Homology classes which are represented by graph links
Homology classes which are represented by graph links
复制标题
由图链接表示的同源类
DOI:
10.1090/s0002-9939-1985-0776213-8
复制
发表时间:
1985
期刊:
影响因子:
--
通讯作者:
Koichi Yano
中科院分区:
文献类型:
--
作者:
Koichi Yano
We give a necessary and sufficient condition for a one-dimensional homology class of a graph manifold to be represented by a graph link. The purpose of this paper is to give a necessary and sufficient condition for a one-dimensional homology class of a graph manifold to be represented by a graph link. For this, we define the Jaco-Shalen-Johannson complex of a graph manifold (?1), using the so-called torus decomposition theorem due to Jaco-Shalen and Johannson (see also Waldhausen [2]), and the main result is stated as follows (?2): THEOREM. Let M be a graph manifold prime to S1 X S2 and p: M -T WM the natural map to the Jaco-Shalen-Johannson complex of M. Then an element a of H1(M; Z) can be represented by a graph link if and only if p*o(a) = 0 in Hl(WM; Z). If the ambient manifold M is not prime to S1 X S2, the homotopy class of the map p: M -T WM is not unique in general, and this makes the statement complicated (?3). The proof of these results is based on the study of global graph links in [3]. 1. Preliminaries. Throughout this paper, manifolds are compact, oriented, of dimension three and with toral boundary, and links are oriented. A manifold M is a graph manifold if there is a family of disjointly embedded tori in M such that each connected component of the manifold obtained by cutting M along these tori is the total space of an S1-bundle over a surface. A link is called a graph link if its exterior is a graph manifold. We say that a manifold is prime to S1 X S2 if its prime decomposition does not contain S1 X S2. See Jaco [1] and Yano [3] for other terminology. To state our result, we need the following theorem due to Jaco-Shalen and Johannson (see Jaco [1] and also Waldhausen [2]). THEOREM 1.1 (JACO SHALEN, JOHANNSON). Let M be a Haken manifold which is either closed or with incompressible boundary. Then there exists a unique Seifert submanifold E c M up to ambient isotopy such that (1) E is maximal, and (2) if E' is a Seifert manifold pair distinct from (S3, 0), (S2 X SI, 0), (D2 X I, aD 2 X I) or (S1 X D2, 0), then every nondegenerate map f: E' -> (M, aM) is homotopic to fo such that fo(E') c E. The E above is called the characteristic Seifert manifold of M. Received by the editors September 19, 1983 and, in revised form, March 14, 1984. 1980 Mathematics Subject Classification. Primary 57M99, 57N10.