[gmx-developers] Implementation of Urey-Bradley

[Mark Abraham]

David van der Spoel wrote:
> Mark Abraham wrote:
> 
>> Hi,
>>
>> I want to use Urey-Bradley functions in my implementation of CHARMM 
>> forcefields for gromacs. Reading the source code I see that calc_bonds 
>> in bondfree.c is where all of the bonded interactions are calculated, 
>> and that the urey_bradley C function will be called when the topology 
>> stipulates a U-B function. However that function will issue a fatal 
>> error because of line 761 of src/gmxlib/bondfree.c:
>>
>>   gmx_impl("Not implemented yet");
>>
>> Having read through the remainder of the function, it appears to be 
>> doing sensible things, but I haven't checked the algebra for the force 
>> decomposition yet.
>>
>> Why does someone think this is not implemented yet? What remains to be 
>> done/tested?
> 
> 
> Maybe I forgot to take it out, please do and recompile. This probably 
> means that the function was never tested, so please do that carefully....

I've now done so. I have numerical agreement of energies (& components 
thereof) calculated with CHARMM27 force field using Accelrys CHARMm, my 
implementation of that force field using gromacs 3.3, and a spreadsheet. 
I can't verify the forces directly, but the gromacs algorithm looks 
correct and the endpoints of energy minimizations look to be the same 
gross state. Thus, I'd suggest removing this comment.

Mark