[论文] Fault-Tolerant Budget Conservation in Distributed Multi-Agent Delegati...

论文概要

研究领域: ML
作者: Genliang Zhu, Chu Wang
发布时间: 2026-10-05
arXiv: 2610.00349

中文摘要

资源限额正成为 AI 智能体跨并发且易故障工作者分发工作的授权边界。父子分配约束、仿射对象和分布式托管本身，在回复丢失、效果超时后完成、消息重复、分支分区或 DAG 汇合混 alias 同一线路时均不能防超支。我们为分布式多智能体委托形式化容错预算守恒。预算是量化资源向量，由专属托管信用表示，在委托 DAG 中流动。分发前分支将信用转为绑在线路、纪元、规范化效果、最大收费、接收者和幂等键的操作预留；以隔离态持久化签名分发许可，网关首次接受前验证。不确定效果保持计费直至认证结算、带围栏权威无效果证明或永久退役。我们在显式调解、持久性、认证、规范化和网关假设下证明所有权分割、账本与效果守恒、后代非放大、至多一次结算、迟到完成安全和分区限制。不可区分性结果表明分区局部可用性需专属预分配。有界 TLA+ 检查、独立 JavaScript 探索器和崩溃注入双进程 SQLite 实验检查声明范围，检出超时退款和历史证书验证变异体。该机制在评估的崩溃、重试、重复、分区、汇合和迟到完成调度下保持已发行预算界限。

原文摘要

Resource limits are becoming an authorization boundary for AI agents that delegate work across concurrent and failure-prone workers. Parent-child allocation constraints, affine objects, and distributed escrow do not by themselves prevent overspend when replies are lost, effects complete after timeout, messages repeat, branches partition, or DAG joins alias one lineage. We formalize fault-tolerant budget conservation for distributed multi-agent delegation. Budgets are quantized resource vectors represented by exclusive escrow credits that move through a delegation DAG. Before dispatch, a branch converts credit into an operation reservation bound to lineage, epoch, normalized effect, maximum charge, receiver, and idempotency key. It persists a signed dispatch permit with quarantine; the gate...

自动采集于 2026-10-05

#论文 #arXiv #ML #小凯