[Gate-users] Sudden drop in recorded coincidences

Martín Belzunce martin.sure at gmail.com
Thu Nov 10 01:37:38 CET 2011 

Hi,
I've had similar problems in long simulations. In my case, I was using
several splits (~100) to run the simulation in a cluster, and a few
splits (<5) generated an abrupt change in count rates. When processing
the outputs, I also found that on those splits, the energy spectrum
was quite different from the rest.
I am also using Mersenne Twister random generator, and tried changing
the seed, but the problem remained.

Best regards,
Martín Belzunce



2011/11/8 Geron Bindseil <geron9 at gmail.com>:
> Hello Ida,
>
> I also have a batch perl script that splits a PET simulation into multiple
> processes at the same time. I have noticed that maybe 6-7% of the time,
> there is a root file with significantly less data than normal. More rare
> are root files with significantly more data than normal. I am using the
> Ranlux64 random number generator with auto generation.
>
> My workaround has been to simply rerun the outlier simulation time
> intervals, and this normally has been successful.
>
> Kind Regards,
> Geron
>
> 2011/11/8 Ida Häggström <ida.haggstrom at radfys.umu.se>
>
>> Hello Uwe and thanks for your reply!
>>
>> Yes, I'm using Mersenne Twister generator with an automatic seed. I
>> completely redid the simulations after the "drop boundary" with new seeds
>> (auto again), and got the same results, i.e. again there was much less
>> coincidences (about a factor 2.4 less). Looking in my raw text files (all
>> 40 files) I also noticed that the scattered fraction (on total) is about a
>> factor 1.8 different from before and after the "drop boundary" (~3% before,
>> ~5.3% after). The same goes for random fraction (~0.9% before, ~1.6%
>> after).  I really have no clue what can be causing this as nothing is
>> changed with the setup. As you also mentioned, I've never seen this
>> phenomenon before when I've only done up to 30 min aquisitions, only now
>> for longer ones.
>> /Ida
>>
>> On Tue, Nov 8, 2011 at 11:54 AM, Uwe Pietrzyk <u.pietrzyk at fz-juelich.de
>> >wrote:
>>
>> >
>> > Hi Ida,
>> >
>> > we are having similar problems, that when running long productions,
>> > there is a sudden decrease in rate and length of root file. Actually, the
>> > run stops.
>> > Starting with the seed given in the GATE output, this can be reproduced.
>> > The effect
>> > appears on 32 and 64 bit systems, as we tested on different MAC-Systems.
>> >
>> > Our suspicion was the random number generator and we are now
>> > switching from Ranlux64 to MersenneTwister, but this needs further
>> > investigations. From your attached files I understand that you used
>> > the MersenneTwister (is this correct??), which we hoped would solve
>> > the problem.
>> >
>> > Kind regards,
>> >
>> > Uwe Pietrzyk
>> >
>> > On 08.11.2011, at 11:05, Ida Häggström wrote:
>> >
>> > Hello!
>> >
>> > I'm simulating a static, homogenous cylinder (F-18) for a 0-3600 sec
>> > acquisition time. I've created a bash script that portions the simulation
>> > onto 40 different processors. I've not included physical decay in my
>> > source, so I simply divide the simulation time into 90 sec acquisitions,
>> > i.e. the first processor simulates acquisition between 0-90 sec, the
>> second
>> > 90-180 sec and so on up to 3510-3600 sec (yielding 40 root-files). My
>> > problem is that between the acquisition 2250-2340 sec to the one 2340-
>> 2430
>> > sec, there's a sudden drop in recorded coincidences (the later root-files
>> > are less than half in size)! I use the exact same setup for everything
>> > (camera, phantom, suorce etc), the only thing I change is the start and
>> end
>> > time of the dynamic aquisition (as well as the output file names of
>> > course).
>> >
>> > I've attached an image of the recorded coincidences vs. time for the
