[gmx-users] Tabulated potentials make newbies crazy

[Mark Abraham]

ms wrote:
> Hi,
> 
> I am really having a hard time figuring out how to use tabulated
> potentials correctly.
> 
> In general, the information on how to use tables is sparse and scattered
> in several points of the manual (ch.5, ch.7, ch.6.7...) -I hope to write
> a short howto in the wiki once I get this working, but now I really
> don't know what to do.

Yeah, that's an unfortunate fact of life. The manual can't be organized 
so that everybody finds all the information they want in one location 
with the detail they need right now. Using tabulated potentials is an 
advanced feature and if the user needs to search the manual diligently 
to find the information, that's probably good for them in the long run. 
There's a good table of contents, index and some cross references, and 
text-searching is also available...

> This is the story, sorry if it is long, but...:
> 
> My problem is conceptually simple:I want to use a tabulated potential
> for the Calpha in my protein chain,to build a CG model.
> 
> I want the Calpha to use the tabulated potential,both between themselves
> and with other atoms, and I want other atoms to behave between
> themselves just as before (or at least as possible as close as before,
> at least for now).
> 
> I first tried therefore to put C-alpha in the energy groups, called the
> table table_C-alpha_C-alpha.xvg, vdw=user , and there it complains
> because one atom cannot belong to two energy groups (C-alpha and
> Protein) at the same time.
> 
> Just to try, therefore, I removed all occurences of energy groups in a
> .ndx files, except for system and C-alpha (I have no solvent). And there
> it complains that atoms belong to different energy groups, but belong to
> the same charge group.
> 
> I understand therefore that energy groups and charge groups must be
> identically defined.

No. You should read in the manual about groups - sections 3.3 and 8.1.1

Get the energy group stuff working before you worry about providing user 
tables. Divide-and-conquer makes troubleshooting much easier.

> Anyway, just for the sake of learning how to get gmx use a tabulated
> potential, I apply the tabulated potential to all the peptide chain,
> (and I rename the table table_protein_protein.xvg). At this point, it
> stops asking me for a "table.xvg" file. I realize that it probably wants
> to know how to define other non bonded interaction. Since there is only
> the protein, I rename the table_protein_protein.xvg as table.xvg and retry.

Group names are probably case sensitive, so your filename must match 
your group names... the above doesn't look right.

> And now I cry because mdrun doesn't start, telling me:
> 
>    Fatal error:
>    Out of range force value 2.44081e+41 in file 'table.xvg'
> 
> The "out of range force value" error does not appear at all on google. I
> understand it's "kinda high" - but it's a very steep repulsion and so it
> is naturally like that at distances close to 0.

That looks like a bug with mdrun trying to read from a non-existent 
file. It is always useful to provide your exact command lines, rather 
than leave us to infer from text.

> Now, my questions are:
> - What is the accepted range of values in tables?

I don't think this is the problem.

> - How do I define a steep repulsion potential correctly?

It's terse, but manual 6.7 seems to have the necessary information.

> - Is there a way to define a table for specific atoms in a chain,
> leaving the other atoms using the standard VdW potential? I guess it
> requires splitting the chain in two energy groups, but I don't
> understand if it a)makes sense/creates problems b)how to deal with the
> fact that the chain should be a single charge group.

Use energygrp-table like you've been trying. See 7.3.12

Mark