InvDesMobility: a reliability-gated first-principles feedback framework for closed-loop materials discovery

2026-06-15 · Source: Artificial Intelligence · Field: Science & Research — Artificial Intelligence & Machine Learning, Engineering & Applied Sciences · Depth: Expert, quick

Summary

InvDesMobility is introduced as a reliability-gated first-principles feedback framework designed for closed-loop materials discovery. This framework integrates multi-agent automated Density Functional Theory (DFT), evidence stratification, generative structure proposal, acquisition ranking, and auditable release. Its core value lies in ensuring that expensive first-principles results, particularly for composite properties like carrier mobility, are independently validated, provenance-recorded, and admitted as feedback only with sufficient evidence. From 516 2DMatPedia-derived candidates, the workflow produced 280 QC-passed materials and 573 retained carrier-direction seed channels. Over multiple iterations, InvDesMobility screened 2.4 x 10^6 structures, submitted 102 candidates for DFT validation, and retained 86 reliability-gated generated channels across 41 formulas. The primary contribution is a transferable feedback contract, making closed-loop inverse design both useful and auditable.

Key takeaway

For research scientists developing closed-loop inverse materials design workflows, InvDesMobility demonstrates a critical need for reliability gating and auditable feedback. Your systems should integrate independent validation and evidence stratification for expensive first-principles calculations, especially for complex properties like carrier mobility. This ensures that only sufficiently evidenced results update generative and acquisition models, enhancing both utility and trustworthiness in materials discovery.

Key insights

The InvDesMobility framework provides a reliable, auditable feedback mechanism for inverse materials design using expensive first-principles calculations.

Principles

• Inverse design value requires prediction performance, independent validation, provenance, and sufficient evidence.
• Composite properties demand careful consideration of intermediate quantities, fit quality, and workflow assumptions.
• A transferable feedback contract enhances closed-loop inverse design utility and auditability.

Method

InvDesMobility integrates multi-agent automated DFT, evidence stratification, generative structure proposal, acquisition ranking, and auditable release, using reliability gating to admit feedback.

In practice

• Implement reliability gating for admitting first-principles results into feedback loops.
• Stratify feedback: relaxed structures for generative models, retained channels for acquisition models.

Topics

• Inverse Materials Design
• Closed-Loop Discovery
• Density Functional Theory
• Carrier Mobility
• Reliability Gating
• Materials Informatics

Code references

• DreamLufei/invDesMobility

Best for: AI Scientist, Research Scientist

Related on AIssential

• LEAP: A closed-loop framework for perovskite precursor additive discovery — Coupling a domain-specialized LLM with active learning and expert review accelerates perovskite additive discovery under low-data conditions.
• FVD: Inference-Time Alignment of Diffusion Models via Fleming-Viot Resampling — FVD uses a Fleming-Viot birth-death mechanism to prevent diversity collapse in diffusion model inference, improving sample quality and efficiency.
• Inverse design of bespoke interatomic potentials via active learning by information-matching — The information-matching (IM) approach tailors interatomic potentials by linking training data selection to specific material property uncertainty targets.
• Physics Guided Conditional Diffusion Framework for Generative Inverse Design of Manufacturable Metasurface based Absorbers — Physics-guided diffusion models can rapidly generate manufacturable metasurface designs with high spectral accuracy and multimodal diversity.
• AutoDFT: A Closed-Loop Multi-Agent Framework for Autonomous DFT Calculations — AutoDFT automates complex DFT calculations via a closed-loop multi-agent LLM framework, adapting dynamically to emergent results.
• De novo molecular generation with optical property preconditioning at the token level — Token-conditioned autoregressive models can generate OLED molecules with targeted optical properties, but control varies by chemical motif.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.