Cornell ILC Global ILC

WWS Home

"TPC signal digitization simulation and reconstruction studies"

Full efficiency for charged particle reconstruction is required for precision particle flow analysis. A TPC for the linear collider must be designed to provide this efficiency over a large solid angle in an environment of high density jets and high noise occupancy. Reconstruction efficiency can be improved with higher readout pad segmentation to the limit of the signal charge width, about 1mm. While it is not even proven that this maximum segmentation would be sufficient to provide full efficiency, there are other limitations to such a high pad density: cost, material, heat, and complexity, that may force a larger pad size. Thus, it is necessary to optimize the readout geometry for reconstruction efficiency.

Simple models of the TPC response are not sensitive to the issues of noise. In the simple models, TPC hits are simulated as 3-dimension space points in the detector as shown in figure 1 on the project page. Charge depositions are treated independently providing no straight-forward way simulate the effects of overlapping hits.

The goal of the this study is to model the charge spreading and signal overlap to provide sensitivity to the effects of overlapping tracks and noise. In the Cornell simulation, charge is spread over neighboring pads (figures 2 and 3), and then accumulated in a simulation of the pulse train (figure 4) observed by the readout electronics, usually a FADC. The pulse train is analyzed to find the unambiguous threshold crossings in the same way that real data is analyzed. Thus, signal overlap is fully simulated. Reconstruction of tracks from the resulting charge/time signals is far more complicated, requiring pattern recognition of time clusters (figure 4) and space clusters (figure 5.

This study uses a modification of the CLEO track reconstruction algorithm to recognize track in this complex environment. Preliminary results were shown at LCWS04, Paris (April 2004). Results indicating the pad segmentation sufficient for full pattern recognition were shown at ALCPG, Victoria (July 2004). Most recently, at ECFA, Vienna (November 2005), results were shown for the noise tolerance (figure 6) for the specific case of random noise hits and a particular pad size.

This study currently uses the older sio data format. Further developments will require adopting the LCIO data format and access routines which will allow access to larger samples, more complicated events and the addition of noise hit distributions derived from beam halo calculations.

In the near future, the simulation program will be modified to interface to the LCIO format directly. The next step, later in the next year, the simulation will be fully integrated into the reconstruction frameworks, such as Brahams-Reco. This will make the simulation accessible to researchers and allow expanded studies of the efficiency dependency on the readout geometry and the various possible noise conditions. The interface to LCIO and integration into the reconstruction frameworks will also provide opportunities to further optimize the pattern recognition.

Support is not sufficient. Up to this point, all development has been done by D. Peterson. Further development will require support for a student to perform the interface and integration work as described above.

Please address the following questions in your statement.

  • What are the goals of this R&D project. How does this R&D project address the needs of one or more of the detector concepts?

  • If there are multiple institutions participating in this project, please describe the distribution of responsibilities.

  • Are there significant recent results?

  • What are the plans for the near future(about 1 year)? What are the plans on a time scale of 2 to 3 years?

  • Are there critical items that must be addressed before significant results can be obtained from this project?

  • Is the support for this project sufficient? Are there significant improvements that could be made with additional support?