On 10/30/14 7:30 AM, Ioanna Styliari wrote:
Thank you both for your prompt replies. Dr Vitaly Chaban: Is there any way of evaluating how many nodes are needed or is it just by a trial-error approach?
If what Vitaly says is true, this would be a very sneaky bug that has not been noticed before. Of course, if there is a test case that shows this problem, it's worth looking at. But from what I've seen, if the model fails, it will eventually fail regardless of the setup. Drastic differences in forces depending on parallelization would be very concerning.
Justin: Although 1fs works, I was aiming for 2fs steps in the MD but it crashes. I was thinking to run the first 2ns with 1fs and then try to continue it with 2fs onwards to see what happens, but it crashed at 700ps of my MD. I am using GROMOS96 53a6. My equilibration "concern" regards the system: should it be equilibrated as a whole or should I equilibrate one part of it at a time, keeping the others fixed? (which was a suggestion I had but I am unsure how to do it)
If you're using a Gromos96 parameter set, you should start by using the right cutoffs: rlist = rcoulomb = 0.9 and rvdw = 1.4.
Restraining portions of the system to allow other elements to relax is a reasonable approach, especially if the failure is occurring so quickly. If the system fails at 700 ps, that indicates some underlying physical instability.
-Justin -- ================================================== Justin A. Lemkul, Ph.D. Ruth L. Kirschstein NRSA Postdoctoral Fellow Department of Pharmaceutical Sciences School of Pharmacy Health Sciences Facility II, Room 629 University of Maryland, Baltimore 20 Penn St. Baltimore, MD 21201 jalem...@outerbanks.umaryland.edu | (410) 706-7441 http://mackerell.umaryland.edu/~jalemkul ================================================== -- 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-requ...@gromacs.org.
