The Ptolemy field of 3-manifold representations (Q2340081)

Let \(M\) be a compact three-manifold with nonempty toric boundary. This paper gives an efficient algorithm to compute the trace field of representations into \(\mathrm{PSL}_2(\mathbb{C})\) of the group \(\pi_1(M)\) from an ideal triangulation \(\mathcal T\) of \(M\). The main theorem in the paper goes roughly as follows: there is a parametrization of representations (with some added data) by points in a so-called ``Ptolemy variety'' which is constructed from the triangulation (assuming the latter has no non-essential edges). The trace field of the representation is then equal to the field generated by the coordinates of a point (i.e. the values at this point of regular functions on the whole variety) in the Ptolemy variety associated to it. Moreover this variety is effectively computable using computer algebra. This rough statement should be enough for people with an interest in the computation of trace fields; in the rest of this review we will try to explain in some detail the steps involved in these computations. Recall that \(\mathcal T\) is an ideal triangulation of \(M\) ; let \(\{1,\ldots,e\}\) and \(\{1,\ldots,v\}\) be the respective sets of indices for the edges and the vertices in \(\mathcal T\). The Ptolemy variety \(P_2(\mathcal T)\) is the affine subvariety of \(\mathbb{C}^e\) given by the equations \[ c_{03}c_{12} + c_{01}c_{23} = c_{02}c_{13} \] for each 3-simplex of \(\mathcal T\), where \((c_{03}, c_{12}), (c_{01}, c_{23})\) and \((c_{02}, c_{13})\) are the pairs of opposite edges in that simplex. To a point on this variety one can associate a decorated representation of \(\pi_1(M)\), that is a representation to \(\mathrm{SL}_2(\mathbb{C})\) together with an equivariant map from the set of peripheral subgroups of \(\pi_1(M)\) (equivalently the set of lifts to the universal cover of the boundary components) to \(\mathrm{SL}_2(\mathbb{C})/P\) where \(P\) is a fixed unipotent subgroup. The representations obtained from the Ptolemy manifold are boundary-parabolic (i.e. images of peripheral subgroups are inside conjugates of \(P\)) and the decorations are ``generic''. Moreover one gets a bijective correspondence this way. Not all \(\mathrm{PSL}_2(\mathbb{C})\)-representations lift to boundary-parabolic \(\mathrm{SL}_2(\mathbb{C})\)-representations: for example the holonomy representation of an hyperbolic 1-cusped manifold never lifts. One can modify the construction of the Ptolemy manifold to includ lifts of such representations in the above picture: there are Ptolemy varieties \(P_2^\sigma(\mathcal T)\) (for \(\sigma\) in a certain finite cohomology group) that parametrize boundary-parabolic \(\mathrm{PSL}_2(\mathbb{C})\)-representations. Then not all such representations are obtained this way, but when the triangulation has no non-essential edge the holonomy representation is. Finally, one associates to each Ptolemy variety its quotient by the diagonal action of \((\mathbb{C}^\times)^v\) : on the representation side this corresponds to taking representation up to conjugacy, and the final result (Theorem 1.14) is that the trace field of a representation (which depends only on its conjugacy class) equals the field generated by the coordinates of a corresponding point on the appropriate reduced Ptolemy variety. Since quotients of varieties are not usable in practice one can replace them by a section of a Ptolemy variety (Proposition 1.16) by normalizing certain edge coordinates to be 1.