Open hadjichris opened 1 year ago
hadjichris
commented
1 year ago Missing $\pmb{DDX}$ decays: We will request the missing decays. Estimate statistics if generated with comparable stats to their analogous decays (eg. MyOtherD*- MyD0 K*+ same stats as the included MyD*- MyOtherD0 K*+).
Trigger emulation: Estimate statistics based on 2016 sample. Normalization mode: $D^{*+}\mu\nu$.
Ghost MC Sample: Confirm sample and statistics.
hadjichris
commented
1 year ago | Missing $\pmb{DDX}$ decays: Reference Prepared a new .dec file (Bd_D0DX,muX=cocktail,ExtraModes,RDstar,TightCut.dec.zip), needs validation check) with just the missing decays. The equivalent Nsim statistics are: | Decay | 2016 | 2017 | 2018 | Total |
|---|---|---|---|---|---|
| B0->D0(Xc->munuX')X (delta) | 15.26M | 15.91M | 20.29M | 54.10M |
Decay B0sig
0.0020 MyOtherD*- MyD0 K+ PHSP;
0.0018 MyOtherD*- MyD0 K*+ PHSP;
0.0033 MyOtherD*- MyD*0 K*+ PHSP;
Enddecay
CDecay anti-B0sigTrigger emulation: Based on table
11574021 $D^{*+}\mu\nu$ normalization mode (FullSim): 13.55M Nsim events for 2017 and 17.29M Nsim events for 2018.
We will use what the Marseille group requested. 150M $D^{ }\mu\nu$ (+150 $D^{ }e\nu$) MC, FullSim - DST - sim-version: 10 sample.
Ghost MC Sample: 11774014 would be preferred (see https://github.com/umd-lhcb/lhcb-ntuples-gen/issues/111#issuecomment-1646872498).
Available samples for 11774014 ($B \rightarrow D^{(*)} X$ cocktail):
/MC/2018/Beam6500GeV-2018-MagUp-Nu1.6-25ns-Pythia8/Sim09l-ReDecay01/Trig0x617d18a4/Reco18/Turbo05-WithTurcal/Stripping34NoPrescalingFlagged/11774014/ALLSTREAMS.DST
/MC/2017/Beam6500GeV-2017-MagUp-Nu1.6-25ns-Pythia8/Sim09l-ReDecay01/Trig0x62661709/Reco17/Turbo04a-WithTurcal/Stripping29r2NoPrescalingFlagged/11774014/ALLSTREAMS.DST
/MC/2016/Beam6500GeV-2016-MagUp-Nu1.6-25ns-Pythia8/Sim09l-ReDecay01/Trig0x6139160F/Reco16/Turbo03a/Stripping28r2NoPrescalingFlagged/11774014/ALLSTREAMS.DST
/MC/2018/Beam6500GeV-2018-MagDown-Nu1.6-25ns-Pythia8/Sim09l-ReDecay01/Trig0x617d18a4/Reco18/Turbo05-WithTurcal/Stripping34NoPrescalingFlagged/Turbo05Filtered/11774014/D02HH.HLTFILTER.MDST
/MC/2018/Beam6500GeV-2018-MagUp-Nu1.6-25ns-Pythia8/Sim09l-ReDecay01/Trig0x617d18a4/Reco18/Turbo05-WithTurcal/Stripping34NoPrescalingFlagged/Turbo05Filtered/11774014/D02HH.HLTFILTER.MDST
/MC/2017/Beam6500GeV-2017-MagDown-Nu1.6-25ns-Pythia8/Sim09l-ReDecay01/Trig0x62661709/Reco17/Turbo04a-WithTurcal/Stripping29r2NoPrescalingFlagged/Turbo04aFiltered/11774014/D02HH.HLTFILTER.MDST
/MC/2017/Beam6500GeV-2017-MagUp-Nu1.6-25ns-Pythia8/Sim09l-ReDecay01/Trig0x62661709/Reco17/Turbo04a-WithTurcal/Stripping29r2NoPrescalingFlagged/Turbo04aFiltered/11774014/D02HH.HLTFILTER.MDST
/MC/2016/Beam6500GeV-2016-MagDown-Nu1.6-25ns-Pythia8/Sim09l-ReDecay01/Trig0x6139160F/Reco16/Turbo03a/Stripping28r2NoPrescalingFlagged/Turbo03aFiltered/11774014/D02HH.HLTFILTER.MDST
/MC/2016/Beam6500GeV-2016-MagUp-Nu1.6-25ns-Pythia8/Sim09l-ReDecay01/Trig0x6139160F/Reco16/Turbo03a/Stripping28r2NoPrescalingFlagged/Turbo03aFiltered/11774014/D02HH.HLTFILTER.MDST| MC_ID | Year | Magnet | Pythia | Sim | Type | EventInputStat | EventStat | Retention(%) |
|---|---|---|---|---|---|---|---|---|
| 11774014 | 2018 | MagUp | Pythia8 | Sim09l-ReDecay01 | ALLSTREAMS.DST | 103000 | 103000 | 100.00 |
| 11774014 | 2017 | MagUp | Pythia8 | Sim09l-ReDecay01 | ALLSTREAMS.DST | 102999 | 102999 | 100.00 |
| 11774014 | 2016 | MagUp | Pythia8 | Sim09l-ReDecay01 | ALLSTREAMS.DST | 103000 | 103000 | 100.00 |
