# [molpro-user] Expanding scf/ks procedure

Jutho Haegeman darkgenius4u at gmail.com
Tue Mar 20 14:22:49 GMT 2007

```As pointed out by Peter's answer below, most of what I'm trying to do
can be done with the matrop facilities, built in in Molpro. I've been
experimenting with matrop the last couple of days, but there's one
thing which I think ain't possible. I'd like to solve the generalized
eigenvalue-problem

[F'] [C] = [S][N][S] [C] [Lambda].

In this equation [F'] a self-adjusted Fock-matrix, by redefining for
example [H0]. On the right side, I'd like to use a matrix [S']=[S][N]
[S], which is not the standard overlap matrix.

For solving a generalized eigenvalue equation, as in the scf-
excercise in the matrop chapter, I diagonalized [S^-1/2][F][S^-1/2],
which was possible because [S^-1/2]=[SMH] is standard provided by
Molpro.

But now I have a different [S']-matrix, and I have no idea of how to
solve the eigenvalue equation with the matrop facilities. There is no
mention in the manual of either solving a generalized eigenvalue
equation with a built-in procedure (like DIAG), or of overwriting [S]
(like is possible with [H0]), after which loading [SMH] would yield
the correct [S'^-1/2], or of any other way to calculate [S'^-1/2].

Any help is greatly appreciated,

Thanks

Jutho Haegeman

Op 13-mrt-07, om 13:05 heeft Peter Reinhardt het volgende geschreven:

> Jutho,
> you may start with the exercice in the manual to write an SCF
> procedure
> with the matrop facilities.
>
> With this you may construct a Fock matrix, diagonalize it, save
> the orbitals, create a density etc.
>
> For instance to calculate the Kohn-Sham matrix from some orbitals
> (at record 2200.1) you just say
>
> matrop
> dens,densu,orb, 5
> save,densu,2200.2,density
>
> dft,b3lyp;
> density,2200.2;
> potential,2900.1
>
> matrop
> coul,cc,densu
> write,ksmatrix,KS.mat,erase
>
> Yours,
>    Peter Reinhardt
>
>
> Jutho Haegeman wrote:
>> Dear Molpro Users,
>>
>> For my master thesis on molecular dissociation, i would like to
>> expand the scf or ks procedure of molpro, by adding a few local
>> density-depending terms in the left site as well as in the right side
>> of the eigenvalue equation. As the different quantum-chemical
>> solution methods in Molpro are implemented as different applications,
>> i guess this should not be too hard to do.  Yet I have no experience
>> with altering a major program like Molpro, so I could use some hints
>> or tips, about where to start and what to keep in mind. For example,
>> the extra terms will contain parameters, which preferably should not
>> be hard-programmed so they can be changed from the input file, for
>> fitting purposes.
>>
>> Thanks in advance for any helpful response,
>>
>> Jutho Haegeman
>>
>
>
> --
>       v  v  v  v  v  v  v  v  v  v  v  v  v  v  v  v  v  v  v  v  v
>>  Peter Reinhardt              |   Maître de Conférences     <
>>  Lab. Chimie Théorique        |        (HDR)                <
>>  Université Paris VI          |                             <
>>  4 place Jussieu, case 137    |  Tél.: +33(0)1 44 27 9657   <
>>  F -- 75252 Paris, France     |  Fax.:               4117   <
>>                 email: Peter.Reinhardt at upmc.fr              <
>>       Web: http://www.lct.jussieu.fr/pagesperso/reinh       <
>       ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^  ^
>

```