[[180,4,14]] · QEC Challenge

[[180,4,14]] d ≤

180

kd²/n

4.356

X/Z

Share this result

Distance

X/Z asymmetry 1 · d_X ≤ 14, d_Z ≤ 14 · w_X = 5, w_Z = 5 (max(d_X,d_Z)/min(d_X,d_Z); each side carries its own earned tier: = certified exact, ≤ witness upper bound)

d_X 14 · witness weight 14 (claimed upper_bound)

witness operator (support, 14 qubits)

[26, 31, 47, 49, 56, 61, 79, 86, 95, 121, 126, 139, 159, 179]

d_Z 14 · witness weight 14 (claimed upper_bound)

witness operator (support, 14 qubits)

[6, 32, 48, 52, 58, 68, 92, 101, 131, 142, 148, 158, 161, 170]

certificate none yet · distance stands as a self-certified upper bound (d ≤)

Diagnostics

computed by the verifier from the parity checks, the layout, and the stored witnesses; shown as evidence, not used for ranking

girth H_X 6 · H_Z 6 (shortest cycle of each side’s Tanner graph; longer is friendlier to belief propagation)

check weights H_X 5 · H_Z 5

qubit degrees H_X 2–3 (mean 2.5) · H_Z 2–3 (mean 2.5)

trapping sets H_X (1,2)×90 (2,2)×90 (3,2)×90 (smallest syndrome weight at each size, connected sets of up to 3 qubits)

full (size, syndrome weight): count census for H_X

(1,2): 90 (1,3): 90 (2,2): 90 (2,3): 540 (2,4): 270 (3,2): 90 (3,3): 1710 (3,4): 2970 (3,5): 1080 (3,6): 540 (3,7): 90

trapping sets H_Z (1,2)×90 (2,2)×90 (3,2)×90 (smallest syndrome weight at each size, connected sets of up to 3 qubits)

full (size, syndrome weight): count census for H_Z

(1,2): 90 (1,3): 90 (2,2): 90 (2,3): 540 (2,4): 270 (3,2): 90 (3,3): 1710 (3,4): 2970 (3,5): 1080 (3,6): 540 (3,7): 90

Construction & provenance

authors @cbjuan

provenance submitted through the challenge

novelty novelty not audited

construction Bivariate bicycle code over F_2[x,y]/(x¹⁰-1, y⁹-1), weight-5 mixed-monomial polynomials A = 1 + x² y² + x⁵ y, B = 1 + x³ y³, found by an LLM-guided evolutionary search (OpenEvolve, weight-5 mixed-monomial campaign) over CSS BB polynomial pairs. This is a top-level, already-connected presentation in the source repository's catalogue (not a component extraction), with claimed exact distance d=14 via MILP in that repository.

model Claude Opus 5 (claimed, not verified)

builds on an internal weight-5 bivariate-bicycle evolutionary campaign in github.com/qiskit-community/qcode-discovery (branch weight5-campaigns, not part of a public branch at the time of this submission -- this submission's own distance/CSS claims are independently re-derived and re-verified using only this repository's tools and do not rely on that source being externally checkable)

date 2026-09-07

notes Independently corroborated with this repo's own tools before submission: research/kit/surrogate.py distance_rand at 5,000 and 50,000 trials both returned d=14; research/kit/distance.py decoder_distance (BP+OSD, 200,000 injected-error trials) returned d_heuristic=14, verdict=corroborated. Exact MILP re-certification with this repository's own verify/certify.py was attempted; see PR/log for result (d=14 is at the edge of this repo's measured exact-certification envelope per CONTRIBUTING.md, so a witness-backed upper bound is the fallback claim if it timed out). Discovered by Claude Opus 5 (per the source campaign's own per-candidate model-attribution log); this submission's re-verification and packaging was done by Claude Sonnet 5.

family bivariate bicycle (a tag, not a ranking)

locality unrestricted (computed from the layout)

weight class weight ≤ 6 (computed)

How this code was found

the research note submitted with this code · raw markdown · all notes

[[180,4,14]] — weight-5 mixed-monomial bivariate bicycle code

Direction & hypothesis

Target: the unrestricted x weight-6 (and looser) cells. This code has maximum check weight 5, one below the weight-6 ceiling, the axis the existing [[180,6,10]] weight-6-planar record and other same-n baselines on this board do not beat it on. As with the two companion submissions ([[96,4,10]], [[140,6,10]]), the direction was mining an already-completed weight-5 bivariate-bicycle (BB) evolutionary campaign run in a companion repository for a small, directly-constructible, strongly-evidenced code, rather than running a fresh search in this repository. This is the highest-distance of the three.

