[gmx-developers] Bug in Ewald Code?

[hessb at mpip-mainz.mpg.de]

>> Message: 5
>> Date: Wed, 1 Nov 2006 08:09:47 +0100 (CET)
>> From: hessb at mpip-mainz.mpg.de
>> Subject: Re: [gmx-developers] Bug in Ewald Code?
>> To: "Discussion list for GROMACS development"
>>         <gmx-developers at gromacs.org>
>> Message-ID:
>>         <62322.84.168.76.209.1162364987.squirrel at lnxsrv9.mpip-mainz.mpg.de>
>> Content-Type: text/plain;charset=iso-8859-1
>>
>> > On Tue, 31 Oct 2006, Dave Rogers wrote:
>> >
>> >>Developers,
>> >>
>> >>Problem:
>> >>  The ewald potential of the input configuration changes from
>> >>sr=0.000000   recip=-0.094927 to sr=-460.603673 recip=-330.965952 for
>> >>a single water molecule (all system files included below) when the
>> >>bond type ([bonds] funct) is changed from 1 (default) to 6 (harmonic)
>> >>for both of the bonds in Voth's water model!  For now I am unable to
>> >>reproduce his numbers for that water model as he used a harmonic bond
>> >>potential.
>> >>
>> >>Thanks for looking into this,
>> >
>> > please add exclusions manually and see if that resolves it. submit a
>> > bugzilla anyway,
>>
>> This is exactly the reason for having the bond type 6.
>> Type 1 and 6 are identical potentials, but type 6 does
>> not produce a chemical bond and therefore no exclusions.
>>
>> Or is the problem that you do not want to use 1 because
>> you want bond constraints for other molecules?
>>
>> Then you will indeed need to use 6 and you can add a chembond,
>> bond type 5, to let Gromacs know that O should be chemically
>> bonded to H. This will automatically generate the exclusions.
>>
>> Berk.
>
> Berk,
>
>   You are right.  I am reading the version 3.3 manual, where section
> 5.3.5 mentions this.
>   Unfortunately, I read through section 4.2 (Bonded interactions),
> where the first potential mentioned is the "Harmonic Potential" and
> the "Fourth Power Potential" contains a prescription for converting
> stretch constants.  Next, I looked at table 5.4 and did not understand
> the difference between bond types 1 and 6, except that 6 explicitly
> says "Harmonic Potential".  Since I was having difficulty reproducing
> the results I was after I tried bond type 6...
>   It would have been helpful for me if table 5.4 included a reference
> to the special role of bond types 6 and 7 and/or a reference to the
> respective potential descriptions in sec. 4.2.  Hopefully changing
> that will lower the number of misguided users such as myself.

The table does include a reference, but only in the form
of a small superscript b for all bonds, except types 6 and 7.
It is difficult to make this clearer, since the table is already
overloaded with information.

If anyone has a suggestion, please tell me.

Berk.