Reproduction Code and Data for "Dispersion Relations Across the Unitarity Boundary"

This dataset contains the numerical code and data supporting the figures and numerical results reported in: K. Liu, "Dispersion Relations Across the Unitarity Boundary," arXiv:2605.00117 (2026). https://doi.org/10.48550/arXiv.2605.00117 The manuscript shows that the Kramers–Kronig (KK) relations, normally treated as binary, are in fact organized into a sharp dichotomy controlled by microscopic unitarity. A single reduced-state transform carries, simultaneously, a zero and a pole in the upper half-plane, and which is realized is decided by the spectral abscissa of the reduced propagator: on the unitary (passive) side the upper-half-plane object is a protected zero (KK holds, yet the zero is directly measurable and forces a "phantom" refractive resonance with no absorptive origin); on the gain-driven non-unitary side the zero becomes a genuine pole, the Blaschke winding number jumps from 0 to 1, and KK acquires a Lorentzian residue correction whose scaling peaks at threshold. The Letter gives the closed-form criteria, a measurable terahertz signature, and the solvable PT-dimer and Jaynes–Cummings models that realize both sides of the boundary. This archive provides:- the full research codebase (Python) used while developing the manuscript figures, including the two scripts that generate the two published figures plus the surrounding exploration / PT-dimer / scaling scans retained for provenance;- all underlying numerical data (pickled NumPy/SciPy results and a JSON source manifest);- a README that maps each published figure to its generator script and gives the SHA-1 hash of the figure PDF exactly as deposited with the paper, so that reproduced output can be verified byte-for-byte. Status of the code. This is a raw dump of the research code as it stood when the manuscript figures were produced; it has not been refactored into a clean, run-out-of-the-box package. The two published-figure scripts resolve paths relative to the script file: make_fig_dichotomy.py (Fig. 1) is fully self-contained, while make_fig_nonunitary.py (Fig. 2) opens data_pole_trajectory.pkl and data_scaling.pkl from its own directory, but in this archive those .pkl files live under data/ — to reproduce Fig. 2 you must either copy the two .pkl files next to the script or edit its HERE-relative paths. Not every script in code/ produces a manuscript figure; only the two listed in the README's figure-to-script table do. See the in-archive README.md for full guidance and the ground-truth hashes. Software environment: Python 3.11 with NumPy, SciPy, Matplotlib, and mpmath; all computations run on a local workstation/laptop. See requirements.txt and the in-archive README for details.