Restore TrackMuon HLT1 Present in Run 2 to Run 3

This merge request aims to reimplement the TrackMuon HLT1 line which was present in Run 2 into Run 3. During Run2 TrackMuon was the line with the highest rate when excluding the general purpose triggers TrackMVA and TwoTrackMVA. See, for example: here and here.

We aim to remplement TrackMuon with the goal of improving LHCb's online selection efficiency to, in the SM, muonic semileptonic decays and, in BSM studies, to mu-coupled BSM particles (e.g. Heavy Neutral Leptons). We target to have this branch merged before the start of 2025 data-taking.

A summary of the proposed selections, physics case, and throughput and bandwidth effect can be found below. A similar summary can also be found in our initial presentation at the RTA WP3 Selections meeting here. The results included below are actively maintained and are thus the most up-to-date.

Proposed Selections

(click to expand)

We propose two possible selections for the reimplementation of TrackMuon: 'Run 2-like TrackMuon' and 'Tight TrackMuon'. Run 2-like TrackMuon implements the same cuts as in 2018-patches, except for a tighter track chi-squared requirement and muon ID. Tight TrackMuon implements harder cuts, reducing the sensitivity of the line to soft muons, but with an associated reduction in rate and HLT2 bandwidth. Both lines trigger almost entirely on muons not from the PV.

The particular cuts we propose are detailed below

Cut variable	Tight TrackMuon	Run 2-like TrackMuon
PT	> 1100 MeV	> 600 MeV
P	> 10000 MeV	> 6000 MeV
Track Reduced chi2	< 2	< 2
Track GhostProb	<0.2	-
Min IP chi2	> 7.4, < 10000	> 7.4
Muon chi2 Corr	< 1.8	<1.8
IsMuon	1	1

Physics Case

(click to expand)

By reimplementing TrackMuon, we improve LHCb's capacity to trigger on soft, displaced muons. This is shown below diagramatically for the Tight TrackMuon proposal in the PT - Min IP chi2 plane of long muons.

For the SM physics program, TrackMuon translates into an improvement in efficiency for many leptonic and semileptonic processes. Picking the example of D0 -> K*- mu+ nu_mu (a decision based on availability of Run 3 MC), we find a five-fold improvement in HLT1 efficiency over TrackMuonMVA. The TOS fraction also increases.

	Efficiency D0 -> K*- mu+ nu_mu	TOS Fraction
Tight TrackMuon	0.107 +/- 0.003	0.851 +/- 0.012
TrackMuonMVA	0.0211 +/- 0.0015	0.72 +/- 0.04

For the BSM physics program, TrackMuon translates into an improvement in efficiency for all new particles associated with muons. Heavy Neutral Leptons (HNLs) are a particularly good example, with HNLs produced alongside a, generally rather soft, muon.

We show below the improvement from Tight TrackMuon for the two dominant production modes of HNLs, semileptonic production B_q -> X mu+ N (dominant for 1.5 - 3.0 GeV HNLs) and leptonic production B_c+ -> mu+ N (dominant for 3.0 - 6.0 GeV HNLs).

Semileptonic production B_q -> X mu+ N

Leptonic production B_c+ -> mu+ N

The efficiency of Tight TrackMuon and Run 2-Like TrackMuon are compared in the table below. Note this is evaluated on HLT2-level variables, and only for semileptonic production with B_u. This explains the small differences with the above plots.

	Semileptonic 1.6 GeV	Semileptonic 3.0 GeV	Leptonic 2.0 GeV	Leptonic 5.0 GeV
Tight TrackMuon	0.295	0.111	0.553	0.229
Run 2-Like TrackMuon	0.481	0.228	0.597	0.386

Throughput and Bandwidth Effect

(click to expand)

Throughput: Neither Tight TrackMuon nor Run 2-Like TrackMuon have any measurable effect on HLT1 throughput.

Bandwidth: Bandwidth impacts (and rates) were evaluated by running HLT1 and HLT2 over 500,000 non-filtered MC minbias events computed under Beam6800GeV-expected-2024-MagDown-Nu7.6-25ns-Pythia8 conditions (production ID 00204940, available in the bookkeeping). HLT1 and HLT2 were on the final 2024-patches branch, with Allen run with the hlt1_pp_forward_then_matching_and_downstream_1200KHz sequence.

The results of the study are given in the table below:

	Reference	Tight TrackMuon	Run 2-Like TrackMuon
HLT1 Rate (KHz)	1200.	1273.	1593.
HLT2 Rate (KHz)	188.	193.	208.
Turbo Bandwidth (MB/s)	3207.	3264.	3466.
Delta Turbo Bandwidth (MB/s)	-	+57.	+259.
Full Bandwidth (MB/s)	11977.	12293.	12907.
Delta Full Bandwidth (MB/s)	-	+316.	+930.