[gmx-users] (MD with constraints giving LINCS error using N cores not on 1 core)
jjhuang0520 at gmail.com
Fri Nov 6 03:20:55 CET 2015
Dear Gromacs users,
Sorry for my incomplete information about my previous problem.
1. It takes around 290 steps for this error to occur.
2. When using particle decomposition in GROMACS 4.6 for another similar system, this error didn’t occur.
3. The system was energy minimized with constraints. And the energy minimization was completed at around 700-900 steps.
4. I also reduce the emtol to 800, 600, and 300, but the errors still occurred, if distance constraint was applied.
5. The box size of my system is 5.9366 nm (cubic box)
6. This error occurred when I use 8, 16 and 32 cores. But when I use one core, I didn’t get the error as I mentioned before.
7. When I reduced the time step to 1 fs, the error still occurred.
8. When I changed the parameters in output control (nstxout/nstvout/nstfout/nstlog=10) to get the first few frames and to visualize these in VMD, I found the distance constraint was failed (the helical structure was deformed and extended very quickly )(But I am sure I've put the distance constraint to the peptides). When I compared these first few frames of this system(with distance constraint) with another same system (without distance constraint), I found helical structure in system WITHOUT distance constraint could stay undeformed (nearly rigid just in the first few frames I guess).
Thank you very much
> On Nov 5, 2015, at 7:54 PM, JINGJING HUANG <jjhuang0520 at gmail.com> wrote:
> Dear Gromaces users,
> I am trying to carry out a mdrun simulation with distance restraint (gmx mdrun -deffnm system). But I got these errors:
> relative constraint deviation after LINCS:
> rms 0.294533, max 23.000101 (between atoms 9932 and 9933)
> bonds that rotated more than 30 degrees:
> atom 1 atom 2 angle previous, current, constraint length
> 9929 9930 90.0 0.1095 2.4655 0.1090
> 9929 9931 49.6 0.1096 0.1069 0.1090
> 9932 9933 90.0 0.1210 2.6160 0.1090
> 9932 9934 32.8 0.0898 0.1018 0.1090
> I’ve read the debugging instructions on line, and also many previous discussions towards this issue (http://www.gromacs.org/Documentation/Errors#LINCS.2fSETTLE.2fSHAKE_warnings <http://www.gromacs.org/Documentation/Errors#LINCS.2fSETTLE.2fSHAKE_warnings>) But I believe for my system, the problem is due to the distance restraint and also the parallel simulation. Because:
> 1. When I removed distance restraint, I didn’t get these errors.However, I need to keep this distance restraint to make peptides rigid for my purpose.
> 2. When I run this mdrun simulation ( gmx mdrun -deffnm system -nt 1) in my laptop (just with one cpu), I didn’t got these errors as well.
> Because this is a inhomogeneous system, except for distance restraint, I guess I also need to do domain decomposition in order to simulate that system in many threads. But I don’t know how to do that. I know in my lab, someone previously did particle decomposition (for a very similar system like this one ) in old version of gromacs, it worked and didn’t got these errors. But now PD was removed in gromacs 5.0.4 (This is version I use now), So I cannot try PD for my system.
> Besides, I am not sure whether my analysis toward these errors is right or not.
> Thank you a lot in advance. I got stuck in this problem for a long time.
More information about the gromacs.org_gmx-users