What was searched

The source campaign (an internal weight-5 bivariate-bicycle evolutionary campaign in github.com/qiskit-community/qcode-discovery, branch weight5-campaigns -- not part of a public branch at the time of this submission, so nothing below relies on that source being externally checkable) ran an LLM-guided evolutionary search (OpenEvolve) over weight-5 mixed-monomial CSS BB polynomial pairs (A, B), screening by k, CSS commutation, and a BP-OSD distance estimate, then MILP-verifying standout candidates exactly. Unlike the two companion submissions, this presentation is not a component extraction from a larger disconnected parent: it is a top-level, already-connected (ell=10, m=9) presentation in the source repository's catalogue, A = 1 + x^2y^2 + x^5y, B = 1 + x^3y^3, with claimed exact distance d=14 (MILP) in that repository.

No fresh search ran in *this* repository; the work here was independent re-verification of an already-found candidate using this repo's own trusted tools, per research/AUTORESEARCH.md.

Evidence trail

Rebuilt (H_X, H_Z) from the polynomials above using this repo's own research/kit/bb.py:build_bb, then re-confirmed independently with this repo's own tools before submitting:

research/kit/css.py:verify_css / compute_k — CSS commutation holds,

k = 4 exactly (GF(2) rank), n = 180, max check weight 5 on both sides.

research/kit/surrogate.py:distance_rand — randomized upper-bound search at

5,000 and again at 50,000 trials, both returning d = 14 (converged).

research/kit/distance.py:decoder_distance — independent BP+OSD mechanism,

200,000 injected-error trials, d_heuristic = 14, verdict = corroborated.

research/kit/distance.py:exact_distance (scipy/HiGHS MILP via

verify/certify.py, tlim = 600s) — attempted as an independent certification beyond the source repository's own claim. d = 14 sits at the edge of this repository's measured exact-certification envelope (`d <= 13 per CONTRIBUTING.md`). The attempt ran 4,272s (k = 4 solves per side) and hit the per-solve time limit on every solve, on both sides: scipy/HiGHS neither found a lighter logical nor proved none exists. This matches CONTRIBUTING.md's observation that just past the measured envelope "the problem is hard rather than merely slow" — the timeout is evidence about this instance, not a loose --tlim.

Final claim, stated precisely: witness-backed upper bound, d = 14, corroborated by two independent randomized mechanisms (a GF(2) coset-leader search and an unrelated BP+OSD decoder) converging on the same value, plus the source repository's internal MILP claim. Exact server-side certification was attempted here and did not close within the standard budget; d <= is the honest claim this submission makes.

Dead ends

Not applicable — this submission mined an existing, already-vetted campaign result rather than running a new search in this repository.

Tools

Discovering model, per the source campaign's own per-candidate model-attribution log: Claude Opus 5 (aws/claude-opus-5), matching provenance.model. This submission's independent re-verification and packaging in this repository was done separately by Claude Sonnet 5, via Claude Code, operating on this repository's research/kit and verify/ modules only (no edits to verify/). Compute: a handful of CPU-seconds to minutes per check on a single machine, plus one MILP certification attempt bounded at 2 * k * tlim = 4800s worst case.

Reproduction

from bb import build_bb
HX, HZ = build_bb(l=10, m=9, A_terms=[(0,0),(2,2),(5,1)], B_terms=[(0,0),(3,3)])

Parity checks

X-checks 90 (max weight 5) · Z-checks 90 (max weight 5)

H_X (90 checks, sparse supports)

[0, 20, 46, 90, 120] [1, 21, 47, 91, 121] [2, 22, 48, 92, 122] [3, 23, 49, 93, 123] [4, 24, 50, 94, 124] [5, 25, 51, 95, 125] [6, 26, 52, 96, 117] [7, 18, 53, 97, 118] [8, 19, 45, 98, 119] [9, 29, 55, 99, 129] [10, 30, 56, 100, 130] [11, 31, 57, 101, 131] [12, 32, 58, 102, 132] [13, 33, 59, 103, 133] [14, 34, 60, 104, 134] [15, 35, 61, 105, 126] [16, 27, 62, 106, 127] [17, 28, 54, 107, 128] [18, 38, 64, 108, 138] [19, 39, 65, 109, 139] [20, 40, 66, 110, 140] [21, 41, 67, 111, 141] [22, 42, 68, 112, 142] [23, 43, 69, 113, 143] [24, 44, 70, 114, 135] [25, 36, 71, 115, 136] [26, 37, 63, 116, 137] [27, 47, 73, 117, 147] [28, 48, 74, 118, 148] [29, 49, 75, 119, 149] [30, 50, 76, 120, 150] [31, 51, 77, 121, 151] [32, 52, 78, 122, 152] [33, 53, 79, 123, 144] [34, 45, 80, 124, 145] [35, 46, 72, 125, 146] [36, 56, 82, 126, 156] [37, 57, 83, 127, 157] [38, 58, 84, 128, 158] [39, 59, 85, 129, 159] [40, 60, 86, 130, 160] [41, 61, 87, 131, 161] [42, 62, 88, 132, 153] [43, 54, 89, 133, 154] [44, 55, 81, 134, 155] [1, 45, 65, 135, 165] [2, 46, 66, 136, 166] [3, 47, 67, 137, 167] [4, 48, 68, 138, 168] [5, 49, 69, 139, 169] [6, 50, 70, 140, 170] [7, 51, 71, 141, 162] [8, 52, 63, 142, 163] [0, 53, 64, 143, 164] [10, 54, 74, 144, 174] [11, 55, 75, 145, 175] [12, 56, 76, 146, 176] [13, 57, 77, 147, 177] [14, 58, 78, 148, 178] [15, 59, 79, 149, 179] [16, 60, 80, 150, 171] [17, 61, 72, 151, 172] [9, 62, 73, 152, 173] [19, 63, 83, 93, 153] [20, 64, 84, 94, 154] [21, 65, 85, 95, 155] [22, 66, 86, 96, 156] [23, 67, 87, 97, 157] [24, 68, 88, 98, 158] [25, 69, 89, 90, 159] [26, 70, 81, 91, 160] [18, 71, 82, 92, 161] [2, 28, 72, 102, 162] [3, 29, 73, 103, 163] [4, 30, 74, 104, 164] [5, 31, 75, 105, 165] [6, 32, 76, 106, 166] [7, 33, 77, 107, 167] [8, 34, 78, 99, 168] [0, 35, 79, 100, 169] [1, 27, 80, 101, 170] [11, 37, 81, 111, 171] [12, 38, 82, 112, 172] [13, 39, 83, 113, 173] [14, 40, 84, 114, 174] [15, 41, 85, 115, 175] [16, 42, 86, 116, 176] [17, 43, 87, 108, 177] [9, 44, 88, 109, 178] [10, 36, 89, 110, 179]

