[CP2K:1905] Re: methane in a water box QM/MM
Teodoro Laino
teodor... at gmail.com
Wed Mar 18 19:06:55 UTC 2009
Hi Marius,
we discussed so many time about the documentation that is almost an
exhausted thread :-).
Anyway.. Axel started with the pages on the google forum and they are
exactly in the state he left them..
Why don't you take over and contribute to them?
Let's just go back now to your question..
When you want to do a PERIODIC QMMM calculation, conceptually speaking
you have to decouple and recouple the QM images with the proper periodicity.
have a look at:
QMMM/QS/regtest-4/crys_per_qmmm.inp, which is true that it is a regtest
but is still ok as a starting point..
First don't forget in all QMMM runs that use quickstep to add
COMMENSURATE to the MGRID section. It is mandatory!
Second if you do fully periodic QMMM you don't have to specify any
poisson section in the DFT section (since the decoupling is handled
directly with the re-coupling).
In the CELL section of the QMMM section you have to enable fully
PERIODIC (i.e. delete possible PERIODIC NONE in both cell sections) and
add to the QMMM section the PERIODIC section:
&PERIODIC
GMAX 0.5 (is a reasonable value.. but depends on the GEEP expansion ..
USE_GEEP_LIB 10 is good enough)
&MULTIPOLE
EWALD_PRECISION 1.0E-6
RCUT CELL/2 (is good enough)
&END
&END
at the moment the only way to decouple and recouple is with the
MULTIPOLE scheme.
You can start from these hints and check the set of parameters better
for you.
And in case you have still time and willingness put down a couple of
pages on how to use the QMMM in cp2k.
Future users will be extremely glad. ;-)
Cheers
Teo
Marius Retegan wrote:
> Hello again,
>
> I was a little hasty with my previous email, since some things are a
> little unclear.
> I looked at the PERIODIC section in QMMM but sincerely that was of
> very little help.
> So if I want to do periodic QMMM, I should remove the decoupling from
> the QM part and add CELL information in the QMMM part. Additionally in
> the QMMM part I should also add PERIODIC. If this is correct, I'm
> stuck since I don't know what parameters to chose next.
>
> Usually, I'm not used to asking so many questions regarding the
> softwares that I used, since I try as much as possible to find the
> answer alone. With cp2k (especially the QM/MM part) this is very
> difficult, mostly due to the its particular nature (plainwaves,
> original techniques for computing QM/MM electrostatic potential, etc...).
> I saw many times being said on the list, that the test are hardly
> something to take as viable production runs, since they are made for
> speed. The manual is OK for constructing your input, but it provides
> very little help in choosing the "right" parameters for your system. I
> understand that parameters depend on the system studied, but some
> heads-up on what is safe to use would be nice. What happend with the
> wiki based documentation proposal? I think that having a great
> software is important, but giving new users a place to start is
> equally important.
>
> Please don't take this email as hostility towards cp2k (very far from
> my intentions), but more as lines coming from a "frustrated" user who
> tries to see the light at the end of the tunnel :).
>
> Thanks
> Marius
>
> On Wed, Mar 18, 2009 at 5:18 PM, Marius Retegan
> <marius.s... at gmail.com <mailto:marius.s... at gmail.com>> wrote:
>
> Thanks a lot for all your comments.
>
> Marius
>
>
> On Wed, Mar 18, 2009 at 5:14 PM, Teodoro Laino
> <teodor... at gmail.com <mailto:teodor... at gmail.com>> wrote:
>
>
> The bug (which was depending on the way of reading natively
> AMBER prmtop
> files) is now fixed in the CVS (thanks for reporting)!
> Just few comments about your QMMM setup.
> 1) The way it is setup you will hardly have a conserved energy
> (you have
> a PERIODIC MM) but you don't use the periodic QMMM version
> (see PERIODIC
> in QMMM). Alternatively you may think about using a spherical
> cutoff for
> the electrostatic (that is tapering the interaction after a
> certain cutoff).
> 2) you may want to specify proper covalent radius for the QMMM
> electrostatic interactions (at the moment you rely on the
> default which
> is 0.78 (if I'm not wrong) for all atoms).
>
> Have a nice QMMM.
> Teo
>
> p.s.: about the xc_deriv and xc_smooth_rho I would not suggest
> you to
> use them unless you really know what you are doing. Most of
> the time you
> don't need to specify those keywords (only for few
> pathological cases.
> CH4 or H2O are definitely not one of them).
>
>
> Marius Retegan wrote:
> > Hi,
> >
> > In my continuing saga of using cp2k for running qm/mm
> calculations I
> > have encounter yet another bump.
> > While using the input below I get this error "FORCEFIELD|
> Missing Bond
> > ( CT, HC)" and angle ...
