Question on HSE06 + SOC Support and MPI Abort in CRYSTAL23

masoud37

Dear CRYSTAL Development Team,

I am currently experiencing consistent MPI_ABORT terminations when attempting to run hybrid functional (HSE06) calculations including spin–orbit coupling (SOC) using CRYSTAL23.

The calculations start correctly, but the code terminates early during execution with an MPI_ABORT message on a specific rank. This behavior persists across a wide range of tested MPI/OpenMP parallelization layouts, including very conservative memory-safe configurations, as well as in reduced MPI mode. The issue appears independent of the batch system or cluster configuration.

Before proceeding further, I would like to kindly ask:

Is the combination of HSE06 and SOC fully supported and considered stable in CRYSTAL23 for periodic systems?

To help with diagnosis, I have attached:

the input file used for the calculation,

the corresponding output file showing the error,

and the batch submission script.

These files are provided only for technical reference and reproducibility. I would greatly appreciate any confirmation regarding the current support status of HSE06 + SOC in CRYSTAL23, as well as any known limitations, restrictions, or recommended settings for such calculations.
Thank you very much for your time and support.

Kind regards,
Masoud

GiacomoAmbrogio

Hi masoud37,
Unfortunately, range separated hybrids (as HSE06) are not supported in two-components calculations (ie SOC), as stated at page 166 of the User manual.

Please consider using a global hybrid instead, as PBE0 or any other "custom" percentage of exact exchange in any GGA functional using the keyword HYBRID (refer to page 135 of the manual).

Hope this helps.

masoud37

GiacomoAmbrogio said in Question on HSE06 + SOC Support and MPI Abort in CRYSTAL23:

two-components

Thank you for your reply.

Kind regards,
Masoud

aerba

Let me just add that we do have a development version of the code for HSE+SOC, which we plan to include in the next release.