Code and Data: Design and Realization of High Performance Textured Lead-Free Piezoelectric Ceramics through Human-AI Collaboration

Published February 8, 2026 | Version 1.0.0

Software Open

Code and data for the paper "Design and Realization of High Performance Textured Lead-Free Piezoelectric Ceramics through Human-AI Collaboration".

This repository contains:

LLM-based composition generation scripts using Claude 3.5 Sonnet for designing new KNN-based lead-free piezoelectric ceramics
Post-hoc statistical analysis code (Leave-One-Out, Conditional Prediction, Incremental Learning) using DeepSeek-V3 via Together AI
Prediction benchmarks across multiple LLMs (Claude, GPT, Gemini, DeepSeek)
All pre-generated data and visualization scripts for reproducing figures in the paper and supplementary information
A Google Colab notebook for one-click reproduction of all plots (no API keys required)

Files

Name	Size	Download all
llm-piezoelectric-ceramics-human-ai-collaboration-main.zip md5:13d546eaec095a72c221b6ce01a97e00	6.0 MB	Preview Download

Office of Naval Research
N00014-22-1-2691
DEVCOM Army Research Laboratory
W911NF-23-2-0229
National Research Foundation of Korea
RS-2024-00448865
Office of Small Business Innovation Research and Small Business Technology Transfer Programs
RS-2025-25465526

Repository URL: https://github.com/bhattadeb34/llm-piezoelectric-ceramics-human-ai-collaboration
Programming language: Python
Development Status: Active