EMC2

Bayesian Image Analysis

hea-www.harvard.edu/AstroStat/EMC2

| Code | References | Usage | How to |

The code

The latest version is ver.11-10-2007

The code is a combination of C and R-based programs. The theory behind it is described in detail in the EMC2 paper (Esch et al., 2004, ApJ, 610, 1213; this is the citation to use if you used this code in your analysis). It has been successfully compiled and run on Mac OS X.4 and various Linux flavors. To install, first download the tar file that contains all the necessary files to a suitable directory (e.g., /path/to/EMC2/installation/directory/) and then type

	cd /path/to/EMC2/installation/directory/
	gunzip -c Bayes-Image-Analysis.tar.gz | tar xvf -
	cd Bayes-Image-Analysis/
	R CMD SHLIB bayes-image-analysis.c

Then, launch R from that directory, and within R, type:

	dyn.load( "bayes-image-analysis.so" )
	img <<- function(){ .C("bayes_image_analysis", map=double(1))
	img()

1. Standard (i.e. Null) model (Galactic Diffuse from GALPROP etc.)
2. Simulated datons from the null
3. Simulated datons from Null + Bright extra discontinuous component
4. Wrong model (squished above the Galactic plane) for Null

References

Connors, A., van Dyk, D.., Chiang, J., & CHASC, 2008, AAS-HEAD, Meeting #10, #3.01

2008HEAD...10.0301C [ADS]

 

Connors, A. & van Dyk, D.A., 2007, SCMA IV, ASP Conf. Series, Eds. G.J.Babu and E.D.Feigelson, Vol. 371, p101

preprint from 2007-aug-17: [.pdf]

2007ASPC..371..101C [ADS]

 

Esch, D.N., Connors, A., Karovska, M., & van Dyk, D.A., 2004, ApJ, 610, 1213

2004ApJ...610.1213E [ADS]

[.pdf]

 

Usage and Syntax

Tutorial

How To

EMC2 is optimized to run as a batch job

all the parameters are input via a text file.

 

Each parameter must be explicitly specified

0                 # 0 for MCMC, 1 for EM
1                 # 1 to fit a scale parameter to background
20                # number of draws/iterations
1                 # 0 for ML (Richardson Lucy if EM), 1 for MS, 2 for MRF
128 128           # size of data
3 4 5 6 7 8 10    # starting values for alpha. should be 7 numbers for 128 x 128
1 0.05            # gamma prior on total cnt from MS component
100               # MS smoothing parameter

FOR ML fitting, the last three lines are ignored

for MRF fitting there is a single smoothing parameter that should appear on line six.

 

Unresolved Questions:

what is the official acronym? TBD

what is the official name? TBD

how are you going to handle version control? BY DATE STAMPS; INCLUDE A PRINT STATEMENT FIRST THING IN CODE SAYING WHAT THE VERSION IS

is it an alpha release? beta? who is the blamee? BETA, BUT INTERFACE IS ALPHA. DIRECT QUERIES TO JASON

what machines has it been run on? MAC (TIGER), LINUX (GENTOO 32-BIT)

tutorial and "How To"s need to be fleshed out TBD

are there more/better references? TBD

| Code | References | Usage | How to |