$L^2$ -eta-invariants and their approximation by unitary eta-invariants (Q4673383)

The main purpose of this paper is to compare different obstructions to a knot being slice. These obstructions are based upon the eta invariants associated to metabelian representations of the knot group. On the one hand, in previous work, the author considered two closely related obstructions based upon finite-dimensional representations, one detecting non-slice and the other non-ribbon knots (of course it is an old conjecture that ribbon = slice). On the other hand Cochran-Orr-Teichner have defined a metabelian \(L^2\)-eta invariant obstruction to sliceness (more precisely 1.5 solvability). T. Kim has given examples of knots whose finite-dimensional metabelian slice obstructions vanish but whose metabelian \(L^2\)-obstruction is non-zero and the author has shown that vanishing of the metabelian \(L^2\)-obstruction does not necessarily imply the vanishing of the finite-dimensional metabelian ribbon obstructions. The main result of the present paper is that the vanishing of the finite-dimensional metabelian ribbon obstruction does imply the vanishing of the metabelian \(L^2\)-obstruction. An important tool in the proof is a theorem of Lück and Schick relating the \(L^2\)-eta invariant of a residually finite group to a limit of finite-dimensional eta invariants of its finite quotient groups.