Akita PCS

Akita is a high-performance, modular lattice polynomial commitment scheme with transparent setup and post-quantum security.

Akita is the public scheme name for this implementation and the intended repository/package name is akita-pcs. The codebase is being decomposed into a focused akita-* crate family rather than remaining a single monolithic package.

The current workspace exposes the main ownership boundaries under crates/:

• akita-field, akita-serialization, and akita-algebra own foundational arithmetic, encoding, NTT, ring, and polynomial utilities.
• akita-transcript, akita-challenges, and akita-sumcheck own Fiat-Shamir transcripts, challenge sampling, and generic sumcheck machinery.
• akita-types owns shared proof, setup, schedule, layout, and commitment data shapes used by both roles.
• akita-config owns concrete runtime config presets and config-backed schedule/SIS policy.
• akita-setup owns config-backed setup construction and optional setup cache persistence.
• akita-verifier owns verifier replay without depending on prover-only polynomial backends.
• akita-prover owns commitment, proving, setup expansion, recursive witness construction, and polynomial backends.
• akita-scheme owns the end-to-end AkitaCommitmentScheme orchestration that wires config, setup, prover, and verifier crates together.
• akita-pcs is the umbrella package that re-exports the broad public surface and hosts examples, benches, and end-to-end integration tests.
• akita-planner owns offline schedule search and proof-size/security planning.

Verifier-only consumers should prefer the slim role crates directly: akita-verifier for verification, akita-types for proof/setup/claim shapes, and akita-config for concrete schedule/config policy. The umbrella akita-pcs package is convenient for examples and end-to-end use, but it also pulls in prover-facing APIs.

Lineage

Akita descends from Hachi and keeps that ancestry explicit, while giving the improved scheme its own name. This is also the line where planned protocol improvements over the original design live: faster verifier-oriented reductions via matrix-claim delegation and tensor-structured challenges, smaller large-field proofs via modulus switching and field-size lowering, and efficient zero-knowledge techniques under the Whiteout design direction.

Contributing

Major features and architectural changes should start with a short spec. See CONTRIBUTING.md and specs/TEMPLATE.md for the review workflow.

Acknowledgements

The CRT/NTT and small-prime arithmetic design in this repository is informed by the Labrador/Greyhound C implementation family. In particular, the pseudo-Mersenne profile uses moduli of the form q = 2^k - offset. Akita provides a Rust-native architecture and APIs, while drawing algorithmic inspiration from those implementations.