Link concordance, homology cobordism, and Hirzebruch-type defects from iterated \(p\)-covers (Q973857)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Link concordance, homology cobordism, and Hirzebruch-type defects from iterated \(p\)-covers
scientific article

    Statements

    Link concordance, homology cobordism, and Hirzebruch-type defects from iterated \(p\)-covers (English)
    0 references
    0 references
    0 references
    26 May 2010
    0 references
    The work of Cochran, Orr and Teichner has produced highly non-trivial filtrations of the classical knot concordance group \(C_1\). Examples illustrating the complexity of these filtrations have relied on ``infection'' by a knot: replacing a regular neighbourhood of a simple closed curve by the exterior of a knot \(K\). Non-triviality of the examples has usually been established by means of suitable signature invariants. However these invariants do not detect infection when the knot \(K\) has finite order in \(C_1\). This paper considers such situations by means of a signature-defect invariant, with values in the Witt group of non-singular hermitean forms over a skew field with involution. In particular, a torsion class in \(\Omega_3(BG)\) and an involution-respecting homomorphism from \(\mathbb{Z}[G]\) to such a skew field \(F\) gives rise to such a Witt class, if \(H_4(G;\mathbb{Z})=0\). For instance, \(G\) could be a cyclic \(p\)-group and \(F\) the field of \(d\)th roots of unity, where \(d\) divides the order of \(G\). The particular case \(p=2\) and \(d=4\) is used to show that an infinite family of rational homology 3-spheres constructed by infecting the connected sum of two lens spaces have distinct homology cobordism types, although they cannot be distinguished by multisignatures, \(\eta\)-invariants or \(L^2\)-signatures. Related arguments are used to show that the Bing double of the figure eight knot is not a slice link. More generally, there are infinitely many amphicheiral knots \(K\) such that no iterated Bing double \(BD_n(K)\) is a slice link. The third main result relates to ``torsion'' in link concordance. The set of link concordance classes is not a group in any natural way, as the notion of sum of links is only well-defined in the presence of extra structure. Nevertheless, one may say that a link \(L\) is 2-torsion if \textit{some} connected sum \(L\#L\) is slice. It is shown that there are infinitely many 2-torsion boundary links representing non-trivial classes at any depth of the Cochran-Orr-Teichner solvable filtration of link concordance.
    0 references
    0 references
    0 references
    0 references
    0 references
    algebraically closed group
    0 references
    Bing double
    0 references
    concordance
    0 references
    link
    0 references
    rational homology sphere
    0 references
    torsion
    0 references
    signature defect
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references