State-of-the-art stellarator optimization code
This page details how to compile the STELLOPT family of codes at Max-Plank Computational Data Facility (formerly Rechenzentrum Garching) (MPCDF). In order to do so you will need an account on their system. These build instructions are for the Intel based compilers
module load git
module load intel/2023.1.0.x
module load mkl/2023.1
module load impi/2021.9
module load hdf5-mpi/1.14.1
module load netcdf-mpi/4.8.1
module load fftw-mpi
module load git
module load intel/21.4.0
module load mkl/2021.2
module load impi/2021.4
module load netcdf-mpi/4.8.1
module load hdf5-mpi
module load fftw-mpi
module load anaconda/3/2021.11
module load git/2.43
module load intel/2024.0
module load mkl/2024.0
module load impi/2021.11
module load netcdf-mpi/4.9.2
module load hdf5-mpi/1.14.1
module load fftw-mpi/3.3.10
module load anaconda/3/2021.11
module load intel
module load mkl
module load netcdf
module load hdf5
module load openmpi
module load petsc
module load slepc
module load nag_flib/intel/mk25
The Raven and Viper clusters are automatically detected by the make.inc script. The IPP-HGW theory cluster requires the user to set the environment variable MACHINE equal to ‘theoryhgw’.
There is an issue with Anaconda where it links against it’s own libraries which are not the same as the system libraries. This doesn’t affect the Fortran parts of the code. The C-Types interface we use in Python to access routines in libstell.so is affected. The knock-on effect is that the benchmarks will fail with error messages about libssh.so. The fix is for the user to add export LD_PRELOAD=/usr/lib64/libcrypto.so.1.1:/usr/lib64/libssl.so.1.1 to your .bashrc (or equivalent). This is an Anaconda/MPCDF issue, not realated to STELLOPT directly.