Abstract:We give a new characterization for the convergence in distribution to a standard normal law of a sequence of multiple stochastic integrals of a fixed order with variance one, in terms of the Malliavin derivatives of the sequence. We extend our result to the multidimensional case and prove a weak convergence result for a sequence of square integrable random variables.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is about the necessary and sufficient conditions for the central limit theorems (CLT) of multiple stochastic integral sequences. Specifically, the authors studied the conditions for the square - integrable random variable sequences $\{F_k\}_{k\in\mathbb{N}}$ belonging to the $n$-th - order Wiener chaos to converge to the standard normal distribution $N(0, 1)$.
### Main problems:
1. **Provide a new necessary and sufficient condition**: By means of Malliavin calculus, the author gives a new necessary and sufficient condition to determine whether these random variable sequences converge to the standard normal distribution.
2. **Prove the main theorem**: Re - prove the main theorem in [7] by Nualart and Peccati using the techniques of Malliavin calculus.
3. **Extend to the multi - dimensional case**: Generalize the above results to the multi - dimensional case and prove a weak convergence result.
4. **Apply to specific examples**: Study the renormalized weak convergence problem of the self - intersection local time of fractional Brownian motion.
### Specific content:
#### 1. New necessary and sufficient conditions
For the random variable sequences $\{F_k = I_n(f_k)\}_{k\in\mathbb{N}}$ belonging to the $n$-th - order Wiener chaos, the author proves that the following conditions are equivalent to the convergence of this sequence to the standard normal distribution $N(0, 1)$:
- $\lim_{k\rightarrow+\infty}E[F_k^4]=3$
- For all $1\leq l\leq n - 1$, $\lim_{k\rightarrow+\infty}\|f_k\otimes_l f_k\|_{H^{\otimes(n - l)}} = 0$
- $\|\mathrm{D}F_k\|^2_{L^2(\Omega)}\rightarrow n$ as $k\rightarrow+\infty$
where $\mathrm{D}$ represents the Malliavin derivative operator, and $\otimes_l$ represents the $l$-th - order contraction of the kernel function $f_k$.
#### 2. Extension to the multi - dimensional case
For multi - dimensional random vectors $\{F_k=(I_{n_1}(f^1_k),\ldots,I_{n_d}(f^d_k))\}_{k\in\mathbb{N}}$, the author proves similar results, including the convergence of the covariance matrix and the independence and normality of each component.
#### 3. Applications
The author applies the above theory to the example of fractional Brownian motion, studies the weak convergence of the normalized sum of its odd - power increments, and uses the ergodic theorem to prove the condition $\|\mathrm{D}F_k\|^2_{L^2(\Omega)}\rightarrow n$.
### Summary:
The core problem of this paper is to provide more concise and powerful necessary and sufficient conditions for the central limit theorems of multiple stochastic integral sequences by introducing new tools of Malliavin calculus, generalize them to the multi - dimensional case, and show their applications in practical problems.