Difference in Acceptance between G3 and G4

[gleasonc]

I see a large difference in acceptance between G3 and G4 (see pic) in cos theta GJ of the eta. I do not see a big difference as a function of eta pi mass. Both these were created with software built on May 6 2019. Let me know if you want any other information

Do the experts want to point me in a direction to begin tracking issues down?

EtaPiMinusMassAcceptance_Compare.pdf

CosThetaAcceptance_Compare.pdf

[jrstevenjlab]

Thanks Colin. To turn this into an actionable issue for the simulation gurus it would be good to narrow this down by looking at the particles in your reaction in the lab frame.

Could you make 2D plots of acceptance vs p and theta in the lab frame for all your final state particles with your signal MC sample for G3 and G4?

[gleasonc]

Sure, I can do that. I will post pictures when I get them made.

One thing I just remembered but forgot to note is that I think the G3 simulation has random triggers and accidentals, while the G4 does not. I will double check this to make sure I'm comparing apples to apples.

[gleasonc]

Here are the acceptance distributions. Both use Random:recon-2017_01-ver03 for background.

EtaPiMinusMassAcceptance_Compare.pdf CosThetaAcceptance_Compare.pdf Theta_Acceptance.pdf Momentum_Acceptance.pdf

Here are the p vs theta for generated and accepted. The columns are proton, pi+, pi-, then eta. The first image is G3, the second is G4. I will make these for data if needed. Momentum_vs_Theta.pdf

Momentum_vs_ThetaG4.pdf

[jrstevenjlab]

Thanks Colin. Can you remind us which version.xml (or individual tagged versions) you are using for the record?

On May 22, 2019, at 11:53 AM, gleasonc notifications@github.com wrote:

Here are the acceptance distributions. Both use Random:recon-2017_01-ver03 for background.

EtaPiMinusMassAcceptance_Compare.pdf https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_JeffersonLab_HDGeant4_files_3208194_EtaPiMinusMassAcceptance-5FCompare.pdf&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=gpvqnMEX-oWdopyBGjNizlAXrQWSrcCazGVGKVi-GJc&s=vW7oH0vRvHqX2e0ZD0L9WTh7ILcNSSmesODK6MT0_f0&e= CosThetaAcceptance_Compare.pdf https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_JeffersonLab_HDGeant4_files_3208195_CosThetaAcceptance-5FCompare.pdf&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=gpvqnMEX-oWdopyBGjNizlAXrQWSrcCazGVGKVi-GJc&s=Lwz_RpI8EfqgQhiDKxfNtNqOVOdHQXoI8vLWjMfIC-A&e= Theta_Acceptance.pdf https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_JeffersonLab_HDGeant4_files_3208196_Theta-5FAcceptance.pdf&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=gpvqnMEX-oWdopyBGjNizlAXrQWSrcCazGVGKVi-GJc&s=K7b9Drdh5gQaRJ3An8KfQzVaKwz6zjT6ta9I6IQU3gc&e= Momentum_Acceptance.pdf https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_JeffersonLab_HDGeant4_files_3208197_Momentum-5FAcceptance.pdf&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=gpvqnMEX-oWdopyBGjNizlAXrQWSrcCazGVGKVi-GJc&s=InJnxoV3VJsGa-DvPe22YA6ohpUEuWyv1ZP9RrLgpxk&e= Here are the p vs theta for generated and accepted. The columns are proton, pi+, pi-, then eta. The first image is G3, the second is G4. I will make these for data if needed. Momentum_vs_Theta.pdf https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_JeffersonLab_HDGeant4_files_3208205_Momentum-5Fvs-5FTheta.pdf&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=gpvqnMEX-oWdopyBGjNizlAXrQWSrcCazGVGKVi-GJc&s=fxEyKJVjLo8l_g-opcOORugt2gew7k8fg8lIwiog5sc&e= Momentum_vs_ThetaG4.pdf https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_JeffersonLab_HDGeant4_files_3208208_Momentum-5Fvs-5FThetaG4.pdf&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=gpvqnMEX-oWdopyBGjNizlAXrQWSrcCazGVGKVi-GJc&s=iAnixr-iA3EXsRah5i10gr7xdtGllNpw0LQM_PHBE2E&e= — You are receiving this because you commented. Reply to this email directly, view it on GitHub https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_JeffersonLab_HDGeant4_issues_111-3Femail-5Fsource-3Dnotifications-26email-5Ftoken-3DADFBEBUXOZXJM5K265WJPMLPWVUARA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODV7QGSA-23issuecomment-2D494863176&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=gpvqnMEX-oWdopyBGjNizlAXrQWSrcCazGVGKVi-GJc&s=46QEWvcdW8CJTJBXR2PWo_5GlS5bkzqFPf_RQa8q1Nc&e=, or mute the thread https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_ADFBEBUGMPX4UTSPAPEXEADPWVUARANCNFSM4HOOD3QA&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=gpvqnMEX-oWdopyBGjNizlAXrQWSrcCazGVGKVi-GJc&s=2Xub3qQFoopHuyoPMmEHhlb7OgI-rYWItWm6fSVrDPo&e=.

[gleasonc]

I am using a build I created on April 23, 2019 using $BUILD_SCRIPTS/my_halld_build_jlab. I'm not sure what exact version of software this builds. The xml file is located at /w/halld-scifs17exp/home/gleasonc/Software/April23_2019/version.xml.

In hindsight, I probably should have used the same reconstruction version between data and MC.

[gleasonc]

Justin and others,

Here are the thrown and reconstructed gamma's from G3 and G4. I scaled the G4 plots to the maximum of the G3 plots for better comparison (not sure if the max or the integral is the better way to do this). There is a small difference in the reconstructed G4 momentum below 2 GeV. There is a large difference in theta between G3 and G4 near the transition region of the BCAL and FCAL.

PvsTheta_Gamma.pdf

ReconnedThrownGamma.pdf

[gleasonc]

Here's things a bit more zoomed it. ReconnedThrownGamma.pdf

[mashephe]

Colin,

This makes sense.

I remember from some very early MC studies (10+ years ago) that the transition region between the BCAL introduced some dramatic variations in acceptances in the GJ frame for this channel (so much so that at one point we even considered, for about ten minutes, a "gap calorimeter".... but that wouldn't have been so useful). This underscores in the importance of fiducial cuts in this region.

