A class of knots with simple \(SU(2)\)-representations (Q2362849)

One recurring theme in the Floer-theoretic study of 3- and 4-manifolds has been the quest to understand geometric and topological consequences of having minimal Floer-theoretic complexity. While most research in this vein has been from the context of Heegaard Floer homology -- e.g. for L-spaces, L-space knots, and ``Floer simple'' knots -- the current paper extends this trend into the realm of instanton Floer homology and representation varieties, but from a new angle. In particular, for knots in \(S^3\), Zentner introduces a notion of simplicity with respect to the \(\mathrm{SU}(2)\)-representation varieties of the knots' complements. He calls a knot \(K \subset S^3\) \textit{\(\mathrm{SU}(2)\)-simple} if every \(\mathrm{SU}(2)\)-representation of \(\pi_1(S^3 \setminus K)\) with traceless meridian class is binary dihedral. All 2-bridge knots are \(\mathrm{SU}(2)\)-simple. Assuming the irreducible \(\mathrm{SU}(2)\)-representations of \(\pi_1(S^3 \setminus K)\) are transversely cut out as solutions of the instanton equations, \(\mathrm{SU}(2)\)-simplicity of \(K\) then implies that both the instanton knot Floer homology \(I^{\natural}(K)\) and its underlying complex \(CI^{\natural}(K)\) are as simple as possible, in that \[ \text{total rank } CI^{\natural}(K)=\text{rank } I^{\natural}(K) = \det(K), \] where we recall that \[ \det(K) := \Delta_K(-1) = | H_1( \Sigma_2(K) )|, \] for \(\Delta_K(t)\) and \(\Sigma_2(K)\) the Alexander polynomial and branched double cover of \(K\), respectively. Thus, if there were known to be a spectral sequence from \(I^{\natural}(K')\) to \(\widehat{HF}(\Sigma_2(K'))\) for arbitrary \(K' \subset S^3\), then \(\mathrm{SU}(2)\)-simplicity of \(K \subset S^3\) would imply that \(\Sigma_2(K)\) is an L-space. It is unknown whether such a spectral sequence exists, but in lieu of that, Zentner produces an infinite family of \(\mathrm{SU}(2)\)-simple alternating knots, which therefore in fact have so-called ``strong'' L-spaces as their branched double covers [\textit{J. E. Greene} and \textit{A. S. Levine}, Algebr. Geom. Topol. 16, No. 6, 3167--3208 (2016; Zbl 1361.57019)]. He furthermore speculates in Question 8.2 on whether \textit{all} branched double covers of \(\mathrm{SU}(2)\)-simple knots are strong L-spaces, or more stringently whether all \(\mathrm{SU}(2)\)-simple knots are alternating. As a primary consequence of this perspective, Zentner establishes in Corollary 9.2 that every graph manifold integer homology sphere admits an irreducible \(\mathrm{SU}(2)\)-representation of its fundamental group. The argument invokes a non-vanishing result of \textit{P. Kronheimer} and \textit{T. Mrowka} [J. Differ. Geom. 84, No. 2, 301--364 (2010; Zbl 1208.57008)] to show, in Proposition 9.1, that the branched double cover of any non-trivial knot of determinant 1 admits an irreducible \(\mathrm{SU}(2)\)-representation of its fundamental group. The corollary's proof is completed with the observation that any graph manifold is the branched double cover of an arborescent link [\textit{F. Bonahon} and \textit{L. C. Siebenmann}, New geometric splittings of classical knots and the classification and symmetries of arborescent knots. \url{https://www-bcf.usc.edu/~fbonahon/Research/Preprints/BonSieb.pdf}] and [\textit{N. Saveliev}, Invariants for homology 3-spheres. Berlin: Springer (2002; Zbl 0998.57001)]. This result -- that there are no graph manifold integer homology sphere ``instanton Floer L-spaces'' -- differs notably from analogous results for either Heegaard Floer homology or \(\mathrm{SL}(2,\mathbb{R})\) representation varieties. That is, \textit{M. Boileau} and \textit{S. Boyer} [J. Topol. 8, No. 2, 571--585 (2015; Zbl 1330.57005)] have shown that precisely one graph manifold integer homology sphere is a Heegaard Floer L-space, namely the Poincaré sphere. Due to further results of \textit{S. Boyer} and \textit{A. Clay} [Adv. Math. 310, 159--234 (2017; Zbl 1381.57003)] constructing \(\widetilde{\mathrm{PSL}}(2,\mathbb{R})\) representatives of real line actions of fundamental groups of certain graph manifolds (retrospectively determined to be the non-L-space graph manifolds [\textit{S. D. Rasmussen}, Compos. Math. 153, No. 5, 1008--1049 (2017; Zbl 1420.57042)]), it follows that the Poincaré sphere is also the unique graph manifold integer homology sphere whose fundamental group fails to admit irreducible representations of \(SL(2, \mathbb{R})\).