arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2205.02128math.PRcs.ITmath.ITmath.STstat.TH
收藏

平滑经验Wasserstein距离的收敛速率

Rate of convergence of the smoothed empirical Wasserstein distance

Adam Block, Zeyu Jia, Yury Polyanskiy, Alexander Rakhlin

AI总结:

本文刻画了平滑经验Wasserstein距离的精确收敛速率,证明K<σ时达到参数化速率n^{-1/2},K>σ时速率变慢,并由此否定T_2传输不等式及相应高斯混合的对数Sobolev不等式,解决了多个公开问题。

摘要:

考虑由来自d维K-次高斯分布$\mathbb{P}$的n个独立同分布样本诱导的经验测度$\mathbb{P}_n$,并令$γ= N(0,σ^2 I_d)$为各向同性高斯测度。我们研究平滑Wasserstein距离$W_2(\mathbb{P}_n * γ, \mathbb{P}*γ) = n^{-α+ o(1)}$的收敛速度,其中$*$表示测度的卷积。对于$K<σ$且在任意维度$d\ge 1$下,我们证明$α= {1\over2}$。对于$K>σ$且维度$d=1$的情形,我们证明其速率更慢,由$α= {(σ^2 + K^2)^2\over 4 (σ^4 + K^4)} < 1/2$给出。这解决了[GGNWP20]中的若干公开问题,特别是精确刻画了获得参数化速率所需的平滑量$σ$。此外,对于任意d维K-次高斯分布$\mathbb{P}$,我们还建立了$D_{KL}(\mathbb{P}_n * γ\\|\mathbb{P}*γ)$在$K<σ$时速率为$O(1/n)$,但在$K>σ$时仅减慢至$O({(\log n)^{d+1}\over n})$。$W_2^2$与KL行为的惊人差异意味着当$σ< K$时$T_{2}$-传输不等式不成立。因此,对于$K>σ$,高斯混合$\mathbb{P} * N(0, σ^{2})$的对数Sobolev不等式(LSI)不可能成立。这解决了[WW+16]中的一个公开问题,他们在条件$K<σ$下建立了LSI,并询问其界是否可以改进。

英文摘要:

Consider an empirical measure $\mathbb{P}_n$ induced by $n$ iid samples from a $d$-dimensional $K$-subgaussian distribution $\mathbb{P}$ and let $γ= N(0,σ^2 I_d)$ be the isotropic Gaussian measure. We study the speed of convergence of the smoothed Wasserstein distance $W_2(\mathbb{P}_n * γ, \mathbb{P}*γ) = n^{-α+ o(1)}$ with $*$ being the convolution of measures. For $K<σ$ and in any dimension $d\ge 1$ we show that $α= {1\over2}$. For $K>σ$ in dimension $d=1$ we show that the rate is slower and is given by $α= {(σ^2 + K^2)^2\over 4 (σ^4 + K^4)} < 1/2$. This resolves several open problems in [GGNWP20], and in particular precisely identifies the amount of smoothing $σ$ needed to obtain a parametric rate. In addition, for any $d$-dimensional $K$-subgaussian distribution $\mathbb{P}$, we also establish that $D_{KL}(\mathbb{P}_n * γ\|\mathbb{P}*γ)$ has rate $O(1/n)$ for $K<σ$ but only slows down to $O({(\log n)^{d+1}\over n})$ for $K>σ$. The surprising difference of the behavior of $W_2^2$ and KL implies the failure of $T_{2}$-transportation inequality when $σ< K$. Consequently, it follows that for $K>σ$ the log-Sobolev inequality (LSI) for the Gaussian mixture $\mathbb{P} * N(0, σ^{2})$ cannot hold. This closes an open problem in [WW+16], who established the LSI under the condition $K<σ$ and asked if their bound can be improved.

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。

↑