Some aspects of multiorthomorphisms on Riesz spaces (Q1999147)

The article explores the notion of an ordered vector space (or Riesz space) from functional analysis, and its goal is to better understand the properties of a certain algebraic operation (similar in definition to the notion of a quadratic form) known as an \(n\)-homogeneous polynomial on such a space. The main body of the paper focuses on examining multiorthomorphisms on Riesz spaces, which are essential to defining appropriately continuous \(n\)-homogeneous polynomials. An \(n\)-orthomorphism on a Riesz space \(E\) is defined as an \(n\)-linear map \(\varphi\) from \(E^n\) to \(E\) which is an orthomorphism on each component, in which case we can define the associated \(n\)-homogeneous polynomial \(P\) on \(E\) by \(P(x)=\varphi(x,x,\dots,x)\). Since all continuous \(n\)-homogeneous polynomials of interest can be thus defined, the article proceeds by establishing important properties of \(n\)-orthomorphisms in the first five sections, then applying these to \(n\)-homogeneous polynomials in the final section. After recalling the basic facts regarding Riesz spaces and their \(n\)-orthomorphisms in Section~2, the article proceeds in Section~3 to explore in detail the properties of the space Orth\(_n(E)\) of \(n\)-orthomorphisms on the Riesz space \(E\). The main result of this section is that Orth\(_n(E)\) is itself a Riesz space. To prove this, one first defines the absolute value \(|T|\) for any operator \(T\) in Orth\(_n(E)\), then decomposes \(|T|\) into a positive and negative part \(T^+\) and \(T^-\), and then proves that all of these are \(n\)-orthomorphisms. This is sufficient to define an ordering on Orth\(_n(E)\), i.e., \(T\leq S\) if \((S-T)^-=0\), which naturally makes it a Riesz space. In Section~4, the article goes on to consider the notion of an \(f\)-algebra, i.e., a Riesz space carrying an associative multiplication which behaves well with respect to infima in the space. The main result of the section is that for any Riesz space \(E\), Orth\(_n(E)\) carries the structure of an \(f\)-algebra. Indeed, for any positive element \(e\) in \(E\), there is a multiplication \(*_e\) on Orth\(_n(E)\) given by \((T*_e S)(x_1,\dots,x_n)=T(e,e,\dots,e,S(x_1,\dots,x_n))\), and provided Orth\(_n(E)\) is non-zero, we can choose \(e\) such that this \(f\)-algebra structure is non-trivial. The section concludes by showing that if \(E\) is itself an $f$-algebra with unit \(1_E\), then (Orth\(_n{(E),+,*_{1_E})}\) is also unital. Section 5 explores some deeper algebraic properties of multiorthomorphism spaces, where it is shown that if \(E\) is a semiprime \(f\)-algebra satisfying certain conditions, then Orth\(_n(E)\) can be considered as a Riesz subspace of Orth\(_{n+1}(E)\), and furthermore if \(E\) is archimedian then Orth\(_n(E)\) is in fact an ideal of Orth\(_{n+1}(E)\). Finally, the article concludes in Section~6 by revisiting \(n\)-homogeneous polynomials. It is shown that the space of polyorthomorphisms on a Riesz space \(E\) of degree \(n\) (the appropriate space of continuous \(n\)-homogeneous polynomials defined by symmetric \(n\)-orthomorphisms) itself has the structure of a Riesz space, and that it is isomorphic to Orth\(_n(E)\). This article should certainly be of interest to most researchers in functional analysis, and all those familiar with Riesz spaces. It should also be of interest within algebra, since it uses the well-developed theory of ordered vector spaces to define and explore new types of algebraic structure, most notably \(f\)-algebras. I would highly recommend this paper to any algebraically minded mathematician, though I would suggest revising some fairly basic notions from functional analysis beforehand, since some assumed definitions and results are not obvious. Overall, this work is a remarkable synthesis of algebra and analysis which is absolutely worth a detailed read.