AMBER Archive (2009)

Subject: RE: [AMBER] dftb_disper problem

From: Ross Walker (ross_at_rosswalker.co.uk)
Date: Fri May 08 2009 - 10:47:39 CDT

Hi Fernando,

It is not based on atom type, simply element and the number of neighbors. In your case with H3O+ you may actually be triggering the 4th hybridization state every now and then. The same might be true of the hydrogens, you may end up with 3 atoms close to one of the hydrogens.

Thus you may need to modify the file to read:

O 0.560 0.560 0.560 0.560 3.8 3.8 3.8 3.8 3.15
H 0.386 0.386 0.386 0.000 3.5 3.5 3.5 3.5 0.8

Of course be aware that this is effectively 'guessing' the parameters here so just keep that in mind when interpreting any results.

Also make sure the code picks up the new file when it runs. The dispersion data should be written to the output file.

Good luck,
Ross

> -----Original Message-----
> From: amber-bounces_at_ambermd.org [mailto:amber-bounces_at_ambermd.org] On
> Behalf Of Fernando Martín García
> Sent: Friday, May 08, 2009 5:43 AM
> To: AMBER Mailing List
> Subject: Re: [AMBER] dftb_disper problem
>
> Hi Gustavo:
> I have changed the parameter that you indicated to me, but the problem
> persist. I was investigating which is the atom that provokes the
> problems and it's a oxygen atom that belongs to a water molecule, that
> needs to be protonated (H3O+)at the end of the reaction I'm simulating.
> These values depend on the type of oxygen atom I'm using?
>
> Thanks
> Fernando
>
> El jue, 07-05-2009 a las 09:01 -0400, Gustavo Seabra escribió:
> > 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
> --
> 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