Abstract:We consider the stochastic heat equation driven by a multiplicative Gaussian noise that is white in time and spatially homogeneous in space. Assuming that the spatial correlation function is given by a Riesz kernel of order $\alpha \in (0,1)$, we prove a central limit theorem for power variations and other related functionals of the solution. To our surprise, there is no asymptotic bias despite the low regularity of the noise coefficient in the multiplicative case. We trace this circumstance back to cancellation effects between error terms arising naturally in second-order limit theorems for power variations.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the conditions for the establishment of the central limit theorem (CLT) of the power variation and other related functionals of the solutions of parabolic stochastic partial differential equations (SPDEs) with multiplicative noise, especially the stochastic heat equation, in high - frequency analysis. Specifically, the paper focuses on how to prove the asymptotic properties of these statistics in the case of noise coefficients with low regularity, and especially how to overcome the technical challenges brought about when the spatial correlation function is given by the Riesz kernel.
### Background and Problem Description
The paper considers the stochastic heat equation in the following form:
\[ \partial_t u(t,x) = \frac{1}{2} \Delta u(t,x) + \sigma(u(t,x)) \dot{W}(t,x), \quad (t,x) \in (0,\infty) \times \mathbb{R}^d, \]
where \(\sigma: \mathbb{R} \to \mathbb{R}\) is a globally Lipschitz function, and \(\dot{W}(t,x)\) is a Gaussian noise that is white in time and homogeneous in space, and its spatial correlation function is given by the Riesz kernel \(F(y) = c_\alpha |y|^{-\alpha}\), where \(\alpha \in (0,1)\).
### Main Objectives
The main objective of the paper is to study the second - order behavior of the normalized power variation \(V_n^p(u,t)\) and prove the central limit theorem in the following form:
\[ \frac{1}{\sqrt{\Delta_n}} \left( V_n^p(u,t) - V^p(u,t) \right) \stackrel{st}{\Rightarrow} Z, \]
where \(Z\) is a certain mixed Gaussian process, and \(\stackrel{st}{\Rightarrow}\) represents stable convergence in the locally uniform topology.
### Technical Difficulties
- **Low Regularity**: Due to the low regularity of the noise coefficient \(\sigma(u(t,x))\) ((1/2 - \alpha/4 - \epsilon)-Hölder continuous in time and (1 - \alpha/2 - \epsilon)-Hölder continuous in space), the traditional methods for proving the central limit theorem are no longer applicable.
- **Cancellation Effect of Error Terms**: The paper discovers that there is a cancellation effect between certain error terms, so that the central limit theorem still holds even in the case of low regularity.
### Solutions
The paper solves the above problems through the following steps:
1. **Step - by - Step Approximation**: Approximate the power variation step - by - step into the form of a sum of martingales, and utilize Jacod's martingale central limit theorem.
2. **Approximation of Conditional Expectations**: Decompose the conditional expectation into multiple parts through Taylor expansion and handle them separately.
3. **Analysis of Error Terms**: Analyze each error term in detail, especially those error terms that do not disappear individually but cancel each other out in the sum.
### Conclusions
The paper successfully proves that the central limit theorem for the normalized power variation \(V_n^p(u,t)\) holds in the case of noise coefficients with low regularity. This result is not only of great theoretical significance, but also of great value in practical applications, such as parameter estimation and statistical testing.