Let $A = (a_{ij})_{1 \leq i,j \leq N}$ and $B = (b_{ij})_{1 \leq i,j \leq N}$ be random matrices, with each $a_{ij}$ and $b_{ij}$ an independent random variable with continuous density function, zero mean and bounded variance. Let $\textbf{c}=(c_j)_{1 \leq j \leq N}$ be a random vector, whose entries are also independent and have a continuous, zero mean, bounded variance distribution.

Define the matrix $D$ by $$ D = \begin{pmatrix} A & 0 \\ BA &A\end{pmatrix}$$ and let $$\textbf{e} = \begin{pmatrix} \textbf{c} \\ B\textbf{c}\end{pmatrix}$$

Let $\textbf{q}_i$ be the generalized eigenvectors of $D$ for $1 \leq i \leq 2\cdot N$. What can we say about $\textbf{q}_i \cdot \textbf{e}$? In particular, is it possible to prove (or disprove) that $P \left( \bigcup_i \textbf{q}_i \cdot \textbf{e} \neq 0\right)=1$? That is, that $\textbf{e}$ is not orthogonal to any of the generalized eigenvectors of $D$ with probability one?


Your Answer

By clicking "Post Your Answer", you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.