Tracking volume comparison: saveall vs longtv
This page compares the two candidate fixes for the MCParticle Lambda bugs (see the campaign index for bug links):
- saveall — runs with
--part.keepAllParticles=True(tracking-volume handler ON, but the flag is supposed to preserve everything). Intended as the physics-correct baseline. - longtv — runs with the new, extended tracking volume proposed in eic/epic#1081. Keeps the handler but makes the Λ decay region fit inside the volume so daughter particles are kept without the save-all crutch.
All plots are produced by analysis/acceptance/plot_lambda_endpoints.py and show primary Λ (lam_is_first == 1) endpoints, grouped by daughter count nd ∈ {0, 1, 2, ≥3, any}.
How to read Use the global selector at the top to switch everything to one energy or variant at once. Use the per-plot dropdown to pick any pair — in particular the
saveall X vs longtv Xcomparisons show saveall first, longtv below, which is the intended reading order.
I. ZX plane (EIC detector overlay)
Scatter — coloured by daughter count
Primary Λ endpoint ZX — coloured by n daughters
Per-event scatter of Λ endpoints in the ZX plane. Colour encodes number of recorded daughters (0, 1, 2, ≥3).
2-D histogram, all categories (any nd)
ZX endpoint density — any nd
Combined density of all primary Λ endpoints in the ZX plane.
2-D histogram, nd = 0 (undecayed)
ZX endpoint density — nd = 0
Primary Λ with no recorded daughters — indicates the particle left the simulated volume without decaying OR that the tracking-volume bug dropped its daughters.
2-D histogram, nd = 1
ZX endpoint density — nd = 1
Primary Λ with a single recorded daughter (partial-keep artefact).
2-D histogram, nd = 2 (canonical two-body)
ZX endpoint density — nd = 2
Canonical two-body Λ decay topology (p π- or n π0).
2-D histogram, nd ≥ 3 (showering)
ZX endpoint density — nd ≥ 3
Showering / re-interacting Λ endpoints.
II. ZY plane
Scatter
Primary Λ endpoint ZY — coloured by n daughters
2-D histograms
ZY endpoint density — any nd
ZY endpoint density — nd = 0
ZY endpoint density — nd = 1
ZY endpoint density — nd = 2
ZY endpoint density — nd ≥ 3
III. ZR plane (EIC detector overlay, R = sqrt(X²+Y²))
Scatter
Primary Λ endpoint ZR — coloured by n daughters
Radial view. Since R ≥ 0, only the upper half of the overlaid detector cross-section receives data points.
2-D histograms
ZR endpoint density — any nd
ZR endpoint density — nd = 0
ZR endpoint density — nd = 1
ZR endpoint density — nd = 2
ZR endpoint density — nd ≥ 3
IV. 1-D Λ endpoint Z distributions
Overlay (all categories on one figure)
Λ endpoint Z — categories overlaid
Step histograms per daughter-count category plus a dashed outline for the combined sample.
Λ endpoint Z — categories overlaid (log y)
Per category, with EIC image as a Z-landmark ribbon
Λ endpoint Z — any nd
Λ endpoint Z — nd = 0
Λ endpoint Z — nd = 1
Λ endpoint Z — nd = 2
Λ endpoint Z — nd ≥ 3
V. Λ polar angle with respect to +Z
The EIC hadron-beam crossing angle (25 mrad) is marked on both plots.
Primary Λ polar angle θ (mrad), categories overlaid
Primary Λ polar angle θ (mrad), log y
Notes
- The interesting regions to watch are (a) the nd = 0 populations — ideally these concentrate far downstream or at world-volume boundaries, not inside the detector, and (b) the nd ≥ 3 populations, which should be rare. A working fix moves events out of
nd = 0andnd ≥ 3intond = 2. - The angular overlay is the cleanest sanity check: all variants must peak at 25 mrad (hadron beam) and show consistent tails. Any big asymmetry between
saveallandlongtvthere indicates a different kinematic pre-selection slipped in, not a tracking-volume effect.
