Brakerski et al. propose a new method for mergeable SNARGs in their paper, supporting unbounded recursive merges with security reduction time scaling only with depth, and apply it to construct the first CCA1-secure multi-hop FHE scheme. Brakerski等人在论文中提出了一种可合并SNARG的新方法,支持无限递归合并且安全归约时间仅与合并深度相关,并应用于构建首个CCA1安全的多跳全同态加密方案。
Notes
First SNARG method supporting unbounded polynomial recursive merges with arbitrary topology
Security reduction time scales only with merge depth, not tree size (potentially exponential)
Method applicable to trapdoor languages; trapdoor used only in security reduction, not proof generation/verification
Constructions from sub-exponential iO yield fully compact proofs; from LWE yield proofs scaling with depth
Application: first CCA1-secure multi-hop fully homomorphic encryption scheme
Application: first adaptive multi-hop aggregate signature scheme from LWE
提出首个支持无限多项式递归合并的SNARG方法,合并拓扑可任意
安全归约时间仅与合并深度相关,不随树大小指数增长
方法适用于陷门语言,陷门仅用于安全归约,不影响证明生成或验证
基于亚指数iO构建完全紧凑证明,基于LWE构建证明大小与深度相关
应用:构建首个CCA1安全的多跳全同态加密方案
应用:构建首个基于LWE的自适应多跳聚合签名方案
零知识证明zkDaily
Q&A Deep Dive 💬今日要点 深入解析 💬
Sat星期六
04.18
2026
Why do we need proof merging? 为什么需要“合并证明”?
In recursive proofs or complex computations, multiple proofs can be compressed into one, reducing verification cost and improving scalability. 在递归证明或复杂计算中,多个证明可以被压缩为一个,从而减少验证成本并提升系统可扩展性。
What are trapdoor languages? 什么是 trapdoor languages?
Trapdoor languages are languages where validity can be efficiently decided given a hidden trapdoor. This trapdoor is used only in the security proof, not in proving or verification. 指存在隐藏 trapdoor 时可高效判定的语言。该 trapdoor 仅用于安全性证明,不参与 SNARG 的生成或验证。
What assumptions are used for constructions? 论文基于哪些假设构造方案?
Using sub-exponential iO yields fully compact SNARGs, while LWE-based constructions have proof size growing with merge depth. 基于 sub-exponential iO 可获得 fully compact SNARG,而基于 LWE 的构造中 proof size 随 merge 深度增长。