Program stops after Edge refining

Tamidor · Post by **Tamidor** » Thu Feb 22, 2007 10:18 am

hello,
I just downloaded the bader analysis program from:
http://theory.cm.utexas.edu/henkelman/r ... /index.php

the bader_lnx.tar file.

When I run it for vasp CHGCAR output with "bader CHGCAR"
the program does the Edge refining iterations but just seems to stop when no more edge points are reassigned.
It does not exit the program but just seems to do not anything then.

Before I used the program:
bader_lnx_pgf.tar, downloaded from: http://theory.cm.utexas.edu/vtsttools/bader/,
where no refining seems to be done, and I only can choose the 4 different output options but no flags are available. But as I unterstood this version is not good when running non orthogonal systems.

Would you please tell me, what I might have done wrong.

Tamidor · Post by **Tamidor** » Thu Feb 22, 2007 10:43 am

I tried compilinge the src, which seemed to work.
But using this bader file, I get a segmentation fault error when trying to run it on a MgO CHGCAR.

andri · Post by **andri** » Fri Feb 23, 2007 12:33 am

What kind of linux distribution and setup are you using? Perhaps you
can try to download the source code from http://theory.cm.utexas.edu/vtsttools/downloads/ and compile it?

Post by **graeme** » Fri Feb 23, 2007 2:52 am

Does this crash using our posted MgO CHGCAR file?

Can you give us some other details, such as which compiler you are using and which version it is.

Tamidor · Post by **Tamidor** » Mon Feb 26, 2007 12:16 pm

Thank you for your answers.
I also tried the example CHGCAR from your site and the program again crashed.

I compiled it using the Makefile.pgf and I also tried the Makefile.pgf_32.
It is than compiled with pgf90 -O3 -fast -c
I don´t know the exact version of the portland compiler on our machine, but I know that it should be up to date.

vozny · Post by **vozny** » Thu Mar 01, 2007 2:39 am

I've seen such a problem for my file if it was quite big.
I've compiled the newest version with intel fortran and the analysis crashed with segmentation error for a slab of 78 atoms after REASSIGNED POINTS: 0 was achieved in about 10 iterations (that were pseudopotential charges).
The I used about half size of the same system and it worked fine, although after REASSIGNED POINTS: 0 was achieved it was thinking very long, so that I initailly thought that the program just hang, but it went through fine.

Hopefully with core charges things would go better.

For your situation, I would try to compile with -O2 and without -fast.

CarstenRostgaard · Post by **CarstenRostgaard** » Thu Apr 26, 2007 7:58 am

I had the same problem when downloading the precompiled version of the code (v0.22c). The program hangs after having "reassigned points". My cube file is quite small (119x71x71 points = 7.6Mb), as I am justing testing on the H2 molecule.

Downloading the source code, and compiling it using "make- f Makefile.pgf" produced the error "pgf90-Error--tp x64 available only on x86-64 systems"

compiling it using "make -f Makefile.ifort" seemed to work, but now I can't use the option "bader -c voronoi". Trying results in the error
"
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
bader 0805B3D5 Unknown Unknown Unknown
bader 080495C9 Unknown Unknown Unknown
bader 0804951A Unknown Unknown Unknown
libc.so.6 00B2DDE3 Unknown Unknown Unknown
bader 08049451 Unknown Unknown Unknown
"

This is not a real problem as I don't need the voronoi charges anyway.

Post by **graeme** » Mon Apr 30, 2007 6:05 am

Thanks for the comments. I've changed the 32 bit linux binary to one built using ifort instead of pgf.

For a 32 bit machine, you can build the program with the portland group compiler using Makefile.pgf_32. The Makefile.pgf is for 64 bit machines (as that error indicates).

It does seem that we have broken the Voronoi calculation option. We'll get this working again, and post back when it is. But, as you mention, this is a pretty useless calculation, except perhaps as a sanity check.