GithubHelp home page GithubHelp logo

jingzhountu / superdiff Goto Github PK

View Code? Open in Web Editor NEW

This project forked from sdkyuanpanda/superdiff

0.0 0.0 0.0 84.7 MB

Implementation of SuperDiff: Diffusion Models for Conditional Generation of Hypothetical New Families of Superconductors

Home Page: https://arxiv.org/abs/2402.00198

License: MIT License

Python 0.44% TeX 2.64% Jupyter Notebook 96.92%

superdiff's Introduction

SuperDiff: Diffusion Models for Conditional Generation of Hypothetical New Families of Superconductors

Authors: Samuel Yuan and S.V. Dordevic

DOI

This is the implementation of SuperDiff, a state-of-the-art method for computationally generating new hypothetical superconductors. SuperDiff is a new method for generating hypothetical superconductors using Diffusion Models and is the first computational superconductor generation method to generate hypothetical new families of superconductors and also the first to have support for conditioning on reference compounds. This repository contains the code, instructions, and model weights necessary to train or directly run a version of SuperDiff. SuperDiff was created by Samuel Yuan and Sasa Dordevic, you may reach us at [email protected] and [email protected].

For ease of replication, pre-trained UNet(s) used for SuperDiff are available in SuperDiff/checkpoints, and outputs contain example raw output data from one experiment with conditional SuperDiff and the four versions of unconditional SuperDiff. Additional output data that support the results of the study are available upon reasonable request to the corresponding author at [email protected].

The folder SuperDiff contains notebooks for some of the experiments we conducted. There, diffusion1d-v3-[VERSION]-SAMPLE.ipynb are the unconditional SuperDiff versions (name corresponds to class—cuprates, pnictides, etc.) and diffusion1d_v4_ilvr_YBa1.4Sr0.6Cu3O6Se0.51.ipynb is conditional SuperDiff trained on cuprates conditioned on YBa1.4Sr0.6Cu3O6Se0.51. Code for the other versions of conditional SuperDiff conditioned on various other compounds (including all conditional SuperDiff results presented in the table of generated new families) is available upon reasonable request to the corresponding author at [email protected].

If you find this work useful, please cite it as

@misc{yuan2024superdiff,
      title={SuperDiff: Diffusion Models for Conditional Generation of Hypothetical New Families of Superconductors}, 
      author={Samuel Yuan and S. V. Dordevic},
      year={2024},
      eprint={2402.00198},
      archivePrefix={arXiv},
      primaryClass={cond-mat.supr-con}
}

superdiff's People

Contributors

sdkyuanpanda avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.