Per-domain orces are stored in rvec *f and global forces in rvec *f_global. Likely you can apply whatever operations you need on *f, which get collected at the end of the MD step.

You may be interested in a similar (or the same?) work done before. Check https://github.com/GiovanniBussi/plumed2/blob/v2.0-hrex/user-doc/tutorials/hrex.txt and the paper http://arxiv.org/abs/1307.5144.

How is the forces stored in f_global? Like for instance if I have 4 replicas, f_global would have 4*natoms elements? I have a different scaling factor for different replicas. 

You'll have to look into the code in repl_ex.c to see how that information is passed, though none of that relies on forces, so you'll have to modify it.

This discussion is better posted to the gmx-developers mailing list. Few of the GROMACS developers pay attention to RG, so it's probably not the best forum for detailed discussions of the code.

Dear all, I scaled the energies and forces, but the output isn't being written into the .edr or .xtc files. How do I do this?

I have rvec* in replex.c which stores forces, and enerd->term[F_EPOT] which stores potential energies

Thanks, 

Manish

You won't find either in an .xtc because it only saves coordinates. Forces are written to the .trr file, if nstfout != 0. Energies have to be written to the .edr file.