Re: How to combine data of different FLUKA runs

From: Stefan Roesler <sroesler_at_mail.cern.ch>
Date: Thu, 23 Apr 2009 19:33:42 +0200

24581_at_smtp1.mi.infn.it>
  <1239975668.17275.71.camel_at_rvext.aria> <Pine.LNX.4.58.0904191342160.1765=
2_at_lxplus219.cern.ch>
  <1240502424.25525.32.camel_at_rvext.aria> <Pine.LNX.4.58.0904231814170.4927=
@lxplus244.cern.ch>
 <1240504220.25525.46.camel_at_rvext.aria>
MIME-Version: 1.0
Content-Type: text/plain; charset=3D"ISO-8859-1"
Content-Transfer-Encoding: 8BIT
X-CERN-Reverse-DNS: cernfe10.cern.ch
X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.0 (smtp=
1.mi.infn.it [192.84.138.69]); Thu, 23 Apr 2009 19:33:48 +0200 (CEST)
Sender: owner-fluka-discuss_at_mi.infn.it

Hi Roger

copy the input file into runx, load the input into Flair (from the runx=20
subdirectory!), modify the RANDOMIZ start seed and then run through Flair.
You'll have to do this for each of the run. Probably its faster to do it=20
directly with rfluka from the command line.

Cheers
Stefan

On Thu, 23 Apr 2009, Roger H=E4lg wrote:

