well. So use pip (after
having installed the necessary build tools as detailed in the instructions).
Install the windows version of R and then install the following packages while running R as "administrator":
> install.packages("mvtnorm")
> install.packages("PolynomF")
> install.packages("fields")
> install.packages("RColorBrewer")
> install.packages("png")
> install.packages("ggplot2")
Note that the period luminosity relation tutorial requires the
installation of R packages related to graph drawing, as well as the rpy2 interface between R and
Python. This can be avoided by using the modified
version of that tutorial.
The recommended installation method for RStan (working with RStudio) can be found here.
To install from source follow the instructions here for the installation of Rstan.
Subsequently make sure that your can run R notebooks by following the instructions on irkernel.github.io. Step 1/2 should be done while running as R as "administrator".
Use anaconda for Mac OsX and follow the instruction above for Linux.
pending...
Use the .pkg file from the CRAN page. Install the packages listed
above in the Linux section.
pending...