[gmx-users] Perturbation Thermodynamic Integration

Mark Abraham mark.j.abraham at gmail.com
Tue May 16 16:11:22 CEST 2017 

On Tue, May 16, 2017 at 3:54 PM Dan Gil <dan.gil9973 at gmail.com> wrote:

> Hi,
>
> The latest version of Gromacs I have is 5.1. Using this version, grompp
> produces no warnings or errors.
>

I know you won't get the new warning from the old code ;-) You'd need the
latest grompp, as I've mentioned twice. If so, then that's a problem that
suggests a smaller time step. (But probably you have bigger problems.)


> I am using OPLS-AA, and I didn't have any issues with it thus far. I
> inherited this research project though 3 years ago. Is there something you
> think I should look at with scrutiny?
>

"No known issues" is not tantamount to "reliable model physics." If schemes
like that have been published and found useful, that's great. Otherwise,
you ought to have the burden of showing that it is a good model physics, or
throwing away the old work and doing something conventional (and if so, use
the verlet scheme and get much better performance).

Mark


> On Tue, May 16, 2017 at 9:42 AM, Mark Abraham <mark.j.abraham at gmail.com>
> wrote:
>
> > Hi,
> >
> > Yes that looks like it is the envelope of the new warning. What does
> grompp
> > from 2016.3 report?
> >
> > (Aside, that looks like an ad hoc non-bonded scheme. As a reviewer, I'd
> > reject that method immediately unless there was evidence that it worked
> > well for a range of observables, and was consistent with other practice
> for
> > that force field. If there is such evidence, for my information, what
> model
> > / force field does it like that?)
> >
> > Mark
> >
> > On Tue, May 16, 2017 at 3:30 PM Dan Gil <dan.gil9973 at gmail.com> wrote:
> >
> > > Sorry, here is the mdp file:
> > >
> > > ;Integration Method and Parameters
> > > integrator               = sd
> > > nsteps                   = 100000
> > > dt = 0.002
> > > nstenergy                = 1000
> > > nstlog                   = 5000
> > >
> > > ;Output Control
> > > nstxout = 0
> > > nstvout = 0
> > >
> > > ;Cutoff Schemes
> > > cutoff-scheme            = group
> > > rlist                    = 1.0
> > > vdw-type                 = cut-off
> > > rvdw                     = 2.0
> > >
> > > ;Coulomb interactions
> > > coulombtype              = pme
> > > rcoulomb                 = 1.0
> > > fourierspacing           = 0.4
> > >
> > > ;Constraints
> > > constraints              = all-bonds
> > >
> > > ;Temperature coupling
> > > tcoupl                   = v-rescale
> > > tc-grps                  = system
> > > tau-t                    = 0.1
> > > ref-t                    = 300
> > >
> > > ;Pressure coupling
> > > pcoupl = parrinello-rahman
> > > ref-p = 1.01325
> > > compressibility = 4.5e-5
> > > tau-p = 5
> > >
> > > ;Free energy calculation
> > > free-energy              = yes
> > > init-lambda-state        = 8
> > > delta-lambda             = 0
> > > fep-lambdas              =
> > > calc-lambda-neighbors    = 1
> > > vdw_lambdas              = 0 0   0   0   0   0   0   0   0   0   0
> > > coul_lambdas             = 0 0   0   0   0   0   0   0   0   0   0
> > > bonded_lambdas           = 0 0   0   0   0   0   0   0   0   0   0
> > > restraint_lambdas        = 0 0   0   0   0   0   0   0   0   0   0
> > > mass_lambdas             = 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
> > > couple-moltype           = HEPT
> > > couple-lambda0           = vdwq
> > > couple-lambda1           = none
> > > couple-intramol          = no
> > > nstdhdl                  = 10
> > >
> > >
> > > On Tue, May 16, 2017 at 1:02 AM, Mark Abraham <
> mark.j.abraham at gmail.com>
> > > wrote:
> > >
> > > > Hi,
> > > >
> > > > What use are you making of constraints? Justin suggested sharing a
> full
> > > mdp
> > > > file, which I think may help. We discovered last year that you can
> get
> > > > equipartition failure for (IIRC) all-bonds constraints for moieties
> > like
> > > > -CH2Cl, and latest grompp now detects this.
> > > >
> > > > Mark
> > > >
> > > > On Tue, 16 May 2017 01:16 Dan Gil <dan.gil9973 at gmail.com> wrote:
> > > >
> > > > > Hello,
> > > > >
> > > > > The last thread was getting too big, and the conversation evolved
> to
> > a
> > > > > topic different from my original question, so I decided to start a
> > new
> > > > > thread.
> > > > >
> > > > > We were discussing thermodynamic integration, and why the
> > mass_lambdas
> > > > > would have any contribution to the derivative of the Hamiltonian.
> > > > >
> > > > > I found a source (link below) which derives the Gibbs free energy
> > > change
> > > > as
> > > > > a function of lambda. I learned that the mass contribution is often
> > > > assumed
> > > > > to be small and negligible, given that the mass difference between
> > the
> > > > two
> > > > > lambda states are small.
> > > > > http://www.tandfonline.com/doi/abs/10.1080/00268970600893060
> > > > >
> > > > > I think that the mass of the two lambda states that equation (14)
> is
> > > > > referring to is the total mass (mass of solvent plus solute). My
> > system
> > > > is
> > > > > 1 solute (~40 atoms) infinitely diluted in solvent (23500). I
> wonder
> > > if I
> > > > > am getting nonzero mass contributions (in my dhdl.xvg output)
> because
> > > of
> > > > > finite-size effects? Would completely neglecting the mass
> > contributions
> > > > be
> > > > > acceptable? Does doing this technically change the system to one
> that
> > > is
> > > > 1
> > > > > solute and an infinite number of solvent molecules where the mass
> > > > > contributions limit is zero?
> > > > >
> > > > > Best Regards,
> > > > >
> > > > > Dan
> > > > > --
> > > > > 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.
> > > >
> > > --
> > > 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.
> >
> --
> 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.
>

More information about the gromacs.org_gmx-users mailing list