Skip to content

Installation

This section describes how to build and install JURASSIC on a Linux system or high-performance computing (HPC) environment. JURASSIC is designed for batch-style execution and is typically compiled from source.

System requirements

JURASSIC is primarily developed and tested on 64-bit Linux systems, including high-performance computing environments. It is built from source using a Makefile-based workflow.

Required software

The following software components are required to build JURASSIC:

  • C compiler: required for auxiliary components and libraries, for example gcc.
  • GNU Make: required to build the bundled libraries and the JURASSIC executables.
  • GNU Scientific Library (GSL): required for numerical kernels and linear algebra support.
  • netCDF-C library: required for netCDF input/output support used throughout the toolchain.
  • HDF5 library: required for netCDF-4/HDF5 input/output workflows.
  • MPI library (optional, retrieval only): required only for MPI-enabled retrieval builds. Supported MPI implementations include OpenMPI and MPICH-derived implementations such as ParaStation MPI. MPI is used exclusively by the retrieval code to distribute independent retrieval tasks across processes.
  • Gnuplot (optional): used by example projects to generate diagnostic plots.

Obtaining the source code

The JURASSIC source code is hosted in the GitHub repository. To obtain the most recent development version from the default branch, clone the repository using:

git clone https://github.com/slcs-jsc/jurassic.git
cd jurassic

Alternatively, download a release archive from the GitHub releases page.

Recommended version

We generally recommend using the current development version from the default branch. Bug fixes are usually applied there and are not routinely backported to older releases.

Configuring the build

JURASSIC uses a Makefile-based build system in src/Makefile. Prior to compilation, you may need to edit that file or set make variables to match your local compiler and MPI setup.

Typical configuration options include:

  • Selection of the C compiler and any optional external toolchains
  • Compiler optimization and debugging flags
  • Enabling or disabling MPI support (retrieval only)
  • Enabling OpenMP parallelization

On HPC systems, it is recommended to load the appropriate compiler and MPI modules before configuring the build.

Building JURASSIC

The dependency build command is run from the libs/ directory. The JURASSIC build, install, and test commands are run from the src/ directory.

Build bundled dependencies

From libs/, compile the bundled third-party libraries:

./build.sh

This populates libs/build/, which is the default include/library location used by src/Makefile.

Using system libraries

If you prefer to use system-provided GSL, HDF5, and netCDF libraries, this bundled-library step can be skipped, but the include and library paths in src/Makefile must point to the corresponding system locations.

Default build (no MPI)

To build JURASSIC without MPI support:

make -j

This builds all executables in serial/OpenMP mode. MPI is not required for this configuration.

Building with MPI-enabled retrieval

To enable MPI support for the retrieval executable, build with:

make -j MPI=1

This will:

  • compile the full tool suite with the MPI compiler wrapper,
  • automatically select mpicc (unless CC is set explicitly),
  • define the MPI preprocessor macro used by the retrieval source code.

MPI-specific runtime behavior is implemented only in the retrieval executable. The other binaries are still rebuilt under the selected MPI toolchain when MPI=1 is used.

Clean rebuild

To perform a clean rebuild:

make clean
make -j

Install binaries

The src/Makefile provides a simple copy-install target. By default, executables are installed to ../bin relative to the src directory:

make install

To install into a custom binary directory:

make install DESTDIR=/path/to/bin

To remove binaries from the same destination:

make uninstall DESTDIR=/path/to/bin

Verifying the installation

After compilation, verify the installation by running the test suite or the example projects described in the Quickstart.

make check

A successful test run or example simulation indicates that JURASSIC has been built correctly and that all required dependencies are working as expected.

Installation on HPC systems

On shared HPC systems, JURASSIC is typically installed in a user workspace rather than system-wide. Recommended practices include:

  • Building JURASSIC with the same compiler and MPI library used for production retrieval runs
  • Enabling MPI only when running MPI-enabled retrievals
  • Using environment modules to manage compiler and MPI versions
  • Testing retrieval scalability with a small number of MPI ranks before large-scale production runs

Further details on MPI execution and performance considerations are provided in Running JURASSIC and HPC workflows.

Troubleshooting

Common issues during installation include:

  • Missing or incompatible compiler versions
  • Using an MPI-enabled build without mpicc or an MPI runtime
  • Mismatches between compile-time and runtime MPI environments
  • Incorrect OpenMP settings
  • Runtime linker errors caused by missing shared-library paths

If you encounter problems, consult the build output carefully and verify that your compiler, library paths, and MPI environment are correctly configured. Additional help may be available through the project maintainers.

Bundled shared libraries

When using the bundled dynamic libraries, executables may need the library directory in LD_LIBRARY_PATH at runtime:

bash export LD_LIBRARY_PATH=/path/to/jurassic/libs/build/lib:$LD_LIBRARY_PATH

The example scripts in projects/ set this path automatically for the default bundled-library layout.

Next steps

Once JURASSIC is installed, proceed to the Quickstart to run your first simulation, or consult the User Manual for detailed information on configuration and usage.