sPHENIX tracking discussion

[Anthony Frawley <afrawley AT fsu.edu>]

Hi All,

I have been experiencing problems with tracking in pp events following the introduction of the new MVTX strobe simulation code. It is not a problem with the strobe simulation code, that seems to be fine.

The problem comes from the fact that we ended up having to record the strobe in the hitsetkey, when we would have preferred to put it in the hitkey. But the hitkey does not have any spare bits. However the strobe simulation frequently generates multiple hits per signal, each has a different strobe ID and therefore each is assigned to a different hitset. Because our code assumes that all clustering is done within a hitset, the presence of multiple copies of a hit distributed across different hitsets plays hell with our entire reconstruction and truth tracing chain. It seems to be especially problematic in pp reconstruction, where the MVTX hits can occur over a many microsecond time range.

I have been looking at how to accommodate the new MVTX "data" stream (and we do have to do that). In the end, the least invasive way to do that seems to be to create a new module that runs on the hits to do the following:

1. Reads all MVTX TrkrHits from the node tree, locates all instances of repeated hits, and chooses one of them to add to a pruned hit list.
2. Reassigns all hits in the pruned hit list to the appropriate hitset, with strobe ID set to zero.
3. Writes the pruned hit list to the node tree.

This effectively discards the strobe ID information from the hits, restoring the expected operation of our code. In future, if (when) we want to capture the strobe information, we can make a table in the new module that preserves the strobe ID information for the MVTX hits for later use.

I am planning to implement this now. If anyone has objections or comments, speak now ....

Cheers

Tony