Matt

On May 23, 2019, at 10:36 AM, gleasonc notifications@github.com wrote:

Justin and others,

Here are the thrown and reconstructed gamma's from G3 and G4. I scaled the G4 plots to the maximum of the G3 plots for better comparison (not sure if the max or the integral is the better way to do this). There is a small difference in the reconstructed G4 momentum below 2 GeV. There is a large difference in theta between G3 and G4 near the transition region of the BCAL and FCAL.

PvsTheta_Gamma.pdf

ReconnedThrownGamma.pdf

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

[gleasonc]

Alex and others, here are the charged particles. There's quite noticeable differences between the pi- reconstruction. The proton and pi+ have differences, but are overall quite similar to each other. ReconnedThrownPiMinus.pdf ReconnedThrownPiPlus.pdf ReconnedThrownProton.pdf PvsTheta_Gamma.pdf PvsTheta_PiMinus.pdf PvsTheta_PiPlus.pdf PvsTheta_Proton.pdf

[gleasonc]

And then here's the reconstructed eta... ReconnedThrownEta.pdf

[jrstevenjlab]

Thanks Colin, just to clarify these are all now plots from monitoring_hists before you do your event selection, right?

I agree presence of the gap between FCAL/BCAL makes sense, but the difference between G3 and G4 is still a little surprising to me. In the monitoring_hists plugin there are also some plots comparing generated and reconstructed values in the Hist_GenReconTrackComparison folder. Could you see if there are any places in those (like DeltaPOverPVsTheta) where the photons appear to be systematically mis-reconstructed in G3 vs G4?

The fiducial cuts will certainly help with this particular discrepancy, but it would be nice to find out if there are other discrepancies we should be concerned with in the simulation.

-Justin

On May 23, 2019, at 10:54 AM, Matthew Shepherd notifications@github.com wrote:

Colin,

This makes sense.

I remember from some very early MC studies (10+ years ago) that the transition region between the BCAL introduced some dramatic variations in acceptances in the GJ frame for this channel (so much so that at one point we even considered, for about ten minutes, a "gap calorimeter".... but that wouldn't have been so useful). This underscores in the importance of fiducial cuts in this region.

Matt

On May 23, 2019, at 10:36 AM, gleasonc notifications@github.com wrote:

Justin and others,

Here are the thrown and reconstructed gamma's from G3 and G4. I scaled the G4 plots to the maximum of the G3 plots for better comparison (not sure if the max or the integral is the better way to do this). There is a small difference in the reconstructed G4 momentum below 2 GeV. There is a large difference in theta between G3 and G4 near the transition region of the BCAL and FCAL.

PvsTheta_Gamma.pdf

ReconnedThrownGamma.pdf

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_JeffersonLab_HDGeant4_issues_111-3Femail-5Fsource-3Dnotifications-26email-5Ftoken-3DADFBEBQZPVZY4E3LIO4KEPTPW2V3HA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWCPSCI-23issuecomment-2D495253769&d=DwMFaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=dNTuqOpjkQbBKzu8uISHoJND9FgvsF5lwgm8s_VV-R0&s=JentfWFFK7f-DgrilKuDJSUnansdW_6tPI1EwhtwX-8&e=, or mute the thread https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_ADFBEBRINYGHVAJLLR3T5D3PW2V3HANCNFSM4HOOD3QA&d=DwMFaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=dNTuqOpjkQbBKzu8uISHoJND9FgvsF5lwgm8s_VV-R0&s=cSF7YPipiLtQZX8uvjJBLpDuWXJJ9GOoBmwjlLx1ADg&e=.

[gleasonc]

just to clarify these are all now plots from monitoring_hists before you do your event selection, right?

That's correct. These are directly out of monitoring_hists prior to any event selection.

I also was not surprised to see a large difference at the transition region. In the acceptance plots I shared earlier, I was cutting between 11 and 12.5 degrees, so some (not much) of this would have gotten cut. I will look into those and see if anything pops up.

[sdobbs]

In his fiducial cut studies, Mark Dalton has shown several differences between G3 and G4 simulations in the transition region. So there is still more understanding needed.

One place to start is on page 14 of https://halldweb.jlab.org/DocDB/0039/003949/002/presentation_fiducialcut_v2.pdf

---Sean

On Thu, May 23, 2019 at 11:38 AM gleasonc notifications@github.com<mailto:notifications@github.com> wrote:

just to clarify these are all now plots from monitoring_hists before you do your event selection, right?

That's correct. These are directly out of monitoring_hists prior to any event selection.

I also was not surprised to see a large difference at the transition region. In the acceptance plots I shared earlier, I was cutting between 11 and 12.5 degrees, so some (not much) of this would have gotten cut. I will look into those and see if anything pops up.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/JeffersonLab/HDGeant4/issues/111?email_source=notifications&email_token=AAJAS2QMLIAFT6WVKRMZ7U3PW226PA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWCT5BA#issuecomment-495271556, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AAJAS2WAAT7BPFW6YTOU3ILPW226PANCNFSM4HOOD3QA.

[jonzarling]

To add to the evidence of differences between geant3 and 4, I looked at this a teensy bit during my photon gun study for FCAL efficiencies too. You can definitely see a difference between G3 and G4 near the beamline. FCAL/BCAL region is a little harder to draw conclusions for. But yes, appears there are real differences between packages at edges.

See page 11 of https://halldweb.jlab.org/DocDB/0040/004025/005/fcal_efficiency.pdf.

Jon

On Thu, May 23, 2019 at 11:48 AM Sean Dobbs notifications@github.com wrote:

In his fiducial cut studies, Mark Dalton has shown several differences between G3 and G4 simulations in the transition region. So there is still more understanding needed.

One place to start is on page 14 of https://halldweb.jlab.org/DocDB/0039/003949/002/presentation_fiducialcut_v2.pdf

---Sean

On Thu, May 23, 2019 at 11:38 AM gleasonc <notifications@github.com mailto:notifications@github.com> wrote:

just to clarify these are all now plots from monitoring_hists before you do your event selection, right?

That's correct. These are directly out of monitoring_hists prior to any event selection.

