[CPMD-list] BLYP instead of B3LYP

[Axel Kohlmeyer]

On Tue, 8 Jan 2008, Reinout Declerck wrote:

RD> This question passed almostly exactly one year ago. See below the reply of 
RD> Axel Kohlmeyer.
RD> 
RD> BLYP is said to reproduce hydrogen bonds rather well, in my view you could 
RD> therefore use BLYP just as well.

...and some parametrizations of HCTH are said to be even better.
but please keep in mind, that hydrogen bonds (in polar liquids 
like water or ammonia) are a very different animal than adsorption 
of (non-polar) hydrogen molecules.

cheers,
   axel.


RD> 
RD> regards,
RD> 
RD> Reinout Declerck
RD> 
RD> 
RD> ----- Original Message ----- 
RD> From: "Axel Kohlmeyer" <akohlmey at cmm.chem.upenn.edu>
RD> To: "Lan-Feng Yuan" <yuanlf at ustc.edu.cn>
RD> Cc: <cpmd-list at cpmd.org>
RD> Sent: Saturday, January 06, 2007 11:18 PM
RD> Subject: Re: [CPMD-list] B3LYP with CPMD?
RD> 
RD> 
RD> > Lan-Feng Yuan wrote:
RD> >>   Dear CPMD people,
RD> >>
RD> >>   I heard that hybrid functionals like B3LYP are terribly slow in CPMD. 
RD> >> While the
RD> >> paper "molecular dynamics simulation of liquid water: hybrid density 
RD> >> functionals"
RD> >> (JPCB, 2006, 110, 3685) adopted a bunch of methods to make B3LYP water 
RD> >> simulation
RD> >> possible, which is very interesting. As they show, hybrid is now only 40 
RD> >> times
RD> >> slower than pure DFT. Are these methods incorporated in current version 
RD> >> 3.11? What
RD> >
RD> > yes. hartree-fock exchange with distance screening is available in 3.11.
RD> > please see the version history in the manual:
RD> >
RD> > - Distance screening for HFX calculation and hybrid funcional for PBC
RD> >
RD> >
RD> >> do you think about the perspective of using hybrid based on plane wave?
RD> >
RD> > this is very much a matter of taste and of how much CPU time you are
RD> > willing to spend on this method. the improvement is undeniable, however
RD> > for most practical applications, you have to compare these improvements
RD> > to other sources of errors, e.g. finite size effects, insufficient 
RD> > statistics, the CP-method itself (when running CP-dynamics, that is). 
RD> > personally, i'd
RD> > rather use the hybrid functional results as an estimate of the error for
RD> > the 'pure DFT' results and invest the additional CPU time to get better
RD> > statistics or larger systems or use a better converged (=larger) basis set
RD> > (note that the authors of the paper point out the 70ry cutoff is at the
RD> > lower end of being converged and that results change with larger basis 
RD> > sets)
RD> > or a combination of it.
RD> >
RD> > cheers,
RD> >      axel.
RD> _______________________________________________
RD> CPMD-list mailing list
RD> CPMD-list at cpmd.org
RD> http://cpmd.org/mailman/listinfo/cpmd-list
RD> 

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.