CRN Paper 3 — Companion Dataset (v2): Coherence-Cost Landscape for Dephasing-Assisted Transport on Connectome Routing Motifs

Companion dataset for "Coherence-Cost Landscape for Dephasing-Assisted Transport on Connectome Routing Motifs: A Markovian Baseline for Non-Markovian Extensions." v2 supersedes v1 (doi:10.5281/zenodo.19340280). The entire evidence base is replaced: - Tier 1 (N=1064, ε=0, T2L): population baseline — 98.87% inverted-U, transport and coherence-cost co-peak at κ=1.0.- Tier 2 (N=50, quintile-stratified, 21κ × 11ε, T2L): transport optimum pinned at κ*=1.0 across all disorder levels; coherence-cost proxy rises 2.18-fold at ε=5.- Tier 3 (N=20, T3L cross-port diagnostic): attenuated, non-monotone cost amplification (1.35-fold at ε=5), indicating regime-dependent entropy geometry.- NM2 (N=50, TCL-2 non-Markovian stress test, 21κ × 11ε × 6τ_c × 10 seeds, T2L): 195,300 cells, 0 CP violations. Transport optimum pinned at κ*=1.0 across all τ_c ∈ [0,1]; R₀ shift <0.7%, ΔC shift <0.5%. Markovian baseline confirmed structurally robust. Total: 340,914 raw simulation cells (145,614 Markovian + 195,300 NM2). GKSL parameters: H = J·L (combinatorial Laplacian, J=0.1), expm_multiply propagator, γ_T=1.0, γ_L=5×10⁻⁴, log1p normalization, T_end=10.