Computing the twisted $L^2$-Euler characteristic
Jacopo G. Chen
TL;DR
This work develops an algorithm to compute Friedl and Lück’s twisted $L^2$-Euler characteristic for suitable $L^2$-acyclic CW complexes by leveraging Oki’s matrix expansion to estimate degrees of Dieudonné determinants within the Linnell skew field $\mathcal{D}(G)$. Grounded in the polytope interpretation of determinants and the universal $L^2$-torsion, the method expresses twists by a character $\phi$ as a single polytope-valued invariant whose degree recovers $\chi^{(2)}(\widetilde{M};\mathcal{N}(G),\phi)$; Lück’s approximation guides the finite-quotient computations, with error controlled by recent quantitative bounds. The paper shows that a truncated, human-assisted version yields accurate results in diverse settings, including hyperbolic link complements, closed census 3-manifolds, free-by-cyclic groups, and higher-dimensional manifolds such as the Ratcliffe–Tschantz fiber, illustrating a practical path to exploring a Thurston-like norm in higher dimensions. Empirically, the approach recovers Thurston and Alexander norm unit balls in several cases and suggests a deep link between twisted $L^2$-invariants and fibering phenomena, with broader implications for understanding geometry via noncommutative determinant theory. The work combines deep noncommutative algebra, $L^2$-invariants, and computational topology to produce an effective toolkit for probing norm-like invariants beyond dimension three.
Abstract
We present an algorithm that computes Friedl and Lück's twisted $L^2$-Euler characteristic for a suitable regular CW complex, employing Oki's matrix expansion algorithm to indirectly evaluate the Dieudonné determinant. The algorithm needs to run for an extremely long time to certify its outputs, but a truncated, human-assisted version produces very good results in many cases, such as hyperbolic link complements, closed census 3-manifolds, free-by-cyclic groups, and higher-dimensional examples, such as the fiber of the Ratcliffe-Tschantz manifold.
