Does there exist a minimal matrix with an analog to single-variable calculus?

41 Views Asked by At

Given square matrices $A, B, C$, let's say we have a function of the form $X \mapsto AX^2 + BX + C$. Is it possible to "optimize" this matrix in any sense, by computing the class of matrices for which the derivative with respect to $X$ has a matrix-norm of $0$, similar to how you'd compute the extrema of the single-variable case $x \mapsto ax^2 + bx + c$? Is there any analog of matrix-valued calculus like this?