No label defined (Q3840316)

In [Topology 34, No. 2, 423--472 (1995; Zbl 0898.57001)] \textit{D. Bar-Natan} first called Vassiliev invariants ``polynomial'' in the following sense: if one thinks of the well-known process of extending an invariant of singular knots with \(k\) crossings to one with \(k+1\) crossings as making a ``derivative'', then Vassiliev invariants are those whose derivatives eventually vanish. Thus it makes sense to think of Vassiliev invariants as the ``polynomials'' among all invariants. In the present article, the author investigates this idea; he states three results: 1. ``Derivation'' of knot invariants satisfies a Leibniz rule parallel to the Leibniz rule for functions. 2. To a knot invariant \(V\) and a knot \(K\) one can -- via a ``twist sequence'' of \(K\) -- attribute a function \(V_k: \mathbb Z\to \mathbb Q\). \textit{J. Dean} [J. Knot Theory Ramifications 3, No. 1, 7-10 (1994; Zbl 0816.57009)] and \textit{R. Trapp} [ibid., No. 3, 391--405 (1994; Zbl 0841.57019)] showed that if \(V\) is a Vassiliev invariant of degree \(n\) then \(V_k\) is a polynomial of degree \(\leq n\); again Vassiliev invariants take the role of polynomials. 3. For a third parallel, the author refers to his article [A combinatorial half-integration from weight system to Vassiliev knot invariant, ibid. 7, No. 4, 519--526 (1998; Zbl 0905.57005)]. He shows there how some success in the fundamental question of integrability of weight systems can be obtained with symmetry considerations which again come from polynomials. For the entire collection see [Zbl 0890.00048].