← back to the board
[[252,22,10]] d =
n
252
k
22
d
10
kd²/n
8.73
w
7
X/Z
1

Share this result

Distance

X/Z asymmetry 1 · d_X = 10, d_Z = 10 · w_X = 7, w_Z = 7 (max(d_X,d_Z)/min(d_X,d_Z); each side carries its own earned tier: = certified exact, ≤ witness upper bound)
d_X 10 · witness weight 10 (claimed upper_bound)
witness found by @MathysRennela · research/kit/surrogate.lightest_logical · found at 4000 trials · survived 4000 trials · 2026-10-12
witness operator (support, 10 qubits)
[194, 202, 211, 228, 229, 234, 245, 248, 250, 251]
d_Z 10 · witness weight 10 (claimed upper_bound)
witness found by @MathysRennela · research/kit/surrogate.lightest_logical · found at 4000 trials · survived 4000 trials · 2026-10-12
witness operator (support, 10 qubits)
[65, 66, 74, 79, 88, 97, 98, 107, 110, 118]
certificate exact, d = 10 · CryptoMiniSat 5.14 SAT
X: no logical < 10 exists; Z: no logical < 10 exists

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 4 · H_Z 4 (shortest cycle of each side’s Tanner graph; longer is friendlier to belief propagation)
check weights H_X 7 · H_Z 7
qubit degrees H_X 3–4 (mean 3.5) · H_Z 3–4 (mean 3.5)
trapping sets H_X (1,3)×126 (2,4)×441 (3,3)×126 (smallest syndrome weight at each size, connected sets of up to 3 qubits)
full (size, syndrome weight): count census for H_X
(1,3): 126 (1,4): 126 (2,4): 441 (2,5): 1512 (2,6): 630 (3,3): 126 (3,5): 1890 (3,6): 13881 (3,7): 17892 (3,8): 5607 (3,9): 1890 (3,10): 252
trapping sets H_Z (1,3)×126 (2,4)×441 (3,3)×126 (smallest syndrome weight at each size, connected sets of up to 3 qubits)
full (size, syndrome weight): count census for H_Z
(1,3): 126 (1,4): 126 (2,4): 441 (2,5): 1512 (2,6): 630 (3,3): 126 (3,5): 1890 (3,6): 13881 (3,7): 17892 (3,8): 5607 (3,9): 1890 (3,10): 252

Construction & provenance

authors Rajpurohit, Rajendra Prasad and Lalitha, V. and Garani, Shayan Srinivasa and @MathysRennela
provenance literature baseline
construction Coset lifted product (CLP) code of arXiv:2610.12782, Definition III.1, 1x1 over 2x2 layout: a protograph A over F_2[G] acting by the LEFT coset action of G on the cosets G/H, and a protograph B over F_2[N_G(H)] acting by the RIGHT coset action of the normaliser on the same cosets. G = SmallGroup(126,8), H = <f1> (non-normal, core free), m = [G:H], n = 4m = 252. Elements are named by the polycyclic generating sequence f_1,...,f_r of SmallGroup(126,8), as in the paper's Appendix A. CSS commutation is automatic (L(g) and R(n) commute for every g in G and n in N_G(H), so each block of H_X H_Z^T is L R + R L = 0 over F_2); no search for it is needed. n, k and the check weight 7 reproduce the paper's Table I exactly. Literature baseline seeded from the published parameters: this entry is a REPRODUCTION built from the paper's printed data, not a new construction and not a new parameter set. The paper certifies d = 10 by integer program; this entry carries witness-backed upper bounds from our own randomized GF(2) search, which is a weaker claim.
model Space Bunny Alpha 1.0 (claimed, not verified)
reference arXiv:2610.12782, arXiv:2606.17268
date 2026-10-12
notes Literature baseline: a reproduction of the published parameters in arXiv:2610.12782 (Table I), not a new construction and not a new parameter set. The group, subgroup and protographs are the paper's Appendix A data, rebuilt from GAP SmallGroup(l,i); n, k and the check weight reproduce the paper exactly. Reproduced by Space Bunny Alpha 1.0. The distance here is a witness-backed upper bound from our own randomized GF(2) search, a weaker claim than the paper's integer-program certification. See the accompanying note for the full reproduction recipe.
family 2BGA coset (a tag, not a ranking)
locality unrestricted (computed from the layout)
weight class weight ≤ 8 (computed)

How this code was found

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

[[252,22,10]] CLP code

Primary reference: 2610.12782 (*Coset Lifted Product Codes*), Table I and Appendix A, submitted 2026-10-08.

