Main idea is using tof hits as a seed to select a regian in TPC to find the first hit for building a track. Because tof is a fast detector, it is helpful to reduce pile up events and reduce the time used in tracking, making it possible to run HLT on TPX machines in a high luminosity situation (pp 500GeV).

The basic method is calculating the (r,phi,z) for tof hits and selecting it for different TPC sectors on tof machine, then passing the seed data to TPX machine for the tracking.

When selection seeds we broaden the phi region of TPC sector by 0.1rad and broaden the z range by 20cm to make it continuous at the border.

The method used to find first hit to build a track is: in phi direction we assume a rcut (which is coming from a pt cut to make sure the track can reach tof) for the helix and calculating the position of the center. Then the cross region of the two circle (red line) is the region where to find the first hits. In Z direction we using vzVpd and 4sigma for the possible position of the vertex and draw lines from them to the tof hit and broaden it by the module width in Z direction.

Using these method we archive an improvement in tracking time by about 50%~60% with a parameters set: rcut = 133, moduleZRange = 7.22, padlocalXRange = 3.5.

Also a test on the lower layer to search the first hits is done. As default the tracking code searches first hits from layer 45(the most out layer of TPC) to layer 5. The upper layer is fixed and we changing the low layer and test the efficiency and tracking time. Using the default first hits searching method, we can get an 55% improvement of tracking time, it is already close our expectation. And the efficiency is about 34%. Considering of the tof matching efficience(~70%), and pile up tracks the efficiency is reasonable.

A checking of nhits in globaltrack:

The two histograms are scale by the their entries.

So we can see there are more long tracks left than short tracks in tracking with TOF seeds.

Also there is a ratio of nhits.

We can see the ratio of tracks with hits>37 is about 0.3. It is consistent with the primary track efficiency. And the ratio of tracks with hits<10 is about 0.12, this is the efficiency we get from global tracks.

A check of tof trigger crossing.This is a plot which shows the tDiff = tdc - triggertime vs the tof tray ID.

We can see for each tray there is only one peak within 100ns, so only one crossing is readed.