I continue here where I left off before in my last post. This is my final post, and it details the results and conclusions of my research:
The correlation values produced by my software make sense because they appropriately reflect the nature of the respective property and main detector. For example, targetY is highly positively correlated with the asymmetry data from main detector three, which is located at the top of the ring of main quartz detectors. This makes sense, because a positive shift in the Y direction would appear as an increase in scattered electrons measured by main detector 3. Additionally, the values of C_k used in the calculation match those of LinRegBlue when correcting for only the 5 beam properties (as LinRegBlue does). The program calculates the initial correlations correctly.
The correction itself is successful, as the residual correlations between targetX, targetY, targetXSlope, targetYSlope, charge, and energy and each of the main detectors go to zero after correction.
The final calculation of the asymmetry will use a mean of the mean values of each of the run segments. While the correction to the mean value is unimportant as of now (the blinding factor is still in place), it reduces the width of the distribution of the data for each of the main detectors, eventually contributing to a higher degree of precision in the final result.
Many variables were chosen to compare with the main detector data after the initial correction. These include linephase (a measure of the phase of the A/C power source), raster position (the position where the beam was ‘painted’ on the target to prevent it from burning through), main detector nine (an additional main detector used to measure background outside of the ring of eight primary main detectors), bare photomultiplier tubes (used to measure background), beam position monitors, luminosity monitors, and finally, the initial 5+1 properties that were corrected for (to check that the correction was done correctly). The most striking remaining correlations are those that exist with main detector 9 and the bare photomultiplier tubes that remain across all runs and run segments. For individual runs, some beam position monitors (especially those sensitive to energy) remained correlated with the main detectors (although the correction for targetX and targetY removes most of the correlations with the beam position monitors).
Residual correlations produced by this software may indicate beam properties that should be the subject of future corrections.
Early in the process of writing this software, I experimented with correcting for several beam position monitors and main detector 9 for the particular run I was using (QwPass3 10055.000). The correction made a slight reduction in the width of the distribution of main detector data.
Preliminary results were posted to the Qweak electronic logbook, which contains the most up to date research of members of the Qweak collaboration. For the purposes of presenting results to the Qweak collaboration, my program was run for additional runs, including QwPass3 10055, 10056, 10057, and QwPass5beta 11700. Correlations vary over runs and run segments, something that will be the subject of future study.
After posting some of my preliminary results to the group’s electronic logbook, I have been in contact with several other Qweak collaborators who have requested that I run my software for additional runs and for additional variables. With more study, this information may be used to correct for more beam effects in the final data, contributing to a higher level of precision in the final experimental result. As a rising sophomore, I have found this opportunity to have this level of involvement in research early to be gratifying. These results and conclusions are preliminary and I will continue work on this research through the fall of 2012.
This research accomplished its main initial purpose: software has been created to correct Qweak asymmetry data for beam properties and plot the remaining correlations with many other beam properties. However, the full extent of its application to many runs and many more beam properties has not yet been fully explored. Additionally, we want to be able to explain why there are remaining correlations with some beam properties after the correction is completed. If we can understand why these correlations exist, we will want to study whether they should be the subject of future corrections, and if this should be done through linear regression or some other method.