On the joint distribution of the maximum and its location for a linear diffusion (Q1090006)

Let X be a regular one-dimensional diffusion and let M be its maximum value before time t, and T (almost surely unique) time at which \(X=M\) before t. A general formula is given for the joint distribution of \(X_ t\), M, and T, in terms of the hitting-time densities, the scale function, and the speed measure of X. Two proofs are given; one is probabilistic and the other is analytical. Various examples are discussed, including the diffusion representation of Brownian local time evaluted at an exponential time.