Документ взят из кэша поисковой машины. Адрес оригинального документа : http://hea-www.harvard.edu/AstroStat/Stat310MM_VI_VII/hl_20070410.pdf Дата изменения: Tue Apr 10 13:00:24 2007 Дата индексирования: Tue Oct 2 05:30:43 2012 Кодировка:

Introducing BLoCXS and Using it to Estimate Calibration Uncertainties Hyunsook Lee Warnings Flow Charts Calibration Beyond ARF

Introducing BLoCXS and Using it to Estimate Calibration Uncertainties
Hyunsook Lee
CfA

April 10, 2007


Introducing BLoCXS and Using it to Estimate Calibration Uncertainties Hyunsook Lee Warnings Flow Charts Calibration Beyond ARF

Acknowledgement and Warnings
· Vinay Kashyap and BLoCXS developers! · Very erroneous and unclear in many asp ects, since I

am NOT yet a
· computer scientist/engineer · bayesian statistician · high energy astrophysicist.

· Interrupt me for questions, clarifications, corrections,

suggestions, and lessons from your expertise.


Introducing BLoCXS and Using it to Estimate Calibration Uncertainties Hyunsook Lee Warnings Flow Charts Calibration Beyond ARF

Outlines

Main Objective: Introducing Bayesian Low Count X-ray Spectral Analysis (BLoCXS) 1 BLoCXS Flow Charts 2 Planned calibration uncertainty studies with BLoCXS


Introducing BLoCXS and Using it to Estimate Calibration Uncertainties Hyunsook Lee Warnings Flow Charts Calibration Beyond ARF

BLoCXS Flow Charts

· Input / Output · EM and MCMC

Change the screen


Introducing BLoCXS and Using it to Estimate Calibration Uncertainties Hyunsook Lee Warnings Flow Charts Calibration Beyond ARF

Uncertainty Calibration with BLoCXS

Changing ARFs in different ways · Process with many ARFs (Multiple Imputation,?) · Random ARFs within EM/MCMC · Model ARFs (Rima's PCA) and use model based ARFs within EM/MCMC Go Back to the Flow Charts


Introducing BLoCXS and Using it to Estimate Calibration Uncertainties Hyunsook Lee Warnings Flow Charts Calibration Beyond ARF

Previous Study, Drake et.al.


Introducing BLoCXS and Using it to Estimate Calibration Uncertainties Hyunsook Lee Warnings Flow Charts Calibration Beyond ARF


Introducing BLoCXS and Using it to Estimate Calibration Uncertainties Hyunsook Lee Warnings Flow Charts Calibration Beyond ARF

Current Goal

Developing Calibration Algorithms within BLoCXS! Example: q0458.pha, q0458.arf, q0458.rmf, q0458.bkg.pha (from ANETA) Ten peteXXXX.arf 's (from Vinay) - one arf file produced errors (no changes in parameters and loglikelihoods)


Introducing BLoCXS and Using it to Estimate Calibration Uncertainties Hyunsook Lee Warnings Flow Charts Calibration Beyond ARF q0458.arf pete0001 pete0034 pete0068 pete0192 pete0315 pete0317 pete0664 pete0667 pete0895

Summaries on Photon Index Draws

mean 1.5954 1.6096 1.6172 1.5707 1.6477 1.6182 1.5435 1.6012 1.5711 1.6337

var 0.00098 0.00103 0.00100 0.00102 0.00102 0.00099 0.00096 0.00102 0.00094 0.00104

4 5 6 5 8 8 6 2 0 7

5% 1.54305 1.5569 1.5658 1.5184 1.5952 1.5673 1.4928 1.5484 1.5202 1.5807

25% 1.57421 1.58806 1.59583 1.54907 1.62628 1.5965 1.52292 1.57995 1.55062 1.61187

50% 1.59523 1.60964 1.61717 1.57052 1.64789 1.61848 1.54332 1.60129 1.57118 1.63331

75% 1.61688 1.63098 1.63883 1.59209 1.66911 1.63917 1.56426 1.62239 1.59197 1.65556

95% 1.6467 1.66276 1.66975 1.62435 1.70106 1.67065 1.59538 1.65373 1.62151 1.68721


Introducing BLoCXS and Using it to Estimate Calibration Uncertainties Hyunsook Lee Warnings Flow Charts Calibration Beyond ARF

PhoIndex: w/ q0458.arf

0 400 1000

Frequency

1.50

1.55

1.60

1.65

1.70

PhoIndex: w/ 9 random arfs

Frequency

0

4000

1.5

1.6

1.7

1.8


Introducing BLoCXS and Using it to Estimate Calibration Uncertainties Hyunsook Lee Warnings Flow Charts Calibration Beyond ARF

norm: w/ q0458.arf

Frequency

0
0.000125

400 800

0.000135

0.000145

0.000155

norm: w/ 9 random arfs

0 4000 10000
0.00012

Frequency

0.00013

0.00014

0.00015

0.00016

0.00017


Introducing BLoCXS and Using it to Estimate Calibration Uncertainties Hyunsook Lee Warnings Flow Charts Calibration Beyond ARF

Help Wanted!

· Sensible Implementation of these schemes requires

feedbacks from statisticians and astronomers. · Prior to this, thorough understanding BLoCXS is a must. · Uncertainties in RMF, though challenges lay in matrix computation, should come along soon.