RE: Fluka Abort Problem

From: Chris Theis <Christian.Theis_at_cern.ch>
Date: Mon, 29 Sep 2008 10:56:38 +0200

Dear Jim,

I quickly looked at your input and there are a couple of things that
came to my attention. The first point is that you are using a number of
cylinders (e.g. envelopo, envelopi, ringo, ringi etc.) which share
coplanar base-planes. In general this is a situation which should be
avoided because it might easily lead to numerical problems and
ambiguities that cannot always be resolved by numerical CSG algorithms.
I'd suggest that you change this in your input to avoid any problem
there.

Furthermore, you're making quite extensive use of the OR operator in
connection with overlapping bodies/sub-regions and parentheses. For
example:

rAnodSlv 5 + ( +AndSlvO1 | +AndSlvO2 | +AndSlvO3 ) - ( +AndSlvI1
| +AndSlvI2 | +AndSlvI3 )

Even though your usage is mathematically legitimate and correct there
are some subtle issues connected to this with respect to tracking. In
addition you should keep in mind that FLUKA's CSG kernel internally
requires the rewrite of regions described with parentheses as it can
only deal with a so called "normalized CSG tree" that does not contain
nested structures. Depending on the case the normalized region
description might become significantly more complex and demanding also
in terms of memory and efficiency. I'm not really sure but I think you
might find the final normalized form in the output file.
In case you still have geometry problems reported by FLUKA, I would try
to change these regions with the union and the parentheses.

> I noticed that other people have experienced the same problem. I used
> SimpleGeo to build the geometry and also used it check for any voids
> and overlaps.

In general one should not forget that a geometry in which a debugger did
not report any errors is not necessarily void of them! The debugger
(being it FLUKA's or SimpleGeo's visual debugger) tests whether certain
points are defined without ambiguities. In case you use a regular grid
of points it is obvious that, depending on the specification of your
grid, some places will never be sampled and errors in those areas cannot
be found by definition. Thus, you can either use several iterations of
different grid parameters or select one of SimpleGeo's QMC approaches
(e.g. low-discrepancy), which overcome the problem of undersampling.=20

Hope that helps
Chris

------------------------------------------------------------------------
Chris Theis
CERN/SC-RP - European Organization for Nuclear Research
1211 Geneva 23, Switzerland
e-mail: Christian.Theis@cern.ch www: http://www.cern.ch/theis
------------------------------------------------------------------------

> -----Original Message-----
> From: owner-fluka-discuss_at_mi.infn.it [mailto:owner-fluka-
> discuss_at_mi.infn.it] On Behalf Of segalaj_at_egr.uri.edu
> Sent: 25 September 2008 18:33
> To: fluka-discuss_at_fluka.org
> Subject: Fluka Abort Problem
>=20
> I am getting the following error when I run the attached input file:
>=20
> Abort called from G1 reason 3RD: NRLTGG > 2000 Run Stopped!
> STOP 3RD: NRLTGG > 2000
>=20
> I noticed that other people have experienced the same problem. I used
> SimpleGeo to build the geometry and also used it check for any voids
> and overlaps.
>=20
> I also attached the SimpleGeo file.
>=20
> Can you please take a look at this and let me know what I am doing
> wrong.
>=20
> Thanks,
> Jim Segala
> University of Rhode Island Physics Department
Received on Mon Sep 29 2008 - 12:18:54 CEST

This archive was generated by hypermail 2.2.0 : Wed Oct 01 2008 - 08:36:35 CEST