Sub-Space-Matrix is not hermitian

Most common reasons:

unreasonable input geometry (e.g. after a relaxation step)

Solution: fix geometry

Error encountered when initializing a parallelized ISIF=3 relaxation

Solution: Replacing e.g. NPAR=4 with NCORE=16 has in some cases solved the problem

problems during reading of the CHGCAR file (e.g. complaint that the dimensions have changed after a change in k-point grid)

Solution: remove CHGCAR file and start density from scratch

problems during initialization altogether

Solution: try completing the run and restarting with from the CHGCAR and WAVECAR; if that does not work: try the same with IALGO=48 (may give related warnings before restarting)

From the VASP forum: If the error persist although you switched to the Davidson algorithm:

on some architectures (especially SGI) some LAPACK routines are not working

properly. However, it is possible to avoid the usage of the ZHEGV subroutine by commenting the line #define USE_ZHEEVX in davidson.F, subrot.F, and

wavpre_noio.F and recompiling VASP.

--->If VASP is not well-compiled for the architecture being used, reducing NCORE might solve the problem altogether.

if nothing else works: try changing PREC Normal to Accurate