H_Z (90 checks, sparse supports)

[0, 69, 90, 143, 169] [1, 70, 91, 135, 170] [2, 71, 92, 136, 162] [3, 63, 93, 137, 163] [4, 64, 94, 138, 164] [5, 65, 95, 139, 165] [6, 66, 96, 140, 166] [7, 67, 97, 141, 167] [8, 68, 98, 142, 168] [9, 78, 99, 152, 178] [10, 79, 100, 144, 179] [11, 80, 101, 145, 171] [12, 72, 102, 146, 172] [13, 73, 103, 147, 173] [14, 74, 104, 148, 174] [15, 75, 105, 149, 175] [16, 76, 106, 150, 176] [17, 77, 107, 151, 177] [18, 87, 97, 108, 161] [19, 88, 98, 109, 153] [20, 89, 90, 110, 154] [21, 81, 91, 111, 155] [22, 82, 92, 112, 156] [23, 83, 93, 113, 157] [24, 84, 94, 114, 158] [25, 85, 95, 115, 159] [26, 86, 96, 116, 160] [6, 27, 106, 117, 170] [7, 28, 107, 118, 162] [8, 29, 99, 119, 163] [0, 30, 100, 120, 164] [1, 31, 101, 121, 165] [2, 32, 102, 122, 166] [3, 33, 103, 123, 167] [4, 34, 104, 124, 168] [5, 35, 105, 125, 169] [15, 36, 115, 126, 179] [16, 37, 116, 127, 171] [17, 38, 108, 128, 172] [9, 39, 109, 129, 173] [10, 40, 110, 130, 174] [11, 41, 111, 131, 175] [12, 42, 112, 132, 176] [13, 43, 113, 133, 177] [14, 44, 114, 134, 178] [24, 45, 98, 124, 135] [25, 46, 90, 125, 136] [26, 47, 91, 117, 137] [18, 48, 92, 118, 138] [19, 49, 93, 119, 139] [20, 50, 94, 120, 140] [21, 51, 95, 121, 141] [22, 52, 96, 122, 142] [23, 53, 97, 123, 143] [33, 54, 107, 133, 144] [34, 55, 99, 134, 145] [35, 56, 100, 126, 146] [27, 57, 101, 127, 147] [28, 58, 102, 128, 148] [29, 59, 103, 129, 149] [30, 60, 104, 130, 150] [31, 61, 105, 131, 151] [32, 62, 106, 132, 152] [42, 63, 116, 142, 153] [43, 64, 108, 143, 154] [44, 65, 109, 135, 155] [36, 66, 110, 136, 156] [37, 67, 111, 137, 157] [38, 68, 112, 138, 158] [39, 69, 113, 139, 159] [40, 70, 114, 140, 160] [41, 71, 115, 141, 161] [51, 72, 125, 151, 162] [52, 73, 117, 152, 163] [53, 74, 118, 144, 164] [45, 75, 119, 145, 165] [46, 76, 120, 146, 166] [47, 77, 121, 147, 167] [48, 78, 122, 148, 168] [49, 79, 123, 149, 169] [50, 80, 124, 150, 170] [60, 81, 134, 160, 171] [61, 82, 126, 161, 172] [62, 83, 127, 153, 173] [54, 84, 128, 154, 174] [55, 85, 129, 155, 175] [56, 86, 130, 156, 176] [57, 87, 131, 157, 177] [58, 88, 132, 158, 178] [59, 89, 133, 159, 179]

Code ID 180-4-14 · download JSON · raw on GitHub