>> entire
>> > 0-3600 aquisition where the drop is obvious. I also attached the raw
>> image
>> > text files, and logs for the Gate-run for the two border aquisitions. Any
>> > input is appreciated! Thanks!
>> > Ida
>> >
>> > ----------- Aliases I send to the macro from my bash-script -------
>> > /control/alias StartTime "2250"
>> >       <----- For the other file, this is "2340"
>> > /control/alias EndTime "2340"
>> >      <----- For the other file, this is "2430"
>> > /control/alias RootOutputName "rootOutput_Prompts_t2250-2340" <----- For
>> > the other file, this is "..._t2340-2430"
>> > /control/alias RawOutputName "rawOutput_t2250-2340"                <-----
>> > For the other file, this is "..._t2340-2430"
>> > /control/alias unstableflag "false"
>> > /control/alias activity "1000000"
>> > /control/alias halfz "125"
>> > /control/alias fullz "250"
>> > /control/alias radius "100"
>> >
>> > ----------- Part of main macro
>> > -------------------------------------------------
>> > #  S T A R T   A C Q U I S I T I O N
>> > #
>> > /gate/application/setTimeSlice     1.  s
>> > /gate/application/setTimeStart     {StartTime}.  s
>> > /gate/application/setTimeStop      {EndTime}. s
>> > /gate/application/startDAQ
>> >
>> > ----------- Source macro
>> > --------------------------------------------------------
>> > #     S O U R C E
>> > #
>> > /gate/source/addSource twogamma
>> > /gate/source/twogamma/setActivity {activity} becquerel
>> > /gate/source/twogamma/setType backtoback
>> > /gate/source/twogamma/gps/particle gamma
>> > /gate/source/twogamma/gps/energytype Mono
>> > /gate/source/twogamma/gps/monoenergy 511. keV
>> > /gate/source/twogamma/setForcedUnstableFlag {unstableflag}
>> > /gate/source/twogamma/setForcedHalfLife 6586.2 s #Fluor-18
>> >
>> > /gate/source/twogamma/gps/type Volume
>> > /gate/source/twogamma/gps/shape Cylinder
>> > /gate/source/twogamma/gps/radius {radius} mm
>> > /gate/source/twogamma/gps/halfz {halfz} mm
>> > /gate/source/twogamma/gps/centre 0. 0. 0. mm
>> > /gate/source/twogamma/gps/angtype iso
>> > /gate/source/twogamma/gps/mintheta 0. deg
>> > /gate/source/twogamma/gps/maxtheta 180. deg
>> > /gate/source/twogamma/gps/minphi     0. deg
>> > /gate/source/twogamma/gps/maxphi   360. deg
>> > /gate/source/twogamma/gps/confine NULL #phantom_P
>> >
>> > #/gate/source/twogamma/visualize 5000 red 2
>> > /gate/source/list
>> > <Gate.rar>_______________________________________________
>> > Gate-users mailing list
>> > Gate-users at lists.opengatecollaboration.org
>> > http://lists.opengatecollaboration.org/mailman/listinfo/gate-users
>> >
>> >
>> > /----------------------------------------------------------------------
>> > Prof. Dr. Uwe Pietrzyk
>> > Diplom-Physiker
>> > Institut für Neurowissenschaften und Medizin / INM-4
>> > Forschungszentrum Jülich GmbH
>> > 52425 Jülich
>> > Tel: +49-2461-61 4763
>> > Fax: +49-2461-61 2820 + 8229
>> > Mobil: +49-162-103-4871
>> > eMail: U.Pietrzyk at fz-juelich.de
>> > http://www.fz-juelich.de/inm
>> > &
>> > Bergische Universität Wuppertal
>> > Fachbereich C - Physik / D.08.01
>> > Gaußstr. 20
>> > 42097 Wuppertal
>> > Tel: +49-(0)202-439-3523
>> > Fax: +49-(0)202-439-2811
>> > eMail: Uwe.Pietrzyk at physik.uni-wuppertal.de
>> > http://www.medizinphysik.uni-wuppertal.de
>> > ----------------------------------------------------------------------/
>> >
>> >
>> _______________________________________________
>> Gate-users mailing list
>> Gate-users at lists.opengatecollaboration.org
>> http://lists.opengatecollaboration.org/mailman/listinfo/gate-users
>>
> _______________________________________________
> Gate-users mailing list
> Gate-users at lists.opengatecollaboration.org
> http://lists.opengatecollaboration.org/mailman/listinfo/gate-users
>

More information about the Gate-users mailing list