[gmx-developers] single point energy for QM/MM.

Berk Hess hess at kth.se
Tue Dec 3 16:50:26 CET 2013


Yes, that would be a good starting point.
Note that this function takes an em_state_t iso a normal t_state state, 
since for EM we need to be able to recall the forces and total energy.

I hope you need to call this function multiple times, otherwise the 
setup overhead is going to be enormous.
For dynamics you should decide to do pair-search and possibly domain 
decomposition now and then.



On 12/03/2013 04:29 PM, David van der Spoel wrote:
> Hi,
> I am building an interface to run gromacs as a slave calculator from 
> another driver, for QMMM (in the qmmm repo). As such I need to 
> implement computing the energy and forces for a given coordinate set 
> (and charges). In mdlib/minimize.c there is a routine
> static void evaluate_energy(FILE *fplog, t_commrec *cr,
>                             gmx_mtop_t *top_global,
>                             em_state_t *ems, gmx_localtop_t *top,
>                             t_inputrec *inputrec,
>                             t_nrnb *nrnb, gmx_wallcycle_t wcycle,
>                             gmx_global_stat_t gstat,
>                             gmx_vsite_t *vsite, gmx_constr_t constr,
>                             t_fcdata *fcd,
>                             t_graph *graph, t_mdatoms *mdatoms,
>                             t_forcerec *fr, rvec mu_tot,
>                             gmx_enerdata_t *enerd, tensor vir, tensor 
> pres,
>                             gmx_large_int_t count, gmx_bool bFirst)
> would that be a good starting point?
> It would have to become non-static.
> I do have a tpr file with all information needed.
> Cheers,

More information about the gromacs.org_gmx-developers mailing list