论文标题
图形幂的集团因素
Clique factors in powers of graphs
论文作者
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。
The $k$th power of a graph $G$, denoted $G^k$, has the same vertex set as $G$, and two vertices are adjacent in $G^k$ if and only if there exists a path between them in $G$ of length at most $k$. A $K_r$-factor in a graph is a spanning subgraph in which every component is a complete graph of order $r$. It is easy to show that for any connected graph $G$ of order divisible by $r$, $G^{2r-2}$ contains a $K_r$-factor. This is best possible as there exist connected graphs $G$ of order divisible by $r$ such that $G^{2r-3}$ does not contain a $K_r$-factor. We conjecture that for any 2-connected graph $G$ of order divisible by $r$, $G^r$ contains a $K_r$-factor. This was known for $r \le 3$ and we prove it for $r = 4$. We prove a stronger statement that the vertex set of any 2-connected graph $G$ of order $4k$ can be partitioned into $k$ parts of size $4$, such that the four vertices in any part are contained in a subtree of $G$ of order at most 5. More generally, we conjecture that for any partition of $n = n_1+n_2+\cdots+n_k$, the vertex set of any 2-connected graph $G$ of order $n$ can be partitioned into $k$ parts $V_1,V_2,\ldots,V_k$, such that $|V_i| = n_i$ and $V_i \subseteq V(T_i)$ for some subtree $T_i$ of $G$ of order at most $n_i+1$, for $1 \le i \le k$.