On S-equivalence of Seifert matrices (Q2563188)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 3424173
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | On S-equivalence of Seifert matrices |
scientific article; zbMATH DE number 3424173 |
Statements
On S-equivalence of Seifert matrices (English)
0 references
1973
0 references
A Seifert matrix is a square matrix \( V \) of integers such that \( V+\varepsilon V'\) is unimodular, where \( {V}^{\prime} \) is the transpose of \( {V} \) and \( \varepsilon \) is \(\pm 1\). With any smooth embedding of \( S^{2 n-1} \) in \( S^{2 n-1} \) there is associated a family of Seifert matrices, with \( \varepsilon=(-1)^{n} \). The members of such a family satisfy a relation known as \(S\)-equivalence, which may be characterized as follows \textit{J. Levine} [Comment. Math. Helv. 45, 185--198 (1970; Zbl 0211.55902)]. \( {V} \) and \( {W} \) are (integrally) congruent if \( {W}={PVP} \) for some integral unimodular \( {P} \). \( {W} \) is a row enlargement of \( {V} \) if it is obtained from \( {V} \) by adjoining two rows and columns to \( {V} \) (at the left and top) so that the first row is 0 and the first column is 0 except for a 1 in the second row. \(W\) is a column enlargement of \( V \) if \( W^{\prime} \) is a row enlargement of \( {V}^{\prime}\), \({V} \) is a reduction of \( {W} \) if \( {W} \) is an,enlargement of \( {V} \). Then \(S\)-equivalence is the equivalence relation generated by congruence, enlargement, and reduction. Let \( F=Q(t) \) be the field of rational functions, with the conjugation defined by \( E=t^{-1} \). Define \( z=(1-t)^{-1} \) and note that \( \bar{z}=1-z=-t z \). Let \( \Lambda \) be the subring of \( F \) generated by \( t, t^{-1} \) and \( z \), and define \( C \) as the \( \Lambda \)-module \( F / \Lambda \). A \( \Lambda \)-module \(A\) with a C-valued \( (-\varepsilon) \)-hermitian form is associated with any Seifert matrix \( V \) as follows. Let \( M=\overline{z V}-\varepsilon z V^{\prime} \). Then \( A \) is the module of column vectors over \( \Lambda \) modulo the columns of \( M \). If \( u, v \) in \( A \) are represented by column vectors \( a, b \), then the value of the form on \( u, v \) is \( u \cdot v=\bar{b} \cdot 1 v^{-1} a \), evaluated in \( F \) and reduced modulo \( \Lambda \). The main result of the paper is an algebraic proof of the theorem that Seifert matrices are \(S\)-equivalent if and only if the associated Seifert forms are isometric. (The result follows independently by combining the work of Levine already cited, in which \(S\)-equivalence is shown to classify certain codimension 2 knots, with recent work of \textit{C. Kearton} [Bull. Am. Math. Soc. 79, 952--955 (1973; Zbl 0276.57006)] which shows that the Seifert forms solve the same classification problem.) Any Seifert matrix is \(S\)-equivalent to a non-singular one. If \( A \) is the module associated with a non-singular \( V \) of rank \( r \) then \( A \) embeds in \( Q A=Q \otimes A \), which is a vector space of dimension \( r \) over \( Q \). Integer congruence classes of Seifert matrices associated with forms isometric to \(A\) correspond to certain integer lattices on \( Q A \) which generate \( A \) as a \( \Lambda \)-submodule of \( Q A \). The proof of the main theorem proceeds by translating the effect of an enlargement reduction step in terms of such lattices. The resulting problem about lattices is shown to be solvable \(p\)-adically for all \( {p} \), and standard arguments show that this implies the full result. The paper also contains some miscellaneous results on the relation between \(S\)-equivalence and integral congruence, as follows. Let \( V \) and \( W \) be non-singular Seifert matrices. If \( \operatorname{det} V \) is prime then \( W \) and \( V \) are \( S \)-equivalent only if they are congruent. If det \( V \) is square-free then if \( V \) and \( {W} \) are \( {S} \)-equivalent, they are congruent over the \(p\)-adic integers for all \( {p} \). If the minimal polynomial of \( {V}\left({V}+\varepsilon {V}^{\prime}\right)^{-1} \) has a repeated factor with constant term not \(\pm 1\) then the \(S\)-equivalence class of \( V \) contains infinitely many congruence classes. An example is given of a matrix not equivalent to its transpose (which implies the existence of a non-invertible knot). Another example shows that matrices may be congruent over the rationals and all \(p\)-adic integers but not \(S\)-equivalent.
0 references