[molpro-user] start,atdens not working in optg (with startcmd or proc)

H. -J. Werner werner at theochem.uni-stuttgart.de
Thu Sep 28 07:16:53 BST 2006

you are doing nothing wrong. During geometry optimizations
the START directive is ignored, and the orbitals from the
last step are used as a starting guess. There is currently
no option to modify this behaviour. If I find some time,
I'll look into your convergence problem.
Best regards
Joachim Werner

On Mi, 27 Sep 2006, Benj FitzPatrick wrote:

>I'm trying to optimize the geometry for the allyl radical
>using rccsd (molpro 2006.1).  My jobs would keep crashing out
>in one of the rhf-scf tasks where one of the z-matrix
>variables was being
>changed (it would never get convergence in SEWARD).  I got
>some advice that maybe I should try keeping the hf steps from
>reading in the orbitals and just generate the atomic densities
>each time (using start,atdens and startcmd).  My input file is
>below, and I've also tried using proc=foo in place of
>startcmd=hf, where foo is then defined below optg.  I get the
>initial rhf jobs to use start,atdens, but as soon as it goes
>into optg it reverts back to reading the orbitals from the
>previous step.  Please let me know what I am doing wrong.  I
>have also attached an output file in case it might help. 
>Note, for the listed output file I actually have 2 hf lines
>just to make sure that start,atdens wasn't being discarded
>after the first calculation (and it isn't).
>Also, I did the geometry optimization using ump2 and it seemed
>to do just fine (though it did take quite a few steps).  I've
>also tried using different geometries (including the one I got
>from the ump2 optimization) and all terminated in the same way.
>On a separate note, are atden and atdens both ok?  They both
>crop up in the 2006.1 manual, but both seem to give me the
>same end result.
>Benj FitzPatrick
>University of Chicago
> ***,optimization of allyl***
> memory,100,M;
> ch2=        1.0798000
> hc3=        1.077400
> hch3=       117.93
> cc4=        1.373000
> cch4=       120.5600
> dih4=       180.000
> hc5=        1.082800
> hcc5=       117.860
> dih5=       179.990
> cc6=        1.373000
> ccc6=       124.250
> dih6=        -0.0150
> hc7=        1.07980
> hcc7=       120.560
> dih7=         0.0140
> hc8=        1.07740
> hcc8=       121.4800
> dih8=       179.8900
> geometry
> ang;
>  h;
>  c,   1, ch2;
>  h,   2, hc3,        1, hch3;
>  c,   2, cc4,        1, cch4,         3, dih4;
>  h,   4, hc5,        2, hcc5,         1, dih5;
>  c,   4, cc6,        2, ccc6,         1, dih6;
>  h,   6, hc7,        4, hcc7,         2, dih7;
>  h,   6, hc8,        4, hcc8,         2, dih8;
> end
> {hf; wf,23,1,1; start,atdens}
> {rccsd; wf,23,1,1}
> {optg,startcmd=hf}

Prof. Hans-Joachim Werner
Institute for Theoretical Chemistry
University of Stuttgart
Pfaffenwaldring 55
D-70569 Stuttgart, Germany
Tel.: (0049) 711 / 685 64400
Fax.: (0049) 711 / 685 64442
e-mail: werner at theochem.uni-stuttgart.de

More information about the Molpro-user mailing list