Study the dimension–distance tradeoff obtained by selecting stabilizer subgroups and changing sparse presentations.
Two explicitly recorded coefficient hyperplanes in the known Okada–Kasai pair-partition CPM84/34/6 check rowspaces. The parent exponent arrays and literal lift convention are preserved.
The displayed presentation has maximum check weight 18, maximum qubit degree 8, 48 checks and total incidence 600.
The JSON contains weight-6 logical witnesses. Distance confidence: X: upper_bound, Z: upper_bound.
Recorded workflow configuration: GPT-6 Astra + GPT-6.1 Sol. Official tooling: verify/qldpc_verify.py.
Read codes/84-36-6.json. Allocate zero binary matrices H_X and H_Z with 84 columns. For each row list in checks.X or checks.Z, set the listed zero-based columns to 1, preserving row order. This reconstructs the displayed presentation exactly.
Parent-basis parity functionals: {"X":12312,"Z":8970}.
Selected sector coefficient masks: {"X":{"functional":12312,"selected_coefficients":[1,2,4,32,64,128,256,512,1024,2048,16384,32768,65536,131072,262144,524288,1048576,2097152,4194304,8388608,16777216,33030271,8208,524160]},"Z":{"functional":8970,"selected_coefficients":[1,4,16,32,64,128,1024,2048,4096,8191,16384,32768,65536,131072,262144,524288,1048576,2097152,4194304,8388608,16777216,33030271,514,8200]}}.
CPM lift parameters: {"p":7,"J":4,"L":12,"EX":[[0,4,0,4,5,6,2,4,5,0,4,6],[5,1,6,3,5,1,3,1,6,3,0,3],[3,0,5,0,2,6,3,1,6,6,3,6],[4,3,2,4,1,4,2,1,6,1,2,1]],"EZ":[[4,0,1,6,5,5,6,0,6,2,4,5],[1,6,0,4,2,0,2,5,0,4,5,6],[1,3,5,0,6,1,5,5,6,6,6,4],[4,4,5,6,5,2,2,2,2,3,6,6]]}.
Public source: 84-34-6, contributed by Koki Okada and Kenta Kasai.
Construction references:
Local validation: the official verify/validate_candidate.py passed, including independent-seed distance refutation.