MEM --- The Maximum Entropy Method
INTRODUCTION
- Developed by Jaynes (1957), (1978)
- Founded on probability theory (Gull & Skilling)
- Confidence testing possible in a rigorous way (in principle, if not practice)
- Other uses are rich and diverse; a few listed here:
- Two versions in the RHESSI release software
- MEM_SATO --- based on HXT MEM program (J. McTiernan)
- MEM_VIS --- Uses "visibilities" A. Conway
- Problems
- Super resolution --- brightness is not constrained
- Relative weight of Entropy & Chisquare not optimized (Lnorm)
- MEM_VIS has problems with multiple rotations
- Confidence testing not fully implemented
- Bayesian derivation of MEM
- Using MEM_SATO
REFERENCES FOR THE CURIOUS
- Jaynes, E.T., 1957, Phys. Rev., 106, 620-630.
Paper
- Review of E.T. Jaynes: Papers on Probability,
Statistics and Statistical Physics, chap. 10, 210-314, R.D.
Rosenkrantz, ed.
- Gull & Skilling
MEMSys5 Manual
Last modified: Sun Oct 17 11:01:39 EDT 2004