Dear Wenge Han:

I neither have experience with Lshift nor with ADF2.3 but I noticed that
you are using a rather strong damping together with "DIIS cyc=20".
However (at least with ADF1999), an additional value for ok (in the DIIS
key) seems strongly recommendable. As is easily seen, the DIIS in both
of your jobs starts already in Cycle 4, because of the default limit of ok
(0.5 I think). I recommend something like ok=0.005. Hope that helps!

By the way, are you dealing with transition metal complexes? I might be
interested in the proper usage of Lshift, too, since it helped me a lot in
Gaussian but confuses me a little bit in ADF.

Regards, Christian

> Dear all:
>
> Because of the SCF convergence problem, I need to use Lshift in
> my calculations. The Lshift on ADF2.3 works very well for my problem.
> But now when I change to ADF2000 (I guess the Lshift in ADF99 is the
> same as that of ADF2000), I found there are more parameters for the option
> of Lshift, and I could not get the SCF converge. My question is if some of
> you know what is the diference between the Lshift in ADF2.3 and ADF2000 (or
> ADF99), and how I can use the Lshift in ADF2000 just the same as that in
> ADF2.3?
>
> Thank you very much in advance for your kind help!
>
> Wenge Han
> e-mail: wengehan_at_scripps.edu
>
> ---------------------------------------------------------------------------
> ----- For your reference, in the input for ADF2.3, I have:
>
> SCF
> Lshift 0.3
> CONVERGE 1e-4
> MIXING 0.04
> ITERATIONS 250
> DIIS cyc=20
> END
>
> and the following is the logfile for the ADF2.3 single-point calculation:
>
> <Jun24-00> <12:12:23> >>>> CYCLE
> <Jun24-00> <12:51:55> 1
> <Jun24-00> <14:08:33> 2 ErrMat 2.50780195 MaxEl -0.13932085
> <Jun24-00> <15:27:27> 3 ErrMat 2.30261351 MaxEl -0.12970959
> <Jun24-00> <16:50:07> 4 ErrMat 2.34128993 MaxEl -0.31206635

> ---------------------------------------------------------------------
> For the same job but with ADF2000:
>
> I changed the input regarding to the Lshift:
>
> SCF
> Lshift 0.3 err=0.0000001 cyc=1
> CONVERGE 1e-4
> MIXING 0.04
> ITERATIONS 250
> DIIS cyc=20
> END
>
> And the logfile of the ADF2000 output:
> <Jun27-2000> <12:35:40> >>>> CYCLE
> <Jun27-2000> <13:19:05> 1
> <Jun27-2000> <14:29:52> 2 ErrMat 2.50797755 MaxEl -0.13937020
> <Jun27-2000> <15:25:39> 3 ErrMat 2.30274022 MaxEl -0.12976119
> <Jun27-2000> <16:23:51> 4 ErrMat 5.48013945 MaxEl -0.69824973
Received on 2000-06-30 10:08:29