[gmx-users] Hamiltonian replica exchange in gromacs 4.6.7

Mark Abraham mark.j.abraham at gmail.com
Sat Mar 10 13:54:44 CET 2018


You certainly want the code to take the path where lambda varies, but if
you comment out the check for unequal lambda and avoid the check that T is
unequal, things should work fine.


On Thu, Mar 8, 2018 at 2:59 AM Carlo Martinotti <
carlo.martinotti at postgrad.curtin.edu.au> wrote:

> Hello everybody,
> So i am trying to set up a replica exchange with solute tempering
> simulation for my membrane-drug systems. We modified the gromacs source
> code in a way that allows the modification of the hamiltonian of singular
> pairs of interactions (eg solute-water, solute-solute, ion-solute etc).
> Until now I tested this part on single simulations and it's now working as
> intended. Now though i have to switch on the replica exchange part and
> allow them to exchange.
> Right now i tried to set up a test system with 2 replicas at the same
> temperatures, but with different scaling.
> Again, remember this is an in house modification of the code, not the
> classical REST methodology.
> I launch the classical mdrun -multidir -replex and the system is
> complaining that there is nothing to exchange cause the systems are the
> same. I assume that this is because the temperature of two systems are
> actually the same and he is using equation 3.141 from manual 4.6.7 part
> 3.13.
> So here is my question:
> From my understanding of manual 4.6.7 part 3.13, to allow the exchange as
> per equation 3.142 i MUST use the lambda routine. My idea is then to try to
> trick the program inserting the same values of lambdas in both of the
> replicas so to force gromacs tu use the equation 3.142 for the evaluation
> of the probability of exchange.
> I am pretty sure that the program is going to complain about that, but in
> the case i can eliminate the check for identical lambdas in the source code.
> Do you think this would suffice? Or do you see things that i am missing
> out ?
> Of course i know that without knowing what exact tweaks of the code we did
> you can't answer accurately, but assume that when a single tempered
> simulation is run the functions to compute the forces and the vdw terms are
> changed in the beginning of the simulations and they stay changed for the
> whole of the simulation, so that when gromacs will compute the cross terms
> in ((U1(x2) − U1(x1)) + (U2(x1) − U2(x2)) it should be assumed to use the
> right hamiltonian.
> Thanks in advance for the time and effort!
> --
> 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