Fixed points of surface diffeomorphisms (Q1325282)

Let \(f : M \to M\) be a homeomorphism of a compact surface. In 1927, Jakob Nielsen defined a homotopy invariant lower bound for the number of fixed points of \(f\), that is now called the Nielsen number \(N(f)\), and he conjectured that \(f\) is isotopic to a homeomorphism with exactly \(N(f)\) fixed points. In 1981, the first author announced [Bull. Am. Math. Soc., New Ser. 5, 176-178 (1981; Zbl 0479.55003)] that the Nielsen-Thurston classification theory for homeomorphisms of surfaces could be used to verify Nielsen's conjecture, for closed surfaces. The principal contents of this very important paper consist of a detailed proof of the announced result and in fact of somewhat more: the homeomorphism with \(N(f)\) fixed points can be chosen to be smooth. In 1984, the first author published an example to demonstrate that, for maps of surfaces, there need not be a map in the homotopy class with only \(N(f)\) fixed points. For a more detailed discussion of these developments, and related subsequent discoveries, see the reviewer's survey paper [Contemp. Math. 152, 9-21 (1993; Zbl 0816.55001)]. If \(M\) has a nonempty boundary, the fact that homeomorphisms must map the boundary onto itself makes it difficult to formulate a corresponding result in terms of the Nielsen number. Motivated by this difficulty in the 1981 announcement, \textit{H. Schirmer} [Pac. J. Math. 122, 459-473 (1986; Zbl 0553.55001)] introduced a Nielsen number \(N(f; X,A)\) for maps of pairs \(f : (X,A) \to (X,A)\). A survey of the extensive development of relative Nielsen theories that quickly followed can be found in Schirmer's paper [Contemp. Math. 152, 291-309 (1993; Zbl 0805.55001)]. These theories include a Nielsen number \(N (f;X - A)\) for the complement due to \textit{X.-Z. Zhao} [Lect. Notes Math. 1411, 189-199 (1989; Zbl 0689.55008)]. The result in the present paper, for homeomorphisms of surfaces with boundary, states that there is an isotopy to a diffeomorphism with exactly \(N(f;M, \partial M)\) fixed points, of which exactly \(N(f;M - \partial M)\) are in the interior of \(M\). The proof of the main theorem uncovered a relationship between the Nielsen number \(N(f)\) of a homeomorphism \(f : M \to M\) of a surface, with or without boundary, of negative Euler characteristic \(\chi (M)\) and its Lefschetz number \(L(f)\), namely, \(|L (f) - \chi (M) |\leq N(f) - \chi (M)\). The authors ask whether this inequality holds also for maps of such surfaces.