| name | studying-qldpc-codes-with-qldpc |
| description | Guides construction, inspection, and experiment planning for quantum LDPC codes using the qLDPC package, while keeping a clear boundary between code-design analysis and later executable circuit or decoder workflows. |
| version | 0.1.0 |
| author | QEC Research Skills |
| license | MIT |
| tags | ["QEC","QLDPC","qLDPC","Codes","Analysis"] |
| dependencies | ["qec-research"] |
Studying QLDPC Codes With qLDPC
What This Skill Is For
Use this skill when the main task is code construction or inspection around quantum LDPC families:
- build or inspect candidate QLDPC constructions,
- reason about parity-check structure, logical operators, or distance estimates,
- plan downstream memory or threshold experiments from a code-design starting point,
- decide whether
qLDPC is the right package or whether another toolkit fits better.
When To Use It Versus Alternatives
Use this skill when the work starts from a code family or parity-check description.
Use building-stim-circuits when the task is already about executable stabilizer circuits and detector-model artifacts.
Use decoding-with-ldpc when registered detector-model or trajectory artifacts already exist and the task is decoder execution rather than code construction.
Do not present this skill as proof that the repo already ships a qLDPC adapter. The current lane is documentation-first.
Required Artifacts And Assumptions
Expected inputs:
- a code family or construction idea,
- target parameters such as block length, rate, distance regime, or field size,
- the intended downstream question such as structural analysis, decoder benchmarking, or manuscript support.
Expected outputs:
- a code-construction brief,
- a list of package-relevant capabilities and caveats,
- explicit next artifacts needed before any executable benchmarking claim.
Assumptions to make explicit:
- whether the task is binary CSS, non-binary, subsystem, or more general stabilizer work,
- whether the output needs parity-check matrices, logical operators, distance estimates, or circuit hooks,
- whether the environment can support
qLDPC and its optional dependencies,
- whether the request is really better served by
PyMatching, PanQEC, qecsim, tesseract-decoder, or another toolkit.
Standard Workflow Checklist
- Read
research-state.yaml to confirm the current repo phase and toolkit assumptions.
- Read
references/qldpc-package-fit.md before choosing qLDPC.
- State the target code family, parameter regime, and downstream research question.
- Separate code-design outputs from later execution outputs so the plan does not blur construction with benchmarking.
- Record any environment caveat, especially Python-version or optional-tooling requirements, before recommending implementation work.
- If the request actually centers on graphlike detector-model decoding, route to a decoder-oriented package instead of forcing
qLDPC.
Validation Loop
Before closing work:
- confirm the code family and parameter regime are explicit,
- confirm
qLDPC is being used for a construction or analysis task it actually fits,
- confirm any downstream circuit or decoder step is labeled as planned rather than already integrated,
- confirm environment caveats are visible if they affect installability or reproducibility.
Common Issues And Fixes
Construction-execution blur
Problem: a plan jumps from parity checks to executable benchmark claims without defining the intermediate artifacts.
Fix: add the missing artifact handoff explicitly, such as code summary, circuit synthesis plan, or decoder task specification.
Package mismatch
Problem: qLDPC gets chosen for a task that is really graph matching, threshold sweeping, or topological-code simulation.
Fix: reroute using the package-selection notes in references/qldpc-package-fit.md.
Hidden environment constraints
Problem: the repo starts to imply qLDPC is available in the current baseline even when the environment does not satisfy its packaging requirements.
Fix: keep the lane documentation-first until an adapter and health-check story exist.
References