I also was not surprised to see a large difference at the transition region. In the acceptance plots I shared earlier, I was cutting between 11 and 12.5 degrees, so some (not much) of this would have gotten cut. I will look into those and see if anything pops up.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub< https://github.com/JeffersonLab/HDGeant4/issues/111?email_source=notifications&email_token=AAJAS2QMLIAFT6WVKRMZ7U3PW226PA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWCT5BA#issuecomment-495271556>, or mute the thread< https://github.com/notifications/unsubscribe-auth/AAJAS2WAAT7BPFW6YTOU3ILPW226PANCNFSM4HOOD3QA>.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/JeffersonLab/HDGeant4/issues/111?email_source=notifications&email_token=AEC6Z4BZUNDJVA2BMRC3YJDPW24GJA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWCU6YA#issuecomment-495275872, or mute the thread https://github.com/notifications/unsubscribe-auth/AEC6Z4CSCNMKFFGSPJFMTYDPW24GJANCNFSM4HOOD3QA .

[zisis0]

Hi Colin:

what do you think about gathering all these plots somewhere? For example, a logbook entry, or dedicated Wiki page or DocDB document, as long as they are accompanied by some text (eg what you wrote in your emails). It would be easier (than email) to find them in the future.

Cheers, Zisis...

On May 23, 2019, at 10:23 AM, Jonathan Zarling notifications@github.com wrote:

To add to the evidence of differences between geant3 and 4, I looked at this a teensy bit during my photon gun study for FCAL efficiencies too. You can definitely see a difference between G3 and G4 near the beamline. FCAL/BCAL region is a little harder to draw conclusions for. But yes, appears there are real between packages at edges.

See page 11 of https://halldweb.jlab.org/DocDB/0040/004025/005/fcal_efficiency.pdf.

Jon

On Thu, May 23, 2019 at 11:48 AM Sean Dobbs notifications@github.com wrote:

In his fiducial cut studies, Mark Dalton has shown several differences between G3 and G4 simulations in the transition region. So there is still more understanding needed.

One place to start is on page 14 of https://halldweb.jlab.org/DocDB/0039/003949/002/presentation_fiducialcut_v2.pdf

---Sean

On Thu, May 23, 2019 at 11:38 AM gleasonc <notifications@github.com mailto:notifications@github.com> wrote:

just to clarify these are all now plots from monitoring_hists before you do your event selection, right?

That's correct. These are directly out of monitoring_hists prior to any event selection.

I also was not surprised to see a large difference at the transition region. In the acceptance plots I shared earlier, I was cutting between 11 and 12.5 degrees, so some (not much) of this would have gotten cut. I will look into those and see if anything pops up.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub< https://github.com/JeffersonLab/HDGeant4/issues/111?email_source=notifications&email_token=AAJAS2QMLIAFT6WVKRMZ7U3PW226PA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWCT5BA#issuecomment-495271556>, or mute the thread< https://github.com/notifications/unsubscribe-auth/AAJAS2WAAT7BPFW6YTOU3ILPW226PANCNFSM4HOOD3QA>.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/JeffersonLab/HDGeant4/issues/111?email_source=notifications&email_token=AEC6Z4BZUNDJVA2BMRC3YJDPW24GJA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWCU6YA#issuecomment-495275872, or mute the thread https://github.com/notifications/unsubscribe-auth/AEC6Z4CSCNMKFFGSPJFMTYDPW24GJANCNFSM4HOOD3QA .

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/JeffersonLab/HDGeant4/issues/111?email_source=notifications&email_token=ADANEOR3C2RN2AY2OD7CEVLPW3AHZA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWCYDKY#issuecomment-495288747, or mute the thread https://github.com/notifications/unsubscribe-auth/ADANEOV2QFSRKPLEUMKQ2JDPW3AHZANCNFSM4HOOD3QA.

Dr. Zisis Papandreou Professor of Physics, Ph.D., P.Phys.

Department of Physics
University of Regina
3737 Wascana Parkway Regina, SK S4S 0A2 CANADA

Phone: (306) 585-5379 Fax: (306) 585-5659 Email: zisis@uregina.ca mailto:zisis@uregina.ca Website: http://www.uregina.ca/science/physics/people/faculty-research/zisis-papandreou/index.html http://www.uregina.ca/science/physics/people/faculty-research/zisis-papandreou/index.html

[gleasonc]

@zisis0 I can definitely do this. I will hopefully get this done sometime tomorrow.

[sdobbs]

Hi all,

I think archiving some of these results in a different form is useful, but please do not forget that this discussion is archived on the web: https://github.com/JeffersonLab/HDGeant4/issues/111

This GitHub issue mechanism has been decided to be the preferred forum for discussing these differences.

Cheers, Sean

On Thu, May 23, 2019 at 4:10 PM zisis0 notifications@github.com wrote:

Hi Colin:

what do you think about gathering all these plots somewhere? For example, a logbook entry, or dedicated Wiki page or DocDB document, as long as they are accompanied by some text (eg what you wrote in your emails). It would be easier (than email) to find them in the future.

Cheers, Zisis...

On May 23, 2019, at 10:23 AM, Jonathan Zarling notifications@github.com wrote:

To add to the evidence of differences between geant3 and 4, I looked at this a teensy bit during my photon gun study for FCAL efficiencies too. You can definitely see a difference between G3 and G4 near the beamline. FCAL/BCAL region is a little harder to draw conclusions for. But yes, appears there are real between packages at edges.

See page 11 of https://halldweb.jlab.org/DocDB/0040/004025/005/fcal_efficiency.pdf.

Jon

On Thu, May 23, 2019 at 11:48 AM Sean Dobbs notifications@github.com wrote:

In his fiducial cut studies, Mark Dalton has shown several differences between G3 and G4 simulations in the transition region. So there is still more understanding needed.

One place to start is on page 14 of

https://halldweb.jlab.org/DocDB/0039/003949/002/presentation_fiducialcut_v2.pdf

---Sean

On Thu, May 23, 2019 at 11:38 AM gleasonc <notifications@github.com mailto:notifications@github.com> wrote:

just to clarify these are all now plots from monitoring_hists before you do your event selection, right?

That's correct. These are directly out of monitoring_hists prior to any event selection.

