The Causal Spine of Discrete Spacetime: Gravitational Wave Dispersion and LISA Falsifiability in Causal Dynamical Triangulations (GRU v2.3)

Version 2.3 of GRU (Geometría Radial Unitaria). New central results: εA = 4.1816% ± 0.0006% — universal invariant, verified across 5 independent seeds and scaling law εA ∝ 1/√N confirmed with slope = −0.500 exact. CMB quadrupole bug corrected. What's new in v2.3: εA = 4.1816% ± 0.0006% — Universal Invariant: Measured by gravitational wave injection experiment on GRU causal mesh (N=20 shells, 2560 nodes). CV = 0.015% across 5 seeds. Zero frequency dependence → purely geometric effect. Scaling law εA ∝ 1/√N verified: slope log-log = −0.500 exact (N=10→5.91%, N=20→4.18%, N=50→2.65%, N=100→1.87%, N=200→1.32%). Confirms analytic Mewes formula. CMB bug corrected: v2.2 used spurious denominator 1+(κ/2.5)² with no physical derivation, giving Dℓ=3.78 µK² (34% tension). v2.3 canonical formula R2(κ)=(1−e−√6·κ/2)² gives Dℓ=6.34 µK² (+0.67σ from Planck 5.77±0.85 µK²) ✅ Fisher updated: σn=0.0027 (1.13× improvement), N=1 SMBHB event for 5σ unchanged. 13 files in GRU_v2_3_scripts.zip (4 Python scripts, 4 result logs, 3 JSON/CSV data files, 1 txt log, 1 README.md with decision tree C1→C2 and replication instructions). 3 new references: Ambjørn & Loll 2026 (arXiv:2604.05641), Gong et al. 2023 (PRD 107, 124015), Wang et al. 2025 (PRD 111, 084064). v2.2 results confirmed: λcorr=2.14±0.79 (21/32 CDT in LISA window), Afinal=1.46×10−16 ✅, nGRU=0.0564, factiva=1.0, ηint=0.169±0.038. Concept DOI (always latest): 10.5281/zenodo.20352929