|
|
|
Novel Methods for the Analysis of Photon-Limited Data
|
|
|
| tatistical techniques, and efficient algorithms to implement them, will be developed for analysis of event data (a.k.a. point data)
in astrophysics, space and Earth science, and other applications. These methods, integrated with existing standard ones, will be
described in a Handbook of Statistics for Event Data, providing scientists with practical guidance on use and interpretation. This
choice of problem area is based on my experience and expertise, and deliberately focuses on an important class of NASA missions.
Event data are largely photon detections: photometry, imaging, timing, and spectrophotometry. Most important in high-energy
astrophysics, where arrival direction and time, plus the energy of individual photons can be accurately measured, the methods will also
have broader applicability. The specific problems addressed include repeated measurements; detection, characterization, and upper
limits (for signals consisting of spatial sources, time variability, or spectral lines); and analysis of image and other structure. Algorithm
design will focus on the practical issues of user comprehension of inputs and processing, plus understanding and interpretation of the
outputs. Statistically rigorous modeling will be combined with robust applicability to diverse contexts, ease of use, transparency of
parameter adjustment, validity and applicability of approximations, unification of novel and old methods, and computation efficiency.
Matlab algorithms (with adaptations to other high-level systems in common use, such as IDL, R, S, and Python, as well as low-level
compiled languages) will posted in the AISRP code repository, with documentation there and in the accompanying Handbook. |
|
|
