[gmx-developers] Re: [gmx-users] dihedral restraints documentation

[Mark Abraham]

On 6/10/2011 6:45 AM, David Mobley wrote:
> Just a note concerning dihedral restraints documentation -- the 
> function and default behavior of dihre-fc needs to be documented 
> somewhere in the manual. I just spent a couple hours (and my student a 
> couple of days!) trying to track down the source of some crashes which 
> ended up being due to the difference between specifying
>
> dihre-fc = 1
>
> in our topology file, and not specifying anything at all.
>
> THis how-to documents what dihre-fc DOES 
> (http://www.gromacs.org/Documentation/How-tos/Dihedral_Restraints) but 
> it is not mentioned anywhere in the gromacs manual at all. It turns 
> out dihre-fc defaults to 1000, so the difference between dihre-fc = 1 
> and omitting dihre-fc is a factor of 1000 in dihedral restraint 
> strength. This is enough (since it's a multiplier!) to make the 
> difference between simulations reliably crashing and simulations not 
> crashing at all...
>
> This clearly should be documented in the manual since (with this 
> behavior) knowing what value of dihre-fc one is using is critical to 
> knowing what dihedral restraint strength one is using.

Thanks for the report. I have noted this need in a Redmine documentation 
issue for 4.5.6.

Mark