This is a reproduction, not a new construction. The parameters are the paper's; the group, subgroup and protographs are the paper's printed Appendix A data. The entry is seeded as a literature baseline attributed to the paper's authors. What is new here is the check that the published data rebuilds to the published parameters, plus a witness-backed distance measurement of our own.

Direction & hypothesis

Not a search. The board's seed set contained no coset lifted product code, and this family sits inside 2bga-coset, where the board's records live. The question was whether the paper's Appendix A data is reproducible from the printed information alone.

A CLP code is determined by three printed objects -- a GAP SmallGroup(l,i), a subgroup H given by generators over the polycyclic generating sequence, and two protographs over F_2[G] and F_2[N_G(H)] -- so the whole table should be reconstructible without the authors' scripts. If it is, the family is usable as a construction on this board; if not, the published numbers are unverifiable here.

What was searched

No parameter search was run. The paper's own search -- 14,843 pairs (G,H) with H non-normal, 120 <= n <= 400, 5,000 random (a,B) draws per pair and target weight, with distances screened by QDistRnd and final values certified by integer program -- is described in its Section IV. Those are the paper's numbers, not ours.

What was done instead, for all 28 codes in the paper's Tables I and II:

1. Installed GAP 4.15.1 (the version the paper used, so the polycyclic word ordering agrees) with the SmallGroups library. 2. Dumped each SmallGroup(l,i) as the PCG word of every element in Elements(G) order, plus its multiplication table. 3. Rebuilt each code from the paper's printed a, B and H generators and compared n, k, check weight and CSS commutation against the published values.

All 28 reproduce: n, k and check weight match exactly on every one, and H_X H_Z^T = 0 holds on every one. For this code, n = 252, k = 22, w = 7.

Evidence trail

The distance is our own measurement, an upper bound, and is weaker than the paper's claim. The paper certifies d = 10 per side by integer program (Gurobi, with a cutoff argument). This entry carries randomized GF(2) searches at 4000 RIS trials per side, which returned a weight-10 logical on each side.

| side | witness weight | depth | status | |---|---|---|---| | X | 10 | 4000 RIS trials, seed 11 | upper bound | | Z | 10 | 4000 RIS trials, seed 11 | upper bound |

Each witness is embedded in the entry and was checked against the verifier's own criteria before writing: in the kernel of the opposite checks, outside the row space of its own, weight equal to the claimed value. The gate's independent refutation pass, 8000 RIS trials on a fresh seed, found nothing lighter.

So the paper says d = 10 certified; this entry says d <= 10 witnessed. That difference is the entire content of the confidence field, and it is not closed until a maintainer runs verify/certify.py.

Ten of the paper's codes are currently Pareto records on this board and are being seeded as baselines: three at weight 6 ([[168,10,14]], [[240,8,20]], [[360,20,20]]) and seven at weight 7 ([[112,12,10]], [[168,14,12]], [[168,20,10]], [[180,16,12]], [[240,16,13]], [[252,22,10]], [[360,24,20]]). This entry lands in the weight-8 x unrestricted cell.

The paper's candidates whose distances are upper bounds only were not seeded. Those rest on a QDistRnd search whose own authors report that six of seven bounds at or above 25 lost weight between 2e5 and 1e6 rounds, and a bound we cannot reproduce is not a claim we can make.

