Approximating It\^o Processes in Adapted Wasserstein Distance with Finite Support Laws
Abstract
Motivated by finite-scenario surrogates in stopping and control problems, we study the approximation of It\^o diffusion laws by laws supported on $M$ paths in an adapted Wasserstein distance, which we term adapted functional quantization. Under the strict adapted Wasserstein distance, no finite-support approximation can achieve vanishing error, while even under its relaxed counterpart, naive empirical measures still fail to converge. To resolve this, we develop a recursive conditional sampling algorithm and derive constructive upper bounds on the adapted quantization error. In one dimension, this yields the Brownian rate $(\log M)^{-1/2}$ for every diffusion with continuous coefficients of linear growth and pathwise uniqueness. In several dimensions, a quantitative Osgood stability condition with exponent $\beta$ yields the rate $(\log M)^{-\eta_p}$, where $\eta_p=\min\{1/2,\beta/(\beta+4p-2)\}$. Without the Osgood condition, we construct a two-dimensional diffusion with smooth bounded coefficients whose adapted quantization error decays more slowly than $(\log M)^{-\gamma}$ for any $\gamma>0$.