Jean-Baptiste Juin's astrophysics project

Title:
	"Galaxy cluster/point sources separation in Sunyaev-Zel'dovich wide field surveys"
Data:
	Official simulated "Atacama Cosmology Telescope" (ACT) maps
	My data are available at: http://www.astro-udec.cl/juin/
Contact:
	Jean-Baptiste, jbjuin@astro-udec.cl
	(+56)412207248
	Univesidad de Concepcion, Concepcion, Chile

Context:
	In few months/years wide field surveys in millimeter wavelength
	will be carried out to observe galaxy cluster through their SZ signal
	and constraint cosmological models.
	Main issue of such surveys will be to separate SZ signal from galaxy
	cluster from contaminants: radio and infra-red point sources.
	
	Galaxy cluster SZ signal take the form of extended sources (from few pixels
	to more that hundred) with some ellipticity, elongation, orientation etc.
	Radio and IR point sources are point sources convolved by the beam of
	the instrument. They are supposed to be round and of the typical size
	of the instrument beam.

Purpose of the project:
	Find a solution to separate point sources from extended galaxy clusters
	and isolate clusters. A "blind" approach would be great but a "trained" method
	(neural network, SVM, etc.) could be strongly considered if quite robust and efficient
	(you have to prove it !) Of course I can provide more data with different parameters
	to test the robustness of the algorithm.
	I provide different data set Astro79 to Astro89 to work with (more details in Data.README).
	Take the data sets, look at the catalog of sources and their appearance in the corresponding map 
	(you have the coordinate to find them). You are free to define any other relevant parameters 
	than the ones provided in the catalogs (you have the map -all information is in- to do it !) 
	even the a full sub-map	of each source could be used as a input 
	of a trained algorithm (Neural network for instance).
	
	With other words: 
	if you plot for instance the catalog flux versus catalog concentration you will clearly
	see the two population with some overlap... you have to provide a parameters that 
	will allow to separate better than that (my definition of the concentration) the 2 populations.

	I'm completely disponible for more interaction mail & phone: 0056412207248.
	Give your number if you are short of money to call international I can do it.