Implement distance correlation (also known as centred kernel alignment) described in Appendix A of the above paper. Equation (A.1) gives the population expression while the empirical version is given by
Tr(KHLH) / sqrt( Tr(KHKH)*Tr(LHLH) )
where K and L are respectively Gram matrices on Xs and Ys and H is the centering matrix.
This is part of the GSoC project Testing and Measuring Variable Interactions With Kernels
See Equivalence of distance-based and RKHS-based statistics in hypothesis testing.
Implement distance correlation (also known as centred kernel alignment) described in Appendix A of the above paper. Equation (A.1) gives the population expression while the empirical version is given by Tr(KHLH) / sqrt( Tr(KHKH)*Tr(LHLH) ) where K and L are respectively Gram matrices on Xs and Ys and H is the centering matrix.
Target: @lambday Contacts: @karlnapf, @sejdino