| 11774014 | 2018 | MagDown | Pythia8 | Sim09l-ReDecay01 | D02HH.HLTFILTER.MDST | 103994623 | 5060985 | 4.87 |
| 11774014 | 2018 | MagUp | Pythia8 | Sim09l-ReDecay01 | D02HH.HLTFILTER.MDST | 103855833 | 5036767 | 4.85 |
| 11774014 | 2017 | MagDown | Pythia8 | Sim09l-ReDecay01 | D02HH.HLTFILTER.MDST | 92997154 | 5223754 | 5.62 |
| 11774014 | 2017 | MagUp | Pythia8 | Sim09l-ReDecay01 | D02HH.HLTFILTER.MDST | 93614023 | 5270957 | 5.63 |
| 11774014 | 2016 | MagDown | Pythia8 | Sim09l-ReDecay01 | D02HH.HLTFILTER.MDST | 88915273 | 5392824 | 6.07 |
| 11774014 | 2016 | MagUp | Pythia8 | Sim09l-ReDecay01 | D02HH.HLTFILTER.MDST | 88879300 | 5371380 | 6.04 |
afernez
commented
1 year ago Here are the specs/stats for the inclusive $J/\psi$ MC used for the $R(J/\psi)$ run2 ghost misID efficiencies (Emily tells me they add the simulation versions together)-
EventType: 24142001
Specs: 6.5TeV, Nu1.6, 25ns, Pythia8, FullSim
Note: all MC is flagged, so Nsim=Ndisk
| Year | Polarity | Sim. Ver. | Nsim [M] |
|---|---|---|---|
| 2016 | MU | Sim09b | 21.9 |
| 2016 | MU | Sim09j | 1.0 |
| 2016 | MU | Sim09k | 1.0 |
| 2016 | MU | Sim09l | 20.1 |
| 2016 | MD | Sim09b | 21.4 |
| 2016 | MD | Sim09j | 1.0 |
| 2016 | MD | Sim09k | 1.0 |
| 2016 | MD | Sim09l | 20.0 |
| 2016 | $\Sigma =87.4$ | ||
| 2017 | MU | Sim09g | 20.0 |
| 2017 | MU | Sim09l | 2.0 |
| 2017 | MD | Sim09g | 20.0 |
| 2017 | MD | Sim09l | 2.0 |
| 2017 | $\Sigma =44.0$ | ||
| 2018 | MU | Sim09g | 20.0 |
| 2018 | MU | Sim09l | 2.0 |
| 2018 | MD | Sim09g | 20.0 |
| 2018 | MD | Sim09l | 2.0 |
| 2018 | $\Sigma =44.0$ |
As a note: out of laziness, I made this table by hand, copying numbers from the bookkeeping.
Comparing to the EventInputStat numbers above (what I'm calling Nsim here) for the ReDecay $D^{*}$ cocktail 11774014, 2016: 207.9M $D^{*}$ vs 87.4M $J/\psi$, 2017: 186.6M $D^{*}$ vs 44.0M $J/\psi$, 2018: 177.8M $D^{*}$ vs 44.0M $J/\psi$. So--I guess if we don't come up with a reason why the MC being ReDecay would pose a problem for us--this naively seems fine for us for stats.
afernez
commented
9 months ago Christos locally generated a small amount of FullSim MC of 11774014 for us to study, and I've now attempted reconstructing and getting true ghost counts for the sample (recall [comment]: the ghost number we're comparing to is R(J/psi) run2's 2,279,991)
Notes about the reco:
GHOST flag on. Without editing the reco script, this turns off the mu stripping cuts (our stripping line def can be found here) as well as the mu truth-matchingmu or md for the CondDBtag. Now, I didn't know if what Christos produced was intended to be MU or MD, so I just tried both, and surprisingly (to me) found both produced identical results... (ie. the number below were the same whether I specified mu or md)mu stripping cuts on and off (to keep them on while keeping the mu truth-matching off, I just commented the relevant lines out)| With these notes in mind, the stats I find from the small sample: | mu Stripping on/off |
$D^0$ Reco Evts | $D^0$ mu_TRUEID==0 |
$D^{*}$ Reco Evts | $D^{*}$ mu_TRUEID==0 |
|---|---|---|---|---|---|
| Off | 419 | 39 | 277 | 27 | |
| On | 258 | 10 | 173 | 7 |
Naively taking the mu Stripping On $D^{*}$ reco number and comparing to R(J\psi) run2's true ghost number, this suggests we'd need to request a sample 2,279,991 / 7 = 325,713x larger than Christos' locally produced sample.
