Re: [Towhee-users] maxextrema exceeded

[Greg L. <gs...@ps...>]

I am replying to my own email for 2 reasons:  1) to share my solution with
others who may have similar problems, and 2) to check with more experienced
users that this solution is valid.

In reading through other emails in the archive, the idea of using an
alternate force field style for the torsions dawned on me.  Initially,
modifying the custom towhee_ff file created during the LAMMPS import
process, I changed the torsion style from 3 (AMBER) to 21 (Multiple Rigid
Dihedrals) (and made the appropriate changes throughout the rest of
towhee_ff) got rid of the error "ATD_CBMC_SETUP:maxextrema exceeded" but
introduced a new error "overlap in initial structure".  I presume this is
because many of the dihedral angles were not at their precise equilibrium
positions.

As seen in the benzene2_ff from another email in the archive, I tried using
the torsion style of 8 (nonbonded terms only) with One-Four Nonbond Logical
set to false.  This sets the torsional energy within the large complex
molecule to 0, thereby avoiding the "overlap in initial structure".  I think
this will work because I am not performing any MC moves that involve
changing the molecule conformation, only in simulating the adsorption of
simple gasses onto this structure.

Thanks,
-Greg

On Wed, May 13, 2009 at 10:11 AM, Greg Larsen <gs...@ps...> wrote:

> Hello,
> I am attempting to import a complex structure (many fused aromatic rings)
> from LAMMPS into Towhee.  While most of the importing process seems to be
> working great, I get the following error:
>
> Verifying input structures are consistent
> Determining Autofit Gaussian bond parameters
> Determining Autofit Gaussian bend A parameters
> Determining Autofit Gaussian bend B parameters
> Determining Autofit Gaussian dihedral parameters
>  ATD_CBMC_SETUP:maxextrema exceeded
>
> Any suggestions on how to overcome this?
>
> maxextrema is defined in atd_cbmc_setup.F as: maxextrema =
> 2*(CBMC_MAXTORPEAKS+CBMC_MAXBENDPEAKS).  I tried doubling the values of
> CBMC_MAXTORPEAKS and CBMC_MAXBENDPEAKS in preproc.h (and then recompiling),
> but get the same error.
>
> I am guessing this problem is due to difficulties in using CBMC methods on
> a complex structure.  If it helps, I do not need to use CBMC methods on this
> structure, I am only intersted in using Towhee to model adsorption onto this
> structure.
>
> Thanks,
>
> -Greg Larsen
>