Quantitative normal approximations for the stochastic fractional heat equation (Q2125631)

This paper considers a stochastic fractional heat equation driven by a multiplicative Gaussian noise. The main results are the existence and uniqueness of the mild solution and a quantitative central limit theorem for the spatial average of the solution. The existence and uniqueness of the mild solution, written in terms of the fractional Green kernel, is provided under the assumption that a fractional version of Dalang's condition on the spectral measure associated to the spatial covariance holds. It is also remarked that the condition is automatically satisfied for many choices of the operator and the noise. The paper gives a central limit theorem for the spatial average of the solution that is quantitative as it provides an explicit bound on the total variation distance between the average on a ball and a standard normal random variable, with the proof relying on Malliavin-Stein approach. The statement is provided for balls, but can be extended to other sets, with changes showing up only in the normalisation constant. Similarly, one can also notice that the limiting normal distribution, the renormalisation rate and the total variation distance are the same as in the heat kernel case, up to constants. The fact that the limiting distribution is Gaussian and not something else (such as a stable Lévy process, as the choice of operator might suggest) as the nature of the limiting distribution depends on the choice of the noise.