Extension of the two-variable Pierce-Birkhoff conjecture to generalized polynomials (Q1959924)

The famous Pierce-Birkhoff conjecture states that a continuous and piecewise polynomial function is a supremum of infimums of finitely many polynomials (see [\textit{G. Birkhoff} and \textit{R. S. Pierce}, ``Lattice-ordered rings'', Anais Acad. Brasil. Ci. 28, 41--69 (1956; Zbl 0070.26602)]). The (finitely many) ``pieces'' are semialgebraic sets. The conjecture was proven 1984 for \(n=2\) by \textit{L. Mahé} [``On the Pierce-Birkhoff conjecture'', Rocky Mt. J. Math. 14, 983--985 (1984; Zbl 0578.41008)]. The Pierce-Birkhoff conjecture for \(n>3\) remains unproved and unrefuted despite great effort of many mathematicians. In the paper under review the author considers so-called generalized polynomials (also called signomials). These are real polynomials with arbitrary real exponents. Their natural domain of definition is the positive orthant. The Pierce-Birkhoff conjecture is extended to generalized polynomials: A continuous and piecewise generalized polynomial function (on the positive orthant) is a supremum of infimums of finitely many generalized polynomials. Here the (finitely many) ``pieces'' are generalized semialgebraic sets. Following the reasoning of Mahé the Pierce-Birkhoff conjecture for generalized polynomials is shown in the case \(n=2\).