sPHENIX tracking discussion

[hachiya <hachiya AT rcf.rhic.bnl.gov>]

Hi Jamie and all,

I am in charge to manage INTT software development the INTT team.
I sent 1st version of the hotdeadmap to Chris for the offline QA.
This will be available for the production quickly.
It is good for the offline QA, and not the final version.
The calibration module will be PR soon.

Best regards,
Takashi



> Hello Genki (cc same),
>
> Thank you for the email.   However, this is not "advice" but
> effectively an order.     The data will all be re-processed, so there
> is no "danger" regarding hits removed.    We need immediate
> information on the tracking performance and have been directly asked
> by the ALD.    Please make the commit now.
>
> Sincerely,
>
> Jamie
>
> ||------------------------------------------------------------------------------------------
> || James L. Nagle
> || Professor of Physics, University of Colorado Boulder
> || EMAIL:   jamie.nagle AT colorado.edu
> || SKYPE:  jamie-nagle
> || WEB:      http://spot.colorado.edu/~naglej [2]
> ||------------------------------------------------------------------------------------------
>
> On Thu, Jul 11, 2024 at 8:21 AM Genki Nukazuka
> <genki.nukazuka AT riken.jp> wrote:
>
> > [External email - use caution]
> >
> > [External email - use caution]
> >
> > Dear Dennis,
> >
> > Thank you for your advice.
> > This discussion was already done in the INTT meeting a few weeks
> > ago, actually. Hot channel rejection can be dangerous as a part of
> > the data is thrown away, so we decided to take a reasonable time to
> > provide codes to sPHENIX that satisfy enough quality. Instead of the
> > latest one, a dummy module was put into the software repository so
> > that people could start (pretend) using it. For further
> > implementation and tests of the codes, we need to take some time,
> > but not too long. We already took a moderate and reasonable way to
> > go. Don't worry.
> > The assignment of INTT software tasks should be led by Takashi
> > since he is the INTT software coordinator, by the way.
> >
> > Best regards,
> > Genki
> >
> > ----
> > Genki Nukazuka
> > Special Postdoctoral Researcher
> > RIKEN/RBRC
> > ----
> >
> > -------------------------
> >
> > 差出人: Perepelitsa, Dennis <dvp AT bnl.gov>
> > 送信日時: 2024年7月10日 12:34
> > 宛先: Genki Nukazuka <genki.nukazuka AT riken.jp>
> > CC: Hao-Ren Jheng <hrjheng AT mit.edu>; sphenix-intt-l AT lists.bnl.gov
> > <sphenix-intt-l AT lists.bnl.gov>; Osborn, Joseph <josborn1 AT bnl.gov>;
> > Nagle, James <Jamie.nagle AT colorado.edu>; Joe Osborn
> > <osbornjd91 AT gmail.com>; Anthony Frawley via sPHENIX-tracking-l
> > <sphenix-tracking-l AT lists.bnl.gov>
> > 件名: Re: [[Sphenix-intt-l] ] Question about silicon seeds/INTT QA
> >
> >
> > Hi Genki and all,
> >
> > I and many others appreciate the hard work done being done by groups
> > working on the INTT.
> >
> > However, I have been following this thread, and I am becoming
> > concerned about delays in putting critical INTT code into the
> > official sPHENIX reconstruction.
> >
> > There are time-critical checks that the Collaboration is trying to
> > do, and it would be most helpful to put in the best map we have into
> > the official production, ASAP, essentially now. I think it is OK if
> > it is not the final version, and analyzers will understand that.
> >
> > Can the INTT group identify a responsible person who will focus on
> > this? What is the prospect of having an initial module like this by
> > tomorrow?
> >
> > Dennis
> >
> > On Jul 10, 2024, at 9:29 AM, Genki Nukazuka
> > <genki.nukazuka AT riken.jp> wrote:
> > Dear Hao-Ren,
> >
> > It's good to add the hot channel rejection process to official data
> > production. Joseph uploaded a dummy module, which will be replaced
> > by another module, for the implementation of the production chain,
> > in my understanding. I don't know who works on adding the process to
> > the official production chain. Also, we, the INTT group, need to
> > finish the hot channel rejection procedure. It needs to be
> > optimized. Since the rejection needs to be done before clustering,
> > updating the rejection process too many times is not a good idea
> > concerning computing resources, in my opinion.
> >
> > Best regards,
> > Genki
> >
> > ----
> > Genki Nukazuka
> > Special Postdoctoral Researcher
> > RIKEN/RBRC
> > ----
> >
> > -------------------------
> > 差出人: Hao-Ren Jheng
> > 送信: 2024 年 7 月 10 日 (水曜日) 8:57
> > 宛先: Genki Nukazuka
> > Cc: sphenix-intt-l AT lists.bnl.gov; Osborn, Joseph; Nagle, James; Joe
> > Osborn; Anthony Frawley via sPHENIX-tracking-l
> > 件名: Re: [[Sphenix-intt-l] ] Question about silicon seeds/INTT QA
> >
> >
> > Dear Genki and Jaein,
> >
> > Thanks for your feedback.
> > These offline QA are produced with the sPHENIX production and
> > software, so a follow-up question would be: is there a plan to
> > implement the hot channel mask for Run2024, or is it already an
> > on-going task, to improve the seeding aside from tuning the
