[Wien] Problem with WIEN2k_07 on the HP-UX (solution)

[Ivas Toni]

Hi Peter,

Thank you for your input,
I didn't solve the problem with WIEN2k_07 but I have installed the
WIEN2k_8
(this probably solves the problem with wien2k v.7 too),
And done this changes to properly compile the code:

-compile rotdef.f with +O1 optimization
-copy cputim_dec.f cputim.f
-recompile
-change Makefile in SRC_mixer to include line:
F.o.:
    same as line with f.o.:

You need to watch out for all files which are named rotdef*.f because
they issue
An error: this intrinsic function don't support complex*20 and real*10
types... 
Recompling with +o1 solves that problem. 

Copy cputim_dec.f to cputim.f instead of cputim_hp.f to cputim.f,
because it seems that on HP-UX B.11.23 U ia64
It doesn't find "times" intrinsic function, dec version works.

Change all Makefiles to include rule F.o.: with same rule as f.o.,
otherwise it will not compile files with .F extension
This is the case with SRC_mixer directory.

This two script do most of the work, but still maybe you will need to
Do some recompiling :)

for i in `find . -name "cputim_dec.f"`
do
echo `echo $i | sed s/_dec//`
cp $i `echo $i | sed s/_dec//`
done 
 
for i in `find . -name "rotdef*.f"`
do
d=`dirname $i`
cd $d
make
f90 +O1 +U77 +source=free -c `basename $i`
# recompile
make
cd ..
done

There is the problem with startup script "x" on this platform in the new
version 8.1, 
the problem is that "cut" utility is not properly started. 
Workaround is to use old one from version 7. 
Here is diff file between x.7 and x.8:
c28
#set cmplx
.
a50
unset super                     # for dstart, creates new.clmsum
.
a158
  case -super:
    set super
    shift; breaksw  
.
a177
if !($?cmplx) then
set cmplx=`cut $file.struct -b -6|\
awk 'BEGIN{c="c"};{if ($0 == "-1 0 0") {\
getline; {if ($0 == " 0-1 0"){\
getline; {if ($0 == " 0 0-1"){c= ""}}}}}};\
END{print c}'` 
endif
.
a178

.
d246
d248 258
d266
a268
   if ($updn == '') then
     if ($?super) then
       echo " 51,'new_super.clmsum',    'unknown','formatted',0">>$def
     else
       echo " 51,'$file.clmsum',    'unknown','formatted',0">>$def
     endif
   else
     if ($?super) then
       echo " 51,'new_super.clm$updn',    'unknown','formatted',0">>$def
     else
       echo " 51,'$file.clm$updn',    'unknown','formatted',0">>$def
     endif
.
c850
 6,'$file.outputm_vresp',   'unknown','formatted',0
.
a1530
-super->        runs dstart and creates new_super.clmsum (and not
$file.clmsum)
.
c1559
echo 'x dstart [-up/-dn -c -fft -super]'

Thank you,
Toni Ivas
Department of Materials
ETH Zurich

-----Original Message-----
From: wien-bounces at zeus.theochem.tuwien.ac.at
[mailto:wien-bounces at zeus.theochem.tuwien.ac.at] On Behalf Of Peter
Blaha
Sent: Friday, January 25, 2008 10:07 AM
To: A Mailing list for WIEN2k users
Subject: Re: [Wien] Problem with WIEN2k_07 on the HP-UX

First thing to check:

Open your struct files  (like tic.struct) in an editor and check if the
number of symmetry operations is .gt. zero

   48      NUMBER OF SYMMETRY OPERATIONS

If not, than you may have missed a step in the initialization.
------------
Check your struct file by copying the whole directory to Linux and run
dstart.
------------
If your struct file is ok, recompile dstart without optimization.
And search for the compiler bug ....

Ivas Toni schrieb:
> Hi,
> 
> I have problem running the WIEN2k_07  on the HP-UX B.11.23 U ia64 I 
> compiled the WIEN2k with HP f90 v3.1 and standard flags:
>      Compiler options:        +O3 +U77 +source=free
>      Linker Flags:               -Wl,-L/usr/local/lib +U77
>      Preprocessor flags:      '-DParallel'
>      R_LIB (LAPACK+BLAS):     -llapack -lblas 
> After setting the case in w2web and running of lapw0 I get error:
> 
>  'ROTDEF' - no symmetry operation found.
>  'ROTDEF' - for jatom, indexi, lattice 2 2
>  'ROTDEF' - atomposition of jatom   0.5000000   0.5000000   0.5000000
>  'ROTDEF' - atomposition of index   0.5000000   0.5000000   0.5000000
> 
> This error comes from missing symmetry operations in
SRC_lapw0/rotdef.f.
> I have tried this with several case files including also tutorial case

> TiC and I allways get the same error. By the way all this cases work 
> fine on my Linux notebook.
> Changing line 92 in rotdef.f to write relevant parameters:
> WRITE(ERRMSG,9000) jatom, index,
> lattic(1:1),lattic(1:2),lattic(1:3),MAXVAL(ABS(X1))
> Gives back:          
> 'ROTDEF' - for jatom, indexi, lattice 2 2 F F F     0.5000000
> Which means that MAXVAL(ABS(X1)) is always bigger then TOLER2=1.5*1D-7

> in the rotdef.f Probably the error comes from some bug in compiler 
> implementation, but I don't know anything about wien2k code, so I 
> can't make a workaround (hacking:))
> 
> Thank you,
> Toni Ivas
> Department of Materials,
> ETH Zurich
> 
>   
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien

-- 

                                       P.Blaha
------------------------------------------------------------------------
--
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW:
http://info.tuwien.ac.at/theochem/
------------------------------------------------------------------------
--
_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien