Pointsource Sensitivity Prediction

1
Point-source Sensitivity Prediction

• Anagroup meeting
• Toby Burnett
• 19 July 04
• Updated from May presentation
• See LAT DOCS AM-04369 and references

2
Post DC1 IRF

• Use GlastRelase v4r2
• Major difference Tracker Iterative Recon
• Second iteration of track fit fixed many bad
  energy estimates
• Heathers all_gamma production
• ag-v4r2-merituple-180GeV-prune-20040427-fulltup.ro
  ot
• Apply standard DC1 cuts.
• All analysis code in package users/burnett/IRF v10

3
The fit function
So, two parameters ? scale factor,
corresponds to rms for gaussian ? exponent
4
The function, cont.
5
Example fits
6
Parametrization of average Aeff

• Integrate over theta out to 66 degrees.
• Divide by 4pi for average
• logistic fit is useful, but actually use an
  interpolation

7
Parametrization of energy dependence

• Scaling function applied to data before fitting
  (form suggested by Bill Atwood)
• Parameters are empirical, rough fits so that the
  sigma parameter is approximately 1

8
Energy dependence of ?
9
Application to point-source sensitivity

• How does one measure the strength of a point
  source?
• No background just count, rms error is Poison
  sqrt(N) for N ? 5 sigma corresponds to 25 events
• Background (assume constant) Use maximum
  likelihood to fit assumed PSF
• For N signal events, the expected variance is now
  less than N by a factor depending on the PSF
  shape and the background the factor is the
  integral

b is the ratio of background in 2??2 to the total
signal
10
Application to point-source sensitivity, cont
The variance derating factor
gaussian
?2.5
? 2.0
?1.5
11
Source and background assumptions

• Functions are EdF/dE
• Background analysis of Sreekumar et al.
• Use as an estimate of the actual background
  surrounding most signals.
• Signal
• Index of typical EGRET blazar
• Factor is flux in gamma/cm-2s-1 for Egt100 MeV.

12
Detection efficiency the effect of background

• ratio of effective events to total, for energy,
  and different signal strength levels

13
Result time for 5? or 25 events
14
What energies are important?
Contributions to the sensitivity integral (adding
up to 25) vs. energy, for front and back. (One
year, 7e-9 flux)
15
Bottom line, and To Do

• the 5? sensitivity is
• 1 year 6.6E-9
• 5 years 2.4E-9 (proposal 1.5E-9)
• Verify with likelihood applied to actual
  simulation, especially with Galactic diffusewhat
  is the effect of source confusion?
• See how much error matrix improves the PSF