Re: [cclib-devel] [cclib-users] Gaussian Log Parser AssertionError

[Karol M. L. <kar...@gm...>]

Ditto. Gaussian throws away or includes basis set functions all the time
for efficiency, and the explenation Thomas gave is correct. I've seen this
many times before myself, in fact, and had the impression cclib deals with
it somehow. Perhaps we just haven't encountered INCREASING NBsUse before.

- Karol

On Dec 07 2013, Noel O'Boyle wrote:
> We'll do something about this. I've seen this before. We just need to
> think at our end about how (or if) this affects other attributes. In
> the meanwhile, just try removing the assertion.
> 
> - Noel
> 
> On 6 December 2013 14:57, Yohann Morille <yoh...@un...> wrote:
> > Hi Adam,
> >
> > below you will find the answer of Thomas, the chemist for who I do some
> > developments,
> >
> > Yohann
> >
> > Dear Yohann and Adam,
> >
> > Here is my explanation/opinion :
> >
> > For this calculation NBasis do not change as it is defined by the basis set
> > chosen. But Gaussian do not use all basis set functions as there is a
> > minimal overlap criteria.
> > During the optimization, when the geometry changes the overlap between basis
> > set can slightly change and NBsUse can increase or decrease. See our example
> >  NBasis=  1192 RedAO= T  NBF=   596   596
> >  NBsUse=  1181 1.00D-06 NBFU=   590   591
> > --
> >  NBasis=  1192 RedAO= T  NBF=   596   596
> >  NBsUse=  1182 1.00D-06 NBFU=   591   591
> >
> > I agree that NBasis should be constant. If it is not the case, the whole
> > basis set has changed  = problem. But a changing NBsUse is not relevant (it
> > can not in any case be over NBasis). Maybe a drastic change (10%) should
> > issue a warning [a case never seen by myself]
> >
> > Cordially
> > Thomas Cauchy
> > The chemist working with Yohann
> >
> >
> > On 06/12/2013 02:38, Adam Tenderholt wrote:
> >
> > Hi Yohann,
> >
> > Thanks for bringing this to our attention. I'm cc'ing the developer list
> > since that is the more appropriate place for this discussion.
> >
> > Is there a reason the number of basis used changes? It looks like it uses an
> > extra basis function (out of 1192), perhaps because of symmetry breaking?
> >
> > It seems to me like NBasis should be constant, and we should just issue a
> > warning if NBsUse changes. Thoughts?
> >
> > Adam
> >
> >
> >
> > On Wed, Dec 4, 2013 at 2:18 AM, Yohann Morille
> > <yoh...@un...> wrote:
> >>
> >> Dear all,
> >>
> >> first of all thank you for the development of cclib package that will be
> >> very useful for me !
> >>
> >> However, I have a problem to parse a Gaussian log file (in attachment).
> >> Indeed, the calculation seems to have converged ("Stationary point found")
> >> but
> >> when I try to parse this log file, I receive an AssertionError from line
> >> 915 of  gaussianparser.py
> >> because the number of basis used (NBsUse) has changed (from 1181 to 1182)
> >> between 2 steps.
> >>
> >> Can we imagine a less restrictive condition like to keep the last NBsUse
> >> if the difference is not too large ?
> >>
> >> thanks,
> >> Yohann
> >>
> >> --
> >>
> >> Yohann Morille - [yoh...@un... Tel: +33 (0)2 4173 5346]
> >> MOLTECH-Anjou  - [http://moltech-anjou.univ-angers.fr]
> >>
> >> MOLTECH-Anjou/CNRS - UFR Sciences Bat.K - 2 Bd Lavoisier - 49045 ANGERS
> >>
> >>
> >>
> >> ------------------------------------------------------------------------------
> >> Sponsored by Intel(R) XDK
> >> Develop, test and display web and hybrid apps with a single code base.
> >> Download it for free now!
> >>
> >> http://pubads.g.doubleclick.net/gampad/clk?id=111408631&iu=/4140/ostg.clktrk
> >> _______________________________________________
> >> cclib-users mailing list
> >> ccl...@li...
> >> https://lists.sourceforge.net/lists/listinfo/cclib-users
> >>
> >
> >
> > --
> >
> > Yohann Morille - [yoh...@un... Tel: +33 (0)2 4173 5346]
> > MOLTECH-Anjou  - [http://moltech-anjou.univ-angers.fr]
> >
> > MOLTECH-Anjou/CNRS - UFR Sciences Bat.K - 2 Bd Lavoisier - 49045 ANGERS
> >
> >
> > ------------------------------------------------------------------------------
> > Sponsored by Intel(R) XDK
> > Develop, test and display web and hybrid apps with a single code base.
> > Download it for free now!
> > http://pubads.g.doubleclick.net/gampad/clk?id=111408631&iu=/4140/ostg.clktrk
> > _______________________________________________
> > cclib-devel mailing list
> > ccl...@li...
> > https://lists.sourceforge.net/lists/listinfo/cclib-devel
> >
> 
> ------------------------------------------------------------------------------
> Sponsored by Intel(R) XDK 
> Develop, test and display web and hybrid apps with a single code base.
> Download it for free now!
> http://pubads.g.doubleclick.net/gampad/clk?id=111408631&iu=/4140/ostg.clktrk
> _______________________________________________
> cclib-devel mailing list
> ccl...@li...
> https://lists.sourceforge.net/lists/listinfo/cclib-devel

-- 
written by Karol M. Langner
Mon Dec  9 20:39:46 EST 2013