[Sbs_software] SBS Simulation Meeting Wednesday

[Andrew Puckett]

Hi Seamus,

I'll give a brief update at the simulation meeting tomorrow on some 
improvements I've made/am making to the "under-the-hood" machinery of 
g4sbs:

1) Option to turn on rejection sampling for the built-in event 
generators, which produces events distributed according to the 
calculated differential cross section; the code generates a 
user-specified number of "pre-events" thrown flat in phase space, that 
aren't tracked through the setup, in order to determine the maximum 
event weight within the user-defined generation limits. This process is 
fast, and leads to reasonably good efficiency for the rejection 
sampling, provided the differential cross section does not vary by too 
many orders of magnitude within the event generation limits. Then it 
uses this maximum event weight for a rejection sampling to produce 
events distributed according to the model cross section. This option is 
useful in particular for physics analyses where one wishes to avoid the 
complications of estimating statistical uncertainties for weighted Monte 
Carlo statistics. This modification is finished, tested, and pushed to 
the uconn_dev branch of the github repo.

2) Modifying the sensitive detector and hit classes to define a hit time 
window and energy threshold for "calorimeter" type; i.e., showering 
detectors (or scintillators when we don't want to turn on the optical 
photon machinery). This modification improves on the existing machinery 
in that the existing machinery only allows for one "hit" per detector 
channel per simulated event; i.e., all energy deposition in a given 
sensitive volume in a given simulated event is summed into a single 
"hit", regardless of the timing of those energy depositions. The "hit 
timing" calculated using this machinery would sometimes be distorted by 
averaging in low-energy secondaries that are significantly delayed 
relative to the primary particle hits.

The new machinery allows more than one hit per sensitive detector 
channel per simulated event by defining a "hit timing window" analogous 
to a gate for charge integration in an ADC and/or a sampling window for 
a Flash ADC type digitization. Starting with the time of the earliest 
energy deposition in a channel, all energy depositions arriving within a 
user-configurable timing window of the "hit start time" will be 
accumulated in the first hit, and any energy depositions occurring later 
than the hit start time plus the gate width will start a new hit in the 
same detector channel. For now I do not consider pulse shape/pileup, but 
I simply assume that the hit timing window will be chosen suitably wide 
to encompass >~ 99% of the primary signal. Note that the simulation only 
includes the time dependence of the energy deposition itself, and does 
not account for the pulse shaping effects of the PMT multiplication or 
process and/or amplification/shaping by front-end electrons.

This modification gives a more accurate measure of the hit timing and 
time resolution, and the energy resolution, for simulated hits. I also 
now allow the user to define an energy threshold for sensitive detector 
hits on a per-detector basis. Currently, a zero threshold is used, 
meaning that any non-zero energy deposition in a sensitive volume (that 
is not an optical photon detector) produces a sensitive detector hit. 
Allowing the user to define a threshold allows for a more realistic view 
of detector background rates/occupancies/efficiencies (assuming the 
threshold is defined in a sensible way) and can in principle reduce the 
size on disk of simulation output files. But it could also be somewhat 
redundant with the external digitization libraries and give misleading 
results if sensible values are not used.

I have already finished coding and testing this change for "calorimeter" 
type detectors that measure energy deposition, but I'm still working on 
the appropriate machinery for the other detectors, including "GEMs" and 
"ECALs". I have tried to define sensible default values for each of the 
detectors in the various experiments, and will add the relevant UI 
commands for user customization. Time permitting (I have to catch a 
flight at 1 PM tomorrow), I will show a few slides about my progress on 
these developments.

Best regards,

Andrew


On 1/30/18 9:56 AM, Seamus Riordan wrote:
> Hi everyone,
> We will have our next meeting regarding the SBS simulation and DAQ, Wednesday, at 9AM ET.
>
> You can call in at:
>
> +1-888-240-2560 <tel:%2B1-888-240-2560>
> (US and Canada) and seeinternational numbers <http://bluejeans.com/numbers>
>
> Access number:  9989030149
> https://bluejeans.com/9989030149
>
> TED 2561B has been reserved for those attending locally at the lab.
>
> An agenda can be found here:
>
> http://hallaweb.jlab.org/12GeV/SuperBigBite/sbssimwg/20180131/
>
> Please add anything you want to share to this week's page from the CUE
>
> /group/halla/www/hallaweb/html/12GeV/SuperBigBite/sbssimwg
>
> or email me slides or plots that you have to show before the meeting!
>
> Best,
> Seamus
> -- 
> Seamus Riordansriordan at anl.gov
> Argonne National Laboratory             Office: (630) 252-4101
> Assistant Physicist                     Fax:    (630) 252-3903
>
> Physics Division Building 203           Room C265
> Argonne National Laboratory
> 9700 S. Cass Ave
> Argonne, IL  60439
>
>
> _______________________________________________
> Sbs_software mailing list
> Sbs_software at jlab.org
> https://mailman.jlab.org/mailman/listinfo/sbs_software

-- 
Andrew Puckett
Assistant Professor, Department of Physics, University of Connecticut
2152 Hillside Road, U-3046
Storrs, CT 06269-3046
E-mail: puckett at jlab.org, puckett at phys.uconn.edu
Office phone: 860-486-7137

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/sbs_software/attachments/20180130/7043ba8e/attachment.html>