[gmx-developers] question about init_em

Carlo Camilloni carlo.camilloni at gmail.com
Wed Sep 28 15:14:56 CEST 2016

Dear GMX developers,

I was looking into the minimisation code (gromacs 5.1.4) 
and I found a difference between the setup of the MD vs the setup of EM that
I don’t understand and I was wondering if this could be a bug

in do_md the MDATOMS are setup either inside the domain decomposition or inside do_md and after that in both
cases there the call to 

update_mdatoms(mdatoms, state->lambda[efptMASS]);

in the case of init_em

    if (DOMAINDECOMP(cr))

        /* Distribute the charge groups over the nodes from the master node */
        dd_partition_system(fplog, ir->init_step, cr, TRUE, 1,
                            state_global, top_global, ir,
                            &ems->s, &ems->f, mdatoms, *top,
                            fr, vsite, NULL, constr,
                            nrnb, NULL, FALSE);
        atoms2md(top_global, ir, 0, NULL, top_global->natoms, mdatoms);
        update_mdatoms(mdatoms, state_global->lambda[efptFEP]);

        if (vsite)
            set_vsite_top(vsite, *top, mdatoms, cr);

shouldnt’ the call to  update_mdatoms(mdatoms, state_global->lambda[efptFEP]);

be place after this if block instead and removed from the “else” block? (the problem is that as it is now many masses are set to zero when using
domain decomposition).

Furthermore shouldn’t it be:

update_mdatoms(mdatoms, state_global->lambda[efptMASS]);
instead of
update_mdatoms(mdatoms, state_global->lambda[efptFEP]);


More information about the gromacs.org_gmx-developers mailing list