LAMMPS allocating lammps->atom->f

Mahdi_Tavakol1 · August 24, 2024, 9:40pm

Dear LAMMPS users

I hope you are all doing fine.

I have a question about the LAMMPS source code. Where is in the source code the lammps->atom->f allocated?
I tried to use gdb to watch that variable but I had not luck.

Best regards
Mahdi

akohlmey · August 24, 2024, 10:05pm

github.com

lammps/lammps/blob/c0770715dd63c7fe4d3a277375070df53214d424/src/atom_vec.cpp#L220


      
            nmax = MAX(n, nmax);
          atom->nmax = nmax;
          if (nmax < 0 || nmax > MAXSMALLINT) error->one(FLERR, "Per-processor system is too big");
          
          tag = memory->grow(atom->tag, nmax, "atom:tag");
          type = memory->grow(atom->type, nmax, "atom:type");
          mask = memory->grow(atom->mask, nmax, "atom:mask");
          image = memory->grow(atom->image, nmax, "atom:image");
          x = memory->grow(atom->x, nmax, 3, "atom:x");
          v = memory->grow(atom->v, nmax, 3, "atom:v");
          f = memory->grow(atom->f, nmax * comm->nthreads, 3, "atom:f");
          
          for (int i = 0; i < ngrow; i++) {
            pdata = mgrow.pdata[i];
            datatype = mgrow.datatype[i];
            cols = mgrow.cols[i];
            const int nthreads = threads[i] ? comm->nthreads : 1;
            if (datatype == Atom::DOUBLE) {
              if (cols == 0)
                memory->grow(*((double **) pdata), nmax * nthreads, "atom:dvec");
              else if (cols > 0)

Mahdi_Tavakol1 · August 24, 2024, 10:13pm

Dear Axel

Thanks for the quick reply.

Best regards
Mahdi