Asymptotic analysis for extreme eigenvalues of principal minors of random matrices (Q2075335): Difference between revisions

The current paper focuses on random matrices. More precisely, the authors consider white Wishart matrices \(W=(w_{ij})_{1\leq i,j\leq p}=X^TX\), where \(X=(x_{ij})_{1\leq \leq n, 1\leq j\leq p}\) and \(x_{ij}\) are independent standard normally distributed random variables. For a subset \(S\) of the index set \(\{1,\ldots, p\}\) with \(|S|=k\), the \(S\)-principal minor of \(W\) is denoted by \(W_S=(w_{ij})_{i,j\in S}\). If one denotes by \(\lambda_1(W_S)\geq \ldots \geq \lambda_k(W_s)\) the eigenvalues of \(W_S\) in descending order, the quantities of interest during the current work are the largest and the smallest eigenvalues of all the \(k \times k\) principal minors of \(W\) in the setting that \(n, p,\) and \(k\) are large but \(k\) relatively smaller than \(\min \{n,p\}\). For the extreme eigenvalues \[ \lambda_{\max}(k)=\max_{1\leq i\leq k,S\subset \{1,\dots, p\}, |S|=k }\lambda_i(W_S) \] and \[ \lambda_{\min}(k)=\min_{1\leq i\leq k,S\subset \{1,\dots, p\}, |S|=k }\lambda_i(W_S) \] the asymptotic behaviour is investigated. The results are then applied in the construction of compressed sensing matrices.