Hydrogen mapping

[pbuslaev]

In the develop python3-adapted branch, I am a bit puzzled with the logics behind this line
To give an example, if I try to modify *-C into *-CC, hydrogens of methyl group are never mapped to hydrogens of the first carbon of ethyl group. Thus, efficiently in the resulting hydrid topology carbon will be bonded to 5 hydroges (three real ones and two dummy ones). This actually blocks using the hmr with GROMACS. But it also looks a bit strange to me in general.

@vgapsys Could you maybe share the idea behind this line? I tested internally and with commenting out this condition I am getting reasonable mappings which I can use with hmr. The following code maps hydrogens to minimize rmsd, which makes sense to me and allows runs with hmr.

[pbuslaev]

@vgapsys I tested this change on our internal compounds and I don't see any reason why not to do this modification.

[vgapsys]

My concern is that such mappings might introduce internal strain due to the bonded interactions via dummy atoms. The cleanest way to avoid this is by avoiding the mapping itself.

Regarding hmr, if we are to use two topologies with hmr and map them with the current mapping scheme, all should work properly. I wonder why internal gmx hmr implementation cannot do this