guiLLeMot

Photo: Bill Dix / Audubon Photo Award

LLM and AI-assisted explorations into fitting of experimental diffraction data, coming at you from Team datalab for the 2025 LLM Hackathon for Applications in Materials Science and Chemistry.

Setup

1. Ensure you have uv installed

2. Install dependencies:

   uv sync --all-extras --dev

3. Set up your environment variables in .env:

   AI_MODEL=google-gla:gemini-2.5-flash-lite
   GEMINI_API_KEY=your_api_key_here

Usage

Run the chat application:

uv run guillemot

Functionality

Given experimental X-ray diffraction data, guillemot will set up and run TOPAS refinements using a natural-language interface. Guillemot uses multimodal image input to inspect refinement outputs, and has tools for retrieving structural information from OPTIMADE providers and outputting refinement plots to images.

Key Tools

• Local file paths automatically parsed as multimodal input (used by agent to read refinement plots)
• save_topas_inp — Create and save a TOPAS .inp refinement input file; returns the path to the generated .inp.
• run_topas_refinement — Run a TOPAS refinement using a provided .inp and diffraction data; returns paths to result files and logs.
• get_optimade_structures — Query OPTIMADE providers for crystal structures and provide them in context to the agent.
• print_structure — Produce a concise human-readable summary of a single structure for quick inspection.
• print_structures — Summarize multiple structures in a compact tabular/list form with basic metadata (ID, formula, space group, lattice, source).
• plot_refinement_results — Plot observed vs calculated pattern and residuals, optionally annotate HKL ticks, save PNG, and return the image filepath and binary content.
• get_sample and get_samples — Download sample metadata from the configured datalab to find uploaded XRD patterns.

License

This hackathon project is released under the terms of the permissive MIT License - see LICENSE file for details.