[gmx-users] polarizable ff and free energy calculation

[Mark Abraham]

On 18/04/2012 10:32 PM, Tom Kirchner wrote:
> Of my first mail, only the attachment was printed in the mailing list. 
> I am sorry for the inconvenience. The attachment can be sent by request.

They came through.

>
> ----------
>
> Hi all,
>
> Recently, I posted a problem concerning the usage of polarizable ff in 
> conjunction with free energy calculation. Sadly I got no answer, but 
> after some research, I hope I have isolated the problem.

Thanks for the clear report.

>
> Using a polarizable ff without free energy makes no problems. As soon 
> as free energy is turned on, Gromacs gives the error:
>
> /Program mdrun_d, VERSION 4.5.5
> Source code file: [...]/src/gromacs-4.5.5/src/mdlib/shellfc.c, line: 365
>
> Fatal error:
> polarize can not be used with qA != qB/
>
> This error occurs, even if the atom and the shell particle have the 
> same charge or are unaltered by the free energy option, e.g. 
> lambda=0.0 and multiplication with (1-lambda). Because of this, I 
> guess there is a problem with different charge variables being single 
> and double precision due to multiplication with lambda.

... well, likely some kind of floating-point-algebra failure, anyway.

>
> The error occurs in this segment of shellfc.c:
>
>         case F_POLARIZATION:
>           if (qS != atom[aS].qB)
>         gmx_fatal(FARGS,"polarize can not be used with qA != qB");
>
> If anybody knows, how to correct this, I would be very grateful.

I've uploaded a draft fix to https://gerrit.gromacs.org/#/c/746/

> Moreover I'm wondering about the physical meaning of the if-statement.

No idea.

Mark
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20120419/9e4b1c8e/attachment.html>