Research

Research overview

Quantum error-correcting codes that combine explicit structure, useful rate, large girth, and practical decoding.

The coding-theoretic challenge

Useful quantum computation requires many reliable logical qubits, yet present systems need large numbers of physical qubits to protect a much smaller logical space. Engineering constraints matter, but code design itself also faces fundamental limitations.

Quantum low-density parity-check codes are attractive because sparse checks can support scalable syndrome processing. The central challenge is to obtain several desirable properties at once: useful coding rate, large girth, strong finite-length performance, controlled error floors, and low-complexity decoding.

Research objective. Recover the design freedom of classical LDPC codes while satisfying quantum commutativity and preserving the correlations relevant to quantum noise.

Two complementary directions

Binary LDPC-based quantum codes

Explicit CSS constructions that control commutativity without forcing all of the graph structure into restrictive orthogonality patterns. The work emphasizes regular sparse matrices, large girth, and reproducible finite-length instances.

Kasai codes →

Non-binary LDPC-based quantum codes

Constructions and decoding over larger finite fields, expanded to binary CSS codes on qubits. The aim is to use algebraic structure and degeneracy-aware decoding to improve the low-error-rate regime.

Non-binary codes →

Construction families

Kasai codes

Large-girth regular CSS LDPC constructions designed around a controlled orthogonality barrier.

Details →

PP codes

Pair-partition constructions with CPM-based verification records and explicit instances.

Details →

Other QLDPC constructions

Affine-coset, finite-field two-branch, and high-rate constructions maintained separately from the two main families.

Details →

From construction to evidence

The site separates conceptual explanations from computational evidence. Construction pages describe the code families; the software and data page records reusable implementations and archives; the distance pages state what is known through benchmarks, lower bounds, and explicit upper-bound witnesses.

Decoding perspective

The decoding experiments use sum-product BP formulations that retain the local correlation between Pauli error components, together with lightweight post-processing where stated. Decoder labels and assumptions are given on the relevant paper, software, or data page.