 |
 |
Hi Pelle, first of all I must confess that I should have waited a little more to send the plots around, they're way too preliminary, but that day I was kind of thrilled and I wanted to share... There are some bias in the plots I'm aware of and they should be just taken as an example of how things will work. Let me summarize the main idea, then I'll try to answer your question and after that I'll tell you what still needs to be decided and done. 1)Simulations: I don't have at hands data simulated at commissioning run conditions. I don't have the complete "know-how" at the moment, but instead of spending time on it I'd rather work on the calibration machinery. the calibration is based on the fit of the SeedHits energy distributions of full energy electrons. The selection of these events will be covered answering your questions. I have 2.2 GeV data and to obtain some values usable for our data I just made a simple propotion: 2.2CalibPoint : 2.2 Gev Beam = 1.92 CalibPoint : 1.92 GeV beam. It assumes the linearity of the energy deposition in the seed, and I think it's fair enough, and can be refined taking into account the longitudinal sampling fraction. Since at both 2.2 and 1.92 GeV FEE clusters involve more than one crystal, I think the value I use it's good within a few percents of precision, enough to test the calibration. Il 11/01/2015 21:25, Hansson Adrian, Per Ola ha scritto: > Hi Luca, > > nice. Some simple questions to get up to speed on your study below. > Sorry if this was discussed in some meeting last week, I’ve been away > until this week. > > The “uncalibrated” and “simulation” has a title "GTP cluster" but the > calibrated plot say “IC Cluster”. Is the "GTP cluster" using the 3x3 > window energy used in the trigger selection and “IC cluster” the > relatively new (real) clustering algorithm recently added to the > reconstruction? GTP and IC clusters are those readable in the reconstructed data, I used the seedhit distribution of the GTP clusters to set the calibration point and then multiplyed the IC data (both seed hits and clustes energy) by that value: CalibratedEnergy=(FitPointFromSimulation/FitPointFromData)*RawEnergy. This is a bias, I know! Keep in mind that seedhit energy distribution for FEE is independent from the clustering algorithm. Anyway GTP and IC clusters are what you say they are. > > If so, do we save the energy that was used in the trigger decision in > the evio file or is this something you recompute offline? If you > recompute, did we verify that we can recompute the trigger decision > with this implementation? If the clustering algorithms are different > in the calibrated and uncalibrated plots, is there a reason for this? The only thing I did is read the clusters and seedhit enrgies and dump them on txt file for analysis, I didn't write back anything in the slcio file. Everything is offline, now. What happens after i set the new gain values, is up to us. > > I assume that you are trying to select Coulomb scattered beam > electrons at full beam energy? What was the event selection? Studying the simulations i showed that, at 2.2Gev, selecting clusters with E>1.5 is enough to see the FEE peak. It doesn't need to be precise because we are interested in the high energy tail and cut off, so it's just a threshold. For this run I just scaled the cut proportionally > > What part of the ECal is shown in these plots? Both top and bottom? It the plots it's just one crystal. > > There seems to be a cutoff at exactly 1GeV (1.5GeV) in uncalibrated > data (simulation). Is that an artificial cut you apply? It's artificial, cutting out low energy stuff keep the SeedHit energy distribution clearer and easier to fit. > > Can you explain a little more how the "calibration point” is defined > and also what you mean exactly with calibrated? I assumed that this > was after applying a correction to each crystal (“gain calibration”) > and without a sampling fraction correction but I’m a little confused > what the "calibration point" mean in this context... I fit the seed hit energy distribution of simulated data with a complementary error function multiplied by a third grade polinomial. The calibration point is the half-height slope. I do the same with the real data and their ratio is the "gain". At this point it isn't right, yet, because I haven't plotted raw energy but "cosmic calibrated" energy. See comments later > > The peak in simulation is close the full beam energy by construction. > What was the peak position in simulation without scaling? If I > understand what you are plotting there should be a sampling fraction > that is not taken into account here. To compare with your calibrated > cluster in data, shouldn’t the scaling be to 1.92*(sampling fraction) > i.e. some ~20% lower? This is a good point and I'm not 100% sure. I've tried to ask during past months but didn't get clear answers. I've been plotting the energy distributions of clusters blindly trusting who wrote the simulation drivers, but i think, since energy deposition in crystal is simulated in SLIC, it doesn't give you the complete energy of the particle, just the fraction depositated in the crystal so both real and simulated data should be in the same condition, i.e., not-sampling fraction corrected. > > Also, it looks like the seed hit energy of the simulation is much > broader than the data. Is that due to some effect of the scaling from > 2.2 to 1.92 or some other reason? I noticed it, and I'm pretty sure it is due to the fact that seed threshold for run 3261 is set to 500 MeV. It doesn't affect the calibration. Now some consideration: This is done on top of holly's cosmics calibration. What I want to do now is divide out the gain values and plot the raw energy and obtain some independent values. If things work right, a fit of my values and holly's values should point to the pedestal values. I actually need to understand how to deal with pedestals, it's something I wanted to ask during last software meeting but we got lost talking about talks. I'll discuss it today at ecal meeting. It should be fine and easy. There is a lot more to do, but I think the outlook is good. Bests Luca > > Thanks! > > /Pelle > > > On Jan 10, 2015, at 3:37 PM, Luca Colaneri > <[log in to unmask] <mailto:[log in to unmask]>> wrote: > >> No, as setting points I've used the points obtained from 2.2 GeV beam >> simulations and proportionally scaled to 1.9. It is good enough to >> test and let me put up all the machinery for the calibration. >> The red vertical line in central and left histrgrams is the >> calibration point. >> >> L. >> >> Il 09/01/2015 20:39, Stepan Stepanyan ha scritto: >>> Luca, >>> >>> Is your simulation with 1.92 GeV electrons. >>> I am trying to understand why peak for correct distribution >>> is at ~1.7 GeV while simulation is at 1.95 GeV. >>> >>> Stepan >>> On 1/9/15 12:34 PM, Luca Colaneri wrote: >>>> There is a lot more to do, this is very very preliminary, but it is >>>> happening... >>>> >>>> Cheers! >>>> >>>> Luca >>>> >>> >> >> >> -- >> PhD Student, >> Università degli Studi di Roma Tor Vergata >> INFN sez. Roma2 >> Via della Ricerca Scientifica 1, 00133 Rome, Italy >> +39 0672594562 >> >> ######################################################################## >> Use REPLY-ALL to reply to list >> >> To unsubscribe from the HPS-SOFTWARE list, click the following link: >> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=HPS-SOFTWARE&A=1 > > > ------------------------------------------------------------------------ > > Use REPLY-ALL to reply to list > > To unsubscribe from the HPS-SOFTWARE list, click the following link: > https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=HPS-SOFTWARE&A=1 > ######################################################################## Use REPLY-ALL to reply to list To unsubscribe from the HPS-SOFTWARE list, click the following link: https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=HPS-SOFTWARE&A=1