AKAIKE MODEL SELECTION CRITERION APPLIED TOSUPERNOVAE DATA

摘要

In this contribution we apply model selection approach based on Akaike cri-terion as an estimator of Kullback-Leibler entropy. In particular, we presentthe proper way of ranking the competing models based on Akaike weights (inBayesian language - posterior probabilities of the models). This important in-gredient is missing in alternative studies dealing with cosmological applicationsof Akaike criterion. Out of many particular models of dark energy we focus on four:quintessence, quintessence with time varying equation of state, brane-worldand generalized Chaplygin gas model and test them on Riess' Gold sample. As a result we obtain that the best model - in terms of Akaike Criterion - isthe quintessence model with evolving equation of state. The odds suggest thatalthough there exist differences in the support given to specific scenarios by su-pernova data all models considered receive similar support. One can also noticethat models similar in structure i.e. ACDM, quintessence and quintessence withvariable equation of state are closer to each other in terms of Kullback-Leiblerentropy. Models having different structure i.e. Chaplygin gas or brane-worldscenario are more distant (in Kullback-Leibler sense) from the best one.