(possible bug) Van der Waals functional fails for graphene with PSML pseudopotential

Using van der Waals functional LMKLL for graphene leads to failure of siesta calculation when a psml pseudopotential from pseudo-dojo is used. No problem with a psf pseudopotential. No problem for some other compilations of the same version of siesta.

Error message in siesta output: mesh1D locate: ERROR: bisection not converged It is shown before writing the 'Single-point calculation' header.

Code version: SIESTA versions: hybrid branch, dev, 5.4, 5.2

System information: The system is Rocky Linux 8. The following modules are used for the example of the calculation enclosed: module load openmpi/4.1.5-gcc-8.5.0-pz4uman module load openblas/0.3.23-gcc-8.5.0-4ubyqcj module load netlib-scalapack/2.2.0-gcc-8.5.0-ik7e4j5 module load fftw/3.3.10-gcc-8.5.0-ysk7al7 module load libxc/6.1.0-gcc-8.5.0-uysrsjr module load netcdf-c/4.9.2-gcc-8.5.0-guav4os module load netcdf-fortran/4.6.0-gcc-8.5.0-q4ctbun

The compilation includes ELSI support as well.

Steps to reproduce: Any calculation of siesta for a carbon system with psml pseudopotential. Can fail or not depending on compilation.

Relevant logs and/or screenshots C.1.psml siesta.fdf siesta.out

Possible fixes Basic compilations I did previously without fftw, ELSI, FLOOK worked fine but I'm not sure which one of those libraries may cause problems.