AMBER Archive (2009)

Subject: Re: [AMBER] dftb_disper problem

From: Gustavo Seabra (gustavo.seabra_at_gmail.com)
Date: Fri May 08 2009 - 11:34:39 CDT

Hi Fernando,

What Ross said is exactly true here. The changes I suggested only set
the variable for an oxygen with 3 neighbors, and with H3O+, if
something comes close enough, you may be having a fourth neighbor
there. It may also be some other atom (not that same oxygen as in the
original message) getting a hybridization state for which it has not
been parameterized, so check the message carefully to know exactly
what to change.

I just want to add one note here: remember that:
1. The parameters there were derived for studies of DNA base stacking,
and may not be ideal for all purposes.
2. The fix we suggested here is a "quick-and-dirty" fix, and the only
reason we can "kinda" trust it is because we don't expect this
configuration to happen too often. Notice that this fix assumes the
polarizability of that atom won't change with the presence of the
extra neighbor.

Having said that, it looks to me that the H3O+ plays an important role
in your calculations, meaning that the "O with 3 neighbors" situation
is the norm instead of the exception for you. In this case, I would
really recommend that you consider following the steps in that article
to generate better parameters, specific for your system.

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

2009/5/8 Fernando Martín García <fmgarcia_at_cbm.uam.es>:
> 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