sPHENIX tracking discussion

[Anthony Frawley <afrawley AT fsu.edu>]

Hi All,

I wanted to summarize for posterity the results of our discussion this morning on truth evaluation for tracking. Let me know if I got anything wrong.

The motivation is that we want to get rid of the existing map (TrkrHitTruthAssoc) that stores the associations between reconstructed hits and g4hits. It is a very large map (in memory and disk usage) that only very indirectly carries the information we want. After a long discussion we decided that the following approach makes more sense. This summary is focused on the TPC, but can be generalized to include the silicon detectors.

In the electron drift code, modify the existing loop over g4particles and their associated g4hits so that:

  -- For a given g4particle the charge deposited in each layer is calculated.

  -- The loop over g4hits is converted to a loop over layers.

For each g4particle and layer:

  -- The undistorted energy and position of the charge in the layer is recorded - this is the ultimate cluster truth information.

  -- After diffusion and distortion of the primary electrons:

      -- The distorted_truth information is calculated by combining all pad readouts for this g4particle and layer, and it is recorded.

      -- The reconstructed hits are accumulated over all g4particles and layers, as now.

The resulting map of (g4particle, layer, undistorted cluster position, distorted cluster position) is then stored as the truth tracking information. It is used after reconstruction by a redesigned evaluator to create maps of:

  -- Cluster-truth cluster

  -- Cluster-distorted_truth cluster

  -- Track - g4particle

The working groups will use the track-g4particle map for evaluation.

The cluster truth maps will be used by tracking experts.

By eliminating hit-g4hit associations, which are never used directly anyway, the memory requirement for storing truth information is greatly reduced.

Cheers

Tony