# TRACE: Transition-Aware Residual Control for Multi-Objective Materials Discovery

arXiv: https://arxiv.org/abs/2608.23631 · HTML: https://mlsanity.com/paper/2608.23631/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Medical and scientific domain agents, one of 16 directions in LLM agents and tool use.
- Builds on 19 of its 38 references that are in the map.
- First posted on arXiv 23 Aug 2026.

## Placed in

- LLM agents and tool use / [Medical and scientific domain agents](https://mlsanity.com/agents/medical-and-scientific-domain-agents/)

## Builds on (most cited first, 19 of 19)

- [Deep Reinforcement Learning in a Handful of Trials using Probabilistic Dynamics Models](https://mlsanity.com/paper/1805.12114/) (2018)
- [When to Trust Your Model: Model-Based Policy Optimization](https://mlsanity.com/paper/1906.08253/) (2019)
- [ChemCrow: Augmenting large-language models with chemistry tools](https://mlsanity.com/paper/2304.05376/) (2023)
- [Crystal Diffusion Variational Autoencoder for Periodic Material Generation](https://mlsanity.com/paper/2110.06197/) (2021)
- [Differentiable Expected Hypervolume Improvement for Parallel Multi-Objective Bayesian Optimization](https://mlsanity.com/paper/2006.05078/) (2020)
- [Symmetry-adapted generation of 3d point sets for the targeted discovery of molecules](https://mlsanity.com/paper/1906.00957/) (2019)
- [Sample Efficiency Matters: A Benchmark for Practical Molecular Optimization](https://mlsanity.com/paper/2206.12411/) (2022)
- [Crystal Structure Prediction by Joint Equivariant Diffusion](https://mlsanity.com/paper/2309.04475/) (2023)
- [MARS: Markov Molecular Sampling for Multi-objective Drug Discovery](https://mlsanity.com/paper/2103.10432/) (2021)
- [Sample-Efficient Optimization in the Latent Space of Deep Generative Models via Weighted Retraining](https://mlsanity.com/paper/2006.09191/) (2020)
- [Learning Multimodal Graph-to-Graph Translation for Molecular Optimization](https://mlsanity.com/paper/1812.01070/) (2018)
- [Efficient Evolutionary Search Over Chemical Space with Large Language Models](https://mlsanity.com/paper/2406.16976/) (2024)
- [LLMatDesign: Autonomous Materials Discovery with Large Language Models](https://mlsanity.com/paper/2406.13163/) (2024)
- [A Deep Generative Model for Molecule Optimization via One Fragment Modification](https://mlsanity.com/paper/2012.04231/) (2020)
- [ChemReasoner: Heuristic Search over a Large Language Model's Knowledge Space using Quantum-Chemical Feedback](https://mlsanity.com/paper/2402.10980/) (2024)
- [Large Language Models for Intent-Driven Session Recommendations](https://mlsanity.com/paper/2312.07552/) (2023)
- [Bayesian Federated Learning: A Survey](https://mlsanity.com/paper/2304.13267/) (2023)
- [LLEMA: Evolutionary Search with LLMs for Multi-Objective Materials Discovery](https://mlsanity.com/paper/2510.22503/) (2025)
- [Robust Machine Unlearning for Quantized Neural Networks via Adaptive Gradient Reweighting with Similar Labels](https://mlsanity.com/paper/2503.13917/) (2025)

## Abstract (arXiv)

Multi-objective materials discovery with LLM agents is often limited not only by how many candidates can be proposed, but by how effectively each costly property evaluation informs the next search step. Existing agents mainly store evaluated candidates and their scores, so they know which materials succeeded but not which executable edits caused useful property changes. This makes local refinement difficult when objectives compete and an edit that improves one property may damage another. We propose TRACE, a transition-aware residual control framework that treats evaluated edits as the basic unit of feedback. TRACE records each local refinement as a parent-edit-child transition with observed property deltas, aggregates transition evidence to estimate reusable edit effects, and ranks future edits by their predicted ability to reduce the current candidate's remaining constraint violations while avoiding damage to already satisfied objectives. In a controlled same-backbone comparison, TRACE improves over LLEMA, the state-of-the-art LLM-agent baseline, raising macro-average hit rate from 18.13% to 25.96%.