> Hi Stefan
>=20
> The goal is to run several fluka runs in parallel, in order to use a
> multiprocessor machine in an optimal way. The only difference of the
> runs will be the initial random seed.=20
>=20
> To use the new possibility of recursively combine the data of different
> runs in flair (0.7.6, "data" tab, "type" field), Vasilis recommended to
> start the parallel runs in subfolders of the main directory. This means
> that they can have the same name and with activated option "recursive"
> the data from all runs would be combined in the summary files. This
> would be an easy and clearly arranged way of handling parallel runs.=20
>=20
> So I wonder, how can I start a run in a subfolder?
> Thanks for your help!
>=20
> Greetz
>=20
> Roger
>=20
> On Thu, 2009-04-23 at 18:15 +0200, Stefan Roesler wrote:
> > Hi Roger
> >=20
> > If I understand your file structure correctly then the input should be =
in=20
> > =20
> > ~/flukawork/test
> >=20
> > and not in the subdirectory run1.
> >=20
> > Cheers
> > Stefan
> >=20
> >=20
> >=20
> > On Thu, 23 Apr 2009, Roger H=E4lg wrote:
> >=20
> > > Hi Stefan
> > >=20
> > > Thanks for your hints.
> > > Yes, I did check these files. The problem is, that fort.11, fort.12,
> > > fort.15 and fort.2 are dead links:
> > >=20
> > > ~/flukawork/test/fluka_10140 $ ls -lh
> > > total 0
> > > lrwxrwxrwx 1 rhaelg rhaelg 27 Apr 22 18:08 elasct.bin
> > > -> /usr/local/fluka/elasct.bin
> > > lrwxrwxrwx 1 rhaelg rhaelg 27 Apr 22 18:08 fluodt.dat
> > > -> /usr/local/fluka/fluodt.dat
> > > lrwxrwxrwx 1 rhaelg rhaelg 45 Apr 22 18:08 fort.1
> > > -> /home/rhaelg/flukawork/test/ranneutron_rw3001
> > > lrwxrwxrwx 1 rhaelg rhaelg 18 Apr 22 18:08 fort.11 -> neutron_rw3001.=
out
> > > lrwxrwxrwx 1 rhaelg rhaelg 44 Apr 22 18:08 fort.12
> > > -> /usr/local/fluka/libec_thihecufealw_10t.pemf
> > > lrwxrwxrwx 1 rhaelg rhaelg 18 Apr 22 18:08 fort.15 -> neutron_rw3001.=
err
> > > lrwxrwxrwx 1 rhaelg rhaelg 17 Apr 22 18:08 fort.2 -> ranneutron_rw300=
2
> > > lrwxrwxrwx 1 rhaelg rhaelg 35 Apr 22 18:08 neuxsc.bin
> > > -> /usr/local/fluka/neuxsc-ind_260.bin
> > > lrwxrwxrwx 1 rhaelg rhaelg 28 Apr 22 18:08 nuclear.bin
> > > -> /usr/local/fluka/nuclear.bin
> > > lrwxrwxrwx 1 rhaelg rhaelg 28 Apr 22 18:08 sigmapi.bin
> > > -> /usr/local/fluka/sigmapi.bin
> > > lrwxrwxrwx 1 rhaelg rhaelg 27 Apr 22 18:08 xnloan.dat
> > > -> /usr/local/fluka/xnloan.dat
> > >=20
> > > nohup.out has the following content when just running the run in the
> > > subfolder:
> > >=20
> > > $TARGET_MACHINE =3D Linux
> > > $FLUPRO =3D /usr/local/fluka
> > > $PEMF =3D /usr/local/fluka/libec_thihecufealw_10t.pemf
> > >=20
> > > Initial seed copied from /usr/local/fluka
> > > Running fluka in /home/rhaelg/flukawork/test/fluka_10140cp: cannot st=
at
> > > `run1/neutron_rw3.inp': No such file or directory
> > >=20
> > >=20
> > > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
 Running FLUKA for cycle # 1
> > > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> > > /usr/local/fluka/flutil/rfluka: line 280: neutron_rw3.inp: No such fi=
le
> > > or directory
> > >=20
> > > But the stated file exists:
> > >=20
> > > ~/flukawork/test $ ls -lh run1/
> > > total 8.0K
> > > -rw-r--r-- 1 rhaelg rhaelg 7.4K Apr 22 18:08 neutron_rw3.inp
> > >=20
> > > I guess it is a problem with the paths. Any further suggestions?
> > >=20
> > > Greetz
> > >=20
> > > Roger
> > >=20
> > > On Sun, 2009-04-19 at 13:45 +0200, Stefan Roesler wrote:
> > > > Hi Roger,
> > > >=20
> > > > Did you check the content of fort.11 (the standard FLUKA output fil=
e) and=20
> > > > fort.15 (the error file)? Also see nohup.out which should have been=
=20
> > > > created in ~/flukawork/test and which contains the printouts from r=
fluka.=20
> > > > In case of doubt please send me those files.
> > > >=20
> > > > Cheers
> > > > Stefan
> > > >=20
> > > >=20
> > > > On Fri, 17 Apr 2009, Roger H=E4lg wrote:
> > > >=20
> > > > > Hi Vasilis
> > > > >=20
> > > > > Thank you for your answer. Your second way works fine. But I was =
not
> > > > > able to get your first suggestion to work.=20
> > > > >=20
> > > > > I can duplicate my standard run in the Run Fluka tab. I save the =
input
> > > > > file in an already created subfolder of the standard run. I enter=
 a
> > > > > Number for the random seed and select the executable.=20
> > > > > When I start the run by clicking Run, flair tries to attach to th=
e run.
> > > > > After a few minutes the Status changes to Finished OK, without st=
arting
> > > > > to calculate anything. The standard run in the main directory fin=
ishes
> > > > > successfully. So I guess it is not because of the input file.=20
> > > > >=20
> > > > > I tried to copy some eventually needed files to the subdirectory,
> > > > > without any success.
> > > > >=20
> > > > > I see the following commands in the flair output:
> > > > >=20
> > > > > Cmd: /usr/bin/nohup /usr/local/fluka/flutil/rfluka -e myfluka -M =
3 neutron_rw3=20
> > > > > /usr/bin/nohup: ignoring input and appending output to `nohup.out=
'
> > > > > =20
> > > > > Cmd: /usr/bin/nohup /usr/local/fluka/flutil/rfluka -e myfluka -M =
3 run1/neutron_rw3=20
> > > > > /usr/bin/nohup: ignoring input and appending output to `nohup.out=
'
> > > > >=20
> > > > > A temporary directory is created for the run in the subfolder and=
 it isn't deleted after finishing the run. It contains the following files:
> > > > >=20
> > > > > ~/flukawork/test $ ls fluka_17313
> > > > > elasct.bin fort.1 fort.12 fort.2 nuclear.bin xnloan.dat
> > > > > fluodt.dat fort.11 fort.15 neuxsc.bin sigmapi.bin
> > > > >=20
> > > > > The manually created subdirectory only contains the input file.=
=20
> > > > >=20
> > > > > Any suggestions what I am doing wrong?
> > > > >=20
> > > > > Best Regards
> > > > >=20
> > > > > Roger
> > > > >=20
> > > > > On Tue, 2009-04-07 at 18:10 +0200, Vasilis Vlachoudis wrote:
> > > > > > Hi Roger,
> > > > > >=20
> > > > > > there are two ways in flair.
> > > > > >=20
> > > > > > 1. Create the runs in sub-folders. From version 0.7.6 you have =
this=20
> > > > > > possibility
> > > > > > and then in the Data Process frame you mark the "Recursive"=
=20
> > > > > > checkbox. Then it will
> > > > > > scan all sub-folders for the same pattern.
> > > > > > =20
> > > > > > 2. In the Data frame select the master run and select also the =
detector(s).
> > > > > > You click the "fitler" button right to the file display.
> > > > > > Flair will popup a dialog to enter the pattern to search. It=
 may look=20
> > > > > > a bit confusing in the begging but the most important rules are=
 written=20
> > > > > > in the syntax text box.
> > > > > > Lets say that the auxiliary runs you created them by adding a=
=20
> > > > > > numerical suffix 2,3,... then modify the rule to
> > > > > > +^\I\d\d\d\d_fort.\U$
> > > > > > The first +/- is to include or remove a file,
> > > > > > ^ means that the pattern should match the beginning of the stri=
ng
> > > > > > \I is substituted by flair with the filename of the input file =
(without=20
> > > > > > the .inp)
> > > > > > \d is any digit from 0-9 (I've added an extra one from the defa=
ult three=20
> > > > > > digits corresponding to the cycle added by fluka)
> > > > > > \U is the unit number
> > > > > > $ means the pattern should match also the end of the string
> > > > > >=20
> > > > > > click ok and process
> > > > > >=20
> > > > > > WARNING you can have different patterns for each Usrxx detector=
=2E But if=20
> > > > > > you select several then only from the pattern of the first sele=
cted=20
> > > > > > detector will appear.
> > > > > >=20
> > > > > > Best Regards
> > > > > > Vasilis
> > > > >=20
> > > > >=20
> > > > >=20
> > > > >=20
> > > >=20
> > >=20
> > >=20
> > >=20
> >=20
>=20
>=20
>=20

--=20
___________________________________

Stefan Roesler
CERN, SC/RP
CH-1211 Geneva 23
Switzerland

Phone: +41-22-7679891
Fax: +41-22-7669639
E-mail: Stefan.Roesler_at_cern.ch
Received on Thu Apr 23 2009 - 22:44:08 CEST

This archive was generated by hypermail 2.2.0 : Thu Apr 23 2009 - 22:44:11 CEST