AMBER Archive (2007)Subject: Fwd: Re: AMBER: Fwd: SCC convergence failure in QM-MM DFTB
From: Francesco Pietra (chiendarret_at_yahoo.com)
Date: Mon Aug 27 2007 - 01:24:33 CDT
Carrying out preliminary min/md and then a min_qm longer than before, did not
convince md_qm to proceed for any longer than about 3000 steps.
NSTEP = 3800 TIME(PS) = 7.600 TEMP(K) = 294.20 PRESS = 0.0
Etot = -24736.9673 EKtot = 3507.7503 EPtot = -28244.7176
BOND = 0.0000 ANGLE = 0.0000 DIHED = 0.0000
1-4 NB = 0.0000 1-4 EEL = 0.0000 VDWAALS = 2798.8552
EELEC = -19490.8468 EHBOND = 0.0000 RESTRAINT = 4.1669
DFTBESCF= -11556.8929
EAMBER (non-restraint) = -28248.8845
------------------------------------------------------------------------------
vlimit exceeded for step 3830; vmax = 21.2787
vlimit exceeded for step 3832; vmax = 38.1763
vlimit exceeded for step 3833; vmax = 21.4165
vlimit exceeded for step 3835; vmax = 25.4013
vlimit exceeded for step 3836; vmax = 20.7452
SANDER BOMB in subroutine eglcao <qm2_dftb_eglcao.f> :
SCC Convergence failure - MAXITER exceeded.
Exiting
Again, changes in DFTBESCF from step to step were minimal, while changes in the
other energy terms were even larger that without preliminary classical runs.
Scripts used:
Classical minimization
&cntrl
imin=1, maxcyc=500, ncyc=200,
cut=25.0, ntb=0, ntc=2, ntf=2,
ivcap=0, fcap=10
/
300K constant temp classical MD
&cntrl
imin=0, ntb=0,
cut=25.0, ntc=2, ntf=2,
tempi=300.0, temp0=300.0,
ntt=3, gamma_ln=1.0,
nstlim=10000, dt=0.001,
ntpr=100, ntwx=100
/
Initial min for qmmm with dftb
&cntrl
imin=1, maxcyc=1000, ncyc=4000,
cut=25.0, ntb=0, ntc=2, ntf=2,
ivcap=0, fcap=10,
ifqnt=1
/
&qmmm
qmmask=':1',
qmcharge=0,
qmtheory=7,
qmshake=0,
qm_ewald=0
/
300K constant temp for qmmd with dftb
&cntrl
imin=0, ntb=0
cut=25.0, ntc=2, ntf=2,
tempi=300.0, temp0=300.0,
ntt=3, gamma_ln=1.0,
nstlim=20000, dt=0.002,
ivcap=0, fcap=10,
ntpr=100, ntwx=100,ifqnt=1
/
&qmmm
qmmask=':1',
qmcharge=0,
qmtheory=7,
qmshake=0,
qm_ewald=0
/
Reminded here that the initial pdb was for a structure minimized by MMFF94
global conformational space search, which agreed with the minimum structure
from amber simulated annealing in vacuum. This was followed by DFT minimization
with the most recent functionals in NWChem (were no crash ever occurred).
I could now certainly carry out
(a) QM-MM DFTB in vacuum; probably diagnostically useful.
(b) Preliminary QM-MM AM1; I disfavor this because AM1 (like PM3) has an
adversarial impact on the solute conformation; these semiempiricals are well
known for their inability to deal with conformations, even in a comparison with
dftb: Kurt W. Sattelmeyer, Julian Tirado-Rives, and William L. Jorgensen J.
Phys. Chem. A, 110 (50), 13551 -13559, 2006.
Probably it will not surprise that the exited md run can be restarted from
previous md.rst. I could write a script so that the md is carried out in
batches shorter than 3000 steps.
Should no mistake arise from what I have done (which is very unlikely, I
admit), I dare suggesting: why not provide a patch for DFTB/PME in order to use
solvatebox with Amber9, while delaying next Amber release until the alluring
plans by Ross Walker are going on supported by a grant that is well deserved?
The latter proposal has already appeared recently on this mailing list.
Thanks
francesco pietra
--- Francesco Pietra <chiendarret_at_yahoo.com> wrote:
> Date: Sun, 26 Aug 2007 09:28:14 -0700 (PDT)
> From: Francesco Pietra <chiendarret_at_yahoo.com>
> Subject: Re: AMBER: Fwd: SCC convergence failure in QM-MM DFTB
> To: amber_at_scripps.edu
>
> The last MD (second addendum) was carried out as initially, according to
> minin
> amd mdin shown at the start of the thread. The only difference (second
> addendum) was a (a) larger Cap and (b) DFT minimization with NWChem of the
> structure used before, which came from MMFF94 global conformational search in
> vacuum plus amber simulated annealing in vacuum; the two converged to the
> same
> minimum, which I took for absolute minimum.
