arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2408.14618math.STstat.TH
收藏

Daniell平滑周期图的Marchenko-Pastur定律

Marchenko-Pastur laws for Daniell smoothed periodograms

Ben Deitmar

AI总结:

本文证明在高维情形下,Daniell平滑周期图的特征值满足Marchenko-Pastur定律,并揭示即使维度远小于样本量,高维效应仍可导致该估计量失去一致性。

摘要:

给定来自d维平稳时间序列$(X_t)_{t \in \mathbb{Z}}$的样本$X_0,...,X_{n-1}$,在给定频率$\theta\in [0,2\pi)$处,谱密度矩阵$F(\theta)$最常用的估计量是Daniell平滑周期图 $$S(\theta) = \frac{1}{2m+1} \sum\limits_{j=-m}^m I\Big( \theta+ \frac{2\pi j}{n} \Big) \\ ,$$ 它是对轻微扰动频率处的$2m+1$个周期图取平均。我们证明,当$d$和$m$随$n$增长且满足$\frac{d}{m} \rightarrow c>0$以及$d\asymp n^α$(其中$α\in (0,1)$)时,Marchenko-Pastur定律对$S(\theta)$的特征值在$\theta\in [0,2\pi)$上一致成立。这表明,即使维度$d$远小于样本量$n$,高维效应也可能导致$S(\theta)$失去一致性。值得注意的是,我们不假设时间序列的$d$个分量相互独立。因此,Marchenko-Pastur定律对Daniell平滑周期图成立,即使对同一过程的样本自协方差矩阵不一定成立。

英文摘要:

Given a sample $X_0,...,X_{n-1}$ from a $d$-dimensional stationary time series $(X_t)_{t \in \mathbb{Z}}$, the most commonly used estimator for the spectral density matrix $F(θ)$ at a given frequency $θ\in [0,2π)$ is the Daniell smoothed periodogram $$S(θ) = \frac{1}{2m+1} \sum\limits_{j=-m}^m I\Big( θ+ \frac{2πj}{n} \Big) \ ,$$ which is an average over $2m+1$ many periodograms at slightly perturbed frequencies. We prove that the Marchenko-Pastur law holds for the eigenvalues of $S(θ)$ uniformly in $θ\in [0,2π)$, when $d$ and $m$ grow with $n$ such that $\frac{d}{m} \rightarrow c>0$ and $d\asymp n^α$ for some $α\in (0,1)$. This demonstrates that high-dimensional effects can cause $S(θ)$ to become inconsistent, even when the dimension $d$ is much smaller than the sample size $n$. Notably, we do not assume independence of the $d$ components of the time series. The Marchenko-Pastur law thus holds for Daniell smoothed periodograms, even when it does not necessarily hold for sample auto-covariance matrices of the same processes.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