Re: negative argument in ALOG bug in FEFF825 on CRAY PVP machines

[Matt Newville <newville@cars.uchicago.edu>]

Hi John, Jim, Alex,

>    The negative argument in ALOG  bug appears to have come indirectly
> from negative values of the muffin tin radius rmt(1), which the code
> reads by mistake from the packed ascii intermediate .bin files with the
> subroutine rdpadd. The packed ascii utilities were introduced in FEFF
> by Matt Newville to facilitate portability (instead of using architecture
> specific binary files), and generally work quite well. According to the
> notes in wrpadd they are designed to pack double precision arrays.
>  These utilities are in padlib.f and are found the COMMON subdirectory
> of the FEFF825 distribution.
>
> ...
>
>   I'll cc this to Matt in case he has a comment on this problem.
> He has mentioned already that we should update the padlib.f library
> to the version in his gifeffit codes. Possibly there is a simple
> switch (?).

Hmmm, that's weird.  A sign error bug in pad/unpad??

I think I missed some of the story.  This is running feff on a
CRAY, where "double precision" means real*16?  I'm sure I never
tested pad/unpad on such a system.

I'd probably want to run some tests on this machine.  Is it
possible to get access to it -- if not, can I write some test
programs to get the results from?

Thanks,

--Matt