Crates.io | ark-mpc |
lib.rs | ark-mpc |
version | 0.1.2 |
source | src |
created_at | 2023-10-10 21:00:39.369086 |
updated_at | 2023-11-08 00:45:21.562242 |
description | Malicious-secure SPDZ style two party secure computation |
homepage | https://renegade.fi |
repository | https://github.com/renegade-fi/ark-mpc |
max_upload_size | |
id | 999475 |
size | 514,297 |
ark-mpc
ark-mpc
provides a malicious secure SPDZ style framework for two party secure computation. The circuit is constructed on the fly, by overloading arithmetic operators of MPC types, see the example below in which each of the parties shares a value and together they compute the product:
use ark_mpc::{
algebra::scalar::Scalar, beaver::SharedValueSource, network::QuicTwoPartyNet, MpcFabric,
PARTY0, PARTY1,
};
use ark_curve25519::EdwardsProjective as Curve25519Projective;
use rand::thread_rng;
type Curve = Curve25519Projective;
#[tokio::main]
async fn main() {
// Beaver source should be defined outside of the crate and rely on separate infrastructure
let beaver = BeaverSource::new();
let local_addr = "127.0.0.1:8000".parse().unwrap();
let peer_addr = "127.0.0.1:9000".parse().unwrap();
let network = QuicTwoPartyNet::new(PARTY0, local_addr, peer_addr);
// MPC circuit
let mut rng = thread_rng();
let my_val = Scalar::<Curve>::random(&mut rng);
let fabric = MpcFabric::new(network, beaver);
let a = fabric.share_scalar(my_val, PARTY0 /* sender */); // party0 value
let b = fabric.share_scalar(my_val, PARTY1 /* sender */); // party1 value
let c = a * b;
let res = c.open_authenticated().await.expect("authentication error");
println!("a * b = {res}");
}
Unit tests for isolated parts of the library are available via
cargo test --lib --all-features
The bulk of this library's testing is best done with real communication; and so most of the tests are integration tests. The integration tests can be run as
./run_integration.zsh
or more directly as
docker compose up