A characterization of Heegaard diagrams for the 3-sphere (Q5903914): Difference between revisions

[This review replaces the earlier one by the same reviewer in Zbl 0626.57003.] I am sorry to inform that my previous review of this paper contains a mistake that invalidates the review. The error consists in considering the set of meridians w as a system of generators for the group presentation, while the author considered it to be a system of relators. The main result of the paper under review is the following very interesting characterization of a Heegaard diagram of \(S^ 3\). Theorem 2. The Heegaard diagram \((F_ g;v,w)\) represents \(S^ 3\) if and only if there exists a system \(w'\) of meridians (``relators'') such that the cyclically reduced group presentation arising from \((F_ g;v,w')\) taking v as generators is the standard one \[ | v_ 1,...,v_ g:v_ 1,...,v_ g|. \] The author proves the ``necessity'' part. The ``sufficiency'' follows from a well-known result of Zieschang [see also \textit{T. Kaneto}, Proc. Am. Math. Soc. 86, 551-552 (1982; Zbl 0511.57006)]. Indeed, since the algebraic length of \(w'\) with respect to v is minimal, it is possible to minimize the geometric length as well, by modifying the system of generators v. This would give the standard diagram of \(S^ 3\). Note that the author only claims that the system \(w'\) exists. If his proof were constructive, the above mentioned result of Zieschang (which is constructive) would give the solution of the old problem of the algorithmical recognition of \(S^ 3\). The proof of Theorem 2 follows from the following results. Result 1 (follows from section 4 of \textit{T. Kaneto}'s: ``On presentations of the fundamental group of the 3-sphere associated with Heegaard diagrams'' [J. Math. Soc. Jap. 33, 147-158 (1981; Zbl 0466.57001)]). Assume that the set v of meridians of \((F_ g;v,w)\) is changed into \(f(v)=v'\), where f is an automorphism of the handlebody V inducing the identity in \(\pi_ 1(V)\). Then the cyclically reduced presentations \(R=| v_ 1,...,v_ g:w_ 1,...,w_ g|\) and \(R'=| v':w|\) are related by a finite sequence of the following moves: 1. Replace relator \(r=x_ 1...x_ k\) by \(r'=x_ 2...x_ kx_ 1\). 2. Replace relator r by \(r^{-1}\). Notice that if \(R'=| v':v'|\) then \(R=| v:v|.\) Result 2 (Theorem 1 in the paper under review). If \((F_ g;v,w)\) represents \(S^ 3\) then it is possible to change v, w by \(f(v)=v'\), \(w'\), where the automorphism f of V induces identity in \(\pi_ 1(V)\), and \((F_ g;v',w')\) is standard. This is proved using in the argument \textit{S. Suzuki}'s: ``On homeomorphisms of a 3-dimensional handlebody'' [Can. J. Math. 29, 111-124 (1977; Zbl 0339.57001)]. The author concludes also the following interesting Corollary: Any cyclically reduced presentation of \(\pi_ 1(S^ 3)=1\) coming from a Heegaard diagram comes also from a Heegaard diagram of \(S^ 3\), \((F_ g;a,b')\), where a is the standard system of meridians in the standard handlebody V in \(S^ 3\).