I also was not surprised to see a large difference at the transition region. In the acceptance plots I shared earlier, I was cutting between 11 and 12.5 degrees, so some (not much) of this would have gotten cut. I will look into those and see if anything pops up.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<

https://github.com/JeffersonLab/HDGeant4/issues/111?email_source=notifications&email_token=AAJAS2QMLIAFT6WVKRMZ7U3PW226PA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWCT5BA#issuecomment-495271556 , or mute the thread<

https://github.com/notifications/unsubscribe-auth/AAJAS2WAAT7BPFW6YTOU3ILPW226PANCNFSM4HOOD3QA .

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub < https://github.com/JeffersonLab/HDGeant4/issues/111?email_source=notifications&email_token=AEC6Z4BZUNDJVA2BMRC3YJDPW24GJA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWCU6YA#issuecomment-495275872 , or mute the thread < https://github.com/notifications/unsubscribe-auth/AEC6Z4CSCNMKFFGSPJFMTYDPW24GJANCNFSM4HOOD3QA

.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub < https://github.com/JeffersonLab/HDGeant4/issues/111?email_source=notifications&email_token=ADANEOR3C2RN2AY2OD7CEVLPW3AHZA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWCYDKY#issuecomment-495288747>, or mute the thread < https://github.com/notifications/unsubscribe-auth/ADANEOV2QFSRKPLEUMKQ2JDPW3AHZANCNFSM4HOOD3QA .

Dr. Zisis Papandreou Professor of Physics, Ph.D., P.Phys.

Department of Physics University of Regina 3737 Wascana Parkway Regina, SK S4S 0A2 CANADA

Phone: (306) 585-5379 Fax: (306) 585-5659 Email: zisis@uregina.ca mailto:zisis@uregina.ca Website: http://www.uregina.ca/science/physics/people/faculty-research/zisis-papandreou/index.html < http://www.uregina.ca/science/physics/people/faculty-research/zisis-papandreou/index.html

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/JeffersonLab/HDGeant4/issues/111?email_source=notifications&email_token=AAJAS2WJU425S73BB6GXQPTPW324FA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWDLGSY#issuecomment-495366987, or mute the thread https://github.com/notifications/unsubscribe-auth/AAJAS2VCXSH6EH2EETIQZLDPW324FANCNFSM4HOOD3QA .

[gleasonc]

Here are some of the plots I shared a couple days ago, this time with more G4 MC and no cuts in the DSelector. The only cuts that would appear here come from the analysis library when making the analysis tree.

Acc vs. cos theta GJ: CosThetaAcceptance_Compare.pdf

Acc vs eta pi mass EtaPiMinusMassAcceptance_Compare.pdf

Acc vs momentum in lab frame Momentum_Acceptance.pdf

Acc vs theta in lab frame Theta_Acceptance.pdf

[T-Britton]

Did you by chance swap an index in pi minus and eta? Both lab theta and momentum seem to agree better if you swap them

[gleasonc]

That may be- I'll dig into the code after the holiday, but I don't think these the case. Here's plots from the same DSelector that was used to create the above plots. Both G3 and G4 had files generated from the same DSelector. Here, you can see at this level (with the same code) things appear opposite each other.

G3:

G4:

[markito3]

Yes and yes. My geant3 simulation was crashing with halld_sim 4.9.1. I don't think there's a fix yet, but its on my to do list.

On Mon, Nov 18, 2019 at 4:02 PM Mark Ito marki@jlab.org wrote:

Colin,

Remember hdgeant4 issue #111:
https://github.com/JeffersonLab/HDGeant4/issues/111 ?

Is this still an issue? I would guess so. Could it be repeated with more
modern software?

   -- Mark

[gleasonc]

I have not forgotten about this issue. I am in the process of making more MC, with software version 4.11 acting as the base. I had to use the nightly build from 12/03/2019 as the 4.11 build led to crashes in the Geant 3 MC. I have attached 2 plots comparing the acceptance in cos theta_GJ and eta pi mass. Geant 4 still shows a larger acceptance at cos theta->1 than what we see with Geant 3. This is only 1/3 of the MC that I am generating. I will produce all previous plots when my jobs finish, which will hopefully be by the end of the week.

EtaPiMinusMassAcceptance_Compare.pdf CosThetaAcceptance_Compare.pdf

[gleasonc]

I generated 1M eta pi- Delta++ events using gen_amp. The eta pi angular distribution is flat and no resonances were included. The config file is: /w/halld-scifs17exp/home/gleasonc/Hall-D/Simulations/Amps/TwoPSAngles/Flat/Random_geant4/4.11/EtaPiMinusDeltaPP.cfg.

Output files are located at /volatile/halld/home/gleasonc/pippimeta/Simulation/2017-01_4.11.00 for G4 and _G3 for G3.

Geant3: 93,372 events pass DSelector. Geant4: 91511 events pass DSelector.

I only have analysis library cuts, Chi2/NDF on kinfit < 10, unused energy < 0.1 GeV, and Egamma > 8GeV. There are some other minor cuts (eg calorimeter fiducial, Delta++ selection) that play a minor role. There are currently no cuts on t, but I can limit this range (things do not change much). For the acceptance plots, I have chi2<6, 8<Egamma<9 GeV. The acceptance vs eta pi mass for G3 and G4 have great agreement. There is a big difference, as seen before, in the acceptance as a function of cos theta GJ.

CosThetaAcceptance_Compare.pdf EtaPiMinusMassAcceptance_Compare.pdf

Here are the plots from the monitoring hists plugin: Thrown and reconstructed particles:

ReconnedThrownGamma.pdf ReconnedThrownPiMinus.pdf ReconnedThrownPiPlus.pdf ReconnedThrownProton.pdf

Reconstructed etas: ReconnedThrownEta.pdf

You can see some minor differences between G3 and G4. Probably the largest difference is the increased yield in pi- at small lab momentum for G4.

Momentum vs theta for thrown and reconstructed particles. There are not any major differences.

PvsTheta_Gamma.pdf PvsTheta_PiMinus.pdf PvsTheta_PiPlus.pdf PvsTheta_Proton.pdf

And here are the acceptances as a function of lab theta and momentum. I checked to see if there was a swap between the eta and pi- as Thomas suggested, but I did not see one (I will dive a little deeper into the details later). Momentum_Acceptance.pdf Theta_Acceptance.pdf

[mashephe]

I relayed this to Colin, but wanted to write here so there is a record...

