The royal road to automatic noncommutative real analyticity, monotonicity, and convexity (Q2166325)

The primary goal of this article is to simplify the proofs of the multivariable generalizations of Löwner's theorem on matrix monotone functions and Kraus' theorem on matrix convex functions. A function \(f:(a,b)\to\mathbb{R}\) is matrix monotone if, for any self-adjoint matrices \(A,B\) of the same size with spectrum in \((a,b)\), we have \(A\leq B\ \Rightarrow \ f(A) \leq f(B)\). Löwner's theorem gives a complete characterization of matrix monotone functions in one variable in terms of analytic continuation to the upper half-plane. Subsequent work has provided multivariable versions of Löwner's theorem (see the introduction of the article for references). The first such results were proven for functions of commuting families of matrices, but the proofs are difficult. As explained by the present authors, ``the difficulty is a symptom of the fact that the variety of commuting tuples of matrices is full of holes -- that is, it is not convex and, thus, unnatural for understanding monotonicity.'' Subsequent results were obtained for functions of non-commuting matrices, with proofs relying on the commutative case. In this article, the authors give a direct proof of the non-commutative case, without passing by the commutative case. The key philosophy of the article is that this direct proof is in fact more natural than the case of commuting variables, thus providing the ``royal road'' of the title. Analogous results are also proven for matrix convex functions of non-commutating variables.