# Chain of Unit-Physics: A Primitive-Centric Approach to Scientific Code Synthesis

arXiv: https://arxiv.org/abs/2512.01010 · HTML: https://mlsanity.com/paper/2512.01010/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Web, search, and research agents, one of 16 directions in LLM agents and tool use.
- Builds on 18 of its 42 references that are in the map.
- First posted on arXiv 30 Nov 2025.

## Placed in

- LLM agents and tool use / [Web, search, and research agents](https://mlsanity.com/agents/web-search-and-research-agents/)

## Builds on (most cited first, 18 of 18)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [gpt-oss-120b & gpt-oss-20b Model Card](https://mlsanity.com/paper/2508.10925/) (2025)
- [The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery](https://mlsanity.com/paper/2408.06292/) (2024)
- [Chain-of-Thought Reasoning Without Prompting](https://mlsanity.com/paper/2402.10200/) (2024)
- [CacheGen: KV Cache Compression and Streaming for Fast Large Language Model Serving](https://mlsanity.com/paper/2310.07240/) (2023)
- [An Empirical Evaluation of Using Large Language Models for Automated Unit Test Generation](https://mlsanity.com/paper/2302.06527/) (2023)
- [ChatUniTest: A Framework for LLM-Based Test Generation](https://mlsanity.com/paper/2305.04764/) (2023)
- [Automated Unit Test Improvement using Large Language Models at Meta](https://mlsanity.com/paper/2402.09171/) (2024)
- [Test-Driven Development for Code Generation](https://mlsanity.com/paper/2402.13521/) (2024)
- [Learning to Generate Unit Tests for Automated Debugging](https://mlsanity.com/paper/2502.01619/) (2025)
- [BatchPrompt: Accomplish more with less](https://mlsanity.com/paper/2309.00384/) (2023)
- [Rocks Coding, Not Development--A Human-Centric, Experimental Evaluation of LLM-Supported SE Tasks](https://mlsanity.com/paper/2402.05650/) (2024)
- [A Reliable Knowledge Processing Framework for Combustion Science using Foundation Models](https://mlsanity.com/paper/2401.00544/) (2023)
- [AMReX: Block-Structured Adaptive Mesh Refinement for Multiphysics Applications](https://mlsanity.com/paper/2009.12009/) (2020)
- [Steering Conceptual Bias via Transformer Latent-Subspace Activation](https://mlsanity.com/paper/2506.18887/) (2025)

## Abstract (arXiv)

Agentic large language models are proposed as autonomous code generators for scientific computing, yet their reliability in high-stakes problems remains unclear. Developing computational scientific software from natural-language queries remains challenging broadly due to (a) sparse representation of domain codes during training and (b) the limited feasibility of RLHF with a small expert community. To address these limitations, this work conceptualizes an inverse approach to code design, embodied in the Chain of Unit-Physics framework: a first-principles (or primitives)-centric, multi-agent system in which human expert knowledge is encoded as unit-physics tests that explicitly constrain code generation. The framework is evaluated on a nontrivial combustion task, used here as a representative benchmark for scientific problem with realistic physical constraints. Closed-weight systems and code-focused agentic variants fail to produce correct end-to-end solvers, despite tool and web access, exhibiting four recurrent error classes: interface (syntax/API) hallucinations, overconfident assumptions, numerical/physical incoherence, and configuration fragility. Open-weight models with chain-of-thought (CoT) decoding reduce interface errors but still yield incorrect solutions. On the benchmark task, the proposed framework converges within 5-6 iterations, matches the human-expert implementation (mean error of 3.1×10-3 %), with a ~33.4 % faster runtime and a ~30 % efficient memory usage at a cost comparable to mid-sized commercial APIs, yielding a practical template for physics-grounded scientific code generation. As datasets and models evolve, zero-shot code accuracy will improve; however, the Chain of Unit-Physics framework goes further by embedding first-principles analysis that is foundational to scientific codes.