Absent any other information, the reconstructed photon distributions would motivate me to look very closely at the material description in the region between the central detector and the FCAL. A plausible explanation for what Colin sees are more photon conversions in the G3 simulation than the G4 simulation.

Matt

On Dec 9, 2019, at 12:04 PM, gleasonc notifications@github.com wrote:

I generated 1M eta pi- Delta++ events using gen_amp. The eta pi angular distribution is flat and no resonances were included. The config file is: /w/halld-scifs17exp/home/gleasonc/Hall-D/Simulations/Amps/TwoPSAngles/Flat/Random_geant4/4.11/EtaPiMinusDeltaPP.cfg.

Output files are located at /volatile/halld/home/gleasonc/pippimeta/Simulation/2017-01_4.11.00 for G4 and _G3 for G3.

Geant3: 93,372 events pass DSelector. Geant4: 91511 events pass DSelector.

I only have analysis library cuts, Chi2/NDF on kinfit < 10, unused energy < 0.1 GeV, and Egamma > 8GeV. There are some other minor cuts (eg calorimeter fiducial, Delta++ selection) that play a minor role. There are currently no cuts on t, but I can limit this range (things do not change much). For the acceptance plots, I have chi2<6, 8<Egamma<9 GeV. The acceptance vs eta pi mass for G3 and G4 have great agreement. There is a big difference, as seen before, in the acceptance as a function of cos theta GJ.

CosThetaAcceptance_Compare.pdf EtaPiMinusMassAcceptance_Compare.pdf

Here are the plots from the monitoring hists plugin: Thrown and reconstructed particles:

ReconnedThrownGamma.pdf ReconnedThrownPiMinus.pdf ReconnedThrownPiPlus.pdf ReconnedThrownProton.pdf

Reconstructed etas: ReconnedThrownEta.pdf

You can see some minor differences between G3 and G4. Probably the largest difference is the increased yield in pi- at small lab momentum for G4.

Momentum vs theta for thrown and reconstructed particles. There are not any major differences.

PvsTheta_Gamma.pdf PvsTheta_PiMinus.pdf PvsTheta_PiPlus.pdf PvsTheta_Proton.pdf

And here are the acceptances as a function of lab theta and momentum. I checked to see if there was a swap between the eta and pi- as Thomas suggested, but I did not see one (I will dive a little deeper into the details later). Momentum_Acceptance.pdf Theta_Acceptance.pdf

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

[jonzarling]

@mashephe do you mean material differences in the fiducial (i.e. non-edge) FCAL region? I'n not sure that bears itself out in simulation. Here's an efficiency plot that may be relevant from firing a photon gun in MC at 6 degrees theta. At high energies geant3 and geant4 are in good agreement, which I think would rule out any major differences due to material mismatches. Efficiency differences at intermediate energy are interesting though. Run 30300 which I think matches the run period Colin is examining.

[mashephe]

Jon,

You raise a good point. Attached is the plot of Colin's that got my attention. Specifically the lower right, which seems to show a clear G3/G4 difference in the number of photons reconstructed in the FCAL. I suspect other kinematic variables are mildly correlated with photon angle, and this is the reason for the trends that have been observed. (I was trying to create a plausible explanation, and arrived at perhaps the material model because I know a lot of work was done there.)

Indeed, this difference (independent of my hypothesis for the cause) does not agree with your photon gun studies.

Could you and Colin compare notes and be sure you both using exactly the same versions of both G3 and G4 simulations?

Matt

[cid:79DEF3FD-044D-40BB-8ADC-C9BD292574D8@indiana.edu]

Matthew Shepherd, Professor Department of Physics, Indiana University, Swain West 265 727 East Third Street, Bloomington, IN 47405

Office Phone: +1 812 856 5808

On Dec 9, 2019, at 4:16 PM, Jonathan Zarling notifications@github.com<mailto:notifications@github.com> wrote:

@mashephe do you mean material differences in the fiducial (i.e. non-edge) FCAL region? I'n not sure that bears itself out in simulation. Here's an efficiency plot that may be relevant from firing a photon gun in MC at 6 degrees theta. At high energies geant3 and geant4 are in good agreement, which I think would rule out any major differences due to material mismatches. Efficiency differences at intermediate energy are interesting though. Run 30300 which I think matches the run period Colin is examining.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

[jonzarling]

Sure thing! Colin and I will discuss and make sure we're making like comparisons.

[gleasonc]

Okay, so Richard mentioned in the geant meeting today that the material between G3 and G4 is basically identical. At the end of the meeting, it was discussed that other particles in the reaction may be the cause of the difference in acceptance. The thing to check in this is if timing for charged particles agrees between G3 and G4. This is a long standing issue. I will produce these plots soon.

In the mean time, I've corrected the ReconnedThrown plots. I've grabbed the histograms from "Independent/Hist_ThrownParticleKinematics/Photon/Momentum" etc rather than using "pippimeta__B4_M17/Hist_ParticleComboKinematics/". Here is what the plot for the photons looks like. ReconnedThrownGamma.pdf And with detected momenta: Detected.pdf

Reminder, the "ReconnedThrown" distributions plot the reconstructed events with their thrown momenta. Now we see that reconstructed photons with thrown theta in the FCAL get lost when we plot their detected momentum (bottom right of ReconnedThrownGamma.pdf compared to right side of Detected.pdf). There's also more photons detected in the BCAL for G3 when compared to G4.

Here's the other particles. ReconnedThrownPiMinus.pdf ReconnedThrownPiPlus.pdf ReconnedThrownProton.pdf DetectedPiMinus.pdf DetectedPiPlus.pdf DetectedProton.pdf

[rjones30]

Colin, can you remind me, what is the meaning of the plots in the top row? -Richard

On Tue, Dec 17, 2019 at 3:53 PM gleasonc notifications@github.com wrote:

Okay, so Richard mentioned in the geant meeting today that the material between G3 and G4 is basically identical. At the end of the meeting, it was discussed that other particles in the reaction may be the cause of the difference in acceptance. The thing to check in this is if timing for charged particles agrees between G3 and G4. This is a long standing issue. I will produce these plots soon.

