Closed fiedl closed 6 years ago
Earlier tests indicated that the number of hits scales with the flasher-pulse-width setting:
Source: https://github.com/fiedl/hole-ice-study/issues/57#issuecomment-384267878
I'm trying the same flasher simulation with different brightness settings.
I'm trying Martin's best parameters with full width and brightness:
[2018-06-01 12:07:18] fiedl@wgs16 /afs/ifh.de/group/amanda/scratch/fiedl/hole-ice-study/scripts/FlasherSimulation
▶ qrsh -l gpu -l tmpdir_size=10G -l h_rss=100G -l s_rt=6:00:00
▶ qrsh -l gpu -l tmpdir_size=10G -l h_rss=50G -l s_rt=6:00:00
[2018-06-01 12:13:48] fiedl@kepler20 ~
▶ $ICESIM/env-shell.sh
[2018-06-01 12:14:44] fiedl@kepler20 /afs/ifh.de/group/amanda/scratch/fiedl/hole-ice-study/scripts/FlasherSimulation
▶ rm -r tmp
▶ ./run.rb --hole-ice=simulation --width=127 --brightness=127 --effective-scattering-length=0.05 --absorption-length=100 --hole-ice-radius-in-dom-radii=0.5
This worked. Plot it:
[2018-06-01 14:50:43] fiedl@fiedl-mbp ~/hole-ice-study/results/flasher_pulse_widths master ⚡
▶ python ~/hole-ice-study/scripts/lib/plot_flasher_hits_vs_simulation.py ~/hole-ice-study/scripts/FlasherSimulation/results/Run-2018-pei9Vaqu
Source: https://github.com/fiedl/hole-ice-study/issues/59#issuecomment-393673790
I'm repeating the same simulation with brightness = 24:
[2018-06-01 14:57:25] fiedl@wgs16 ~
▶ qrsh -l gpu -l tmpdir_size=10G -l h_rss=50G -l s_rt=6:00:00
[2018-06-01 14:59:54] fiedl@kepler22 ~
▶ $ICESIM/env-shell.sh
[2018-06-01 15:00:40] fiedl@kepler22 /afs/ifh.de/group/amanda/scratch/fiedl/hole-ice-study/scripts/FlasherSimulation
▶ ./run.rb --hole-ice=simulation --width=127 --brightness=24 --effective-scattering-length=0.05 --absorption-length=100 --hole-ice-radius-in-dom-radii=0.5
[2018-06-01 15:40:30] fiedl@fiedl-mbp ~/hole-ice-study/scripts/lib master ⚡
▶ python ~/hole-ice-study/scripts/lib/plot_flasher_hits_vs_simulation.py ~/hole-ice-study/scripts/FlasherSimulation/results/Run-2018-iep7guuC
# I3CLSimMakeHits
:param UnWeightedPhotons:
Enabling this setting will disable all optimizations. These
are currently a DOM oversize factor of 5 (with the appropriate
timing correction) and a biased initial photon spectrum that
includes the DOM spectral acceptance. Enabling this setting
essentially means that all photons that would be generated
in the real detector *will* actually be generated. This will siginificatly
slow down the simulation, but the optional ``PhotonSeries``
will contain an unweighted sample of photons that arrive
at your DOMs. This can be useful for DOM acceptance studies.
:param DOMOversizeFactor:
Set the DOM oversize factor. To disable oversizing, set this to 1.
# I3CLSimMakePhotons
:param UnWeightedPhotons:
Enabling this setting will disable all optimizations. These
are currently a DOM oversize factor of 5 (with the appropriate
timing correction) and a biased initial photon spectrum that
includes the DOM spectral acceptance. Enabling this setting
essentially means that all photons that would be generated
in the real detector *will* actually be generated. This will siginificatly
slow down the simulation, but the optional ``PhotonSeries``
will contain an unweighted sample of photons that arrive
at your DOMs. This can be useful for DOM acceptance studies.
:param UnWeightedPhotonsScalingFactor:
If UnWeightedPhotons is turned on, this can be used to scale
down the overall number of photons generated. This should normally
not be touched (it can be used when generating photon paths
for animation). Valid range is a float >0. and <=1.
:param DOMOversizeFactor:
Set the DOM oversize factor. To disable oversizing, set this to 1.
Claudio has confirmed that the brightness and width settings do not scale the number of photons linearly. But one could introduce a scaling where the flashes are converted into photons in I3CLSimLightSourceToStepConverterFlasher
.
From the number of generated photons onwards it’s all linear Somewhere in the flasher parameterization it will decide on a mean number of photons You can scale that if you want to
Alternatively, I've introduced a thinning mechanism in #85.
In a scenario where timing information is ignored and only the number of hits in each dom is used, calculation time can be saved by simulating less photons:
(a) Use a thinning parameter for the photon propagation in clsim (b) Use a smaller pulse width setting (c) Use a smaller brightness setting
(b) and (c) are only possible if the total number of photons and the number of hits scale with these settings, i.e. if increasing the brightness from 63.5 to 127 also doubles the number of photons, et cetera.