Welcome to the resource topic for 2024/1592
Title:
DART: Distributed argument of knowledge for rough terrains
Authors: Steve Thakur
Abstract:We describe a fully distributed KZG-based Snark instantiable with any pairing-friendly curve with a sufficiently large scalar field. In particular, the proof system is compatible with Cocks-Pinch
or Brezing-Weng outer curves to the the widely used curves such as secp256k1, ED25519, BLS12-381 and BN254.
This allows us to retain the fully parallelizable nature and the O(1) communication complexity of Pianist ([LXZ+23]) in conjunction with circumventing the huge overhead of non-native arithmetic for
prominent use cases such as scalar multiplications and/or pairings for Bitcoin (secp256k1), Cosmos (Ed25519) and Ethereum PoS (BLS12-381) signatures.
As in [LXZ+23], we use a bivariate KZG polynomial commitment scheme, which entails a universal updatable CRS linear in the circuit size. The proof size is constant, as are the verification time -
dominated by three pairings - and the communication complexity between the Prover machines. With a 9-limb pairing-friendly outer curve to Ed25519, the proof size is 5 KB. With the same curve, the communication complexity for each worker node is 5 KB and that of the master node is 5 KB per machine.
The effective Prover time for a circuit of size T ·M on M machines is O(T · log(T)+M · log(M)). The work of each Prover machine is dominated by the MSMs of length T in the group G1 and a single sum of univariate polynomial products computed via multimodular FFTs1 of size 2T. Likewise, the work of the master node is dominated by the MSMs of length M in the group G1 and a single sum of univariate polynomial products via multimodular FFTs of size 2M.
ePrint: https://eprint.iacr.org/2024/1592
See all topics related to this paper.
Feel free to post resources that are related to this paper below.
Example resources include: implementations, explanation materials, talks, slides, links to previous discussions on other websites.
For more information, see the rules for Resource Topics .