In the mean time, I've corrected the ReconnedThrown plots. I've grabbed the histograms from "Independent/Hist_ThrownParticleKinematics/Photon/Momentum" etc rather than using "pippimeta__B4_M17/Hist_ParticleComboKinematics/". Here is what the plot for the photons looks like. ReconnedThrownGamma.pdf https://github.com/JeffersonLab/HDGeant4/files/3975413/ReconnedThrownGamma.pdf And with detected momenta: Detected.pdf https://github.com/JeffersonLab/HDGeant4/files/3975412/Detected.pdf

Reminder, the "ReconnedThrown" distributions plot the reconstructed events with their thrown momenta. Now we see that reconstructed photons with thrown theta in the FCAL get lost when we plot their detected momentum (bottom right of ReconnedThrownGamma.pdf compared to right side of Detected.pdf). There's also more photons detected in the BCAL for G3 when compared to G4.

Here's the other particles. ReconnedThrownPiMinus.pdf https://github.com/JeffersonLab/HDGeant4/files/3975365/ReconnedThrownPiMinus.pdf ReconnedThrownPiPlus.pdf https://github.com/JeffersonLab/HDGeant4/files/3975366/ReconnedThrownPiPlus.pdf ReconnedThrownProton.pdf https://github.com/JeffersonLab/HDGeant4/files/3975367/ReconnedThrownProton.pdf DetectedPiMinus.pdf https://github.com/JeffersonLab/HDGeant4/files/3975368/DetectedPiMinus.pdf DetectedPiPlus.pdf https://github.com/JeffersonLab/HDGeant4/files/3975369/DetectedPiPlus.pdf DetectedProton.pdf https://github.com/JeffersonLab/HDGeant4/files/3975370/DetectedProton.pdf

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/JeffersonLab/HDGeant4/issues/111?email_source=notifications&email_token=AB3YKWAKQOGCETIFI346CMLQZE34FA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEHD5KHI#issuecomment-566744349, or unsubscribe https://github.com/notifications/unsubscribe-auth/AB3YKWDSLVTUOXY7SYL6HOLQZE34FANCNFSM4HOOD3QA .

[gleasonc]

Those are the thrown theta and momenta prior to any reconstruction. I included them to show how the thrown distributions compare between G3 and G4. There is not much we learn from this top row other than the thrown distributions look similar.

[rjones30]

Colin,how can these not be identical? -Richard

On Tue, Dec 17, 2019 at 4:06 PM gleasonc notifications@github.com wrote:

Those are the thrown theta and momenta prior to any reconstruction. I included them to show how the thrown distributions compare between G3 and G4. There is not much we learn from this top row other than the thrown distributions look similar.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/JeffersonLab/HDGeant4/issues/111?email_source=notifications&email_token=AB3YKWG347IWG5QSBN22HBLQZE5OZA5CNFSM4HOOD3QKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEHD6P3A#issuecomment-566749164, or unsubscribe https://github.com/notifications/unsubscribe-auth/AB3YKWD3XL2BZQVXTSLQUNLQZE5OZANCNFSM4HOOD3QA .

[gleasonc]

Good question. I generated 1M events for G3 and 1M for G4 for eta pi- Delta++. Same configuration file for gen_amp. The generated data looks identical. However, the number of entries in the G3 thrown photon is 3.19M compared to 3.08M for G4 (pi+ has 1.28M for G3 and 1.21 for G4). Maybe this difference is in the random triggers that I included? I would think these get implemented in the same way in G3 and G4 but maybe that is not the case?

[aaust]

Random triggers are merged in in mcsmear, after the simulation step. The only difference between the 2 generated samples is the random seed, but this should not change the total number of thrown photons.

[gleasonc]

Here's a follow up to the BCAL timing for pi+. I see that data (only 1 run from 2017) agrees fairly well with G3 (all 2017), but not with G4. The biggest disagreement is in the tails like Alex sees. I should note that each distribution in the right panel of the figure below is scaled to have the maximum match. PiPlus_DeltaTVsP_HistPID_BCAL.pdf

[gleasonc]

Updated timing plots, now using runs from Fall 2018. Regular timing:

Wide timing:

Acceptances regular timing:

Acceptances wide timing:

Comparing acceptances for wide and regular timing cuts with G3:

and G4:

The acceptance agreement gets worst when loosening the timing cuts on the pi+ and pi-. This s due to the fact that opening the timing cuts does not really affect the G4 simulation. Within the statistics, the two distributions are quite similar.

If we go back to one of the original concerns we see there are more photons, and thus etas, detected in the FCAL in G4 when compared to G3.

Using the regular timing:

Photon timing:

[gleasonc]

I updated the plots in the above post that we discussed Tuesday. Working on creating a data file with wide timing cuts.

[gleasonc]

I just updated the distributions with the recent changes to the control.in card. I am using halld_sim and reconn version 4.16. I run into similar issues as Nacer with selecting the proper recon, anallysis, and sim version, so I just select this one with out comparing to data. I think I understand how to do this using the MCWrapper submit form, but I am not generating many events so it doesn't make sense for me to use the osg. Here is 10M events, but I can probably get away with 1M events for these studies. I used 1M events in previous studies.

General comments: 1)G4 now has a (much) larger acceptance than G3 for both cos theta GJ and eta pi- mass. 2) The G4 acceptance is fairly symmetric around cos theta=0 3) The G3 acceptance is a bit lower? I have 10x more statistics than the study from the previous comments, so it may just be statistical. The shape is unchanged. 4) While G4 still has a larger acceptance than G3 for photons in the FCAL, the distributions match better.

I probably missed some things, but this was meant to just be a follow up to the recent developments.

[rjones30]

Hello all,

I any comparisons between G3 and G4, can I request that you run G3 twice, once with HADR=1 and once with HADR=4. Basically G3 has these two options, and it is important in the comparison to see that G3 does not speak with one voice. It has 2. There is no obvious choice of why one is better than the other.

-Richard

On Wed, May 6, 2020 at 2:36 PM gleasonc notifications@github.com wrote:

I just updated the distributions with the recent changes to the control.in card. I am using halld_sim and reconn version 4.16. I run into similar issues as Nacer with selecting the proper recon, anallysis, and sim version, so I just select this one with out comparing to data. I think I understand how to do this using the MCWrapper submit form, but I am not generating many events so it doesn't make sense for me to use the osg. Here is 10M events, but I can probably get away with 1M events for these studies. I used 1M events in previous studies.

