[Wien] Re: DOS Problems and tetra using intel ifc8.1 -r8 option

[Jorissen Kevin]

Thanks for fixing the wrong use of the arrays.
 
** Personally, I feel it would be simplest to have all of wien in real*8.  But what are memory requirements?  The biggest array in the tetra program is FC, with dimension (number of bands) x (number of inequivalent k-points) x (number of DOS to calculate).
 
Typically, this last number is of the order 10, and very complicated cells with many bands are usually large and don't require too many k-points.
In the very worst case, I would have a system with many bands (say 2000, a reasonable value for NUME), I would need 273 dos cases (for H telnes), and would be using a large full k-mesh (say 1500 points).  Assuming y bytes per real, that means    y x 150 Mb, or 600 Mb for real*4 and 1200 Mb for real*8.
For the normal user, who only wants 10 dos components, that shrinks down to 20 Mb and 40 Mb.
Personally, I have never had such a large case.  But it's possible.
 
** Anyway, I believe Peter's main argument for not switching to real*8 is that the numbers in the qtl-file have only a few digits, and are therefore already accurately described by real*4; and any additional 'accuracy' would be a lie - it would basically boil down to extra zeroes ;-).
 
 
** We could be a little more efficient on memory use if we took mf=mg out of param.inc and replace it by nndos in the allocate statements.  A user who wants only total dos wasts a factor of 20 on memory with current settings (however, this amount of memory is still quite small).
 
 
 
 
Kevin Jorissen
 
EMAT - Electron Microscopy for Materials Science   (http://webhost.ua.ac.be/emat/)
Dept. of Physics
 
UA - Universiteit Antwerpen
Groenenborgerlaan 171
B-2020 Antwerpen
Belgium
 
tel  +32 3 2653249
fax + 32 3 2653257
e-mail kevin.jorissen at ua.ac.be
 

________________________________

Van: wien-admin at zeus.theochem.tuwien.ac.at namens Peter Blaha
Verzonden: zo 31-10-2004 8:56
Aan: wien at zeus.theochem.tuwien.ac.at
Onderwerp: Re: [Wien] Re: DOS Problems and tetra using intel ifc8.1 -r8 option



My suggestions:            Please don't use the -r8 switch.

It is indended for programs written in single precision, which should run
in double precision.
WIEN2k is written (mostly) in double precision, only in a few parts, where
this is not necessary (like in TETRA) it is in real*4.

Nevertheless I made some changes in TETRA, and declare the REAL*4 arrays
also in the subroutines as real*4, not just with "dimension".

> Following up on Gerhard Fechers observations, i have come to the conclusion
> (from debugging with idb, and comparing case.dos* outputs) that the
> following changes to SRC_tetra are both required and sufficient:
> 1)  in tetra.f and arbdos.f, drop the "*4" qualifier from the declarations
>     RNUMB,DENSTY,FERMDEN,EBS,FC,QTL;XQTL,EHELP,DENBR and SNUMB.
> 2)  add a new member function doreallocate_r8_d3() in reallocate.f (a simple
>     clone of doreallocate_r4_d3 from the same file with real*8 instead of *4)
>
> Without these, tetra would segfault in ADOS() for high band numbers
> (due to misallocation of the FC array) and/or print spurious *** or ???
> values (probably ifc's way of saying NaN).
> A context diff is appended for convenience.
>
> Best regards,
> Martin
> --
> Dr. Martin Kroeker            martin at ruby.chemie.uni-freiburg.de
> c/o Prof.Dr. Roehr
> Institut fuer Anorganische und Analytische Chemie der Universitaet Freiburg
>
> _______________________________________________
> 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


-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 8624 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20041105/45963217/attachment.bin