[gmx-users] Re: Re: Unexplained crashes with Gromacs-4.0 & Neighbor Searching

[Mark Abraham]

Inon Sharony wrote:
> 
> 
> I dumbed down the system (molecule) to keep the potential pitfalls to a 
> minimum. One of the three atoms is just a deadweight (molecular weight 
> 9999999.9) . Could this be the problem? If so, how? I'll try simulating 
> a lighter molecule to see if that takes care of it.
> Though this is "a commonly-reported problem", so far I failed in finding 
> past reference to it in the archives -- at least I can't differentiate 
> my problem with the one you reported, and another instance which turned 
> out to be because the protein topology was completely out of order (from 
> the Hydrogen at home people).

This "commonly-reported problem" is treated briefly here 
http://wiki.gromacs.org/index.php/Errors#Range_Checking_error.

If you've got most of your system coupled to 300K, and a single atom 
coupled to 0K, it's no surprise that your integration is unstable. 
Thermostat algorithms work well in thermodynamic limits, not for 
handfuls of atoms.

Mark