Central limit theorems for linear spectral statistics of large dimensional \(F\)-matrices (Q424702): Difference between revisions

The author establishes the central limit theorem for the linear spectral statistic of a large Fisher matrix \(\mathbf F\) with explicit expressions of the asymptotic means and covariance functions. As a consequence, a similar result is shown for a beta matrix that is of the form \(({\mathbf I} + d \cdot \mathbf F)^{ - 1}\), where \(d\) is a constant and \(\mathbf I\) is an identity matrix. The Fisher matrix \(\mathbf F\) is defined as \(\mathbf F=\mathbf S_1 \mathbf S_2^{-1} \), where \(\mathbf S_1\) and \(\mathbf S_2\) are sample covariance matrices from the two independent variable arrays of dimensions \(p \times n_1\) and \(p\times n_2\) respectively, either real or complex valued.