S.J. Bailey, D.J. Prindle, S. Schönfelder* and T.A. Trainor
The large data volume recorded in December, 1994 was used for robust testing of the tracking software for the NA49 Main Time Projection Chambers (MTPCs). During 1995 these data were used to fine-tune the tracking parameters, test for distortions and inefficiencies, and make improvements in the software. In October of 1995 this software was used to begin production analysis of the 94 data.
During and after the analysis of the 94 data a number of improvements were made to the analysis chain. First, new data structures were introduced to allow a more global approach to the analysis. These allow the various TPCs to be treated simultaneously rather than segregating each one as a separate physics analysis. It is still more efficient to do the track finding process for the Vertex TPCs (VTPCs) separately from the MTPCs, but the resulting data files can be merged so that physics analysis may be done for all TPCs together.
The new data structures were also reorganized to reduce memory requirements. During the 94 production lack of memory per CPU was a perpetual problem and reduction of the size of the structures has been crucial for the stability of the software within the limited batch processing facilities available. The script for running the software was also entirely re-written in order to provide more robust error handling, flexibility, and debugging capabilities.
As part of the continued development of the tracking software we modified several algorithms and were able to improve the MTPC tracking speed by a factor of five without a loss of tracking efficiency. This established a precedent, and significant timing improvements were also made in gain calibrations, the MTPC momentum module, and VTPC tracking. The entire MTPC analysis chain for both MTPCs now takes less than 1.5 minutes per event whereas previously the tracking for one MTPC alone took 3-4 minutes.
During recent development of the MTPC software a number of general-use modules were created. These modules were initially written only for the MTPCs, but because of their utility they were converted for use by the VTPCs as well. The prime example is a drift speed module. Knowing the exact drift speed is crucial to the correct reconstruction of the y position (and thus momentum) of tracks. In NA49 numerous hardware methods were devised for determining the drift speed in the TPCs, including laser calibrations and drift speed monitors. But the most accurate method of determining the drift speed comes from looking at the TPC charge distribution itself. By counting the number of timebins in which data occurs and knowing the exact vertical dimensions of the TPCs, the drift speed can be determined with a precision of a few parts in 104. This module has become the standard against which other drift speed monitors are compared. Its use in the analysis chain has significantly improved the quality of the reconstructed tracks.