On the Pierce-Birkhoff conjecture in three variables (Q995607): Difference between revisions

The Pierce-Birkhoff conjecture is concerned with continuous semi-algebraic functions on \(\mathbb{R}^n\) that are piecewise polynomial; the ring of these functions is denoted by \(\text{PWP}(n)\). Every function that is obtained from polynomials via finite suprema and infima clearly belongs to \(\text{PWP}(n)\). In fact, these functions also form a ring, which is denoted by \(\text{SID}(n)\). The Pierce-Birkhoff conjecture says that \(\text{PWP}(n)= \text{SID}(n)\) for each \(n\). This is practically trivial for \(n= 1\) and has been proved 25 years ago for \(n= 2\). Since then there has not been much progress. The logical next step is to study the case \(n= 3\). In the present paper the author proves: Theorem. If \(h\in\text{PWP}(3)\) then there is a finite set \(X+h\subset\mathbb{R}^3\) such that for each \(\varepsilon> 0\) there is a function \(h_\varepsilon\in\text{SID}(3)\) with \(h(z)= h_\varepsilon(z)\) for all \(z\not\in\bigcup_{x\in X_h}B(x; \varepsilon)\), where \(B(x;\varepsilon)\) is the ball about \(x\) with radius \(\varepsilon\). Theorem. Every function \(h\in\text{PWP}(3)\) can be obtained from rational functions with finitely many poles via finite suprema and infima. The theorems do not decide the Pierce-Birkhoff conjecture for \(n= 3\). However, if the conjecture should be false in three dimensions then these are probably the best results that are to be expected.