[image: EtaPiMinusMassAcceptance_Compare] https://user-images.githubusercontent.com/11827432/81214186-785da400-8fa5-11ea-86dd-2c40738c7480.png [image: CosThetaAcceptance_Compare] https://user-images.githubusercontent.com/11827432/81214219-84e1fc80-8fa5-11ea-802e-2de6674dd03f.png [image: Theta_Acceptance] https://user-images.githubusercontent.com/11827432/81214392-c8d50180-8fa5-11ea-8abb-5e130a7365ab.png [image: GammaTheta] https://user-images.githubusercontent.com/11827432/81214473-e6a26680-8fa5-11ea-8e3a-c5abbe8f2a6d.png

General comments: 1)G4 now has a (much) larger acceptance than G3 for both cos theta GJ and eta pi- mass. 2) The G4 acceptance is fairly symmetric around cos theta=0 3) The G3 acceptance is a bit lower? I have 10x more statistics than the study from the previous comments, so it may just be statistical. The shape is unchanged. 4) While G4 still has a larger acceptance than G3 for photons in the FCAL, the distributions match better.

I probably missed some things, but this was meant to just be a follow up to the recent developments.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/JeffersonLab/HDGeant4/issues/111#issuecomment-624818499, or unsubscribe https://github.com/notifications/unsubscribe-auth/AB3YKWHSQMRGOIFEKAH6N7LRQGU27ANCNFSM4HOOD3QA .

[jrstevenjlab]

Hi Colin,

Thanks for the updated plots. Could you also post the updated pi+/- DeltaT distributions you showed previously in this issue, comparing G3/G4/Data? One of the questions with this cuts card is how big an impact it has on our current timing cuts and this would help answer that question.

-Justin

On May 6, 2020, at 2:36 PM, gleasonc notifications@github.com wrote:

I just updated the distributions with the recent changes to the control.in card. I am using halld_sim and reconn version 4.16. I run into similar issues as Nacer with selecting the proper recon, anallysis, and sim version, so I just select this one with out comparing to data. I think I understand how to do this using the MCWrapper submit form, but I am not generating many events so it doesn't make sense for me to use the osg. Here is 10M events, but I can probably get away with 1M events for these studies. I used 1M events in previous studies.

https://urldefense.proofpoint.com/v2/url?u=https-3A__user-2Dimages.githubusercontent.com_11827432_81214186-2D785da400-2D8fa5-2D11ea-2D86dd-2D2c40738c7480.png&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=Z9aQ3Slara4SdfBhb1WVvdJ7xSQToRDa5Z4wzC_0dgM&s=_YC28F8lrNM4qUgSW2hYVkovRlfD4qma9ZX2BvBwVKI&e= https://urldefense.proofpoint.com/v2/url?u=https-3A__user-2Dimages.githubusercontent.com_11827432_81214219-2D84e1fc80-2D8fa5-2D11ea-2D802e-2D2de6674dd03f.png&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=Z9aQ3Slara4SdfBhb1WVvdJ7xSQToRDa5Z4wzC_0dgM&s=3c1j_0GzrlX8EsfS_V1EQkN-QZiaPxT0TIrvL_Dlsc0&e= https://urldefense.proofpoint.com/v2/url?u=https-3A__user-2Dimages.githubusercontent.com_11827432_81214392-2Dc8d50180-2D8fa5-2D11ea-2D8abb-2D5e130a7365ab.png&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=Z9aQ3Slara4SdfBhb1WVvdJ7xSQToRDa5Z4wzC_0dgM&s=kLK_aDqXmA3MhtiSG61Bjr1Kk1wQQ_1fjfaRqbaJ7Jo&e= https://urldefense.proofpoint.com/v2/url?u=https-3A__user-2Dimages.githubusercontent.com_11827432_81214473-2De6a26680-2D8fa5-2D11ea-2D8e3a-2Dc5abbe8f2a6d.png&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=Z9aQ3Slara4SdfBhb1WVvdJ7xSQToRDa5Z4wzC_0dgM&s=m7qs67fd1FssW8flThBZTfL1z7ygqjGOwKSxHSFPcm8&e= General comments: 1)G4 now has a (much) larger acceptance than G3 for both cos theta GJ and eta pi- mass. 2) The G4 acceptance is fairly symmetric around cos theta=0 3) The G3 acceptance is a bit lower? I have 10x more statistics than the study from the previous comments, so it may just be statistical. The shape is unchanged. 4) While G4 still has a larger acceptance than G3 for photons in the FCAL, the distributions match better.

I probably missed some things, but this was meant to just be a follow up to the recent developments.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_JeffersonLab_HDGeant4_issues_111-23issuecomment-2D624818499&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=Z9aQ3Slara4SdfBhb1WVvdJ7xSQToRDa5Z4wzC_0dgM&s=AHBCouS7BxHiJyBT61pPXVrGqzKvk85i_y0L9cgFCpk&e=, or unsubscribe https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_ADFBEBTQ3YK2Z5W7HG4EVILRQGU3BANCNFSM4HOOD3QA&d=DwMCaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=nTWoLox3qePKkU3jkN4zKWpZiD5wwsYi6X23ArE2v3s&m=Z9aQ3Slara4SdfBhb1WVvdJ7xSQToRDa5Z4wzC_0dgM&s=fi-kGt8KM7LgPQom0Gh4NgM-KiPFygII6WfdpUaUKno&e=.

[gleasonc]

Richard, I am reproducing these simulations with the latest halld_sim for two G3 simulations, HADR 1 and HADR 4, and one with G4. This should be done today.

Here are the timing plots for the pi+ and pi- in the fcal. I have comparisons between G3, G4, and data, but the reconstruction for the simulations is with halld_recon 4.16, so not the same as the 2017 v3 recon launch.

