cronokirby

(2026-01) Jindo; Practical Lattice-Based Polynomial Commitment for Zero-Knowledge Arguments

2026-01-12

Abstract

We present Jindo\textsf{Jindo}, a new lattice-based polynomial commitment scheme (PCS) readily available for compiling polynomial interactive oracle proofs (PIOP) into zero-knowledge arguments of knowledge (ZKAoK).
Jindo\textsf{Jindo} improves upon the previous lattice-based PCSs CELPC\textsf{CELPC} (CRYPTO' 24) and Greyhound\textsf{Greyhound} (CRYPTO' 24) by seamlessly integrating their strengths.
Specifically, we incorporate the coefficient encoding method of CELPC\textsf{CELPC} with a new evaluation protocol inspired by Greyhound\textsf{Greyhound}, inheriting only the advantages of both schemes. As a result, Jindo\textsf{Jindo} enjoys native support for a large base field, a large challenge set, the evaluation hiding property, and efficient batched evaluation.

Our implementation shows that Jindo\textsf{Jindo} improves CELPC\textsf{CELPC} with 1.3x faster proof generation, 3.7x faster verification, and 4.8x smaller proof size when evaluating polynomials of degree 2192^{19} over a 255-bit prime field.
Furthermore, Jindo\textsf{Jindo} outperforms CELPC\textsf{CELPC} with 3.5x faster proof generation, 9.7x faster verification, and 12.3x smaller proof size when compiling PIOP from Buckler\textsf{Buckler} (CCS' 25) to prove the validity of an RLWE sample, demonstrating its practical efficacy in ZKAoK construction from PIOP.