>
> The suggestion about WAT rearrangement is something that I also vaguely
> conceived for the same reasons that Bud made explicitly. Comforted by this,
> I'll try (starting from the DFT minimized structure) first classical minin
> and
> mdin, then repeating dftb minin and mdin as before. If not reported further,
> it
> means that "classical" has solved the problem.
>
> Thanks
>
> francesco
>
>
>
> --- "M. L. Dodson" <mldodson_at_houston.rr.com> wrote:
>
> > Francesco Pietra wrote:
> > > Second addendum to the original (Subject) issue:
> > >
> > > Using a larger solventCap (25.0) and starting from pdb file for the
> solute
> > > which had been minimized by "ab initio" DFT with the most recent
> > functionals
> > > with NWChem, sander bombed for the same reason somewhat earlier than with
> > cap
> > > 20.0 and stating from pdb minimized by global conformational space MMFF94
> > and
> > > Amber9 simulated annealing. I add also the two last steps in the
> > presumption
> > > they may help. DFTBESCF is about the same as for Cap 20.0, while all
> other
> > > energy values are higher (for 1951 WAT residues added for Cap 25.0, with
> > > respect to 947 residues for Cap 20.0). If useful, I add that DFT as said
> > above
> > > never crashed.
> > >
> > >
> > > NSTEP = 2600 TIME(PS) = 5.200 TEMP(K) = 299.15 PRESS =
>
> > 0.0
> > > Etot = -24120.4790 EKtot = 3566.8407 EPtot =
> > -27687.3196
> > > BOND = 0.0000 ANGLE = 0.0000 DIHED =
> > 0.0000
> > > 1-4 NB = 0.0000 1-4 EEL = 0.0000 VDWAALS =
> > 2754.0573
> > > EELEC = -18926.0584 EHBOND = 0.0000 RESTRAINT =
> > 9.0684
> > > DFTBESCF= -11524.3870
> > > EAMBER (non-restraint) = -27696.3881
> > >
> >
>
------------------------------------------------------------------------------
> > >
> > >
> > > NSTEP = 2700 TIME(PS) = 5.400 TEMP(K) = 300.31 PRESS =
>
> > 0.0
> > > Etot = -26475.7499 EKtot = 3580.5889 EPtot =
> > -30056.3388
> > > BOND = 0.0000 ANGLE = 0.0000 DIHED =
> > 0.0000
> > > 1-4 NB = 0.0000 1-4 EEL = 0.0000 VDWAALS =
> > 2856.2068
> > > EELEC = -21370.9526 EHBOND = 0.0000 RESTRAINT =
> > 8.8248
> > > DFTBESCF= -11550.4178
> > > EAMBER (non-restraint) = -30065.1636
> > >
> >
>
------------------------------------------------------------------------------
> > >
> > > SANDER BOMB in subroutine eglcao <qm2_dftb_eglcao.f> :
> > > SCC Convergence failure - MAXITER exceeded.
> > > Exiting
> > >
> > >
> >
> > To incorporate an answer to Professor Roitberg within a reply to
> > you, sander in amber9 cannot do box/PME simulations.
> >
> > Now on to your simulation. I think we need some preliminary
> > answers, first. How did you get the structure you started with?
> >
> > Professor Roitberg's suggestion to start with a structure that had
> > been minimized with AM1 seems to me a good one. Then you might
> > try a preliminary VACUUM SCC-DFTB minimization. If you cannot get
> > a SCF in vacuum, you are unlikely to get one with a water cap,
> > either.
> >
> > The output given seems to indicate that the energy is dropping by
> > ~2500 over a hundred steps, and coupled with the lack of major
> > changes in the QM energy, would seem to me to indicate the solvent
> > is doing some rearrangement. This would also seem to me to be
> > unrelated to the inability to achieve SCF, but I am open to
> > other suggestions by the QMMM experts that might read this posting.
> >
> > This output is from an MD run. Did you not minimize the system
> > and do classical MD to relax the system to the force field? At
> > the very least you need some time with the classical FF to relax
> > the solvent voids created by LEaP. We need more information on
> > what you have done. In detail. Otherwise we are all just wasting
> > time.
