Causal spinfoam vertex for 4d Lorentzian quantum gravity

We introduce a new causal spinfoam vertex for $4$d Lorentzian quantum gravity. The causal data are encoded in Toller $T$-matrices, which add to Wigner $D$-matrices $T^{(+)}+T^{(-)}=D$, and for which we provide a Feynman $\mathrm{i}\varepsilon$ representation. We discuss how the Toller poles cancel in the EPRL vertex, how the Livine-Oriti model is obtained in the Barrett-Crane limit, and how spinfoam causal data are distinct from Regge causal data. In the large-spin limit, we show that only Lorentzian Regge geometries with causal data compatible with the spinfoam data are selected, resulting in a single exponential $\exp(+\mathrm{i}, S_{\mathrm{Regge}}/\hbar)$ and a new form of causal rigidity.

💡 Research Summary

The paper introduces a novel causal spinfoam vertex for four‑dimensional Lorentzian quantum gravity, extending the widely used Engle–Pereira–Rovelli–Livine (EPRL) model by incorporating explicit causal data. The key innovation is the use of Toller T‑matrices, which are meromorphic functions on the Lorentz group SL(2,ℂ) characterized by simple poles (the “Toller poles”) in the complex ρ‑plane. These matrices satisfy the additive relation T⁺ + T⁻ = D, where D denotes the usual Wigner D‑matrix that appears in the EPRL vertex. The authors provide a Feynman‑type iε prescription for the T‑matrices, making the causal structure manifest at the level of the vertex amplitude.

In the spinfoam path integral, each edge of a 4‑simplex is assigned an orientation σₐ = ±1 (ingoing/outgoing). For a wedge (ab) formed by two edges a and b, the product κ_{ab}=σₐσ_b determines whether the wedge is “thick” (co‑chronal) or “thin” (anti‑chronal). The causal vertex amplitude is built by inserting a T‑matrix with sign ± determined by κ_{ab} into the usual product of group elements g_a⁻¹g_b. Explicitly, the amplitude reads ⟨A(σₐσ_b) v| = ∫∏a dg_a ∏{a<b} T^{(κ_{ab})}(γj_{ab},j_{ab})(g_a⁻¹g_b). This construction makes the amplitude depend explicitly on the combinatorial causal data σₐ, unlike the original EPRL vertex which is independent of σₐ.

Two important properties follow from the iε representation. First, the sum of the two T‑matrices reproduces the Wigner D‑matrix, so the EPRL vertex can be written as an unconstrained sum over all possible wedge signs κ_{ab}=±1. Second, the T‑matrices admit a coherent‑state expansion that mirrors the one used in the EPRL asymptotic analysis, allowing the authors to carry over saddle‑point techniques. The pole structure of the T‑matrices is such that, when both signs are summed, the poles cancel, leaving a regular d‑matrix; this explains why the EPRL model, which effectively sums over all causal configurations, does not exhibit any explicit causal dependence.

The paper also demonstrates that in the limit γ→∞ (Barbero–Immirzi parameter going to infinity) with fixed areas ρ=γj, the causal vertex reduces to the Livine–Oriti causal version of the Barrett–Crane model. In this limit the reduced T‑matrices become simple exponential functions that match the edge amplitudes proposed by Livine and Oriti, confirming that the new vertex interpolates smoothly between the EPRL and Barrett–Crane frameworks.

The authors then turn to the large‑spin (j ≫ 1) asymptotics. They consider a semiclassical boundary state peaked on a non‑degenerate Lorentzian 4‑simplex, described by spins j_{ab} and spinors ζ_{ab} that encode the geometry of the five boundary tetrahedra. The Regge geometry is specified by timelike normals N^I_a to each tetrahedron, their volumes V_a, and causal signs s_a=±1 indicating whether each normal is future‑ or past‑pointing. The Regge action reads S_Regge = ∑{ab} A{ab} 8πG s_a s_b β_{ab}, where β_{ab} is the hyperbolic dihedral angle between tetrahedra a and b.

In the standard EPRL analysis, the action appearing in the exponent of the amplitude is S = ∑{ab}(γj{ab}B_{ab}+j_{ab}Φ_{ab})+i∑{ab}j{ab}Q_{ab}. Stationary‑phase conditions produce two critical points g^{(±)}a, with B{ab}^{(±)} = ± s_a s_b β_{ab}. Consequently the on‑shell action becomes S^{(±)} = ± S_Regge/ħ + Υ, leading to a cosine of the Regge action in the asymptotic expression.

When the causal T‑matrices are used, the same coherent‑state machinery yields an identical functional form for the exponent, but now the causal signs σₐσ_b are fixed by the combinatorial data. Consistency between the spinfoam causal data and the Regge causal data requires σₐσ_b = s_a s_b for every wedge; this is the “causal rigidity” condition. It is satisfied only for transitions such as 1 → 4, where one tetrahedron lies on the past boundary and four on the future boundary of the 4‑simplex. Under this restriction the global sign ± in B_{ab} is no longer summed over; only one of the two critical points survives. The resulting amplitude therefore contains a single exponential exp(+i S_Regge/ħ) rather than a cosine, demonstrating that fixing the causal structure selects a unique orientation of the Regge geometry.

In summary, the paper achieves three major results: (1) it provides a mathematically rigorous way to embed causal information into spinfoam vertices via Toller T‑matrices and an iε prescription; (2) it shows that the causal vertex reduces to known models (EPRL and Livine–Oriti Barrett–Crane) in appropriate limits, establishing continuity with existing literature; (3) it proves that in the semiclassical regime the causal vertex enforces a compatibility condition between combinatorial and geometric causal data, leading to a single‑exponential asymptotics and introducing the notion of causal rigidity. These findings open a clear pathway toward causal spin‑network dynamics, suggest a bridge between spinfoam quantum gravity and causal‑set approaches, and provide a concrete tool for future investigations of time‑oriented quantum geometries.