“double free or corruption (out)” error in TDDFT + EFP job with large number of fragments

Hi all,

I’m running TDDFT + EFP single-point calculations in Q-Chem 6.1.1 to compare excitation energies for a QM solute in two environments:

  • An 8 Å solvation shell (fewer EFP fragments)

  • A complete solvent box (many more EFP fragments)

The same $rem setup runs fine when I use ~410 EFP fragments (8 Å shell), but it crashes when I increase the EFP count to 1302 (bulk-like model).

Error:

This is a multi-thread run using 10 threads
/usr/local/qchem-6.1.1/exe/qcprog.exe .so3sq_choline_geranate-extract.pdb.86_allsolvEFP_tddft.inp.748747.qcin.1 /scratch/senal/qchem748747/
double free or corruption (out)
Abort (core dumped)
Error: in the serial run

What I’ve tried:

  • Increasing memory (MEM_TOTAL)
  • Changing NUM_THREADS
  1. Is there a known practical/technical limit on the number of EFP fragments in EFP jobs?

  2. Does this error suggest a memory problem for very large EFP systems?

I’m unable to attach the full outputs here due to post body-length limits, but I can provide them if needed.

Please contact Q-Chem support.