mirror of
https://github.com/zoo-apps/papers.git
synced 2026-11-02 02:29:15 +00:00
65 lines
2.7 KiB
TeX
65 lines
2.7 KiB
TeX
\section{The PQZ Algorithm}
|
|
|
|
|
|
Post-Quantum Zero-Knowledge (PQZ) combines three primitives into a unified framework:
|
|
|
|
\subsection{Poseidon2 Commitments}
|
|
|
|
We replace Pedersen commitments (which rely on discrete log hardness) with Poseidon2 \cite{grassi2023poseidon2}, a ZK-friendly algebraic hash function secure under collision resistance assumptions that do not require elliptic curve hardness.
|
|
|
|
\begin{definition}[Poseidon2 Commitment]
|
|
For message $m \in \mathbb{F}_p^n$ and randomness $r \in \mathbb{F}_p$:
|
|
$$\text{Commit}(m, r) = \text{Poseidon2}(m \| r)$$
|
|
\end{definition}
|
|
|
|
\textbf{Performance.} Benchmarks on Zoo Network validators (NVIDIA H100):
|
|
|
|
\begin{table}[h]
|
|
\centering
|
|
\begin{tabular}{lrrr}
|
|
\toprule
|
|
\textbf{Operation} & \textbf{Poseidon2} & \textbf{Pedersen} & \textbf{Speedup} \\
|
|
\midrule
|
|
Commitment & 29 ns & 92 $\mu$s & 2000$\times$ \\
|
|
Note commit & 50 ns & 99 $\mu$s & 1980$\times$ \\
|
|
Merkle path (depth 32) & 928 ns & 2.9 ms & 3125$\times$ \\
|
|
Gas cost & 800 & 6,000 & 7.5$\times$ \\
|
|
\bottomrule
|
|
\end{tabular}
|
|
\caption{Poseidon2 vs Pedersen on GPU-accelerated validators.}
|
|
\end{table}
|
|
|
|
\subsection{STARK Proofs over Goldilocks}
|
|
|
|
For the core proving system, PQZ uses STARKs over the Goldilocks field ($p = 2^{64} - 2^{32} + 1$), which provides:
|
|
|
|
\begin{itemize}
|
|
\item \textbf{Transparency}: No trusted setup. Verification relies only on hash function collision resistance.
|
|
\item \textbf{Post-quantum security}: Security reduces to hash collision resistance (Poseidon2), not discrete log or factoring.
|
|
\item \textbf{Efficient FRI}: The Goldilocks field admits $2^{32}$-th roots of unity, enabling efficient Fast Reed-Solomon Interactive Oracle Proofs.
|
|
\end{itemize}
|
|
|
|
\subsection{ML-DSA Receipt Signing}
|
|
|
|
After STARK verification, the Z-Chain validator set signs the resulting receipt using ML-DSA (FIPS 204) threshold signatures coordinated through K-Chain:
|
|
|
|
\begin{enumerate}
|
|
\item Prover submits STARK proof to Z-Chain.
|
|
\item Z-Chain validators independently verify the proof.
|
|
\item Upon $\geq 67\%$ agreement, validators produce a threshold ML-DSA signature on the receipt.
|
|
\item The signed receipt is stored on-chain and indexed by receipt hash.
|
|
\end{enumerate}
|
|
|
|
\subsection{Groth16 Export for External EVMs}
|
|
|
|
For interoperability with Ethereum and other EVM chains that only support BN254 pairings, PQZ provides a Groth16 export:
|
|
|
|
\begin{enumerate}
|
|
\item Take a verified STARK receipt from Z-Chain.
|
|
\item Generate a Groth16 proof that ``I know a STARK proof that was verified and signed by the Zoo validator set.''
|
|
\item Export the Groth16 proof to the target EVM chain.
|
|
\end{enumerate}
|
|
|
|
This wraps the post-quantum STARK in a classical Groth16 for compatibility, while the original STARK receipt remains the source of truth on Z-Chain.
|
|
|