Repository reorganization and the algorithm expansion panel¶
English · 简体中文
This round first reorganizes code and documentation, then expands a batch of algorithms that can actually generate circuits. All work happens inside the oracq repository, preserving existing uncommitted changes.
Stage |
Scope |
Status |
|---|---|---|
R1 |
Layering into infrastructure / algorithms / applications; split up elementary, solvers, differential while keeping legacy import compatibility |
Done |
R2 |
Sphinx + MyST + API documentation; manuals separated from tutorials, historical records archived |
Done |
R3 |
Oracle queries, Fourier arithmetic, search and amplitude amplification, estimation, variational algorithms, quantum walks, number theory, simple error correction |
Done |
R4 |
Mathematical witnesses for new algorithms, regression of old cases, real-backend validation, strict documentation builds, packaging and migration acceptance |
Done |
S1 |
Review-driven revision: architecture-paradigm conventions document, replaceable QSP phase synthesizer, paper text and bibliography updates |
In progress |
The S1 conventions are recorded in architecture-paradigm.
Algorithm files are named by purpose: oracle_algorithms.py, fourier.py, search.py, estimation.py, variational.py, walks.py, number_theory.py, error_correction.py, plus the existing hamiltonian.py, qlss.py, lchs.py, schrodingerization.py, cbmd.py, carleman.py, qham.py.
The first batch of new implementations covers Bernstein–Vazirani; Simon sampling with GF(2) post-processing; QFT addition; general amplitude amplification; Hadamard/Swap tests; standard amplitude estimation; MaxCut QAOA; VQE measurement circuits; parameterized ansatz; the periodic coined walk; modular multiplication for order finding and factor post-processing; and three-bit bit/phase-flip encoding and decoding. Gate-level output and each algorithm’s scope of applicability are recorded separately — the small modular-multiplication-table implementation is not treated as scalable Shor arithmetic.
The documentation states the problem, inputs, steps, and results in complete Chinese sentences. Tutorials are organized around user tasks; specifications give explicit contracts; the API is generated from source. Builds must be free of Sphinx warnings, and tutorials must run. Existing stage reports moved into archive, keeping their references without mixing them into the main reading path.