Reconstructing a matrix exponential from data

54 Views Asked by At

Suppose you have a matrix $A\in\mathbb{R}^{n\times n}$, and you are given vectors $x_i$, $y_i$ and scalars $t_i$ such that $y_i = e^{t_iA}x_i$. If you have $n$ of these vectors you should have sufficient information to solve for $A$, with the exception of certain pathological cases. Is this possible, analytically or algorithmically?