VDC Tracking

P.E. UlmerOld Dominion University

Expert …

or, nobody else wanted to talk about it;

You decide

Hall A Analysis WorkshopDecember 11, 2001

Algorithm description Tracking problems seen in E94-004 My definition of “good” event Rate dependence of tracking

efficiency Hit patterns for a sample of events Possible repair strategies

Thanks to Jeff Templon for providing me with his paper: “VDC Reconstruction in ESPACE”, dated October 22, 2001

Outline

LeCroy 1877 multihit TDC TDC Start: wire TDC Stop: trigger 1.496 usec time window

1st hit has largest TDC value Rough timedistance conversion:

tdc_cut > 0: cut out negative distances tdc_cut > 1: cut out large positive

distances For hits passing cuts:

Accept only largest TDC values Load VDC Wire/TDC structure Find clusters Analyze clusters for tracks

Algorithm Overview

Analysis flow (main VDC routines): scan_file input process_data process_spectrometer spectrometer_reconstruct vdc_presort find_clusters vdc_pattern vdc_tracks find_tracks setup_tree order_paths fill_track

Algorithm Details - 1

spectrometer_reconstruct: vdc_presort:

find_clusters: NGAP_CLUS=2 1 wire gap max

vdc_pattern: Determine “pivot point” per cluster

vdc_tracks: Get rough slope/intercept per cluster find_tracks: Find all tracks

Fill structure with “golden” track

Algorithm Details - 2

setup_tree: Establish all possible links (branches)

between clusters Layers ordered as: u1,u2,v1,v2 Fit for u1/u2 and fit for v1/v2:

intercept + slope * (x(i) +/- t0) 3 parameters: intercept, slope, t0 x(i): distance including separation between

planes

Require 4 or more wires in u1+u2 and v1+v2

Algorithm Details - 3

order_paths: Define chi-square:

Rank paths (roads) by ascending chi-square: “Minimum Spanning Tree”

Algorithm Details - 4

2)2,1()2,1(

2002

)],([min

)]2,1()2,1([

00 vvtuut

vvtuut

fill_track: 1st track: track with min. chi-square Examine track with next lowest chi-

square All clusters distinct from track 1?

Yes: this track becomes track 2 No: go to next track and repeat

Sort surviving tracks in order of ascending average t0.

Lowest t0: “golden” track Fill structure with golden track (done in

spectrometer_reconstruct)

Algorithm Details - 5

clusters/plane = 1 multiplicity/plane: 3 to 7 angle_diff: -0.4 to +0.5 slope:

from 0.4 to 1.0 ( u1, v1, v2) from 0.5 to 0.9 (u2)

tracks/spectrometer = 1 (guaranteed by clusters/plane = 1)

|th_tra| < 0.15 Analysis flags:

tdc_cut=2 tx_cor=1

Cuts: “good” events

Possible Strategies

Multi-plane consistency: Angle from single plane determined to

~ 1 deg Track to other, corresponding chamber:

~ 1.5 cm spot (4 wires) Find appropriate cluster Does this cluster point back to our

starting cluster? Slope test:

Define clusters by group of wires with reasonable delta_t between adj. wires

Consistent with scintillator paddles defining trigger?

Cause for high multiplicities, even at low rate? Hardware solutions?

Figures – from E94-004

H(e,e’p) – Run 2591 e_reactz vs. h_reactz (event_type=5) log plot of eu1_slope (event_type=1)

eu1_mult < 6 eu1_mult > 5 eu1_mult > 6

log plot of e_miss (event_type=5) mult < 6 (all 8 planes) mult < 8 (all 8 planes) no multiplicity cuts

D(e,e’p) production runs E-arm efficiency vs. rate (electron.ps) H-arm efficiency vs. rate (hadron.ps)

Figures of event hit patterns