1 / 21

UNED EUMETSAT

A new approach to the optimization of the extraction of astrometric and photometric information from multiwavelength images in cosmological fields. UNED EUMETSAT. Structure. Data Mining Needs : m ultiwavelength observations Objective 1: preliminary labelling

cormac
Download Presentation

UNED EUMETSAT

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. A new approach to the optimization of the extraction of astrometric and photometric information from multiwavelength images in cosmological fields UNED EUMETSAT GREAT Workshop

  2. Structure GREAT Workshop • Data MiningNeeds: multiwavelengthobservations • Objective 1: preliminarylabelling 2.1. Objective 1: candidatesforisolatedsources 3. Objective 2: thecross-matchingproblem 3.1. Objective 2: theresolution of thecross-matchingproblem 3.2. Objective 2: implementation of theastrometriciterativecross-matching 3.3. Objective 2: addingphotometrytothecross-matchinginference 3.4. Objective 2: astrometric and photometricBayesfactors.

  3. Structure GREAT Workshop 3.5. Objective 2: an extended framework for the cross-matching problem 3.5.1. Objective 2: bayesian Inference for the consideration of non detection 3.5.2. Objective 2: hypothesis of the bayesian inference for the extended framework 3.5.2. Objective 2: resolution of all possible combinations of non detection 3.5.3. Objective 2: generic formalism of the bayesian inference for the extended framework 3.5.4. Objective 2: example of the extended framework for the case of three catalogues. 4. Conclusions and Future Lines of Work.

  4. Data Miningneed: multiwavelengthObservations IRAC FOCAS GREAT Workshop

  5. Objectives FLAG GREAT Workshop • Objective 1: tolabelthesources. • Objective 2: toconstructreliable sets possibleoverlapping Note: in thefollowingwewill use thetermchanneltorefertoanimageobtained in a givenpassband.

  6. Objective 1: preliminary labelling • FITS Images CATALOGUES Sextractor GREAT Workshop

  7. 2.1. Objective 1: candidates for isolated sources Future evolution SVM Maximum margin hyperplane. GREAT Workshop

  8. 3. Objective 2: the cross-matching problem • Identification of thesamesourceacrossmultiplewavelengthobservations. • BayesianastrometriccrossmatchingbasedonBudavari and Szalay (ApJ, 679 , 2008). • Twomutually exclusive hypothesis: • H: all positions correspondto a single source. • K: notall positions correspondto a single source. • H will be representedby a single position onthesky and K will be representedbydifferent celestial coordinates. • HypothesiscomparisonbasedontheBayes factor, implemented as aniterativeprocess GREAT Workshop

  9. 3.1. Objective 2: the resolution of the cross-matching problem • p(m|H): probability that the source is in the position m. • p(xi|m,H): probability that one source of channel i which is in the position m is detected in the position xi D = data composed of the positions measured. m = real position parametrized with a three dimensional normal vector. GREAT Workshop Bayes Factor as the ratio of the two following evidences:

  10. 3.2. Objective 2: implementation of the astrometric iterative cross-matching GREAT Workshop This method computes, for n catalogues in an iterative way, the overall Bayes Factor in every step assuming that all other subsequent catalogues will contribute sources at the best possible position. It establishes a correspondence between each Bayes Factor and a distance cut-off.

  11. 3.2. Objective 2: implementation of the astrometric iterative cross-matching B0 = 5 σ ≤ 0.2’’ GREAT Workshop

  12. 3.3. Objective 2: adding photometry to the cross-matching inference GREAT Workshop The formalism introduced so far is obviously valid when using photometric information. Budavari and Szalay (ApJ, 679 , 2008) have introduced a Bayesian framework to photometric measurements in various passbands. In the simplest case a model can be parametrized by a discrete spectral type T, the redshift z and an overall scaling factor for the brightness α.

  13. 3.4. Objective 2: astrometric and photometric Bayes factors GREAT Workshop Similarly to the astrometric equations, the photometric Bayes factor is given by the ratio: The Bayesian analysis is inherently recursive. A consequence of this is that the combined Bayes factor of the astromeric and photometric measurements is simply the product of the two. Jakob Walcher, Brent Groves, Tamás Budavari and Daniel Dale, 2010

  14. 3.5. Objective 2: an extended framework for the cross-matching problem GREAT Workshop Using a Bayes factor that includes photometry, we can conclude that a n-tuple produces an inconsistent SED if the hypothesis K is more probable but once this conclusion is reached, the SED is weeded out. We propose here a formalism by which we can select which measures of the SED are consistent and therefore an incomplete but useful SED is produced.

  15. 3.5.1. Objective 2: bayesianInferencefortheconsideration of non detection • Photometricinformationistakenintoaccountthrough SED models. • Thedetectionprobabilitydependsdirectlyonthespatial flux density. Forsimplicityweassumehere a simple thresholdontheintegrated flux. • Startingpoint: each of the n-tuplesfromtheastrometriccross-matchingimplementation. GREAT Workshop

  16. 3.5.2. Objective 2: hypothesis of the bayesian inference for the extended framework η: parameters for modelling the SED • Hypothesis: • H1: allthefluxesgicorrespondtothesamesource . • K1: notallthefluxesgicorrespondtothesamesource Cn,1 ; Cn,2; Cn,3;...;Cn,n GREAT Workshop

  17. 3.5.3. Objective 2: generic formalism of the bayesian inference for the extended framework • Hypothesis K1 can be decomposed in allpossiblecombinations of non detections. • Resolution of thecorrespondingcombinatoryproblem. • One new sub-hypothesisisestablished per combinationderivedfromthecombinatoryproblem. GREAT Workshop

  18. 3.5.3. Objective 2: generic formalism of the bayesian inference for the extended framework GREAT Workshop

  19. 3.5.4. Objective 2: example of the extended framework for the case of three catalogues. GREAT Workshop

  20. 4. Conclusions and Future Lines of Work. GREAT Workshop Mainconclusions: • Theproposedframeworkallowsfortheidentification of thechannelsthat produce a consistent SED. • Theselection of prior probabilities has a stronginfluenceontheselectedmodel. Futurelines of workderivedfromcurrentlimitations: • Theconsideration of more thanone non detection per channeldeserves a specificstudy of theproblem. • Theboundariesderivedfromthevoronoitessellation can be improved. • Theconsideration of real astrometric position as a mathematicalpointimplieslimitationsforthe case of extended sources.

  21. Thank you for your attention GREAT Workshop

More Related