> > parameters? (I do not know if this is a valid question or it makes
> > any sense)
> >
> > Best,
> > Hao-Ren
> >
> > *In the MVTX group we recently had an updated set of pixel masks
> > ready and remaining tasks are being worked on, If I am not mistaken
> >
> > On Jul 9, 2024, at 20:54, Genki Nukazuka <genki.nukazuka AT riken.jp>
> > wrote:
> >
> > Dear Hao-Ren,
> >
> > I remember that you showed this plot in the shift change meeting on
> > June/19
> (https://docs.google.com/document/d/1s0KTIgEyfS9PbAy6WVfta86m8bVLP7QNJ0KBexweTK4/edit
> > [1]), and I commented at that time. This is due to some hot channels
> > in the INTT data. You need to remove them before clustering. You can
> > ask Jaein (jaeinhwang213 AT gmail.com) for technical help. If you need
> > INTT cluster data with hot channel removal right now, I can make it
> > quickly.
> >
> > Best,
> > Genki
> >
> > ----
> > Genki Nukazuka
> > Special Postdoctoral Researcher
> > RIKEN/RBRC
> > ----
> >
> > -------------------------
> >
> > 差出人: sphenix-intt-l-request AT lists.bnl.gov
> > <sphenix-intt-l-request AT lists.bnl.gov> が Osborn, Joseph
> > <josborn1 AT bnl.gov> の代理で送信
> > 送信日時: 2024年7月9日 18:24
> > 宛先: sphenix-intt-l AT lists.bnl.gov <sphenix-intt-l AT lists.bnl.gov>;
> > Hao-Ren Jheng <hrjheng AT mit.edu>
> > CC: Nagle, James <Jamie.nagle AT colorado.edu>; Joe Osborn
> > <osbornjd91 AT gmail.com>; Anthony Frawley via sPHENIX-tracking-l
> > <sphenix-tracking-l AT lists.bnl.gov>
> > 件名: Re: [[Sphenix-intt-l] ] Question about silicon seeds/INTT QA
> >
> >
> > Hi all, and adding the tracking list for further comments others
> > might have,
> >
> > Here are a few comments from the tracking side of things.
> >
> > *
> > The seeding algorithm uses wide search windows right now to try to
> > ensure that we get some correlation between the silicon. These
> > windows were blown up based on the previous MVTX ideal geometry,
> > where there was significant layer to layer rotation (i.e. the ideal
> > geometry was wrong compared to how the MVTX was actually
> > constructed). That rotation has now been implemented to reduce the
> > residuals, so these windows likely need to be re-tuned.
> > *
> > The silicon seeding algorithm creates (at a minimum) triplets in the
> > 5 layers. In our simulation, we deploy this algorithm in the MVTX to
> > reduce the presence of duplicates. In the real data, with the
> > misalignment, we found we could not find anything in the MVTX alone
> > (again, before the rotation alignment fix), so we expanded the
> > seeding algorithm to search in the MVTX+INTT. This means you likely
> > get a nontrivial background of fake tracks where 2 layers of the
> > MVTX match to a single layer in the INTT.
> > *
> > Silicon seeds should not be considered as real tracks. We need to
> > make some basic quality criteria to reduce the presence of
> > background, mismatches, etc. One of these that we have been using in
> > alignment studies is that the track must have 5 measurements on it,
> > for exactly this reason.
> >
> > I'm happy to chat about this more. The track reconstruction is
> > going to continuously need to be optimized as we learn more about
> > the alignment, backgrounds, timing correlation between the silicon
> > subsystems, etc. Right now we need to make selections to have high
> > purity at a cost of efficiency, and as we learn more about the data
> > we will be able to re-tune these to improve the efficiency with a
> > cost to purity.
> >
> > Joe Osborn
> >
> > ___________________________
> > Joe Osborn, Ph.D
> > Physics Department
> > Brookhaven National Laboratory
> > josborn1 AT bnl.gov
> >
> > -------------------------
> >
> > From: sphenix-intt-l-request AT lists.bnl.gov on behalf of Hao-Ren
> > Jheng
> > Sent: Tuesday, July 9, 2024 10:28 AM
> > To: sphenix-intt-l AT lists.bnl.gov
> > Cc: Nagle, James; Joe Osborn
> > Subject: [[Sphenix-intt-l] ] Question about silicon seeds/INTT QA
> >
> > Dear INTT group and experts (cc RC, Jamie, and tracking expert,
> > Joe),
> >
> > I have been working on the silicon seeds offline QA and have one
> > question that requires INTT experts’ input.
> > We have one QA plot showing the correlation between number of MVTX
> > clusters v.s number of INTT clusters per silicon-only track, and we
> > are seeing a large fraction of tracks having only 1 INTT cluster
> > (please see the attached plot).
> > Do you have an idea why this is the case and what could be the cause
> > of it?
> >
> > Best,
> > Hao-Ren
>
>  Dennis V. Perepelitsa
> Associate Professor, Physics Department
> University of Colorado Boulder
>
>
> Links:
> ------
> [1] 
> https://urldefense.com/v3/__https://docs.google.com/document/d/1s0KTIgEyfS9PbAy6WVfta86m8bVLP7QNJ0KBexweTK4/edit__;!!P4SdNyxKAPE!GbqURPuWWPwk-Pqk4pmIU5xho_gDilgvsCsVbOnBbcj-9ljOZd5kpzqOA3r8ARI30hY5L09LbCPvjWUKpr-APOPAq9BsgnVwRbb57w$
> [2] 
> https://urldefense.com/v3/__http://spot.colorado.edu/*7Enaglej__;JQ!!P4SdNyxKAPE!HVkoJ-o1vphK5E9PH2Pkf6Sv4RrnYvnLPMZ5ZCN0Rk0ww52xvc_PbGeCl-TGafCaFo0MYnX2RYbR2HCTnprmk4LUjmS89_A8ST7Q$