I do not think we learn anything new when comparing to the plots above.

pi- in FCAL for G3, G4, and comparison

pi+ in FCAL for G3, G4, and comparison

pi- in BCAL for G3, G4, and comparison

pi+ in BCAL for G3, G4, and comparison

photon in FCAL for G3, G4, and comparison

photon in FCAL for G3, G4, and comparison

[gleasonc]

The two simulations for G3 have completed with version set 4.19, and I see vastly different acceptances between HADR1 and HADR4. I only get ~9200/1M events with HADR4 after the analysis library, as opposed to ~79400/1M with HADR1. When I run diff on the control.in files, this is what I get:

ifarm1802.jlab.org> diff Gcontrol.in Gcontrol_HADR4.in 300c300 < HADR 1

HADR 4

Did I miss something else in the control.in file to change? Or any ideas on why there are vastly different outputs?

Here is the cos theta GJ acceptance between the two. HADR1 produces the expected similar result from previous studies. Zooming in on the HADR4 acceptance shows that it is flat, except for a drop off going to -1.

[rjones30]

There is nothing else you should change in the control.in file: just HADR 1 changes to HADR 4. It looks like there are cuts in your analysis that are super sensitive to the details of the hadronic model in the simulation. Either that, or there is something in your analysis chain that you changed between the two runs and forgot about. Run numbers exactly the same? ccdb variation? calib date? You tried to insert a "diff" of your two control.in files but the diff output was somehow screwed up by the markup so I cannot tell what the diff is actually saying, apart from the fact that something changed on line 300.

[rjones30]

Naively, I would say that you just have a factor 10 in your simulated sample size between the two. Would that explain it?

[gleasonc]

I only changed HADR 1 to HADR 4. Everything else is the same. There are still 1 M generated events, over all 2017 runs. When I run diff on the Gcontrol.in, the only difference I get is at line 300: HADR 1 HADR 4 Hopefully that formatting is ok. Everything else should be exactly the same. I will keep digging through things- something weird is going on with the monitoring plots for the HADR4 simulation. The nnumber of thrown particles is much less than with HADR1. Maybe I need to try and rerun everything...

[gleasonc]

Thanks for the fruitful discussion during todays afternoon simulation session. Here are the corresponding plots for the 4.19 version set.

Here are the plots from the monitoring histograms that one gets from running the reaction filter. The "thrown" and "reconned thrown" distributions are nearly identical between G4 and G3 with HADR=1. I haven't included the HADR=4 due to the issue with the simulation.

There are large differences between the detected particles. You can see the major difference in detected photons in the FCAL for photons. There are also some major discrepancies for low momenta pi- and pi+ in the BCAL, from 20-60 degrees. I'm guessing these are correlated with each other in some non trivial ways? The proton agreement between G3 and G4 is pretty good though.

[markdalton]

What does the number of charged pions have to do with the BCAL? This seems like tracking to me.

[markito3]

Again here, for the particles to make it to the reconstructed plots, the entire event must have a successful kinematic fit, correct?

[gleasonc]

Here are the acceptances with the HADR 4 card in geant 3. The acceptance with HADR 4 is much larger than the HADR1 and geant 4 simulations. I will post the monitoring plots soon, and I have a list of other things to check.

[rjones30]

Colin's latest result confirms my earlier observation that when everything in the simulation is working as it should, hdgeant4 interpolates somehow between hdgeant(HADR=1) and hdgeant(HADR=4). The only possible difference between the two G3 simulations is the hadronic interaction model, since NOTHING else in the code references the value of this HADR parameter. Hence I would conclude that hadronic interaction modeling is what drives the remaining difference between the two g3 acceptances and what g4 gives.

-Richard J.

On Mon, May 18, 2020 at 3:28 PM gleasonc notifications@github.com wrote:

Here are the acceptances with the HADR 4 card in geant 3. The acceptance with HADR 4 is much larger than the HADR1 and geant 4 simulations. I will post the monitoring plots soon, and I have a list of other things to check. [image: EtaPiMinusMassAcceptance] https://user-images.githubusercontent.com/11827432/82227426-bb007400-98f5-11ea-9671-97788e75e18e.png [image: CosThetaAcceptance] https://user-images.githubusercontent.com/11827432/82227446-c2278200-98f5-11ea-8b68-b5a457dc28d5.png

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/JeffersonLab/HDGeant4/issues/111#issuecomment-630390261, or unsubscribe https://github.com/notifications/unsubscribe-auth/AB3YKWHEWIC577WGHVLE6QLRSGD47ANCNFSM4HOOD3QA .

[rjones30]

This is not to say that the simulation doesn't need work. Maybe there are ways the simulation handles the energy deposition in calorimeters that makes it artificially sensitive to the hadronic modeling. For example, if we were to turn off neutron energy deposition or slow nuclear recoils (alphas, heavier nuclear fragments) and then renormalize the net energy yield to account for the part we dropped, we might become much less sensitive to hadronic models AND get better agreement with real data response as well. What effect do alphas have in the FCal response? Nothing, more or less.

This requires a program where we decompose the total yield from charged pions in the calorimeters into various mechanisms, and see what might be important in generating this sensitivity. -Richard J.

On Mon, May 18, 2020 at 3:42 PM Richard Jones rjones30@gmail.com wrote:

Colin's latest result confirms my earlier observation that when everything in the simulation is working as it should, hdgeant4 interpolates somehow between hdgeant(HADR=1) and hdgeant(HADR=4). The only possible difference between the two G3 simulations is the hadronic interaction model, since NOTHING else in the code references the value of this HADR parameter. Hence I would conclude that hadronic interaction modeling is what drives the remaining difference between the two g3 acceptances and what g4 gives.

-Richard J.

On Mon, May 18, 2020 at 3:28 PM gleasonc notifications@github.com wrote:

Here are the acceptances with the HADR 4 card in geant 3. The acceptance with HADR 4 is much larger than the HADR1 and geant 4 simulations. I will post the monitoring plots soon, and I have a list of other things to check. [image: EtaPiMinusMassAcceptance] https://user-images.githubusercontent.com/11827432/82227426-bb007400-98f5-11ea-9671-97788e75e18e.png [image: CosThetaAcceptance] https://user-images.githubusercontent.com/11827432/82227446-c2278200-98f5-11ea-8b68-b5a457dc28d5.png

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/JeffersonLab/HDGeant4/issues/111#issuecomment-630390261, or unsubscribe https://github.com/notifications/unsubscribe-auth/AB3YKWHEWIC577WGHVLE6QLRSGD47ANCNFSM4HOOD3QA .