RE: tracking or geometry problem?

From: Chrysostomos Valderanis <>
Date: Sat, 28 Aug 2010 11:36:26 +0200

Dear Francesco,

Thank you very much for your answer! You remark about the increased
accuracy through the use of the FREE card for the ROT-DEFI was really
Unfortunately I can not implement it correctly. In order to write the
cards in free format I filled all the values which where asumed in fixed
format input. Flair seems to interpret correctly the cards. But whem I
look at the Fluka output file, under the coordinate transformations
section, I see different transformations. The net outcome is that in the
free format version of my input nothing works due to geometry errors.
I guess I did something stupid in the way I write the ROT-DEFI cards in
free format, but I cannot locate it.

Any hint?


From: Francesco Cerutti
Sent: Thu 26/08/2010 18:55
To: Chrysostomos Valderanis
Subject: Re: tracking or geometry problem?

Dear Makis,

> 1. Since only tracking is required in my code and no fluctuations exist (ie
> pencil beam), I though that the boundary crossings of two events are going
> to be *exactly* the same. As you can see from the logfile this is not the
> case. What am I missing?

actually the initial step is randomized, in order to avoid unnatural step
effects for transport in a material. We are now going to remove this
randomization for transport in vacuum. Anyway, you can already get quite
identical trajectories by setting the tracking precision through the
MGNFIELD card (WHAT(1-2-3)), and the STEPSIZE card too. See
(from slide 29 onwards)

> 2.The proton is not crossing all the boundaries I was hoping to. As you can
> see from the logfile there is a region succesion that is aborted at line
> 265. the proton is exiting a region I was hoping to enter first. The net
> result of this is that the proton is not seeing the full magnetic length and
> gets a wrong direction. I checked with the geometry and didn't find anything
> unusual. Is there anything else to check?

There are in fact some problems in your geometry. In particular, the Q1TR
transformations do not bring the Q1En container exactly upon the QSLQ0En
prototype. In order to do that, the z-coordinate of the Q1En vertex should
be 3993.96549586196, and not 3994.012495861956, being an accuracy worse
than 1e-5 totally unacceptable (indeed you need much better). With this
correction, you will see the same problem occurring later along your
magnet string, for analogous reasons.
Otherwise, note that you can increase the accuracy of the transformation
definitions by using the free format (only) for the ROT-DEFI cards (for this
purpose, insert a FREE card and a FIXED card just before and after -
respectively - the ROT-DEFI card list).



Received on Sat Aug 28 2010 - 15:30:18 CEST

This archive was generated by hypermail 2.2.0 : Sat Aug 28 2010 - 15:30:18 CEST