Re: Tabulated potentials - regd

From: Jérôme Hénin (jhenin_at_ifr88.cnrs-mrs.fr)
Date: Tue Oct 09 2012 - 11:29:47 CDT

Dear Ramesh,

What I can see here is that in many instances your minimization does
not converge, presumably the system just blows up. When it does
converge, it gives consistent values.

I suppose the algorithms in NAMD are not optimized for crystals or
solids in general, so they could be a little too sensitive due to
tight packing, high symmetry, etc.

I recommend trying:
1) shorter grid spacings, to see if there is convergence for
sufiiciently fine grids
2) different minimizer settings
3) a better starting point for the minimization (eg the result of a
previous successful minimization)

Regards,
Jerome

On 9 October 2012 16:01, Ramesh Cheerla <cheerla.ramesh1_at_gmail.com> wrote:
> Dear NAMD users,
>
>
> I am performing simulations of a polymer crystal in NAMD, the
> forcefield that I have considered for these simulation has the Vdw
> interactions in buckingham format, I am making use of tabulated potentials
> for this purpose.
> I have generated table with a spacing(dr) of 0.01 and tried to run
> minimization but there is no change in the position of atoms and potential
> energy, So to see the dependence of potential energy on spacing(dr) I have
> made simple tests by varying "dr". By doing so I have understood that even
> for a small change of "dr" also there is a large change in potential
> energy.
> Here i am summarizing the potential energies that I have obtained for each
> "dr"
> dr(spacing) PE(kcal/mol)
> 0.010500 0.1385
> 0.010570 0.1404
> 0.010571 9999999999.9999
> 0.010572 9999999999.9999
> 0.010573 0.1394
> 0.010575 9999999999.9999
> 0.010580 0.1372
> 0.010590 9999999999.9999
> 0.010600 0.1431
> 0.010700 9999999999.9999
> 0.010900 9999999999.9999
> 0.011000 0.1426
>
> How can I account for this large change in potenmtial energy even for a
> small change in spacing, I have used table interpretation type as "cubic"
> (tableInterpType cubic) in my configuration file.
>
> Can anybody explain the reason for this large change? Is the tabulated
> potential in NAMD is very sensitive to the spacing or is this a limitation
> in NAMD ?
>
> Thank you in advance.
>
> --
> Regards,
> Ramesh,
>

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:22:09 CST