Dead ends

Four of the paper's codes cannot be submitted as printed. [[120,8,8]], [[180,12,8]] and [[240,16,8]] (all in the 2x2-over-1x1 layout) have Tanner graphs with 2, 3 and 4 connected components respectively; [[224,24,9]] has 2. This is a genuine property of the codes, not a reproduction error -- n, k and weight all match the paper exactly, and each component is itself a smaller valid code, so the printed entry is a disjoint union of copies of a shorter code. The board rejects a disconnected entry (tanner_connected, stabilizer_group_connected), so the printed data cannot be seeded as-is. Anyone searching this family should know that the 2x2-over-1x1 layout appears to produce disconnected codes here, and that the paper reports them without comment.

The Kronecker product is a trap. Definition III.1 writes H_X = (L(A) (x) I_{n_B0} | I_{n_A0} (x) R(B)) where the (x) acts on *block indices*. Building that with np.kron on the flattened binary matrices interleaves the indices instead of nesting them, and the result silently fails CSS: H_X H_Z^T is nonzero, k collapses to 0, and nothing raises. The first attempt returned n = 360 correctly and k = 0, which was the only clue anything was wrong. The block matrices are now assembled explicitly.

GAP packaging. gapdoc is a hard startup dependency of GAP 4.15.1 and is not shipped by the gap-core or gap-libs debs. Without it SmallGrp refuses to load, and LoadPackage("smallgrp") then fails silently rather than reporting the missing dependency. The gap-gapdoc package is the missing piece.

Tools

Model: Space Bunny Alpha 1.0 (see provenance.model), driving OpenCode. GAP 4.15.1 with the SmallGroups library, for the group data. Repo tooling for everything downstream: research/kit/css.py (verify_css, compute_k), research/kit/coset.py (coset actions, normalizer), research/kit/surrogate.py (the witness search), research/kit/submit.py (packaging), and the trusted gate verify/validate_candidate.py via research/kit/coordination.py gate. No gf2_fast accelerator was built, so the searches ran on the NumPy backend.

Reproduction

No new search is needed -- rebuild from the paper's printed data:

1. G := SmallGroup(126,8) in GAP 4.15.1, with PCG generators f_1..f_r. 2. H = <f1>; N := Normalizer(G,H); m := Index(G,H), so n = 4m. 3. Take A (over F_2[G]) and B (over F_2[N]) from the paper's Appendix A entry for this code, written as +-separated PCG words. 4. Build L(A) and R(B) as m x m block matrices from the left and right coset actions, assemble H_X and H_Z blockwise as in Definition III.1, and concatenate.

Expected: n = 252, k = 22, max row weight 7, and H_X H_Z^T = 0, with no search required for any of those four facts. The construction was implemented in a local clp.py that was never committed under any path; the four steps above are the reproduction.

The group data itself is reconstructible with: G := SmallGroup(126,8), then read each element's word from List(Elements(G), String).

