Edgeworth-type expansions for transition densities of Markov chains converging to diffusions (Q2496939)

Consider triangular arrays of homogeneous Markov chains \(X_n(k)\) (\(0 \leq k \leq n\)) that converge weakly to a diffusion process as \(n \to +\infty\). The authors prove Edgeworth-type expansions of order \(o(n^{-1-\delta})\) (\(\delta>0\)) for transition densities of those Markov chains. For this purpose, they apply the so-called paramatrix method to represent the transition density as a nested sum of functionals of densities of sums of independent and identically distributed variables. Then, as the main result, plugging Edgeworth expansions into this representation provides an expansion for the transition density for the Markov chain (see Section 4). Several auxiliary results, a short introduction to the parametrix method for diffusions and a complete proof (see Section 6) of their main result are given as well.