The merge 之后，EL的职责主要 区块验证状态转移函数stf，交易驱动的状态机。 每笔交易都由客户端验证、执行，并将其结果累积到状态树中 其他EIP机制涉及到的状态量，例如 EIP-1559 基本费用、EIP-4844 blob gas、EIP-4844 信标根环缓冲区、信标链提款等。 区块构建 tx pool

dual staking model PoS网络早期可能面临“死亡螺旋”问题，即代币价值下降会削弱网络的安全性，导致总锁定价值（TVL）下降，进一步压低代币价格,进而导致死亡螺旋。 how it works 使用两种代币保护同一PoS网络，其中一种可以是波动性低、流动性深、更易获取的外部网络代币，如以太坊（ETH).同时使用网络native token捕获网络价值 different method #TODO:我们选择哪种模型

The future AI era is a huge decentralized computing network. The demand for on-chain artificial intelligence is increasing day by day. Many projects have begun to provide reliable and trustworthy neutral neural network training and execution solutions. However, a complete AI solution is a model, calculation Power and data are jointly composed. Existing algorithm innovation has become increasingly weak. The ability to master high-quality and diverse data has become a very important part of whether the model can break through the ceiling. According to KPMG’s 2023 research report, the adoption of generative artificial intelligence has huge implications at the data level Risks need to be addressed, including intellectual property issues, personal data sharing, lack of regulatory frameworks and bias in generative AI models, etc. These issues can be well solved by adopting blockchain technology Omnichan Data Network is a modular data network that strives to provide a transparent, reliable and open ecosystem of data models to serve any AI domain usage scenario, and enables revenue distribution to participants through tokenization of data models KEY HIGHLIGHTS data interoperabilitydual chain framework Consensus Layer(cosmos CometBFT consensus engine) + Execution Layer(EVM) ref:https://docs.omni.network/learn/omni/architecture data programmable: onchain data becomes a programmable dataset

问题定义 给定公钥,一组消息,一组基于消息的签名,基于Bls12_381曲线,使用pedersen hash 做哈希字符串到EC point映射, 完整数据集参考.关键函数如下: pub fn verify(pk: G2Affine, msg: &[u8], sig: G1Affine) { let (_, h) = hash_to_curve(msg); assert!(Bls12_381::product_of_pairings(&[ ( sig.into(), G2Affine::prime_subgroup_generator().neg().into() ),