New LGD shower clusterizer at UConn

In order to address different types of the split-offs we concluded that 2 or 3 separate methods are needed in the cluster cleanup procedure. Thus we turned to the new method of making clusters from the hit list. It is iterative method that starts with the list of the LGD-hits ordered by energy. At the begining of each iteration each block, from the pool of blocks that are not already used as cluster-seeds, has been challenged whether it is allowed to become a seed for a new cluster. The criteria for becoming a seed is based on the allowance function that is calculated for all already seeded clusters. The next step within iteration goes over all non-seeded blocks and share block energy among clusters based on the expectation function. Both allowance and expectation functions depend on the distance of the block from a given cluster and the cluster energy and polar angle. The expectation function is quadratic Gaussian, elliptic in u, v space, with the radial width (σu) that increase with angle. The allowed function is sum of quadratic Gaussian and exponential tail that grows logarithmically with shower energy. At the very beginning of each iteration the cluster characteristics (number of hits, total energy, nmax energy, r.m.s) are computed. Those clusters that has changed since the previous iteration have their hits list reset, and hits returned to the pool. If nothing changes the iteration stops.

Here are some fixes of different types of splits with the new UConn clusterizer:

• type 1 fix 1, fix 2
• type 2 fix 1, fix 2 fix 3
• type 3 fix 1

Comparison between UC and IU clusterizer on full real data set with Fully Contained Forward (FCF) cut. Tagging and CPV veto analysis use weights as defined in tech-note [1] and tagging logic. The CPV veto is based on Recoil-CPV time difference (3 γ) 2 γ , and requires no hits inside coincidence window (usually -3,3 ns). Numbers in the table represent signal yields (the integral under Gaussian) deduced from Gaussian + 2-nd polynomial fit, and not the overall statistics.

2 γ	Cuts progression	No cuts	Tagged	Tagged neutral
IU	π° η	10,159,415 1,857,619	5,359,167 1,005,360	2,783,384 (-3,3 ns), 2,173,432 (-5,5 ns) 536,597 (-3,3 ns), 412,092 (-5,5 ns)
UC	π° η, CPV cuts	10,890,916 1,867,016	5,874,974 1,004,718	3,169,960 (-3,3 ns) 2,483,336 (-5,5 ns) 546,721 (-3,3 ns), 422,001 (-5,5 ns)

---

3 γ	Cuts progression	π°γ (0.1<m2(1)<0.18 GeV)	CPV veto	Tagged neutral	Tagged neutral π°γ	Tagged neutral ηγ
IU	3 γ	1,636,492	980,485 (-3,3 ns)	497,552 (-3,3 ns)	371,828(-5,5 ns)
UC	3 γ	1,723,669	1,000,460 (-3,3 ns)	(-3,3 ns)	555,303 (-3,3) 381,358 (-5,5)	5,575 (-3,3 ns)

3 γ	Pair mass	M vs m2	m2 vs m2	Δφ	Out-of-plane pT (pTy)
IU	m2, m2(3)	dot plot, surf plot	dot	Rec-FWD	pTy, pTy(MC)from ω simulation Δ pTy(MC) (measured-generated)
UC	m2, m2(3)	dot plot, surf plot	dot		pTy, pTy tag neutral 2G-fit pTy(mc) from φ simulation pTy(3γ)-pTy(mc)

---

4 γ	Cut progression	Tagged	Tagged neutral
IU	4γ	tagg	(-3,3 ns)
UC	4 γ	tagg	(-3,3) ns

4 γ	Pair mass	M vs m2	m2 vs m2	ππ selection
UC	m2	surf plot	surf	invariant mass pair mass vs invariant mass dot plot and x-projection

---

• fiducial cuts
5 γCut progression TaggedTagged neutral IU 5 γ tag (-3,3 ns) UC 5 γ tag (-3,3 ns) 6γ
• η (3π) IU (235596) vs UC (334218)
• tagged spectra IU (150249) vs UC (208249)
• tagged spectra with CPV veto IU (96869) vs UC (122570)

Comparison with MC data:

• γp -> π° : IU vs UC
• γp -> η(2γ) : IU vs UC
• γp -> ω(π°γ) : IU vs UC
• γp -> η(3π°) : IU vs UC