Welcome to the Phobos Multiplicity Reconstruction Corrections page

This note discusses the multiplicity reconstruction acceptance corrections. If you need information on running the Multiplicity package look here, or here for general Phobos information.

The reconstruction uses the current standard 'tune', that is, settings of the truncation energy, energy deposition and secondary fraction that were created for events from the HIJing generator. In this example, I have reconstructed data from a dataset named

hijing_19981216_1_400_500.root

which has very different characteristics, as the standard HIJing tracks are augmented by the addition of a 'plasma' consisting of a number of high energy pions emitted near zero pseudorapidity. Because these additional particles are not as likely to hit magnet and support materials as the average track, they have a lower secondary fraction than the standard tune and therefore yield a low value of reconstructed multiplicity. Also, the energy deposited by these tracks is less than the average for HIJing events. This difference is quite striking, as can be seen from the standard multiplicity plot here. The lower left hand frame shows the cross section dN/dEta, where the red points are reconstructed and the blue are generated. I use the term 'generated' in this note for values obtained by counting all tracks that had hits on multiplicity sensors; this is not at all the same thing as counting all tracks, as we'll see below. The lower right frame shows one reason; the green circles are the parametric fits to the Total/Primary track fractions, the blue are again generated values, and the low number of generated secondaries stands out clearly in comparison with a plot of normal HIJing events. Similarly, the difference in energy deposition per track can be seen in the upper right hand frame.

The most important question, therefore, is this: since we will not know a priori the average characteristics of RHIC events, how well will our reconstruction do when applied to events that are quite different from the model to which we have tuned our parameters?

To answer this, I present a table summarizing the acceptance tree for 10 events from the plasma-augmented dataset.

These numbers are generated in TPhMul and printed by the macro $PHATHOME/phatmul/macros/tabmul.C, which can be invoked by specifying it as the end-of-event macro when running runmul.C.

Phobos multiplicity reconstruction MC track acceptance for 10 events

........ Number .Charged .* W/hits . Silhits .Mulhits . Active .... Used .. Mapped
This ev . 16641 ... 8993 .* . 7418 .... 7340 ... 6280 ... 6280 .. 6280.0 .. 7022.0
. fract ... 1.0 . 0.5404 .* 0.4458 .. 0.4411 . 0.3774 . 0.3774 .. 0.3774 .. 0.4220
Primary . 13350 ... 6376 .* . 5171 .... 5160 ... 4900 ... 4900 .. 4900.0 .. 5581.0
. fract ... 1.0 . 0.4776 .* 0.3873 .. 0.3865 . 0.3670 . 0.3670 .. 0.3670 .. 0.4181

All evts 162361 .. 86839 .*. 71715 ... 71117 .. 61018 .. 61018 . 61018.0 . 68078.0
. fract ... 1.0 . 0.5349 .* 0.4417 .. 0.4380 . 0.3758 . 0.3758 .. 0.3758 .. 0.4193
Primary. 131207 .. 62156 .*. 50533 ... 50438 .. 47891 .. 47891 . 47891.0 . 54170.0
. fract ... 1.0 . 0.4737 .* 0.3851 .. 0.3844 . 0.3650 . 0.3650 .. 0.3650 .. 0.4129

Reconstructed 6224.8 charged primaries, avge= 5995.1, sum= 59951.4
Generated ... 6233.5 charged primaries, avge= 6043.2, sum= 60432.5

# charged primaries = # reconstructed * 1.008 (tune) * 1.029 (acceptance)

The first four lines are for the most recent event, in this case the 10th, which had 13350 primary and 16641 total tracks. The Phobos Monte Carlo usually saves all primary tracks but only those secondaries that produce hits, and their parents. Each primary track typically produces about 35 secondaries; most of these half-million particles are electrons that are absorbed in the interior of the magnet and coils and play no role in our analysis.

Following along the rows, we see that about half the tracks are charged, and about 17% of those slip through the cracks or out the ends and produce no hits in any sensor. 1% of those that produce hits do not hit a silicon sensor, and 86% of those that hit silicon have at least one hit in a ring, octagon or vertex sensor. No sensors are inactive at present (many will be for the engineering run) and no tracks were rejected for producing more than 30 hits. The column labelled 'Mapped' shows the number of tracks entered into the TPhPadMaps for reconstruction. This is 12% larger than the true number of tracks 'Used' because the reconstruction of necessity double-counts tracks that hit more than one layer. The real data have no way of telling us how they were generated; energy deposited in a pad means a particle impact. This is correctly accounted for in the reconstructed cross section by adding the phase space occupied by each pad to the total phase space.

The next four lines show the same information summed over 10 events.

The following two lines show the reconstructed number of charged primaries, for the last event and summed over all events. We see that the reconstructed multiplicity on the average is low by 3.5%. The second line shows that the average reconstructed multiplicity is lower than that generated by 0.8%. This difference is the error in tuning, that is, the systematic effect of the difference between

The remaining 2.9% can therefore be attributed to geometric acceptance and losses in the beam pipe. I estimate the losses from the high-eta regions to be ~ 1% by extrapolating dN/dEta; this will be done automatically in a future version of the multiplicity package. This leaves 1.9% for losses in the beampipe.

I have purposely not attributed errors to these corrections. The statistical errors are insignificant. We have to make an estimate of the systematic errors in the tune and acceptance corrections, and this is Art, not Science. The next planned step is to carry out this analysis for a range of models, including some quite bizarre ones, and that will provide us with at least a quotable range of corrections.

Summary: the most stringent test we can give the multiplicity reconstruction package is to ask it to calculate the total charged multiplicity for a model very different from the model used to calculate the reconstruction parameters. It passes this test with flying colors.

Back to top

Last edited: Wednesday, March 31, 1999 08:27:40 AM by Robin Verdier [verdier@mit.edu]