[Wien] not enough precision in lapwso

akai akai at po.cc.yamaguchi-u.ac.jp
Fri Aug 8 05:00:32 CEST 2003


Dear, wien2k users

Now I use Wien2k (version 03 release 18/01/2003) with the intel 
compiler 7.1(Build 20030701Z)
and MKL6.0 on Xeron 2.4GHz.

In a non-magnetic case,  an eigen-energy value for a spin up state 
printed in an outputso file  is
  not agreement with that for the spin down state. The following is a 
part of the lapwso result for IrSb3 ;

      K=   0.00000   0.00000   0.00000            1
       MATRIX SIZE=  440   WEIGHT= 1.00
      EIGENVALUES ARE:
      -3.7464251   -3.7462733   -3.7446046   -3.7444988   -3.7435388
      -3.7433911   -3.7430059   -3.7429169   -3.6957735   -3.6957733
      -3.6957733   -3.6957731   -3.6957709   -3.6957709   -3.6957685
.

I think some single precision calculations are included in the lapwso 
routine.
The lapw1 is no problem,  since in a output1 file a up-spin energy 
value is exactly the same as
the down-spin value. So the intel compiler and the MKL may not cause 
the trouble directly.

Our compiler options for lapwso are following;
FC = ifc
CC = icc
FOPT     = -FR -mp -w
FGEN     =
LDFLAGS = -L../SRC_lib -L/opt/intel/mkl60/lib/32 -Vaxlib -static
LIBS   = -lmkl_lapack -lmkl_ia32 -lguide -lpthread
DESTDIR  = .
EXECNAME = lapwso
.

Thank you.

##################################################
         Akai Koji
         Media and Information Technology Center
         Yamaguchi University

         Tokiwadai 2-16-1, Ube, Yamaguchi  755-8611, Japan.

         phone: +81 836 85 9900
         fax: +81 836 85 9901
         e-mail: akai at po.cc.yamaguchi-u.ac.jp
##################################################




More information about the Wien mailing list