Qector Mega: IonQ Superion 256 Walking Cat Decoder (v2.2.1 Frozen Release)
Description
Qector Mega: IonQ Superion 256 Walking Cat Decoder (v2.2.1 Frozen Release)
Distribution Package Manifest & Technical Release Summary
Archive: qector_ionq_v2_frozen_clean.zip (445 entries)
Access Tier: Restricted (Proprietary / NDA Evaluation Only)
Distribution License: LicenseRef-Proprietary (Redistribution Strictly Prohibited)
Overview
| Metadata Attribute | Specification |
|---|---|
| Release Version | 2.2.1 (Canonical Production Freeze) |
| Author | Guillaume Lessard (ORCID: 0009-0000-3465-3753) |
| Affiliation | iD01t Productions |
| Programme | qector-decoder-v3 |
| Target Hardware | IonQ Superion 256 |
| Target Architecture | Walking Cat (arXiv:2604.19481) |
| Python Support | CPython ≥3.9 (x86_64 / AMD64) |
| Repository / Portal | qector.store |
Abstract & Mathematical Invariants
qector-mega v2.2.1 is a frozen production decoder engineered for quantum error correction on the IonQ Superion 256 Walking Cat architecture. The release contains optimized PyO3 Python bindings, high-throughput Rust source code, Lean verification stubs, automated unit tests, and production certification matrices.
Core Verification Invariant
All public decode routines yielding a correction vector c are strictly required to satisfy:
where H is the parity-check matrix and s is the observed syndrome vector. Any syndrome lying outside reachable syndrome spaces terminates immediately with an explicit typed error.
Supported Quantum Codes
The package targets three Walking Cat code configurations, verified via certs/cert_q102_production.json:
| Code Designation | Code Parameters [[n,k,d]] | Configuration / Polynomials | Artifact Hash |
|---|---|---|---|
| Q70 | [[70,6,9]] | Standard parity configuration | 139f6386e5fe0268 |
| Q102 | [[102,22,9]] | A=[22,26,37,50], B=[19,28,29,35] | 9d01bcd0d7c915aa |
| Gross-144 | [[144,12,12]] | Gross bivariate cyclic configuration | 28f2131d00cff8e0 |
Certified Production Performance (v2.2.1)
All metrics below are certified by certs/cert_q102_production.json and certs/cert_math.json using the v2.2.1 production binary:
| Code | Batch Size (B) | Median Latency (p50) | Tail Latency (p95) | Throughput | Exhaustive / Invariant Checks |
|---|---|---|---|---|---|
| Q70 | 2,000 | 0.128 ms | 0.173 ms | 8,644 shots/s | Validated against parity check |
| Q102 | 2,000 | 0.320 ms | 0.558 ms | 3,634 shots/s | Weight-2 (w2) exhaustive: 5,151 / 5,151 |
| Gross-144 | 2,000 | 0.278 ms | 0.308 ms | 4,414 shots/s | Validated against parity check |
- Mathematical Precision (
cert_math.json): ϕ involution maximum numerical deviation is 2.5×10−13 across the recorded test grid. - SIMD Implementation Details: AVX2 dual-YMM path delivers ∼76 μs (p50) on Q70-like row-elimination operations on the reference build host. The AVX-512 backend remains feature-gated due to
rustc 1.75target_featureinstabilities.
Distribution Artifacts & Integrity Hashes
The archive contains six verified binary wheels (qector_mega-2.2.1-*.whl). Legacy naming schemes (qector_ionq-*) are deprecated and absent from this bundle.
Linux Distribution (manylinux_2_34_x86_64)
- SHA-256:
b69584fe57a05232f61193ad953fe0e1486db12504017bd5fbaede678846c4d3wheels/qector_mega-2.2.1-cp312-cp312-manylinux_2_34_x86_64.whl(1,077,916 bytes)
Windows Distributions (win_amd64)
- Python 3.9:
a663651e1396dc3b5d17ff1c85ea64e665ed39133d40097b48f5591f25b57305wheels/qector_mega-2.2.1-cp39-cp39-win_amd64.whl - Python 3.10:
191dfeecb1141ecbd9eac2eef386498e790142f93df039eebb218d7eea1ace9fwheels/qector_mega-2.2.1-cp310-cp310-win_amd64.whl - Python 3.11:
298062b91384ff1c34c69187c22e9695f1833a2d3779f4d6f6c3be82ff1d7e55wheels/qector_mega-2.2.1-cp311-cp311-win_amd64.whl - Python 3.12:
8a0858cc19524c070426f247e923f8015b3c0f17faf607b2e7296729b2742c83wheels/qector_mega-2.2.1-cp312-cp312-win_amd64.whl - Python 3.13:
ac0dcff193f01835242e5d8d0c8db342aa3da92b9447b8a0a93e89a251f02094wheels/qector_mega-2.2.1-cp313-cp313-win_amd64.whl
Archive Internal Provenance & Audit Notice
This release package represents a structured multi-version development tree containing 445 files. When auditing internal documentation, note the following version baselines:
- Active Binaries: All compiled artifacts strictly reflect release 2.2.1.
- Historical Metadata: Certain repository files reflect predecessor iterations and should not be cited as current v2.2.1 metrics:
-
FROZEN_MANIFEST.txtandZENODO_DESCRIPTION.mdcontain legacy headers and references to v1.7.8 wheel packaging. -
CHANGELOG.mdcatalogs intermediate developments (1.7.x→2.0.0→2.1.0). -
v2.0.0 Baseline Benchmarks (
BENCHMARKS.md, 2026-09-14): -
Q70: p95=0.45 ms
-
Q102: p95=1.26 ms
-
Gross-144: p95=0.88 ms
-
Legacy campaign text detailing GPU T4 throughput (62,826/34,402/28,643 shots/s) and CPU Colab benchmarks (7,110/3,022/3,941 shots/s) are superseded by v2.2.1 certification data.
Citation & Licensing
This software is distributed under restricted proprietary licensing for evaluation under non-disclosure agreements only. Unauthorized distribution, extraction, or copying is strictly prohibited.
Citation Format
@software{lessard2026qector,
author = {Lessard, Guillaume},
title = {Qector Mega 2.2.1: IonQ Superion 256 / Walking Cat Decoder (Frozen Restricted Package)},
year = {2026},
publisher = {iD01t Productions / qector.store},
version = {2.2.1},
note = {ORCID: 0009-0000-3465-3753. Proprietary evaluation bundle.}
}
Files
Additional details
Related works
- Is documented by
- Software documentation: 10.5281/zenodo.22684231 (DOI)
- Is source of
- Dataset: 10.5281/zenodo.22678419 (DOI)
Dates
- Submitted
-
2026-09-15Software