No label defined (Q4407585)

The author studies representations of positive polynomials on semialgebraic sets. If the set \(S\subseteq\mathbb{R}^n\) is basic semialgebraic, i.e., \(S=\{x\in\mathbb{R}^n\mid f_1(x)\geq 0, \dots, f_r(x)\geq 0\}\), \(f_1, \dots, f_r\in \mathbb{R}[X_1, \dots, X_n]\), then all polynomials in the cone \[ P(f_1,\dots, f_r)=\biggl\{f \in \mathbb{R}[ X_1,\dots,X_r] \,\Bigl|\, \exists a_\varepsilon\in \sum\mathbb{R}[X_1, \dots,X_r]^2: f=\sum_{\varepsilon\leq \{0,1\}^r}a_\varepsilon \cdot f_1^{\varepsilon_1} \cdot \dots\cdot f_r^{\varepsilon_r}\biggr\} \] are trivially nonnegative on \(S\). (The set \(\sum A^2\) is the sums of squares in the ring \(A.)\) The question is whether, or when, every polynomial that is nonnegative on \(S\) belongs to the cone. Using methods of functional analysis, \textit{K. Schmüdgen} proved [Math. Ann. 289, 203--206 (1991; Zbl 0744.44008)] that if \(S\) is compact and if \(f|_S>0\), then \(f \in P(f_1,\dots,f_r)\). The present author gives the first purely algebraic proof of this remarkable result. His method is to relate compactness of the set \(S\) to the condition that the cone \(P(f_1,\dots,f_r)\) is Archimedean, and then to use the representation theorem of Kadison-Dubois. The author studies these questions in greater generality -- many of his results are true for algebras over subfields of \(\mathbb{R}\) and for preorderings of higher level. Major developments have taken place since the author wrote this thesis. An account of the current state-of-the art was written by \textit{C. Scheiderer} [Positivity and sums of squares: A guide to some recent results, \texttt{http://www.math.univ-rennes1.fr/geomreel/raag01/Surveys.html\#Scheiderer}].