This chapter discusses GYRE installation in detail. If you just want to get up and running, have a look at the Quick Start chapter.


To compile and run GYRE, you’ll need the following software components:

  • A modern (2003+) Fortran compiler

  • The BLAS linear algebra library

  • The LAPACK linear algebra library

  • The LAPACK95 Fortran 95 interface to LAPACK

  • The HDF5 data management library

  • The crlibm correctly rounded math library

  • The crmath Fortran 2003 interface to crlibm

  • An OpenMP-aware version of the ODEPACK differential equation library (optional)

On Linux and MacOS platforms, these components are bundled together in the MESA Software Development Kit (SDK), which can be downloaded from the MESA SDK homepage. Using this SDK is strongly recommended.

Building GYRE


Download the GYRE source code, and unpack it from the command line using the tar utility:

tar xf gyre-master.tar.gz

Set the GYRE_DIR environment variable with the path to the newly created source directory; this can be achieved e.g. using the dirname built-in command:

export GYRE_DIR=$(dirname gyre-master)


Compile GYRE using the make utility:

make -j -C $GYRE_DIR

(the -j flags tells make to use multiple cores, speeding up the build).


To check that GYRE has compiled correctly and gives reasonable results, you can run the calculation test suite via the command

make -C $GYRE_DIR test

The initial output from the tests should look something like this:

TEST numerics (OpenMP)...
TEST numerics (band matrix)...
TEST numerics (*_DELTA frequency units)...
TEST numerics (rotation, Doppler shift)...
TEST numerics (rotation, traditional approximation)...

If things go awry, consult the Troubleshooting chapter.

Custom Builds

Custom builds of GYRE can be created by setting certain environment variables, and/or variables in the file $GYRE_DIR/src/build/Makefile, to the value yes. The following variables are currently supported:


Enable debugging mode (default no)


Enable OpenMP parallelization (default yes)


Enable MPI parallelization (default no)


Use double precision floating point arithmetic (default yes)


Use correctly rounded math functions (default yes)


Use Fortran IEEE floating point features (default no)


Enable floating point exception checks (default yes)


Include HDF5 support (default yes)


Enable experimental features (default no)

If a variable is not set, then its default value is assumed.

Git Access

Sometimes, you’ll want to try out new features in GYRE that haven’t yet made it into a formal release. In such cases, you can check out GYRE directly from the rhdtownsend/gyre git repository on GitHub:

git clone

However, a word of caution: GYRE is under constant development, and features in the main (master) branch can change without warning.