MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

According the the original definition by Smythe, a homology boundary link $L\subseteq S^3$ with $m$ components is a link which sarisfies one of the following equivalent conditions:

(1): The fundamental group $\pi_1 (S^3\setminus L)$ of the link complement surjects onto the free group $F_m$ of rank $m$;

(2): There exists a collection $S_1,\ldots,S_m$ of disjoint oriented surfaces properly embedded in the complement $M$ of a tubular neighbourhood of $L$ such that for every $i=1,\ldots,m$, the boundary of $S_i$ is homologous in $\partial M$ to the $i$-th longitude of $L$.

It is not difficult to show that (2) implies (1): the homological condition implies that $S_1,\ldots,S_m$ are linearly independent as elements of $H_2(M;\partial M;\mathbb{Z})$, so the complement of the collection of surfaces is connected, and we may construct a map on the bouquet of $m$ copies of $S^1$ inducing an epimorphism on fundamental groups.

On the other hand, if (1) holds one may construct a family of disjoint surfaces that represent a basis of $H_2 (M;\partial M;\mathbb{Z})$. However, it could happen that these surfaces do not satisfy (2): for example, if $m=2$ it could happen that $\partial S_1$ is equal to 3 times the first longitude plus 2 times the second one, while $\partial S_24 is equal to 2 times the first longitude plus the second one. How could I trade such a system of surfaces for a system satisfying (2)?

share|cite|improve this question
I think this is a version of Brown representability:… The construction Ryan points out there, which Serre and Thom used, looks to me like it carries through to this context with no change. – Daniel Moskovich Dec 22 '10 at 12:23
up vote 3 down vote accepted

This is easy.

if (1) holds one may construct a family of disjoint surfaces that represent a basis of $H_2(M;\partial M;\Bbb Z)$

Indeed; but in order to do so, one has to identify the free group $F_m$ with the fundamental group of the wedge of $m$ circles. This identification depends on a choice of basis of $F_m$. Given one choice $(x_1,\dots,x_m)$, other choices can be obtained by Nielsen transformations:

a$_{ij}$) $x_i\mapsto x_{ij}$, $x_k\mapsto x_k$ for $k\ne i$;

b$_i$) $x_i\mapsto x_i^{-1}$, $x_j\mapsto x_j$ for $j\ne i$.

One may also renumber the $x_i$'s.

Now the given epimorphism $\pi_1(S^3\setminus L)\to F_m$ descends to an isomorphism $h$ between the abelianizations. The first abelianization, $H_1(S^3\setminus L)$, has a basis given by a choice of meridians of the link; and a basis for the second is given by our basis $(x_i)$ of the free group. Then $h$ corresponds to an $m\times m$ matrix over $\Bbb Z$ with determinant $\pm 1$. Nielsen transformation of the $x_i$'s induce elementary row transformations of this matrix. They suffice to diagonalize the matrix, with $\pm 1$ on the diagonal. The result is that, geometrically, the $i$th meridian goes homologically onto the $i$th circle of the wedge with some orientation. Apply the Pontryagin construction now, and the surfaces will satisfy (2).

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.