AI

A self-learning scientific agent for X-ray diffraction

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 08/10/2026, 12:18:24 UTC.

Briefings

A self-learning scientific agent named Gan Jiang was developed to automate powder X-ray diffraction analysis using a custom ecosystem of four engines. The agent successfully identified five-phase ancient Egyptian cosmetics, tracked lattice evolution in batteries, and compared atomic configurations in catalysts. It achieved single-phase top-1 accuracy of 96.30% on MP500 data, significantly outperforming the strongest comparator at 58.00%. The system validates revisions before reuse without retraining the underlying physical models.

Key points

• Gan Jiang is a self-learning agent built on a custom X-ray diffraction ecosystem consisting of XMatcher, XQueryer, XDecomposer, and WPEM.
• The agent achieved 96.30% single-phase top-1 accuracy on MP500 data, outperforming the strongest comparator at 58.00%.

Context

This research addresses the challenge of turning analytical experience into reusable expertise for scientific agents. The agent integrates four specialized engines to handle phase identification, multiphase decomposition, and physics-constrained modeling, allowing it to diagnose failures and validate skills without retraining the language model.

Limitations

The study relies on simulated and experimental data from the DeltaXRDbench platform. Specific composition details for the ancient Egyptian cosmetic sample are not supplied in the evidence, limiting the ability to fully replicate the exact experimental conditions.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 08/11/2026, 00:01:33 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeiarhcxzkwvaszi2ewfjlrooxx6b6kvi53a4jikh3pqop3gs4vw2aq