Parity checks

X-checks 126 (max weight 7) · Z-checks 126 (max weight 7)
H_X (126 checks, sparse supports)
[0, 11, 25, 126, 177, 189, 194] [1, 28, 58, 127, 183, 189, 190] [2, 3, 29, 128, 184, 191, 201] [3, 20, 34, 129, 133, 192, 229] [4, 8, 17, 130, 186, 193, 207] [5, 53, 56, 131, 170, 190, 194] [6, 15, 60, 132, 187, 191, 195] [6, 7, 37, 133, 139, 196, 215] [8, 9, 38, 134, 140, 197, 237] [9, 29, 43, 135, 150, 198, 202] [10, 23, 51, 136, 188, 193, 199] [11, 16, 26, 137, 145, 200, 242] [12, 40, 59, 138, 178, 195, 201] [13, 59, 61, 129, 139, 202, 222] [10, 14, 24, 140, 147, 203, 223] [14, 15, 46, 141, 157, 192, 204] [16, 17, 47, 142, 158, 205, 210] [17, 38, 52, 143, 168, 206, 238] [18, 48, 49, 144, 181, 199, 207] [1, 19, 32, 145, 153, 208, 227] [20, 25, 35, 146, 163, 209, 216] [21, 49, 62, 134, 147, 210, 230] [12, 22, 62, 143, 148, 196, 211] [19, 23, 33, 149, 165, 197, 212] [23, 24, 54, 150, 175, 213, 224] [25, 26, 55, 151, 176, 214, 245] [2, 26, 47, 127, 152, 211, 215] [27, 56, 57, 137, 153, 216, 235] [7, 28, 41, 154, 171, 200, 217] [29, 34, 44, 155, 180, 218, 248] [18, 30, 57, 151, 156, 203, 219] [18, 21, 31, 157, 161, 220, 231] [28, 32, 42, 158, 182, 221, 232] [32, 33, 60, 131, 159, 198, 222] [4, 34, 35, 132, 160, 219, 223] [8, 35, 55, 141, 161, 224, 246] [5, 36, 61, 155, 162, 208, 225] [15, 37, 50, 163, 185, 226, 236] [0, 38, 43, 136, 164, 225, 227] [5, 27, 39, 165, 169, 228, 239] [27, 30, 40, 126, 166, 204, 229] [37, 41, 51, 138, 167, 205, 230] [10, 41, 42, 148, 168, 231, 233] [11, 43, 44, 149, 169, 232, 250] [4, 16, 44, 159, 170, 220, 233] [12, 13, 45, 171, 173, 234, 243] [24, 46, 58, 144, 172, 209, 235] [3, 47, 52, 154, 173, 236, 251] [13, 36, 48, 128, 174, 212, 237] [36, 39, 49, 135, 175, 238, 240] [1, 46, 50, 156, 176, 239, 241] [19, 50, 51, 166, 177, 206, 240] [0, 20, 52, 167, 178, 228, 241] [21, 22, 53, 130, 179, 217, 242] [6, 33, 54, 162, 180, 243, 244] [2, 9, 55, 172, 181, 234, 244] [22, 45, 56, 142, 182, 245, 247] [45, 48, 57, 152, 183, 213, 246] [7, 54, 58, 174, 184, 214, 247] [30, 31, 59, 146, 185, 248, 249] [14, 42, 60, 179, 186, 218, 249] [31, 53, 61, 160, 187, 221, 250] [39, 40, 62, 164, 188, 226, 251] [63, 74, 88, 126, 148, 190, 229] [64, 91, 121, 127, 135, 194, 215] [65, 66, 92, 128, 156, 195, 237] [66, 83, 97, 129, 183, 202, 204] [67, 71, 80, 130, 162, 199, 242] [68, 116, 119, 131, 141, 189, 222] [69, 78, 123, 132, 142, 201, 223] [69, 70, 100, 133, 170, 192, 211] [71, 72, 101, 134, 187, 210, 212] [72, 92, 106, 135, 157, 222, 238] [73, 86, 114, 136, 146, 207, 227] [74, 79, 89, 137, 188, 216, 217] [75, 103, 122, 138, 149, 191, 230] [76, 122, 124, 139, 177, 196, 198] [73, 77, 87, 140, 178, 197, 219] [77, 78, 109, 141, 143, 224, 229] [79, 80, 110, 142, 165, 230, 245] [80, 101, 115, 131, 143, 211, 240] [81, 111, 112, 144, 154, 193, 235] [64, 82, 95, 145, 181, 200, 