The \(C\)-polynomial of a knot (Q5917534): Difference between revisions

In low-dimensional topology, the \(A\)-polynomial is a strong tool and has been studied intensively from different angles by many researchers. In this paper, by using the \(C\)-polynomial, the author further explores information encoded in the \(A\)-polynomial. The \(C\)-polynomial \(C_{W,K,B}(t)\) is a one-variable polynomial invariant for the triple \((W,K,B)\), \(W\) a connected oriented closed 3-manifold, \(K\) a knot in \(W\), and \(B\) a basis of \(H_1(W_K;{\mathbb Z})\), where \(W_K\) is the exterior of the knot \(K\). This invariant is defined as a reduction of the \(A\)-polynomial for the triple \((W,K,B)\), which is a generalization of that in [\textit{D. Cooper, M. Culler, H. Gillet, D. D. Long} and \textit{P. B. Shalen}, Invent. Math. 118, 47--84 (1994; Zbl 0842.57013)]. Here are more details. The \(A\)-polynomial \(A_{W,K,B}(x,y)\) is constructed basically by the defining polynomial of the \(SL(2,\mathbb{C})\)-character variety of knot group \(\pi_1(W_K)\) projected to that of \(\pi_1(\partial W_K)\). Actually, the resulting variety can be identified approximately with a 1-dimensional variety or the empty set in \({\mathbb C}^2\) through a certain operation. Its defining polynomial is a polynomial in two variables over \({\mathbb Z}\) in the former case, and is defined to be one in the latter case. Now, by the Taylor expansion of \(A_{W,K,B}(x,y)\) at \((\varepsilon_1, \varepsilon_2)\), \(\varepsilon_1, \varepsilon_2 \in \{\pm 1\}\), where \((\varepsilon_1, \varepsilon_2)\) is a solution of \(A_{W,K,B}(x,y)=0\), we have \(A_{W,K,B}(x,y)=\sum_{k=n}^{d} \sum_{i=0}^{k} a_{k,i} (x-\varepsilon_1)^{k-i}(y-\varepsilon_2)^i\). Then pick up the part \(\sum_{i=0}^{n}a_{n,i} (x-\varepsilon_1)^{n-i}(y-\varepsilon_2)^i\) and put \(t^i=(x-\varepsilon_1)^{n-i}(y-\varepsilon_2)^i\) to get a one-variable polynomial \(C_{W,K,B,(\varepsilon_1,\varepsilon_2)}(t)\), called a main factor of the \(C\)-polynomial later. Then the product of \(C_{W,K,B,(\varepsilon_1,\varepsilon_2)}\) over all different solutions \((\varepsilon_1, \varepsilon_2)\) is the \(C\)-polynomial of the triple \((W,K,B)\). If \(A_{W,K,B}(x,y)=0\) has no solution of the form \((\varepsilon_1,\varepsilon_2)\), the \(C\)-polynomial is defined to be zero, and is said to be trivial. Note that the \(C\)-polynomial has integral coefficients. As above, the operation to reduce the \(A\)- to the \(C\)-polynomial is somewhat artificial, however it preserves some information on the representation of the knot group. For example, the \(C\)-polynomial is non-trivial if \(W_K\) is hyperbolic. In particular, it is non-trivial for hyperbolic knots and torus knots in the 3-sphere \(S^3\), a property it shares with the \(A\)-polynomial. Another example is a relationship between the \(C\)-polynomial and Kuga's polynomial [\textit{K. Kuga}, J. Math. Soc. Japan 45, No.1, 67--76 (1993; Zbl 0801.57005)]. In fact, a modified Kuga polynomial is a factor of the \(C\)-polynomial. Some of the main results in this paper are as follows. If \(C_{W,K,B}(t)\) is not monic, then either the 3-manifold \(W\) is not a homology 3-sphere or \(W_K\) contains a closed essential surface or \(W_K\) is a solid torus. Let \(K\) be a non-trivial knot in the 3-sphere whose exterior contains no essential closed surfaces. If there exists a main factor \(C_K(t)=C_{S^3,K,B,(\varepsilon_1,\varepsilon_2)}(t)\) of its \(C\)-polynomial such that \(C_K(1)\neq \pm 1\) and \(C_K(-1)\neq \pm 1\), then \(K\) has Property-P.