> >
> > Bud Dodson
> >
> > >
> > >
> > >
> > > --- Francesco Pietra <chiendarret_at_yahoo.com> wrote:
> > >
> > >> Date: Sun, 26 Aug 2007 00:26:42 -0700 (PDT)
> > >> From: Francesco Pietra <chiendarret_at_yahoo.com>
> > >> Subject: Fwd: SCC convergence failure in QM-MM DFTB
> > >> To: Amber <amber_at_scripps.edu>
> > >>
> > >> Perhaps relevant to add to the issue below:
> > >>
> > >> ---solvateCap EQE TIP3PBOX {coordinates of a solute central atom} 20.0
> > >>
> > >> placed the solute right at the center of the spherical cap; though,
> along
> > the
> > >> longer axis, the 98-atoms solute was not far from the spherical cap
> edge.
> > I
> > >> wonder whether a much bigger cap should be set, and whether 20.0 present
> > one
> > >> has hindered SCC convergence.
> > >>
> > >> --- I was surprised that DFTB MD used as as little memory as AM1 MD (ca
> > 0.1%
> > >> of
> > >> the 4GB available per node). Having not yet read the DFTB theory in
> > detail, I
> > >> am comparing to classical DFT, where modern functionals at modest 6-31G*
> > >> basis
> > >> set level, with the same 98-atoms molecule in vacuum, led to use all the
> > >> available memory (ad there was still much swapping). I understand that
> > DFTB
> > >> working must be profoundly different from classical DFT.
> > >>
> > >> francesco
> > >>
> > >> --- Francesco Pietra <chiendarret_at_yahoo.com> wrote:
> > >>
> > >>> Date: Sat, 25 Aug 2007 10:54:12 -0700 (PDT)
> > >>> From: Francesco Pietra <chiendarret_at_yahoo.com>
> > >>> Subject: SCC convergence failure in QM-MM DFTB
> > >>> To: Amber <amber_at_scripps.edu>
> > >>>
> > >>> Now the 98-atoms solute is well inside TIP3PBOX cap (prepin from
> > >>> antechamber).
> > >>> After successful minimization with:
> > >>>
> > >>> Initial min for qmmm with dftb
> > >>> &cntrl
> > >>> imin=1, maxcyc=500, ncyc=200,
> > >>> cut=20.0, ntb=0, ntc=2, ntf=2,
> > >>> ivcap=0, fcap=10,
> > >>> ifqnt=1
> > >>> /
> > >>> &qmmm
> > >>> qmmask=':1',
> > >>> qmcharge=0,
> > >>> qmtheory=7,
> > >>> qmshake=0,
> > >>> qm_ewald=0
> > >>> /
> > >>>
> > >>> MD with:
> > >>>
> > >>> 300K constant temp for qmmd with dftb
> > >>> &cntrl
> > >>> imin=0, ntb=0
> > >>> cut=20.0, ntc=2, ntf=2,
> > >>> tempi=300.0, temp0=300.0,
> > >>> ntt=3, gamma_ln=1.0,
> > >>> nstlim=20000, dt=0.002,
> > >>> ivcap=0, fcap=10,
> > >>> ntpr=100, ntwx=100,ifqnt=1
> > >>> /
> > >>> &qmmm
> > >>> qmmask=':1',
> > >>> qmcharge=0,
> > >>> qmtheory=7,
> > >>> qmshake=0,
> > >>> qm_ewald=0
> > >>> /
> > >>>
> > >>> sander.MPI (mpirun -np 4) exited because:
> > >>>
> > >>> vlimit exceeded for step 2886; vmax = 26.4356
> > >>> SANDER BOMB in subroutine eglcao <qm2_dftb_eglcao.f> :
> > >>> SCC Convergence failure - MAXITER exceeded.
> > >>> Exiting
> > >>>
> > >>> Much obliged for suggestions what to do best. Thanks
> > >>>
> > >>> francesco pietra
> > >>>
> > >>>
> > >>>
> > >>>
> > >
> > --
> > M. L. Dodson
> > Email: mldodson-at-houston-dot-rr-dot-com
> > Phone: eight_three_two-five_63-386_one
> > -----------------------------------------------------------------------
> > The AMBER Mail Reflector
> > To post, send mail to amber_at_scripps.edu
> > To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> >
>
>
>
>
>
____________________________________________________________________________________
> Moody friends. Drama queens. Your life? Nope! - their life, your story. Play
> Sims Stories at Yahoo! Games.
> http://sims.yahoo.com/
>
____________________________________________________________________________________Ready for the edge of your seat?
Check out tonight's top picks on Yahoo! TV.
http://tv.yahoo.com/
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
|