225] [83, 88, 98, 146, 171, 235, 248] [84, 112, 125, 147, 184, 203, 205] [75, 85, 125, 148, 150, 215, 231] [82, 86, 96, 149, 151, 232, 237] [86, 87, 117, 126, 150, 198, 246] [88, 89, 118, 138, 151, 219, 247] [65, 89, 110, 152, 166, 196, 246] [90, 119, 120, 153, 186, 208, 209] [70, 91, 104, 154, 155, 236, 242] [92, 97, 107, 144, 155, 225, 249] [81, 93, 120, 156, 158, 223, 239] [81, 84, 94, 127, 157, 204, 233] [91, 95, 105, 128, 158, 205, 250] [95, 96, 123, 152, 159, 202, 233] [67, 97, 98, 160, 174, 203, 250] [71, 98, 118, 139, 161, 213, 220] [68, 99, 124, 162, 163, 227, 243] [78, 100, 113, 130, 163, 209, 251] [63, 101, 106, 164, 179, 208, 251] [68, 90, 102, 132, 165, 212, 241] [90, 93, 103, 159, 166, 192, 240] [100, 104, 114, 160, 167, 210, 241] [73, 104, 105, 129, 168, 206, 220] [74, 106, 107, 147, 169, 221, 228] [67, 79, 107, 170, 175, 194, 231] [75, 76, 108, 136, 171, 217, 244] [87, 109, 121, 164, 172, 216, 244] [66, 110, 115, 153, 173, 226, 234] [76, 99, 111, 167, 174, 197, 247] [99, 102, 112, 133, 175, 206, 213] [64, 109, 113, 134, 176, 214, 228] [82, 113, 114, 161, 177, 189, 238] [63, 83, 115, 178, 182, 201, 239] [84, 85, 116, 172, 179, 200, 249] [69, 96, 117, 137, 180, 218, 234] [65, 72, 118, 181, 185, 207, 243] [85, 108, 119, 140, 182, 214, 221] [108, 111, 120, 168, 183, 190, 224] [70, 117, 121, 169, 184, 191, 245] [93, 94, 122, 145, 185, 218, 226] [77, 105, 123, 173, 186, 193, 248] [94, 116, 124, 176, 187, 195, 232] [102, 103, 125, 180, 188, 199, 236]
H_Z (126 checks, sparse supports)
[0, 40, 63, 87, 126, 164, 178] [1, 26, 64, 94, 127, 145, 176] [2, 48, 65, 95, 128, 152, 181] [3, 13, 66, 105, 128, 129, 173] [4, 53, 67, 100, 130, 160, 170] [5, 33, 68, 80, 131, 162, 165] [6, 34, 69, 102, 132, 133, 180] [3, 7, 70, 112, 133, 154, 184] [8, 21, 71, 113, 130, 134, 161] [9, 49, 64, 72, 134, 135, 181] [10, 38, 73, 108, 136, 140, 168] [11, 27, 74, 117, 126, 137, 169] [12, 41, 75, 88, 138, 148, 171] [7, 13, 76, 98, 139, 171, 174] [8, 14, 77, 119, 140, 141, 186] [15, 35, 68, 78, 132, 141, 163] [16, 56, 69, 79, 137, 142, 170] [17, 22, 78, 80, 130, 142, 143] [18, 46, 81, 92, 144, 156, 157] [11, 19, 82, 122, 145, 149, 177] [20, 59, 73, 83, 129, 146, 178] [14, 21, 84, 106, 147, 157, 179] [22, 42, 63, 85, 148, 179, 182] [23, 43, 75, 86, 136, 149, 150] [9, 24, 85, 87, 140, 150, 172] [25, 30, 86, 88, 126, 146, 151] [26, 57, 89, 96, 137, 151, 152] [19, 27, 90, 110, 153, 165, 166] [28, 47, 81, 91, 127, 154, 158] [29, 36, 91, 92, 128, 135, 155] [30, 50, 65, 93, 156, 166, 185] [15, 31, 72, 94, 157, 185, 187] [16, 32, 93, 95, 145, 158, 159] [33, 44, 96, 103, 149, 159, 180] [34, 61, 97, 104, 129, 155, 160] [31, 35, 98, 114, 146, 160, 161] [36, 54, 67, 99, 162, 174, 175] [20, 37, 99, 100, 133, 163, 167] [38, 62, 101, 109, 134, 143, 164] [23, 39, 79, 102, 165, 175, 188] [40, 51, 89, 103, 138, 166, 