[Paper Review] The Universal Perturbative Quantum 3-manifold Invariant, Rozansky-Witten Invariants, and the Generalized Casson Invariant

Nathan Habegger, G.E. Thompson|ArXiv.org|Nov 8, 1999

Geometric and Algebraic Topology35 references14 citations

TL;DR

This paper establishes a conjectural link between the universal perturbative quantum 3-manifold invariant $Z^{LMO}$ and Rozansky-Witten invariants, showing that $Z^{LMO}(M) = 1$ when $b_1(M) > 3$ and is fully determined by the cohomology ring when $b_1(M) = 3$. It further suggests that the generalized Casson invariant can be computed from $Z^{LMO}$, providing a physical and mathematical framework connecting topological quantum field theories and finite-type invariants.

ABSTRACT

Let Z^{LMO} be the 3-manifold invariant of [LMO]. It is shown that Z^{LMO}(M)=1, if the first Betti number of M, b_{1}(M), is greater than 3. If b_{1}(M)=3, then Z^{LMO}(M) is completely determined by the cohomology ring of M. A relation of Z^{LMO} with the Rozansky-Witten invariants Z_{X}^{RW}[M] is established at a physical level of rigour. We show that Z_{X}^{RW}[M] satisfies appropriate connected sum properties suggesting that the generalized Casson invariant ought to be computable from the LMO invariant.

Motivation & Objective

• To establish a conjectural physical and mathematical relationship between the LMO invariant $Z^{LMO}$ and Rozansky-Witten invariants $Z^{RW}_X$.
• To investigate whether the generalized Casson invariant can be computed from $Z^{LMO}$, particularly through connected sum properties.
• To determine the topological significance of $Z^{LMO}$ for 3-manifolds with low first Betti number ($b_1(M) \leq 3$).
• To explore the role of $Z^{LMO}$ as a universal perturbative quantum invariant and its potential to unify quantum invariants via weight systems.
• To provide evidence that $Z^{LMO}$ captures classical topological invariants such as the Casson-Walker-Lescop invariant and possibly higher-order invariants.

Proposed method

• Analyzes $Z^{LMO}(M)$ using Feynman diagram techniques in the space $A(\emptyset)$, modulo IHX and AS relations.
• Applies the LMO invariant's structure to 3-manifolds with $b_1(M) \geq 4$, proving $Z^{LMO}(M) = 1$ via vanishing of relevant diagrams.
• Uses cohomology ring data to fully determine $Z^{LMO}(M)$ when $b_1(M) = 3$, relying on the AS relation and diagrammatic constraints.
• Establishes a heuristic physical correspondence between Rozansky-Witten invariants and $Z^{LMO}$, based on path integral arguments and BRST symmetry.
• Applies the Hilbert space formalism on $S^2$ and derives the connected sum formula for $Z^{LMO}$, linking it to the partition function on $S^3$.
• Uses Berezin integration and curvature tensor decomposition on hyper-Kähler manifolds to express the Euler class as a Grassmann integral, supporting the weight system in $Z^{RW}$.

Experimental results

Research questions

• RQ1 Can the generalized Casson invariant be computed from the LMO invariant $Z^{LMO}$ via a universal perturbative framework?
• RQ2 What is the precise relationship between the Rozansky-Witten invariant $Z^{RW}_X$ and the LMO invariant $Z^{LMO}$ for 3-manifolds with $b_1(M) \leq 3$?
• RQ3 How does $Z^{LMO}(M)$ depend on the cohomology ring of $M$ when $b_1(M) = 3$?
• RQ4 Does $Z^{LMO}$ encode the Reidemeister-Franz torsion or Alexander polynomial through its path integral formulation?
• RQ5 To what extent does $Z^{LMO}$ capture the structure of moduli spaces of flat connections, as suggested by the physical duality with Chern-Simons theory?

Key findings

• $Z^{LMO}(M) = 1$ for all 3-manifolds $M$ with first Betti number $b_1(M) > 3$, indicating triviality of the invariant in high Betti number cases.
• When $b_1(M) = 3$, $Z^{LMO}(M)$ is completely determined by the cohomology ring of $M$, showing a strong topological constraint.
• The Rozansky-Witten invariant $Z^{RW}_X(M)$ is conjectured to satisfy connected sum properties analogous to those of $Z^{LMO}$, suggesting a universal structure.
• A physical path integral argument supports the conjecture that $Z^{RW}_X(M) = \chi_G(M)$, the regularized Euler characteristic of the moduli space of flat $G$-connections.
• The Euler class of a compact hyper-Kähler manifold $X$ of real dimension $4n$ is expressed as a Grassmann integral involving the Riemann curvature tensor and holomorphic forms.
• The paper provides a formal derivation of the Euler class as $\mathbf{e}(TX) = \sqrt{g} \left( \int d\mu(\chi_\alpha) \exp\left( \frac{1}{24} \Omega_{IJKL} \chi^I_\alpha \chi^J_\beta \chi^K_\gamma \chi^L_\delta \epsilon^{\alpha\beta\gamma\delta} \right) \right) d^{4n}x$, linking it to the Rozansky-Witten weight system.