Difference between revisions of "Reconstruction Tasks and Topics for Further Development"
(→Comments from Curtis) |
|||
| Line 22: | Line 22: | ||
With regard to (1), the problems manifest themselves as large chi-squares from tracking and the tracking errors being problematic in a way that cannot be systematically corrected. I would guess that these are correlated with underlying issues in: | With regard to (1), the problems manifest themselves as large chi-squares from tracking and the tracking errors being problematic in a way that cannot be systematically corrected. I would guess that these are correlated with underlying issues in: | ||
| − | + | # pattern recognition where incorrect hits are used. | |
| − | + | # large scattering in material between CDC and FDC. | |
| − | + | # possible errors in the fitting procedure itself. | |
| − | + | # other as yet unknown causes. | |
I think that the only way that we are going to make progress here is to go systematically through the code and try to identify problems and fix them. I believe that Simon, Paul M. and possibly Kei after the beam tests. | I think that the only way that we are going to make progress here is to go systematically through the code and try to identify problems and fix them. I believe that Simon, Paul M. and possibly Kei after the beam tests. | ||
With regard to (2), my feeling from our b1-pi studies is that this is coupled to the BCAL reconstruction. In this case, the big improvements will hopefully come from tuning the algorithm in the the existing code. However, there are some caveats. | With regard to (2), my feeling from our b1-pi studies is that this is coupled to the BCAL reconstruction. In this case, the big improvements will hopefully come from tuning the algorithm in the the existing code. However, there are some caveats. | ||
| − | + | # The BCAL readout segmentation will affect this. We need to settle this asap, or move forward under the 1-2-3-4 assumption. | |
| − | + | # Once segmentation is settled, we can go back and tweak the parameters in the BCAL code. | |
In this case, Matt updated code which is checked in and Will L. is now looking at it. Modulo (i) being decided, I hope that we can move forward fairly quickly here. | In this case, Matt updated code which is checked in and Will L. is now looking at it. Modulo (i) being decided, I hope that we can move forward fairly quickly here. | ||
Finally, for (3), this is a higher level issue that will ultimately require (1) and (2) to be improved.Your posed questions put this in the light more of discussion, rather than underlying software issues. While it may lead to issues as we move forward on the above, it is hard to say at the moment. | Finally, for (3), this is a higher level issue that will ultimately require (1) and (2) to be improved.Your posed questions put this in the light more of discussion, rather than underlying software issues. While it may lead to issues as we move forward on the above, it is hard to say at the moment. | ||
For progress, Paul M. is going through the code looking at the inputs to the PID and trying to make sure that they make sense. | For progress, Paul M. is going through the code looking at the inputs to the PID and trying to make sure that they make sense. | ||
Revision as of 12:52, 7 September 2011
Mark's Draft List
The items listed here are areas that we have identified as needing further development and/or study. Feel free to add your topics or comment on those already listed. Comments can be entered on the "discussion" tab (see above).
- Particle identification
- What is the general scheme (or schemes) that we want to support?
- What is the nature of the user interface into particle ID information?
- χ2 distributions from track fits
- What are the main causes of tracks in the zero probability peak?
- Are hit reconstruction errors calculated correctly? Consistently?
- Re-do checks of error matrices from photon reconstruction
- When photons get split into multiple clusters, is all of the energy detected?
- How do we handle large single Coulomb scattering in tracking?
- How do we handle decays in tracking?
- What is our efficiency for π0 detection? η→γγ?
Comments from Curtis
This looks like a good list. I might suggest that it be reworked a bit. In particular, we have three broad topics with issues:
- tracking
- photon reconstruction
- particle identification.
With regard to (1), the problems manifest themselves as large chi-squares from tracking and the tracking errors being problematic in a way that cannot be systematically corrected. I would guess that these are correlated with underlying issues in:
- pattern recognition where incorrect hits are used.
- large scattering in material between CDC and FDC.
- possible errors in the fitting procedure itself.
- other as yet unknown causes.
I think that the only way that we are going to make progress here is to go systematically through the code and try to identify problems and fix them. I believe that Simon, Paul M. and possibly Kei after the beam tests.
With regard to (2), my feeling from our b1-pi studies is that this is coupled to the BCAL reconstruction. In this case, the big improvements will hopefully come from tuning the algorithm in the the existing code. However, there are some caveats.
- The BCAL readout segmentation will affect this. We need to settle this asap, or move forward under the 1-2-3-4 assumption.
- Once segmentation is settled, we can go back and tweak the parameters in the BCAL code.
In this case, Matt updated code which is checked in and Will L. is now looking at it. Modulo (i) being decided, I hope that we can move forward fairly quickly here.
Finally, for (3), this is a higher level issue that will ultimately require (1) and (2) to be improved.Your posed questions put this in the light more of discussion, rather than underlying software issues. While it may lead to issues as we move forward on the above, it is hard to say at the moment. For progress, Paul M. is going through the code looking at the inputs to the PID and trying to make sure that they make sense.
