DiffGA: Steering Frozen Diffusion Models for Materials Design via Evolutionary Search
Abstract
Steering candidate generation toward user-defined objectives is central to inverse materials design, yet remains difficult when objectives are black-box, changing, or non-differentiable, as in many simulation- and experiment-derived properties. Generative models offer a solution but typically require retraining or fine-tuning for each new objective, demanding labeled data and substantial compute, or rely on differentiable guidance that is brittle to gradient noise and undefined for many practical targets. We present DiffGA, a retraining-free framework that injects evolutionary search into the sampling trajectory of a retrained diffusion model, evolving partially denoised states under an arbitrary black-box oracle. Because oracle evaluation often dominates campaign cost, we couple DiffGA with an on-the-fly Bayesian surrogate that reuses representations from state-of-the-art machine-learned force fields (MLFFs). A Gaussian process is fit during the campaign on the force fields's graph embeddings and, once it demonstrably ranks an unseen generation, replaces expensive property evaluations under an uncertainty-aware acquisition rule, reducing oracle cost with negligible loss of accuracy. Across three crystal-design tasks using a frozen MatterGen backbone, DiffGA outperforms diffusion-only, compute-matched, purely evolutionary, and fine-tuned baselines in oracle-assessed rates of stable, unique, and novel hits. Rather than proposing scientific compounds, we use density functional theory to quantify MLFF oracle fidelity, showing that it is a faithful proxy in key regimes but systematically biased in others, delineating when proxy-driven generative discovery can be trusted. On a representative sample, DiffGA's top hits survive first principles re-evaluation as physically realistic hits, while the surrogate-accelerated pipeline preserves this fidelity at a fraction of the cost.
Related
- Multi-property Feynman-Kac Steering of Crystal Diffusion Models
- Reading and Steering Representations of Materials-Science Mechanisms in an Open-Weight Language Model
- Diffusion-induced instabilities promote cooperation in eco-evolutionary networks
- Diffusion-Based Generative Screening of ABX3 Perovskite Compositions for Bandgap-Targeted Materials Discovery
- Monolithic Carrier-Beam Steering with Milliradian Precision and Single-Voltage Control
BibTeX
@misc{agarwal2026diffga,
title = {{DiffGA: Steering Frozen Diffusion Models for Materials Design via Evolutionary Search}},
author = {Shivang Agarwal and Rodrigo Wang and Kevin Ryczko and Aayush R. Singh},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.14999},
eprint = {2610.14999},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.14999},
note = {CC BY-NC-SA 4.0},
}
RIS
TY - UNPB
TI - DiffGA: Steering Frozen Diffusion Models for Materials Design via Evolutionary Search
AU - Shivang Agarwal
AU - Rodrigo Wang
AU - Kevin Ryczko
AU - Aayush R. Singh
PY - 2026
DA - 2026/10/10
PB - arXiv
DO - 10.48550/arxiv.2610.14999
UR - https://arxiv.org/abs/2610.14999
AB - Steering candidate generation toward user-defined objectives is central to inverse materials design, yet remains difficult when objectives are black-box, changing, or non-differentiable, as in many simulation- and experiment-derived properties. Generative models offer a solution but typically require retraining or fine-tuning for each new objective, demanding labeled data and substantial compute, or rely on differentiable guidance that is brittle to gradient noise and undefined for many practical targets. We present DiffGA, a retraining-free framework that injects evolutionary search into the sampling trajectory of a retrained diffusion model, evolving partially denoised states under an arbitrary black-box oracle. Because oracle evaluation often dominates campaign cost, we couple DiffGA with an on-the-fly Bayesian surrogate that reuses representations from state-of-the-art machine-learned force fields (MLFFs). A Gaussian process is fit during the campaign on the force fields's graph embeddings and, once it demonstrably ranks an unseen generation, replaces expensive property evaluations under an uncertainty-aware acquisition rule, reducing oracle cost with negligible loss of accuracy. Across three crystal-design tasks using a frozen MatterGen backbone, DiffGA outperforms diffusion-only, compute-matched, purely evolutionary, and fine-tuned baselines in oracle-assessed rates of stable, unique, and novel hits. Rather than proposing scientific compounds, we use density functional theory to quantify MLFF oracle fidelity, showing that it is a faithful proxy in key regimes but systematically biased in others, delineating when proxy-driven generative discovery can be trusted. On a representative sample, DiffGA's top hits survive first principles re-evaluation as physically realistic hits, while the surrogate-accelerated pipeline preserves this fidelity at a fraction of the cost.
ER -