[Wien] PORT warnings

[L. D. Marks]

This is only a warning. There is a good chance it will happen during a
minimization if you start far from the solution. If it is happening
continuously, it might be a symptom that there is something seriously
wrong with how the calculation is setup, e.g. you could have RKMAX too
small so the energy and forces are inconsistent, or something else.

A brief, more technical explanation. The PORT routine is a Quasi-Newton
method using part of the port3 library (http://www.netlib.org/port). These
methods update an approximation of the second derivative using, as a
constraint, what is called the "curvature condition". If you are far from
the minimum, the second derivative when you start can be quite different
from what it is when you finish, and the curvature condition can be
violated during the process. The code knows how to handle this. I consider
the the port3 routines to be some of the best, most stable and robust
optimization programs in existence, constructed by real experts in the
field, not me. (Written by David Gay in 1980).

C  ***  REFERENCES  ***
C
C 1.  DENNIS, J.E., GAY, D.M., AND WELSCH, R.E. (1981), ALGORITHM 573 --
C             AN ADAPTIVE NONLINEAR LEAST-SQUARES ALGORITHM, ACM TRANS.
C             MATH. SOFTWARE 7, PP. 369-383.
C
C 2.  DENNIS, J.E., AND MEI, H.H.W. (1979), TWO NEW UNCONSTRAINED OPTI-
C             MIZATION ALGORITHMS WHICH USE FUNCTION AND GRADIENT
C             VALUES, J. OPTIM. THEORY APPLIC. 28, PP. 453-482.
C
C 3.  DENNIS, J.E., AND MORE, J.J. (1977), QUASI-NEWTON METHODS, MOTIVA-
C             TION AND THEORY, SIAM REV. 19, PP. 46-89.
C
C 4.  GOLDFARB, D. (1976), FACTORIZED VARIABLE METRIC METHODS FOR UNCON-
C             STRAINED OPTIMIZATION, MATH. COMPUT. 30, PP. 796-811.


On Wed, 1 Sep 2004, Torsten Andersen wrote:

> Dear L.D.Marks,
>
> I have tried the PORT option for two cases now - one of a pure Cu film,
> which seem to be converging much faster with PORT than with NEWT, the
> other with an Fe layer on both sides of a Cu slab. In the latter case, I
> got in the 3rd optimisation step a warning:
>
> STOP >>  (mini) arrived at end -> exit
>   :WARNING, CURVATURE CONDITION FAILED
>   :WARNING, S.Y WAS -0.158690753693532947E-07
> Fallback to compatibility mode with "old" save_lapw
>
> The calculation continues, but is it something I should worry about? The
> Fe/Cu case is of course spin-polarised, and I am only trying to optimise
> in the direction perpendicular to the film (z).
>
> Best regards,
> Torsten Andersen.
>
> --
> Dr. Torsten Andersen        TA-web: http://deep.at/myspace/
> AG Hübner, Department of Physics, Kaiserslautern University
> http://cmt.physik.uni-kl.de    http://www.physik.uni-kl.de/
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>

-----------------------------------------------
Laurence Marks
Department of Materials Science and Engineering
MSE Rm 2036 Cook Hall
2225 N Campus Drive
Northwestern University
Evanston, IL 60201, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
mailto:L-marks at northwestern.edu
http://www.numis.northwestern.edu
-----------------------------------------------