> >
> > &GLOBAL
> > PROJECT mebox_qmmm
> > RUN_TYPE MD
> > PRINT_LEVEL LOW
> > &END GLOBAL
> >
> > &MOTION
> > &MD
> > ENSEMBLE NVT
> > STEPS 1000
> > TIMESTEP 1.0
> > TEMPERATURE 298
> > &THERMOSTAT
> > REGION GLOBAL
> > TYPE CSVR
> > &CSVR
> > TIMECON 100
> > &END CSVR
> > &END THERMOSTAT
> > &END MD
> > &PRINT
> > &RESTART OFF
> > &END RESTART
> > &RESTART_HISTORY OFF
> > &END RESTART_HISTORY
> > &END PRINT
> > &END MOTION
> >
> > &FORCE_EVAL
> > METHOD QMMM
> >
> > &DFT
> > CHARGE 0
> > BASIS_SET_FILE_NAME GTH_BASIS_SETS
> > POTENTIAL_FILE_NAME GTH_POTENTIALS
> > &MGRID
> > CUTOFF 280
> > COMMENSURATE
> > &END MGRID
> > &SCF
> > MAX_SCF 20
> > EPS_SCF 1.0E-6
> > &OT T
> > PRECONDITIONER FULL_ALL
> > MINIMIZER DIIS
> > &END OT
> > &PRINT
> > &RESTART OFF
> > &END RESTART
> > &RESTART_HISTORY OFF
> > &END RESTART_HISTORY
> > &END PRINT
> > &END SCF
> > &QS
> > EPS_DEFAULT 1.0E-12
> > MAP_CONSISTENT
> > EXTRAPOLATION ASPC
> > EXTRAPOLATION_ORDER 3
> > &END QS
> > &XC
> > &XC_GRID
> > XC_SMOOTH_RHO NN10
> > XC_DERIV NN10_SMOOTH
> > &END XC_GRID
> > &XC_FUNCTIONAL BLYP
> > &END XC_FUNCTIONAL
> > &END XC
> > &POISSON
> > POISSON_SOLVER MULTIPOLE
> > PERIODIC NONE
> > &END POISSON
> > &END DFT
> >
> > &MM
> > &FORCEFIELD
> > VDW_SCALE14 0.5
> > EI_SCALE14 0.8333333
> > SCALE_CUTOFF .FALSE.
> > PARM_FILE_NAME mebox.prmtop
> > PARMTYPE AMBER
> > &SPLINE
> > EMAX_SPLINE 1.0
> > RCUT_NB 10
> > &END
> > &END FORCEFIELD
> > &POISSON
> > &EWALD
> > EWALD_TYPE SPME
> > ALPHA .36
> > GMAX 23 23 23
> > &END EWALD
> > &END POISSON
> > &END MM
> >
> > &QMMM
> > E_COUPL GAUSS
> > USE_GEEP_LIB 7
> > &CELL
> > ABC 8.0 8.0 8.0
> > PERIODIC NONE
> > &END CELL
> > &QM_KIND C
> > MM_INDEX 1
> > &END QM_KIND
> > &QM_KIND H
> > MM_INDEX 2 3 4 5
> > &END QM_KIND
> > &END QMMM
> >
> > &SUBSYS
> > &CELL
> > ABC [angstrom] 23.2782522 22.7141099 23.2326917
> > PERIODIC XYZ
> > &END CELL
> > &TOPOLOGY
> > COORDINATE PDB
> > COORD_FILE_NAME mebox_npt.pdb
> > CONNECTIVITY AMBER
> > CONN_FILE_NAME mebox.prmtop
> > &END
> > &KIND C
> > BASIS_SET DZVP-GTH
> > POTENTIAL GTH-BLYP-q4
> > &END KIND
> > &KIND H
> > BASIS_SET DZVP-GTH
> > POTENTIAL GTH-BLYP-q1
> > &END KIND
> > &END SUBSYS
> > &END FORCE_EVAL
> >
> > If I change QMMM with FIST the input runs.
> >
> > I also have some additional questions, but on somewhat
> different matters.
> >
> > In the latest papers that I could find that use cp2k to do qm/mm
> > calculations (http://dx.doi.org/10.1021/ja076081h and
> > http://dx.doi.org/10.1021/jp074858n) the authors used a
> integration
> > step of 0.5 and 0.48. Unfortunately, they did not discussed
> their
> > choice. I'm used to running qm/mm md with other packages,
> but I've
> > always used an integration step of 1 fs. Is cp2k different :-)?
> >
> > Is there any place where I can find information regarding the
> > parameters xc_deriv and xc_smooth_rho and are "good" choices?
> >
> > Thanks,
> > Marius
> >
> >
> >
> >
> >
> >
> >
> >
> > >
>
>
>
>
>
>
> >
More information about the CP2K-user
mailing list