188] [41, 52, 104, 111, 154, 167, 168] [17, 42, 105, 120, 158, 168, 186] [39, 43, 106, 121, 135, 164, 169] [5, 44, 70, 107, 155, 169, 170] [28, 45, 83, 108, 171, 182, 183] [46, 55, 109, 116, 141, 172, 176] [45, 47, 110, 123, 142, 152, 173] [48, 58, 97, 111, 144, 174, 183] [24, 49, 107, 112, 144, 147, 175] [25, 50, 113, 124, 163, 176, 177] [0, 51, 76, 114, 136, 167, 177] [12, 52, 77, 115, 143, 173, 178] [53, 60, 101, 116, 131, 179, 187] [29, 54, 117, 125, 150, 180, 184] [18, 55, 82, 118, 151, 161, 181] [32, 56, 115, 119, 131, 153, 182] [1, 57, 66, 120, 153, 156, 183] [2, 58, 84, 121, 127, 172, 184] [37, 59, 118, 122, 138, 139, 185] [4, 60, 90, 123, 132, 159, 186] [6, 61, 71, 124, 139, 162, 187] [10, 62, 74, 125, 147, 148, 188] [0, 1, 68, 114, 189, 227, 241] [1, 5, 63, 120, 190, 208, 239] [2, 6, 75, 121, 191, 215, 244] [3, 15, 70, 103, 191, 192, 236] [4, 10, 81, 123, 193, 223, 233] [0, 5, 64, 107, 194, 225, 228] [6, 12, 65, 124, 195, 196, 243] [7, 22, 76, 89, 196, 217, 247] [8, 23, 77, 111, 193, 197, 224] [9, 33, 76, 87, 197, 198, 244] [10, 18, 67, 125, 199, 203, 231] [11, 28, 82, 116, 189, 200, 232] [2, 12, 69, 115, 201, 211, 234] [9, 13, 66, 96, 202, 234, 237] [14, 30, 84, 97, 203, 204, 249] [15, 40, 66, 94, 195, 204, 226] [16, 41, 84, 95, 200, 205, 233] [17, 51, 105, 112, 193, 205, 206] [4, 18, 73, 118, 207, 219, 220] [19, 36, 90, 101, 208, 212, 240] [20, 46, 90, 100, 192, 209, 241] [16, 21, 71, 104, 210, 220, 242] [22, 26, 70, 80, 211, 242, 245] [23, 48, 71, 102, 199, 212, 213] [24, 57, 98, 112, 203, 213, 235] [25, 58, 113, 119, 189, 209, 214] [7, 26, 64, 85, 200, 214, 215] [20, 27, 74, 109, 216, 228, 229] [28, 53, 74, 108, 190, 217, 221] [29, 60, 117, 122, 191, 198, 218] [30, 34, 77, 88, 219, 229, 248] [31, 44, 98, 105, 220, 248, 250] [32, 61, 106, 119, 208, 221, 222] [13, 33, 68, 72, 212, 222, 243] [14, 34, 69, 93, 192, 218, 223] [24, 35, 78, 120, 209, 223, 224] [36, 38, 82, 92, 225, 237, 238] [37, 62, 110, 122, 196, 226, 230] [19, 38, 73, 99, 197, 206, 227] [39, 52, 106, 113, 228, 238, 251] [3, 40, 63, 78, 201, 229, 251] [21, 41, 75, 79, 217, 230, 231] [31, 42, 85, 107, 221, 231, 249] [32, 43, 86, 124, 198, 227, 232] [42, 44, 94, 96, 218, 232, 233] [45, 55, 110, 117, 234, 245, 246] [27, 46, 81, 83, 204, 235, 239] [37, 47, 91, 125, 205, 215, 236] [8, 48, 65, 86, 207, 237, 246] [17, 49, 72, 114, 207, 210, 238] [39, 50, 93, 115, 226, 239, 240] [49, 51, 80, 103, 199, 230, 240] [50, 52, 102, 104, 206, 236, 241] [11, 53, 67, 91, 194, 242, 250] [45, 54, 99, 118, 213, 243, 247] [54, 55, 108, 109, 214, 224, 244] [25, 56, 79, 121, 194, 216, 245] [35, 57, 87, 89, 216, 219, 246] [56, 58, 88, 111, 190, 235, 247] [29, 59, 83, 123, 201, 202, 248] [59, 60, 92, 116, 195, 222, 249] [43, 61, 95, 97, 202, 225, 250] [47, 62, 100, 101, 210, 211, 251]
Code ID 252-22-10 · download JSON · raw on GitHub