论文标题

部分可观测时空混沌系统的无模型预测

Sublinear-Time Algorithms for Max Cut, Max E2Lin$(q)$, and Unique Label Cover on Expanders

论文作者

Peng, Pan, Yoshida, Yuichi

论文摘要

储层计算是预测湍流的有力工具,其简单的架构具有处理大型系统的计算效率。然而,其实现通常需要完整的状态向量测量和系统非线性知识。我们使用非线性投影函数将系统测量扩展到高维空间,然后将其输入到储层中以获得预测。我们展示了这种储层计算网络在时空混沌系统上的应用,该系统模拟了湍流的若干特征。我们表明,使用径向基函数作为非线性投影器,即使只有部分观测并且不知道控制方程,也能稳健地捕捉复杂的系统非线性。最后,我们表明,当测量稀疏、不完整且带有噪声,甚至控制方程变得不准确时,我们的网络仍然可以产生相当准确的预测,从而为实际湍流系统的无模型预测铺平了道路。

We show sublinear-time algorithms for Max Cut and Max E2Lin$(q)$ on expanders in the adjacency list model that distinguishes instances with the optimal value more than $1-\varepsilon$ from those with the optimal value less than $1-ρ$ for $ρ\gg \varepsilon$. The time complexities for Max Cut and Max $2$Lin$(q)$ are $\widetilde{O}(\frac{1}{ϕ^2ρ} \cdot m^{1/2+O(\varepsilon/(ϕ^2ρ))})$ and $\widetilde{O}(\mathrm{poly}(\frac{q}{ϕρ})\cdot {(mq)}^{1/2+O(q^6\varepsilon/ϕ^2ρ^2)})$, respectively, where $m$ is the number of edges in the underlying graph and $ϕ$ is its conductance. Then, we show a sublinear-time algorithm for Unique Label Cover on expanders with $ϕ\gg ε$ in the bounded-degree model. The time complexity of our algorithm is $\widetilde{O}_d(2^{q^{O(1)}\cdotϕ^{1/q}\cdot \varepsilon^{-1/2}}\cdot n^{1/2+q^{O(q)}\cdot \varepsilon^{4^{1.5-q}}\cdot ϕ^{-2}})$, where $n$ is the number of variables. We complement these algorithmic results by showing that testing $3$-colorability requires $Ω(n)$ queries even on expanders.

扫码加入交流群

加入微信交流群

微信交流群二维码

扫码加入学术交流群,获取更多资源