sPHENIX HCal discussion

[kistenev AT bnl.gov]

Hi, Dennis, It is not so much of about improvement, it is of extracting information you are looking for from the data. It is

- electron identification assistance;

- isolated hadron studies;

- jets. 

List is short and based upon the same primitives and those are informative only above the certain threshold. Jet algorithm is the foundation to everything, as soon as object passed one interaction length in calorimeter material it (or its showering components) accumulates transverse momenta due to multiple and nuclear scatterings comparable to its longitudinal momenta, the object emittance at the OHCal entrance definitely above anything what we name single tower.  At that stage paying too much attention to intricacies of town inclinations is wasted time. You have tower CG, its eta/phi corrected for vertex position will be sufficient locator. The difference to classical calorimeter configuration - towers are NOT equally spaced. Angle between vectors pointing to tower CG’s vary,  the sPHENIX HCal design projected on a lego plot is not a chess board. In zero approximation the energy in the tower can be assigned to the bin in eta-phi space which includes tower CG - but - nothing forbids  configurations when two distinct towers will contribute to the same eta-phi bin space (iterations).  

It sounds a bit confusing but it all becomes simple if you look on the picture in calorimeter as the planet Earth picture with heights measured in points on non perfect grid. Being smart  and knowing that particle vector crossed two overlapping towers you may add their energies in a nonstandard way - with CG’s computed for the front part in one and rear in the second. This is in the final iteration and probable future. As of today - read good book on image processing and another one on vector manipulation - this field is exploding. 

Edward

On Dec 11, 2019, at 3:41 PM, Perepelitsa, Dennis <dvp AT bnl.gov> wrote:

Hi Edward and John,

Thanks a lot for the detailed suggestions.

I agree this would be interesting to investigate. However, like John, it is not obvious to me that in HI collisions the improvement would be visible given the large backgrounds and shower shape fluctuations / overlaps (it may well be that there is an improvement, it is just not obvious to me).

I’ll think more about what kind of specific studies we could do, and hopefully we can also talk more in private...

Dennis

On Dec 11, 2019, at 7:34 AM, Lajoie, John G [PHYSA] <lajoie AT iastate.edu> wrote:

Hi Dennis,

 

I may be saying the same thing as Edward in my own different coordinate space, but here’s my take on things.

 

The tilt of the plates in the outer HCAL means that the size and shape of the cluster contains information about the entry angle of the particle into the calorimeter if you cluster in eta and phi. To first order the lateral extent of the cluster will be minimized when the vector of the particle entering the HCAL is aligned with the titled vector of the outer HCAL, and will be asymmetric as it deviates from that vector.

 

You could potentially make use of this. If you continue to cluster in eta and phi then you could use a cluster template library to determine the most probable entry angle for a given cluster, which may help in disentangling cluster overlaps. In particular, when you have a large cluster that results from several charged track overlaps, you could use the template library to help you extract the energy in each cluster, using the tracks themselves to seed the process. 

 

I think this approach would be most helpful in HI collisions, where the particle density is high, if you try to implement a particle flow approach. So, rather than make this a bug, make it a feature by using the additional information contained in the cluster shape.

 

Whether or not this works is a question of degree – it may be that hadronic shower fluctuations are so large that this is all moot. However, if that’s really the case, then I think it would be true that there is nothing to be gained by trying to cluster in different coordinates either.  I think it should be pretty straightforward to set up a test, developing a MC template library with some single-particle events and then trying to apply it. It may not work by itself, but it might also point the way to something that does….

 

My $0.02.

 

John

 

From: sPHENIX-HCal-l <sphenix-hcal-l-bounces AT lists.bnl.gov> On Behalf Of Edouard Kistenev
Sent: Tuesday, December 10, 2019 7:27 PM
To: Perepelitsa, Dennis <dvp AT bnl.gov>
Cc: HCal sPHENIX <sphenix-hcal-l AT lists.bnl.gov>
Subject: Re: [Sphenix-hcal-l] HCAL meeting 12/10

 

Hi Dennis, for many years I tried to explain that so called polar coordinate are rather useless - waste of computer time. If you want something really new - why one of your students will not look into the design of the vector processors (chipsets)  customary used to manipulate images in computer games. Most likely the solutions people use are well documented.

As for the tilted HCal - you have no choice but to image calorimeter towers as 3-vectors pointing to centers of mass (gravity) which were all computed by Richie Ruggiero (sure he has tables) weighted by energy measured in a tower. This is an approximation. But every tower may and shell  be used as a jet seed. This jet will have its zero approximation vector (vector sum of all contributors). Use it to draw the line through the central two or more overlapping (radially) towers in Inner and Outer sections, decide on energy sharing along that axis and now build two or more representative “momenta” which can be used to localize shower CG’s in both section. Next comes global CG and leakage correction. This is just canvas of what can be done with calorimeter data viewed as point (local  СG) deposited energies in the need of a lot of manipulating and correcting. This is not exactly “clustering”, you will still need to find solution to enhance the quality of your image (field built of seeds) while conserving energy. 

Not a single operation in this whole analysis chain needs to know about  azymuthal and/or polar angles.

 

Edward

Sent from my iPhone

On Dec 10, 2019, at 4:35 PM, Perepelitsa, Dennis <dvp AT bnl.gov> wrote:

Dear HCal group, 

 

I just wanted to bring to your attention that there has been an effort within the JS and Sims groups to implement a useful clustering algorithm for the I+OHCal.

 

Here is a pointer to today’s Sims meeting: https://indico.bnl.gov/event/5494/ 

 

One particular item which came up in the Sims meeting is whether the tilted nature of the calorimeter towers suggest that eta-phi may not be the best measure of 3-D topological connection. Perhaps you have some opinion on this.

 

Happy for any feedback,

 

Dennis

 

Dennis V. Perepelitsa
Assistant Professor, Physics Department
University of Colorado Boulder

 

On Dec 9, 2019, at 9:51 AM, Lajoie, John G [PHYSA] <lajoie AT iastate.edu> wrote:

 

Dear HCAL’ers:

 

It’s that time again – our next HCAL meeting will be tomorrow, Tuesday December 10^{tRuss h} at 11AM BNL time:

 

https://indico.bnl.gov/event/7183/

 

In addition to the usual reports on the sector assembly and tile testing, I have also added time for a discussion of the iHCAL cooling needs.  Russ has pointed out that we need to make a plan for any cooling testing we may want to do with the SS310 prototype so we can finalize the iHCAL design in the near future. 

 

Additional contributions are certainly welcome!

 

Regards,

John

 

John Lajoie

Professor of Physics

Iowa State University

 

(515) 294-6952

lajoie AT iastate.edu

 

_______________________________________________
sPHENIX-HCal-l mailing list
sPHENIX-HCal-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/sphenix-hcal-l

 

 

_______________________________________________
sPHENIX-HCal-l mailing list
sPHENIX-HCal-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/sphenix-hcal-l

Dennis V. Perepelitsa
Assistant Professor, Physics Department
University of Colorado Boulder