hi,

 i have been facing a problem of linear dependence which systematically
appears on a type of solid when runnign BAND for a single point
ground state computation. I had never seen it before for any other
compound i have tried, and it affects the runs in parallel at the
IBM-SP4.

this is not the common basis set linear dependence (Kohn-Sham orbital
expansion) but linear dependence of the fit functions used in the
evaluation of effective potential matrix elements.

Apparently the code (BAND) corrects automatically such fault by removing
the appropriate function, accoding to a similar (seems to me) criterium to
that appearing in the manual, regarding the basis set linear dependence.

However, whenever a density of states stage is entered after the single
point energy (either simplified version computed by default at the end
of a single point step, or an explicit density of states
calculation in a second stage), the calculation goes crazy and a huge
number of threads enters the processors.

I have already reported this problem with full details, but have not
had any answers.

i'd appreciate some feedback

thanks

Manolo

PS i could send the input files again provided this was of interest
to anybody
 
###########################################################################
 Manuel Perez Jigato
 Helsinki University of Technology
 Laboratory of Physics
 Room U203
 Otakaari 1M
 Otaniemi, Espoo
 P.O. Box 1100
 FIN-02015 HUT
 Finland

 Tel: +358-(0)9-451 3107
 Fax: +358-(0)9-451 3116
 E-mail: mpj_at_fyslab.hut.fi
###########################################################################
"Nobody tells me what to think, except Mrs. Pauling."
                                                        Linus Pauling
###########################################################################
Received on 2003-10-28 09:15:16