hadjichris
commented
9 months ago These are the conditions I used (Sim10):
LHCbApp().DDDBtag = "dddb-20220927-2016"
LHCbApp().CondDBtag = "sim-20201113-6-vc-mu100-Sim10"I generated 10k events for 2016 mu. This small sample was to check that I can generate things correctly. If we need a larger sample to decide let me know. It only took 2h to generate the current one.
afernez
commented
7 months ago As a note: the table above remained accurate once correctly reconstructing the local sample with the right db tags (which just required rebuilding our DaVinci image).
Christos now generated two more sets of local inclusive $D^{*}$ MC for study. The sample used above used the nominal filtering where the muon is already truth-matched and stripped (L40), so it isn't really surprising that the ghost stats were so low. The two new samples don't have the muon filtering: one runs the filtering script with the muon cuts off (Code = "(ALL)"), and the other is just the Brunel output without any filtering.
Here are the stats for all 3 local samples, now including the counts of true hadrons misidentified as muons:
| Filtered | $\mu$ Matched in Filter | $\mu$ Stripped in Filter | $\mu$ Stripped in Reco (not matched) | $D^0$ Reco Evts | $D^0$ mu_TRUEID == $\mu$ | $D^0$ mu_TRUEID == $e$ | $D^0$ mu_TRUEID == $\pi^-$ | $D^0$ mu_TRUEID == $K^-$ | $D^0$ mu_TRUEID == $p$ | $D^0$ mu_TRUEID == ghost | $D^{*}$ Reco Evts | $D^{*}$ mu_TRUEID == $\mu$ | $D^{*}$ mu_TRUEID == $e$ | $D^{*}$ mu_TRUEID == $\pi^-$ | $D^{*}$ mu_TRUEID == $K^-$ | $D^{*}$ mu_TRUEID == $p$ | $D^{*}$ mu_TRUEID == ghost |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| True | True | True | True | 258 | 207 (80.2%) | 5 (1.9%) | 33 (12.8%) | 3 (1.2%) | 0 (0.0%) | 10 (3.9%) | 173 | 144 (83.2%) | 1 (0.6%) | 18 (10.4%) | 3 (1.7%) | 0 (0.0%) | 7 (4.0%) |
| True | True | True | False | 419 | 208 (49.6%) | 20 (4.8%) | 129 (30.8%) | 17 (4.1%) | 6 (1.4%) | 39 (9.3%) | 277 | 145 (52.3%) | 11 (4.0%) | 79 (28.5%) | 12 (4.3%) | 3 (1.1%) | 27 (9.7%) |
| True | False | False | True | 1063 | 226 (21.3%) | 155 (14.6%) | 519 (48.8%) | 72 (6.8%) | 9 (0.8%) | 82 (7.7%) | 723 | 149 (20.6%) | 99 (13.7%) | 358 (49.5%) | 53 (7.3%) | 4 (0.6%) | 60 (8.3%) |
| True | False | False | False | 2160 | 246 (11.4%) | 232 (10.7%) | 1187 (55.0%) | 165 (7.6%) | 58 (2.7%) | 272 (12.6%) | 1483 | 161 (10.9%) | 147 (9.9%) | 818 (55.2%) | 119 (8.0%) | 42 (2.8%) | 196 (13.2%) |
| False | False | False | True | 1060 | 227 (21.4%) | 154 (14.5%) | 516 (48.7%) | 71 (6.7%) | 9 (0.8%) | 83 (7.8%) | 719 | 149 (20.7%) | 98 (13.6%) | 356 (49.5%) | 52 (7.2%) | 4 (0.6%) | 60 (8.3%) |
| False | False | False | False | 2155 | 247 (11.5%) | 230 (10.7%) | 1185 (55.0%) | 163 (7.6%) | 59 (2.7%) | 271 (12.6%) | 1477 | 161 (10.9%) | 145 (9.8%) | 816 (55.2%) | 118 (8.0%) | 43 (2.9%) | 194 (13.1%) |
Probably most notable things here, with the intention of figuring out what we'll need to request to get good ghost misID efficiencies:
126 $\Lambda_b$ background: The Event Types needed are already produced. We will just generate the corresponding nTuples.
125 Missing $DDX$ decays: In 11894600 three modes are missing:
MyOtherD*- MyD0 K*+,MyOtherD*- MyD*0 K*+, andMyOtherD*- MyD0 K+. Do we only request the missing decays?111 Ghost MC Sample: Use [12875420]() as our RUN2 ghost sample?