[Nauty] Signal 40

[igor]

Thank you sincerely for the help.

I've tried to do my homework, and now I also see that it is the
message from the operating system.  Our local (Slovene) version of
Fedora Linux writes "./a.out signal 40" while my supervisors 
Debian gives even more signals (32,60) on really small graphs
(I'm talking about the automorphisms which fix the first vertex).
But in all cases NAUTY's output are correct automorphisms of my 
graphs.  

Unfortunately, my knowledge of Linux is very weak, so I have not
been able to experiment with *limit commands.  Also nobody at our
local help list had any idea how to use these commands. :(  

So my (newbie) bypass of the problem is a procedure which checks 
if NAUTY's output is correct, and since it is I'm now happy
while unable to check could it be even better.  So I'm writing
just to say thanks or to report more, if you find the problem
interesting for the general applications.

Best regards, igor



> Hi Igor,
> 
> Nauty is not capable of writing a message like "signal 40".  It looks
> more like a message from the operating system, but I don't know what
> it is.  The assignment of signal numbers to events differs from OS
> to OS.  What OS are you using?
> 
> One thing to try (assuming you are on a Unix-like system) is to run
> "limit" and try increasing the limits you see there.
> 
> Anyone else have an idea on this?
> 
> Brendan.
> 
> 
> * igor <igor.mat at uni-mb.si> [050123 07:05]:
> > I'm using NAUTY to compute strong bounds on the chromatic number of
> > a vertex transitive graph.  The list of automorphisms of the graph
> > produced by NAUTY decreases the number of variables and computation
> > times drastically.  NAUTY works on my machie perfectly on graphs of
> > sizes up to almost 3000 (I've set MAXN in a C program based on 
> > nautyex.c to this value), and finds a lot of automorphisms.  However,
> > when instructed to return the automorphisms with the first vertex 
> > fixed (colored with the second color while all other vertices are
> > colored with the same color) NAUTY complains 
> > 
> > signal 40
> > 
> > on graphs with size 1024 (while 512 is still OK). The results seem
> > to be OK nevertheless.
> > 
> > Should I worry about this message?  Is there anything what could
> > improve performance?  (NAUTY is much much faster, and requires much
> > much less space then the rest of my code, so I don't mind, if it runs
> > say 100 times slower.)
> > 
> > Best regards,
> >                  igor
> > 
> > -- 
> > Igor Dukanovic
> > EPF Maribor
> > Slovenia/Europe
>