The effect of UCX+extrapolator on Geant DDC tracks

K. Kainz and W. J. Llope

This page was last updated on November 9, 1999.

First, Geant was used to propagate some simulated protons through the DDC and onto TOF. Two different schemes to derive these simulated proton data are described below. Ntuples were produced for DDC tracks (333), DDC space points (222), and TOF hits (334). The TOF hits ntuple stores the TOF holes where Geant alone puts the tracks.

Next, the digitizer was applied to the DDC ntuples. The digitized tracks were then fitted by UCX, and extrapolated to TOF.

For this test, the default field maps, surveyed TOF geometry, and surveyed (non-rotated) DDC wiremap were used. The DDC Geant geometry file had to be shifted forward in z so that it would line up with the wiremap coordinates, with the upstream face at z = 391, and the downstream face at z = 491. The track-fitting and extrapolation code used is the same as that previously applied to p98 data.

The plots shown below feature the UCX-plus-extrapolator code's placement of the digitized track upon TOF minus Geant's placement, plotted versus Geant's placement. Error bars correspond to error in the mean (left plot) and standard deviation (right plot).

Two sets of simulated protons were used. The first set was derived from p98 data (run 3381), in an attempt to provide "realistic" proton trajectories. The plot is comprised of about 16k tracks.

[Postscript version]

To produce the data in the second set, simulated protons were launched from a single point in the DDC ( (x, y, z) = (5, 0, 410) ). Their momentum distribution was uniform, varying from 2 GeV/c to 6 GeV/c. As for their trajectory, their angle in the xz-plane ranged uniformly from -15 degrees to 45 degrees. This provided full coverage over the TOF array. About 19k events are involved in this plot.

[Postscript version]

If there exists a problem with the way UCX fits the DDC tracks, it should show up in this test; just as the extrapolator undershoots the actual TOF hit when using p98 data, so it should also undershoot when using simulated tracks. No UCX+extrap undershoot is seen in the plots produced; in fact, the variation across the TOF array is easily within one slat width.

However, if the result from UCX refitting agrees well with Geant's propagation of the track through the DDC, then this would rule out the track fitting and extrapolating software as the source of the problem. In the results above, the UCX and Geant extrapolation to TOF agree well, modulo a sidewards shift of the TOF of about 1 cm. This would support the notion that the problem is someplace in the geometry. The TOF geometry was spot tested with a tape measure recently, and those measurements are in agreement with the defualt position of the central wall in software. As for the mapped fields, Trentalange revisited the field positioning and believes that is also correct. We ask for comments.

UCX+extrap minus Geant appears to be quite close to zero for the beam-left and beam-right Rice walls. For the central wall, there appears to be about a 1/2-slat offset present. Incidentally, this is the same order of shift that had to be applied to the central wall in all of the other alignment schemes (field map shifts, TOF shifts, etc.) tried before.

In closing, we note that the the extrapolator undershoot problem is remedied if the non-rotated wiremap is shifted 6 cm toward the target in z. The figure below compares the extrapolator accuracy for p98 data, for the default non-rotated wiremap location (left frame) and after the 6 cm shift (right frame).

[Postscript version]

If the 4-mrad rotated wiremap is used, the extrapolation results before and after the 6 cm shift are much the same as those attained without the rotation, except that the additional central wall shift required would be about 2.2 cm instead of 0.85 cm.

[Postscript version]