QECTOR Decoder v3 (1.0.0) - Full Scientific Verification & Validation Report

This deposit contains the complete author conducted scientific verification and validation package for QECTOR Decoder v3 version 1.0.0 (PyPI: qector decoder v3). Package under test qector decoder v3 1.0.0 (cp311 win_amd64 release wheel) Black box test suite written exclusively against the public API 120 tests × 2 license tiers (Community / Enterprise) × 2 independent runs Verification results Community tier: 105 passed / 0 failed / 15 skipped (87.5 %) Enterprise tier: 119 passed / 0 failed / 1 skipped (99.17 %) Zero residual failures, zero flaky tests Key contents of this deposit Full verification report (PDF + Markdown + JSON + CSV) Logical error rate cross check vs PyMatching (rep d3, p=0.01, 12 000 seeded shots) • qector Blossom LER = 0.00392 • PyMatching LER = 0.00317 • Δ = +0.00075 (passes the acceptance gate Δ ≤ 1e-3) • Wilson 95 % intervals and binomial noise floor analysis CPU and GPU throughput scaling (distance 3 → 63, batches up to 1 000 000 shots) Single shot latency and batch throughput figures for both license tiers Complete SHA 256 provenance manifest for every artifact Charts: LER comparison, statistical proof, scaling curves, verification outcomes, benchmark summary Mathematical formulas (LER estimator, Wilson intervals, acceptance gate, matching definition) Log of three defects identified and fixed before the 1.0.0 release Full black box test suite and verification harness so any third party can re execute the entire procedure Hardware note All measurements performed on a machine with NVIDIA GeForce GTX 1660 Ti (Turing, 2019). GPU figures are indicative; CPU figures are authoritative. Related records This deposit continues the QECTOR benchmark series previously published as: 10.5281/zenodo.21339300 (v0.6.6) 10.5281/zenodo.21760127 (v0.7.0) PyPI: https://pypi.org/project/qector-decoder-v3/Community: QECTOR Research Archive