Almost sure diffusion approximation in averaging: Direct proofs with rough paths flavors

We consider again the fast–slow motions setups in the continuous time [Formula: see text] [Formula: see text] and the discrete time [Formula: see text] [Formula: see text] where [Formula: see text] and [Formula: see text] are smooth matrix and vector functions, respectively, [Formula: see text] is a centered vector stationary stochastic process with weak dependence in time and [Formula: see text] is a big parameter. We obtain estimates for the almost sure approximations of the process [Formula: see text] by certain diffusion process [Formula: see text]. In [P. K. Friz and Yu. Kifer, Almost sure diffusion approximation in averaging via rough paths theory, Electron. J. Probab. 29 (2024) 1–56] and in other recent papers concerning similar setups the results were obtained relying fully on the rough paths theory. Here we derive our probabilistic results as corollaries of quite general deterministic estimates which are obtained with all details provided following somewhat ideology of the rough paths theory but not relying on this theory per se which should allow a more general readership to follow complete arguments.