From: Giuseppe Battistoni (Giuseppe.Battistoni@mi.infn.it)
Date: Sat Feb 10 2007 - 16:56:08 CET
Hi Joseph
First of all there is the question of gdb.
Which linux version are you running? Let us know, probably there can be
some patch to be inserted in order to let gdb work (there is a well known
problem in the interaction between gdb, g77 and many kernel versions)
I say this because it should be of interest for many users.
Second: please change all your rand() function call to the use of
flrndm(xdummy) function (the call to the random number of fluka). A part
from many technical reasons and quality of the pseudo-random algorithm,
in the way you will benefit from the possibility of reproducing
histories starting from the last random seed left by FLUKA.
Also this item should be considered by all FLUKA users.
Third; your specific problem is the following. In source.f you define a
C = (y0-R)*S/(A1*x1) + exp(-aa1/x1)
For reasons that I leave to you to understand, at some time
C happens to be <0 and you go into error (Arithmetic exception) at the
line where:
E2 = -x1*log(C)
In particular, I got:
C = -0.00043633548082916418
y0 = 0.5776799338855626
R = 0.9966654
...
Good luck
Giuseppe Battistoni
On Fri, 9 Feb 2007, Joseph Comfort wrote:
> My application crashes after about 515 events. I use a source.f routine
> to generate a beam profile (for a beam line). The coding has been
> checked in a separate code, and is fine. I also use a mgdraw.f routine
> to select specific boundary crossings and write some information to a
> file. I have been doing these procedures for other situations, and
> things have been fine. This is a new application, with a rather more
> complicated geometry than previously. However, from the first 500
> events, I don't see any problems. The geometry debugger does not show
> any errors.
>
> I have tried the code on 3 different machines with different
> architectures and systems, and all produce the same crash. There are no
> messages in the .out, .log, or .err files. I can read the core dump on
> one machine, and it gives
> Core was generated by `/home/comfort/step1/linefluka'.
> Program terminated with signal 6, Aborted.
> Cannot access memory at address 0x40017504
> #0 0xffffe410 in ?? ()
> (gdb) bt
> #0 0xffffe410 in ?? ()
> Cannot access memory at address 0xbfffe974
>
> I'm attaching my step1_n.inp, source.f, and mgdraw_line.f files. The
> error might be mine, but I have not been able to find it. I am not
> entirely certain about some things in the input file, but they are
> comparable to things I have been doing, I believe successfully.
>
> Thank you,
> Joe Comfort
>
This archive was generated by hypermail 2.1.6 : Sat Feb 10 2007 - 17:07:51 CET