AMBER Archive (2009)

Subject: Re: [AMBER] dftb_disper problem

From: Gustavo Seabra (gustavo.seabra_at_gmail.com)
Date: Thu May 07 2009 - 08:01:42 CDT

Hi Fernando,

This is actually a common problem with the use of dispersion in DFTB.
What is happening here is that, at every step, the program tries to
guess a hibridization state for an atom by counting the number of
atoms within a certain distance threshold, in this case, 1.2A. Then,
from the DISPERSION.INP_ONCHSP, it chooses the polarizability data for
this hibridization. The DISPERSION.INP_ONCHSP file has the following
format:

$ cat DISPERSION.INP_ONCHSP
O 0.560 0.560 0.000 0.000 3.8 3.8 3.8 3.8 3.15
N 1.03 1.03 1.090 1.090 3.8 3.8 3.8 3.8 2.82
C 1.382 1.382 1.382 1.064 3.8 3.8 3.8 3.8 2.5
H 0.386 0.386 0.000 0.000 3.5 3.5 3.5 3.5 0.8
P 1.6 1.6 1.6 1.6 4.7 4.7 4.7 4.7 4.5 !PO4
S 3.0 3.0 3.0 3.0 4.7 4.7 4.7 4.7 4.8 !S not SO2

The first 4 numbers after the atom name are the polarizabilities for
each hibridization state, defined by the number of neighbors. (To
understand the parameterization, look at the original reference: M.
Elstner, P. Hobza, T. Frauenheim, S. Suhai, and E. Kaxiras, J. Chem.
Phys. vol. 114 (2001) 5149 ) Notice that those parameters have been
derived for DNA bases, so there is no guarantee they will be
appropriate for anything else.

Basically, these parameters were derived depending on the number of
neighbors you may expect to find for an atom. However, during an MD
calculation, it eventually happens that an extra atom gets close
enough to confuse the program. In your case you have an Oxygen (atom
2) with 3 neighbors, and there is no polarizability data in the file
for this situation. in this case, sander complains and crashes.

============
WORKAROUND
============
In the Oxygen line, just copy the parameter for 2 neighbors in to the
next column. it will not be precise, but it will preventing sander
from crashing. Line 1 in the DISPERSION.INP_ONCHSP file will will then
look like:
O 0.560 0.560 0.560 0.000 3.8 3.8 3.8 3.8 3.15

Good luck,
Gustavo Seabra
Postdoctoral Associate
Quantum Theory Project - University of Florida
Gainesville - Florida - USA

2009/5/7 Fernando Martín García <fmgarcia_at_cbm.uam.es>:
> Hi all:
>
> I have a problem with the flag "dftb_disper". When i launch my job with
> a value of dftb_disper= 0, the job run well, but if i try to use
> dftb_disper =1, the following message appears:
>
> --------------------------------------------------------------------------------
>   4.  RESULTS
> --------------------------------------------------------------------------------
>
>          *******************************************
>           WARNING: a parameter is 0.0 for atom   2
>                    IZP =   2
>                    nei =   4 ( 3 neighbors.)
>                    hh1 =   0.00
>                    hh2 =   3.80
>                    Ni  =   3.15
>          *******************************************
>  SANDER BOMB in subroutine <dispersion_params>
>  qm2_dftb_dispersion_params.f
>  Exiting.
>
> These are the parameters that i use in this job:
>
>  &cntrl
>  ntx = 7, irest = 1, ntrx = 1,
>  ntxo = 1, nmropt = 1, ntr = 0,
>  ntpr = 100, ntwx = 100,
>  ntf = 1, ntb = 2, dielc = 1.0,
>  cut = 9, nsnb = 10, scnb = 2.0, scee = 1.2,
>  imin = 0, ibelly=0,
>  nstlim = 1000, dt = 0.001,
>  temp0 = 300.0, tempi = 300.0,
>  ntt = 1, vlimit = 20.0,
>  ntp = 1,
>  ntc = 1, tol = 0.00001, pres0=1, comp=44.6,
>  jfastw=0, nscm=1000,
>  ifqnt=1
> /
>  &qmmm
>  iqmatoms=  1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17,
> 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28,
>  qmcharge= 0,
>  qm_theory='DFTB',
>  dftb_disper= 1,
>  dftb_maxiter= 70,
>  qmshake= 0,
>  qmmm_int = 2,
>  peptide_corr = 0,
>  qmcut= 8.0
>
> I've downloaded the parameter file "mio" from www.dftb.org and installed
> in $AMBERHOME/dat/slko, and also the other two files are in the same
> directory (DISPERSION.INP_ONCHSP and CM3). Can anybody tell me where is
> the problem?
>
> Thanks
>
> --
> Fernando Martín García.
> Centro de Biología Molecular "Severo Ochoa".
> C/ Nicolás Cabrera, 1.
> Campus UAM. Cantoblanco, 28049 Madrid. Spain.
>
>
> _______________________________________________
> AMBER mailing list
> AMBER_at_ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber