# Mimir: Physics-Grounded LLM Agents for Long-Horizon Irrigation Control

arXiv: https://arxiv.org/abs/2610.02038 · HTML: https://mlsanity.com/paper/2610.02038/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning.
- Builds on 21 of its 37 references that are in the map.
- First posted on arXiv 1 Oct 2026.

## Placed in

- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)
- LLM agents and tool use / [General LLM agent frameworks and benchmarks](https://mlsanity.com/agents/general-llm-agent-frameworks-and-benchmarks/)

## Builds on (most cited first, 21 of 21)

- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [Do As I Can, Not As I Say: Grounding Language in Robotic Affordances](https://mlsanity.com/paper/2204.01691/) (2022)
- [Voyager: An Open-Ended Embodied Agent with Large Language Models](https://mlsanity.com/paper/2305.16291/) (2023)
- [PaLM-E: An Embodied Multimodal Language Model](https://mlsanity.com/paper/2303.03378/) (2023)
- [Improving Factuality and Reasoning in Language Models through Multiagent Debate](https://mlsanity.com/paper/2305.14325/) (2023)
- [Inner Monologue: Embodied Reasoning through Planning with Language Models](https://mlsanity.com/paper/2207.05608/) (2022)
- [VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Models](https://mlsanity.com/paper/2307.05973/) (2023)
- [Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models](https://mlsanity.com/paper/2510.04618/) (2025)
- [Control Barrier Functions: Theory and Applications](https://mlsanity.com/paper/1903.11199/) (2019)
- [Safe Reinforcement Learning in Constrained Markov Decision Processes](https://mlsanity.com/paper/2008.06626/) (2020)
- [Online Continual Learning For Interactive Instruction Following Agents](https://mlsanity.com/paper/2403.07548/) (2024)
- [Self-Improvements in Modern Agentic Systems: A Survey](https://mlsanity.com/paper/2607.13104/) (2026)
- [The Black-Box Simplex Architecture for Runtime Assurance of Autonomous CPS](https://mlsanity.com/paper/2102.12981/) (2021)
- [Optimizing Irrigation Efficiency using Deep Reinforcement Learning in the Field](https://mlsanity.com/paper/2304.01435/) (2023)
- [Grounding Generative Planners in Verifiable Logic: A Hybrid Architecture for Trustworthy Embodied AI](https://mlsanity.com/paper/2602.08373/) (2026)
- [Hydra: Accurate Multi-Modal Leaf Wetness Sensing with mm-Wave and Camera Fusion](https://mlsanity.com/paper/2508.02409/) (2025)
- [Hydra-Bench: A Benchmark for Multi-Modal Leaf Wetness Sensing](https://mlsanity.com/paper/2507.22685/) (2025)
- [Self-evolving Embodied AI](https://mlsanity.com/paper/2602.04411/) (2026)
- [AIoT-based Continuous, Contextualized, and Explainable Driving Assessment for Older Adults](https://mlsanity.com/paper/2603.00691/) (2026)
- [Toward Self-Adaptive Physical AI: Can LLM Agents Manage Long-Horizon Physical Tasks?](https://mlsanity.com/paper/2609.13436/) (2026)

## Abstract (arXiv)

Large language model (LLM) agents increasingly combine reasoning, tool use, and action, but most evidence comes from episodic tasks with relatively immediate feedback and reset failures. Long-running physical control operates in a different regime: actions alter future states, errors compound across decisions, and an agent must improve from experience without being allowed to rewrite the physical rules that make execution safe. We study this regime through irrigation, where daily decisions interact with soil-water dynamics over entire growing seasons. We present Mimir, a physics-grounded LLM agent organized around two repair timescales. At the fast timescale, a structured physical interface and deterministic simulator turn an LLM output into a proposal that we numerically check, revise, and subject to bounded deterministic action selection before execution. At the slow timescale, recurrent failure patterns are consolidated into persistent contextual principles that condition future proposals, while the physical model, evaluator, and execution constraints remain immutable. Under a common retrospective evaluator across multiple sites, crops, and years, Mimir attains the lowest reported aggregate control cost among the evaluated references and uses about 51% less irrigation than the historical schedule replay. The ablation study show higher control cost when forward simulation, verified revision, or persistent context is removed; model-scale and model-family studies show no monotonic gain from increasing LLM size. The resulting lesson show that persistent physical agents can combine semantic reasoning with bounded, evidence-driven self-improvement while reserving physical truth and actuator authority for explicit numerical mechanisms.