Vacuum polarization in quantum-corrected and effective black hole geometries: A high performance approach

# Data underlying the paper's figures This folder contains the data from which every figure (and every tableentry) of the paper is built. Each case is one complete mode-sumcomputation (Sec. IV of the paper): a radial profile vp(x), withvp = 16 pi^2 M^2 <phi^2>, in units rs = 2M = 1, x = r/rs. The profiles are supplied on the REPORTED window x - x_+ >= 10^-2 only. ## dataset_miyabi/ — the unified production campaign (430 runs) Single-provenance dataset (Miyabi/JCAHPC, 2026-08-03, v2 engine, zerohealth-gate warnings) behind Secs. V and VI: seven metrics(KS, ZLMY-1, ZLMY-2, Bonanno-Reuter, Bardeen, Schwarzschild,Reissner-Nordstrom), field masses mu = 0.025, 0.05, 0.2, 0.5, 1.0,couplings xi = 0, 1/6 (plus the six-point xi scan). - `part_1/` — main parameter grids (geometry x deformation x mass x xi): trends, profiles, overlays, DeWitt-Schwinger tails, TK comparison (Figs. 2-7, 9-13; Tables I-III).- `part_2/` — validation runs: Anderson RN/Schwarzschild benchmarks, classical limits, light-mass anchors (Fig. 1).- `part_3/` — near-extremal study: Bardeen sign map, matched-temperature mass scans for Bardeen/BR/RN (Figs. 14-15; Sec. VI).- `part_4/` — six-point xi scan for KS and BR (Fig. 8). Each phase directory contains `INDEX.csv` (one row per case: geometry,quantum parameter, m, xi, grid size, horizon value vp_H, health-gatestatus, host, runtime, source checksum) and one CSV per case. Per-caseCSVs carry a comment header stating the metric functions f(x), g(x),the line element, the run parameters and provenance, followed by thevp(x) profile (two whitespace-separated columns: x, vp). ## ds_tail.json The parameter-free analytic DeWitt-Schwinger predictionvp_DS = (1/4)[a_2]/m^2 (Eq. (19) of the paper), evaluated from theanalytic curvature invariants of each geometry; these are the dashedcurves of Fig. 13. ## Provenance and error budget Runs were executed on the Miyabi Supercomputing System (JCAHPC);bitwise-class reproducibility versus the development workstation wasverified (max profile deviation 2x10^-12). The absolute noise floor is~3x10^-7 in vp; horizon values carry 0.1% window stability and a<~0.5% method-level systematic (Sec. V of the paper). `src_sha256` inthe INDEX files identifies the exact generated source of each run.