LAT performance for DC2 and the “4-panel plot” Toby Burnett DC2 workshop 28 June 05 - T. Burnett

The proposal, motivated by Steve Update the LAT performance plots now at http://www-glast.slac.stanford.edu/software/IS/glast_lat_performance.htm Use all_gamma updated for DC2 Connect with Science Tools versions of the IRF functions (see Jim’s talk) Question: Can one reliably use IRF functional representations? DC2 workshop 28 June 05 - T. Burnett

My PSF and Point Source Sensitivity Analysis There are three memos in LAT docs on the subject: AN-3455, a proposed PSF functional representation Propose that we formally adopt it AN-3456, analysis of all-gamma with GR v4r2, DC1 cuts, including fits to the PSF and Aeff AN-4369, Implications for point source sensitivity. Two plots from this memo are in the LAT performance page Examine the functional form DC2 workshop 28 June 05 - T. Burnett

The PSF function: 2 parameters Let  be the deviation between actual and measured direction (McDirErr in tuple-speak), and define the scale factor . Then define the scaled difference u = 0.5(/ )2 If both projected distributions are Gaussian, then 2= x2+ y2  is the standard deviation, and the probability distribution for u is exp(-u). What we observe is not exp(-u), but (1+u/)- See the detailed discussion in the memo: note that this is actually Student’s T. DC2 workshop 28 June 05 - T. Burnett

DC2 workshop 28 June 05 - T. Burnett Panel a: the PSF It is normal incidence. I interpret it as the first bin in cos(), so <26 deg. I evaluate in 2 bins per energy decade, using the Atwood function of the generated energy to scale the deviation DC2 workshop 28 June 05 - T. Burnett

DC2 workshop 28 June 05 - T. Burnett Panel a, cont. Example of the results. The Atwood scaling function removes most of the energy dependence, but the shape depends on energy. Each point is then rescaled back to degrees… DC2 workshop 28 June 05 - T. Burnett

DC2 workshop 28 June 05 - T. Burnett Panel a: the new result DC2 workshop 28 June 05 - T. Burnett

DC2 workshop 28 June 05 - T. Burnett Panel a: a footnote Do the 68% (and 95%) values correspond to the simple analytic function? DC2 workshop 28 June 05 - T. Burnett

DC2 workshop 28 June 05 - T. Burnett Panel b This must include the background-suppression cuts, but that is somewhat science-dependent. The cuts actually introduce a “wart”. We expunge this! DC2 workshop 28 June 05 - T. Burnett

DC2 workshop 28 June 05 - T. Burnett Panel b Points are measured for 0-26 degrees, adjusted by 4% to account for drop Lines are simple “logistic function” representations DC2 workshop 28 June 05 - T. Burnett

DC2 workshop 28 June 05 - T. Burnett Panel c This is slightly energy dependent. Don’t know what was used here, I’ll take the results for 1000 GeV, front+back. DC2 workshop 28 June 05 - T. Burnett

DC2 workshop 28 June 05 - T. Burnett Panel c Curve is a simple linear (in cos theta) fit with two segments DC2 workshop 28 June 05 - T. Burnett