Statistical energy as a tool for binning-free, multivariate goodness-of-fit tests, two-sample comparison and unfolding

We introduce the novel concept of statistical energy as a statistical tool. We define statistical energy of statistical distributions in a similar way as for electric charge distributions. Charges of opposite sign are in a state of minimum energy if they are equally distributed. This property is used to check whether two samples belong to the same parent distribution, to define goodness-of-fit tests and to unfold distributions distorted by measurement. The approach is binning-free and especially powerful in multidimensional applications. (C) 2004 Elsevier B.V. All rights reserved.