[gmx-users] Testing gromacs: Thermodynamics integration using tabulated potential

[Mark Abraham]

On Tue, May 27, 2014 at 7:06 PM, Fahimeh Baftizadeh <
fahimeh.baftizadeh at googlemail.com> wrote:

> Dear Mark,
>
> Thanks for your reply.
> Before this test, I have performed several thermodynamics integration
> playing with the number of lambda points and also decoupling first coulomb
> and then VDW or decoupling both at the same time. All my tests gave the
> same results as long as I am not using tabulated form of potential. In all
> of them, I obtained 74kj/mol as solvation free energy!
>

That does not sound healthy - what did g_bar report for the errors in each
case?

Now, my question is that, why using tabulated potential makes that much of
> a difference?
>

Not by design, but the possibility of a bug is non-negligible. I would pay
particular attention to the use and implelementation of dispersion
correction.


> Do you mean that decoupling both Q and VDW can be a problem while using
> tabulated potential?


No, decoupling together is a problem in general. But I am not the expert
here! Anybody who can really help will definitely want to see your .mdp
files and know what your solute is.


> In any case, I will lunch a simulation decoupling them
> separately and will update you.
>

That will help establish where the problem lies.

Mark

I really appreciate your help
> Fahimeh
>
>
>
> On Tue, May 27, 2014 at 12:20 PM, Mark Abraham <mark.j.abraham at gmail.com
> >wrote:
>
> > Hi,
> >
> > Decoupling both q and vdw at the same time is not a good idea. You should
> > observe that at nearly-disappeared lambda, the error bars grow
> > dramatically.
> >
> > Mark
> >
> >
> > On Tue, May 27, 2014 at 5:38 PM, Fahimeh
> > <fahimeh.baftizadeh at googlemail.com>wrote:
> >
> > > Hi everyone,
> > >
> > > I am computing solvation free energy of molecule A in solvent B. I used
> > > two different procedure and I thought that the results should be
> > > comparable, however they are very different.
> > >
> > > Procedure 1:
> > > Thermodynamics integration using 20 lambda points. VDWtype was switch
> and
> > > I used PME coulombtype.  I turned off VDW and coulomb interactions both
> > at
> > > the same time. The free energy of solvation was  74kj/mol
> > >
> > > Procedure 2:
> > > Thermodynamics integration using 20 lambda points. I introduce
> tabulated
> > > potential for the interaction between molecule A and molecule B.
> > therefore
> > > the vdwtype and columbtype was set to User. I provided two files,
> called
> > > table.xvg and table_A_B.xvg which are identical for now as a test.
> These
> > > files contain, r, 1/r, 1/r**2, -1/r**6, -6/r**7, 1/r**12, 12/r**13.
>  The
> > > other parameters for VDW interactions are untouched. The result of this
> > > simulation however is very different ~ 50kj/mol ...
> > >
> > > I am expecting to receive not identical but something comparable ...
> > could
> > > you please help me to understand this?
> > >
> > > Many thanks in advance
> > >
> > > Fahimeh
> > > --
> > > Gromacs Users mailing list
> > >
> > > * Please search the archive at
> > > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> > > posting!
> > >
> > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > >
> > > * For (un)subscribe requests visit
> > > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> > > send a mail to gmx-users-request at gromacs.org.
> > >
> > --
> > Gromacs Users mailing list
> >
> > * Please search the archive at
> > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> > posting!
> >
> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >
> > * For (un)subscribe requests visit
> > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> > send a mail to gmx-users-request at gromacs.org.
> >
>
>
>
> --
>
> ---------------------------------------------------------------------------------------------------
> "If you torture the data long enough, they will confess to anything!" John
> Tukey
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>