Exclusive D02KsHH HLT1 line

An exclusive HLT1 line for selecting D⁰→Kₛ⁰π⁺π⁻ decays.

RTA Approval Information

• Expert/technical code review done by @thboettc
• Physics choices: removed IPChi2 cuts to improve correlation between data and simulation; presented to CharmWG, RTA-WP3 and RTA-DPA meetings (see below)
• Adds 1 Hlt1 lines
• Effect on throughput is <1% on A5000
• Effect on event rate is 7/ 10000, ( 21.00 +/- 7.93) kHz

Talks

• RTA DPA general 16/2/2024
• RTA WP3 6/3/2023
• RTA WP3 9/5/2023
• Charm WG 17/5/2023

Selection

The line is based on the combination of two CompositeParticle objects (V) and the following requirements are made:

1. Each of the CompositeParticle vertex χ² should be 0 < χ² < maxVChi2 (20.)
2. The DOCA of the two vertices is DOCA(V₁V₂) < maxDOCA (50 µm)
3. The invariant mass of the combination should be |m(V₁V₂)-m(D⁰)| < massWindow (100 MeV)
4. The transverse momentum of the combination should be Pₜ(V₁V₂) > minComboPt_D0 (1500 MeV)
5. The proper time of the combination should be cτ(V₁V₂) > minCTau_D0 (0.5 * 0.1229 mm)

The V with the largest z is then assumed as the Kₛ⁰ candidate and further requirements are made

5. m(Kₛ⁰) mass window [455,545] MeV
6. minimum Pₜ (200 MeV) of the Kₛ⁰
7. minimum P (1500 MeV), Pₜ (200 MeV) and IP (0.2mm) among the Kₛ⁰ components
8. the Kₛ⁰ η in [2,5]

The other V is considered as the hh and provides the D⁰ vertex, therefore

9. minP(h₁,h₂) > 1500 MeV;
10. 2 < η(h₁h₂) < 5;
11. minPₜ(h₁,h₂) > 250 MeV;
12. minIP(h₁,h₂) > 0.06 mm;
13. DOCA(h₁h₂) < 0.05mm;

The line also includes a monitoring tuple with the following information:

• m(V₁V₂)
• Pₜ(V₁V₂)
• P(V₁V₂)
• DOCA(V₁V₂)
• cτ(V₁V₂)
• m²(Kₛ⁰h⁺)
• m²(Kₛ⁰h⁻)
• minIPχ²(Kₛ⁰)
• minIPχ²(hh)
• minIP(Kₛ⁰)
• minIP(hh)

Rate

Calculated with $HLTEFFICIENCYCHECKERROOT/scripts/hlt_calculate_rates.py

Line:	 Hlt1D2KshhDecision   Incl:   21.0 +/-  7.9 kHz,	Excl:    12.0 +/-  6.0 kHz

On MB (calculated by @thboettc )

Hlt1D2Kshh:                                         7/ 10000, (   21.00 +/-     7.93) kHz

Efficiency

2.1% (1.0% exclusive) on a sample of D⁰→Kₛ⁰π⁺π⁻ decays with denominator the events where all the pions are Long reconstructable tracks with 2 < η < 5.

Throughput

From the latest pipeline

• hlt1_pp_default (throughput)

  Device	Throughput (kHz)	Reference	Speedup	Change	Status
  NVIDIA RTX A5000	99.34	98.16	1.01x	1.21%	OK
  AMD EPYC 7502 32-Core	11.96	5.57	2.15x	114.58%	OK
  NVIDIA GeForce RTX 2080 Ti	80.61	80.72	1.00x	-0.14%	OK
  NVIDIA GeForce RTX 3090	120.8	120.9	1.00x	-0.08%	OK

• hlt1_pp_forward_then_matching (throughput_full)

  Device	Throughput (kHz)	Reference	Speedup	Change	Status
  NVIDIA RTX A5000	78.51	78.77	1.00x	-0.3%	OK
  AMD EPYC 7502 32-Core	8.35	8.34	1.00x	0.09%	OK
  NVIDIA GeForce RTX 2080 Ti	63.19	62.42	1.01x	1.23%	OK
  NVIDIA GeForce RTX 3090	96.86	96.91	1.00x	-0.05%	OK

Handles for Bandwidth division

For physical reason, it would be better to tune the line with respect to the D⁰ proper time parameter. By default the minimum proper time of the D⁰ candidate is 0.5*0.1229, it can be tightened up to 0.1229 (1 D⁰ lifetime) with a reduction in efficiency of 20%.