Re: FLUKA: USRBDX results and running speed

[Alberto Fasso' <Alberto.Fasso@cern.ch>]

Dear Fan,

I had a quick look at your input file and I think I can answer some of
your questions. Maybe I, or somebody else, will find the time to
do a test run and answer also the last question.

> 1) The execution took >110000 s cpu time on a PII-400 linux machine 
> (with 256 MB memory), for 1E5 events. This is many times slower than
> similar problem running on the same machine with MCNPX.

The cpu time can vary a lot depending on the setting you use. You are
using (most likely without real necessity) three options which are
potentially very time-consuming:
a) Card DEFAULTS: you use option ICARUS, which is meant for a very special
   use (calculations for the ICARUS experiment). Option NEW-DEFA is
   recommended for most applications, unless you have a very good reason
   to use something else. Since you are new to FLUKA, I suggest that
   you begin with it as your basic choice (but it may also be instructive
   to run the same problem with some other set of defaults, to get the
   feeling of how they can affect your results and your CPU time) 
b) Card STEPSIZE: this is rarely needed, unless you work with a magnetic
   field. If this is not the case, the built-in defaults of FLUKA can take
   care of optimizing the step size much better than the average user can
   do.
c) Card GEOBEGIN with WHAT(2) = 0.01: the manual says clearly that
   this option is "used to set the accuracy parameter - reserved for
   program development". You should not touch it unless you know well
   what you are doing (and I don't know many persons who do :-)

> 2) What is the meaning of running cycles in FLUKA? the execution always
>    go through  5 cycles no matter how many events you are tring to to
>    simulate, is this right?

Events per cycle and number of cycles are independent of each other and
must be decided separately. Events per cycle are set in input via the
START card. The number of cycles is set via the rfluka script, and
can be any number you wish. Examples:
  Just one cycle: rfluka -N0 -M1 ... myfile
  two cycles:     rfluka -N0 -M2 ... myfile
  5 cycles:       rfluka -N0 -M5 ... myfile
          or just rfluka ... myfile (5 is the default)
  16 cycles:      rfluka -N0 -M16 ... myfile
You can also run additional cycles later. For instance, supposing you have
already done 5 cycles and you want to add 3 more:
                  rfluka -N5 -M8 ... myfile

> 3) To my surprise the output files are not changes from cycle to cycle,
> i.e. wal005_fort.47 is exactly the same as wal001_fort.47. The same for
> the residual nuclei files.

This is because you have commented off the RANDOMIZE card. In this
case the program starts every time from the default random number seeds,
and of course gives exactly the same results every time .

> 4) looking at the neutron spectra in *_fort.47 and fort.48, it seems
>    there are cut-offs at both low and high energy ends (at ~ 3E-7 and 
>    ~0.5-0.6  GeV respectively). Is this right?
>    if so how to change them?

I can't say, because I don't have your fort.47 and fort.48 files. As I
said, maybe I will try to do a test run. But you have set a lower limit
for scoring at 1.E-7 GeV, which is not very far from "~ 3E-7" (if you
take into account that low-energy neutrons can only be scored within
the FLUKA group structure). And also at the higher limit, starting with
1 GeV protons, I would not expect many neutrons  with more than 
"~0.5-0.6 GeV". Maybe it is only a lack of statistics?

A few side remarks on your input;
- Option OUTLEVEL is obsolete and has no practical effect. I am planning
  to take it out of the manual soon. I hope Alfredo Ferrari will also
  take it out of the sample input distributed with the code. 
  Anyway, it does no harm.
- Option LOW-MAT is not needed when the "neutron-material" chosen
  coincides with the default one, as in your case (and in most cases).
  Again, no harm done, but each unnecessary card increases the
  possibilities of mistakes (I have seen a lot made with LOW-MAT!)
- Option START has WHAT(2) misaligned (with which consequences, I 
  cannot guess). Also WHAT(5) is set = 1., which is likely to produce
  tons of output (2 lines per event, times 100000 events requested...)

Best regards,

  Alberto


> Dear Fluka team,
> 
> I am new to FLUKA and I am trying to use it to evaluate the neutron and
> gamma ray fluxes
> at Tungsten and Aluminium targets with proton incidents. I started with a
> simple geometry:
> 1cm thick w spherical shell with a 2 cm Al shell outside. 1 GeV protons were
> fired isotropically
> from the centre of the sphere, and I am recording the neutron flux (current)
> exiting the W shell as well
> as the flux exiting the Al sphere. Please see the attached .inp file for
> more details. Of course i have also created the correponding .pemf file. The
> code excuted ok, but there are a few things I feel are not right
> I am pretty sure they are due to some misuses of the code. I will appreciate
> any help to identify the cause of the problem.  Here are a few of my
> questions:
> 
> 1) The execution took >110000 s cpu time on a PII-400 linux machine (with
> 256 MB memory), for
>     1E5 events. This is many times slower than similar problem running on
> the same machine
>      with MCNPX.
> 2) What is the meaning of running cycles in FLUKA? the execution always go
> through  5 cycles no
>     matter how many events you are tring to to simulate, is this right?
> 3) To my surprise the output files are not changes from cycle to cycle, i.e.
> wal005_fort.47 is exactly the
>     same as wal001_fort.47. The same for the residual nuclei files.
> 4) looking at the neutron spectra in *_fort.47 and fort.48, it seems there
> are cut-offs at both low and high
>     energy ends (at ~ 3E-7 and ~ 0.5 -0.6  GeV respectively). Is this right?
> if so how to change them?
> 
> Regards
> 
> Fan
> 
> -------------------------------------------------------------------
> Dr. Fan Lei
> Space Department, DERA
> G081, A8 Building
> Farnborough, GU14 0LX
> U.K.
> 
> Tel  . 44-1252-393904
> Fax.  44-1252-396330
> email: flei@scs.dera.gov.uk
> 
> 
> 
> The information contained in this e-mail and any subsequent correspondence 
>is private and is intended solely for the intended recipient(s). For those 
>other than the intended recipient any disclosure, copying, distribution, or 
>any action taken or omitted to be taken in reliance on such information is 
>prohibited and may be unlawful.
> 

-- 
Alberto Fassò
CERN-EP/AIP, CH-1211 Geneve 23 (Switzerland)
Phone: (41 22) 767 2398    Fax: (41 22) 767 9480   Alberto.Fasso@cern.ch