Provides the first unified theoretical formalism for hierarchical memory systems used by long-context language agents.
March 24, 2026
Original Paper
Toward a Theory of Hierarchical Memory for Language Agents
arXiv · 2603.21564
The Takeaway
It decomposes dozens of disparate RAG and agent memory implementations into three formal operators (Extraction, Coarsening, Traversal). This allows researchers to rigorously compare memory architectures and identify self-sufficiency constraints in retrieval strategies.
From the abstract
Many recent long-context and agentic systems address context-length limitations by adding hierarchical memory: they extract atomic units from raw data, build multi-level representatives by grouping and compression, and traverse this structure to retrieve content under a token budget. Despite recurring implementations, there is no shared formalism for comparing design choices. We propose a unifying theory in terms of three operators. Extraction ($\alpha$) maps raw data to atomic information units