on the importance of uncertainties in response functions martin c. weisskopf stephen l. odell allyn...
DESCRIPTION
Changes result in differences less than (TRANSCRIPT
On the Importance of On the Importance of Uncertainties in Response Uncertainties in Response
FunctionsFunctions
Martin C. WeisskopfStephen L. O’Dell & Allyn F. TennantNASA Marshall Space Flight Center
2010 April 142010 April 14 1
We consider two responses for We consider two responses for the Chandra HRMA/LETG/HRC-the Chandra HRMA/LETG/HRC-S.S.
• “Old” response is from 2003.• “New” response is from 2009
December. It includes a correction for higher orders.
• Changes are due to updates in the calibration of this spectrometer.
2
Changes result in differences Changes result in differences less than (<5%) BUT …. less than (<5%) BUT ….
3
The impact of these changes depends upon the spectrum.
4
So you might think that nothing So you might think that nothing much happens.much happens.
5
These parameters characterize These parameters characterize the first model.the first model.
• xsect = vern• abund = wilm• tbvarabs (NH and O free)• dust scattering (2004, ApJ 601,
1050) frozen @ 0.15 • powerlaw
6
7
• Allowing for scattering by the interstellar dust is important• Impacts other parameters @10%
level
Note --- an important aside.Note --- an important aside.
Note --- an important aside.Note --- an important aside.
• I told you most of the relevant details of the model.
• This information is woefully missing in a substantial fraction of published papers.
• Verifying someone else’s results requires knowledge of these details. 8
We fit the pulsar’s spectrum to a We fit the pulsar’s spectrum to a powerlaw with the old response.powerlaw with the old response.
9
We fit the pulsar’s spectrum to a We fit the pulsar’s spectrum to a powerlaw with the new responsepowerlaw with the new response
10
Each response gives an Each response gives an acceptable fit to the data.acceptable fit to the data.
Parameter Old NewNH [1022 cm-2] 0.44 0.17 0.40 0.15O 0.67 0.08 0.78 0.09 1.63 0.02 1.64 0.02Norm [NE @ 1 keV]
0.51 0.09 0.49 0.09
2 () 1460 (1474)
1439 (1474)Errors are for 1 interesting parameter: 1 at
(2min +1).
11
Note --- an important aside.Note --- an important aside.
• I told you all the relevant details of the fit.
• This information is missing in a substantial fraction of published papers.
• Believing the results requires knowledge of these details. Quoted errors for bad fits are generally underestimates.
12
What happens when we add the What happens when we add the thermal component? thermal component?
Case Old New
Low-T limit 0.17 0.20
High-T limit 0.16 0.21
13
Temperatures are 3- upper limits, in keV.
The seeming innocuous refinements The seeming innocuous refinements raised the 3raised the 3 upper limit by upper limit by 24%
14