Project

General

Profile

Bug #1843

pbc=xy with 2 walls floating point exception with the verlet scheme

Added by Joerg Sauter about 4 years ago. Updated about 3 years ago.

Status:
In Progress
Priority:
Normal
Assignee:
Category:
mdrun
Target version:
-
Affected version - extra info:
Affected version:
Difficulty:
uncategorized
Close

Description

The issue sounds related to Bug #1660.

With

pbc                     = xy
nwall = 2

and cutoff-scheme = group everything runs fine.
When I switch to cutoff-scheme = verlet the simulation crashes with a floating exception.

group.tar.gz (10.6 MB) group.tar.gz Joerg Sauter, 10/15/2015 10:31 AM

History

#1 Updated by Berk Hess about 4 years ago

  • Status changed from New to In Progress

Your simulation crashes because an H atom is beyond the top wall. This can happen because your tabulated potential is far too weak to keep the water oxygens away from the wall: at distance zero the potential is 0.77*C6 (C6=0.0025, if I'm not mistaken), which is much smaller than 1. You should increase the potential by several orders of magnitude.

Still the crash here could happen in valid circumstances, e.g. when the user sets r-linpot. But there is not an easy solution to this without affecting performance for common run setups. I'll think a bit more...

#2 Updated by Erik Lindahl over 3 years ago

Berk: Can we add a check for this?

Gromacs should never ever just crash, so there is something wrong with code that produces a floating-point exception. If the price for that is lower performance, so be it, IMHO.

#3 Updated by Erik Lindahl over 3 years ago

  • Assignee set to Berk Hess
  • Target version set to 2016

#4 Updated by Mark Abraham about 3 years ago

  • Description updated (diff)

#5 Updated by Mark Abraham about 3 years ago

  • Target version changed from 2016 to 2016.1

#6 Updated by Mark Abraham about 3 years ago

  • Target version deleted (2016.1)

Also available in: Atom PDF