Open research towards the discovery of room-temperature superconductors.
Discover ways to transform this asset
POST /speech/from-post
Automated recap of the latest activity in #superconductors, created by @hermes.
Discover other posts like this one
So far a really interesting paper. Published in 2018. Adding some informal notes and interesting findings here. Finding out how much literature is based on this study.
Using what we learned when trying to use the MLFF's latent space for Tc prediction, there's a way we can simplify things for the prediction model and give it a better change of picking up on the signa
In this next experiment, I'm going to try to build more of an intuition and physical understanding of some of our most important latent features, as they relate to importance of predicting the superc
The below animation shows a selection of important features to superconductivity and how they evolve as the materials are heated up to their critical temperature. Notice how for most features, there i
Over the past week, the Superconductors team has made significant strides in both computational modeling and strategic planning towards the discovery of room-temperature superconductors. This period highlighted advancements in machine learning models, insightful feature analyses, and the establishment of a clear roadmap for future research and development.
Machine Learning Model Enhancements
Model Accuracy and Data Quality: Improvements in the superconducting state classifier model have led to better prediction accuracy and more reliable transitions from superconducting to non-superconducting states. This is primarily due to:
Increased Temperature Range and Data Points: Expanding the temperature range and increasing the number of data points around higher transition temperatures have enhanced the model's robustness.
Enhanced Simulation Parameters: Utilizing larger supercell sizes and enabling periodic boundary conditions have resulted in more accurate crystal simulations.
Over the last week or so, I've been working on making some upgrades to the superconducting state classifier model. See the first attempt here:
Feature Importance and Analysis
Latent Feature Insights: Detailed analyses using SHAP values and feature importance metrics have identified key latent features that influence superconducting probability. Notably:
Key Features: Features 63, 91, and 220 play pivotal roles in predicting superconductivity, with specific behaviors under temperature variations.
Sweet Spot Phenomenon: Superconductivity arises from a delicate balance of multiple properties, necessitating neither maximization nor minimization of individual features alone.
By looking at our superconducting state classifier model feature importance, we can understand what features we should be looking at. From there, we can start to study how these features change across
Strategic Roadmapping
Defined Milestones: A comprehensive roadmap has been established to guide the team towards achieving room-temperature superconductivity, with clear milestones set for May 2025, Fall 2025, and Early 2026. These milestones focus on platform development, material discovery, and scaling simulation capabilities.
As we move towards potential commercial viability or try and build some credibility in the space, it's important for us to set some goal posts and aim for them.The discovery a room temperature superco
Integration of Computational and Experimental Efforts
The advancements in machine learning models provide a robust foundation for guiding experimental synthesis efforts. By identifying critical latent features that influence superconductivity, the team can better target material properties during synthesis.
Strategic Importance of Roadmap Milestones
Achieving the outlined milestones will not only advance the research but also build credibility, attract investments, and foster valuable collaborations with academia and industry partners.
Challenges in Correlating Latent Features with Physical Properties
While significant progress has been made in identifying important latent features, establishing direct correlations with tangible physical properties remains a complex task. Future efforts will need to focus on bridging this gap, potentially through the integration of additional feature sets like Magpie.
Correlation Between Latent Features and Physical Attributes
Establishing clear links between latent vectors used in machine learning models and actual physical properties of materials remains challenging. This is crucial for translating computational insights into actionable synthesis strategies.
Access to Synthesis Facilities
Securing resources and access to advanced synthesis and characterization facilities is essential for experimental validation of computational predictions. Navigating institutional resources and potential collaborations will be key.
Generalization Beyond Dataset Biases
Ensuring that the model's predictions are generalizable beyond the current dataset's biases is necessary for discovering a diverse range of superconducting materials.
Enhance Feature Correlation Studies
Continue exploring correlations between latent features and physical properties, perhaps by integrating additional datasets or leveraging advanced feature engineering techniques.
Strengthen Experimental Collaborations
Actively pursue partnerships with research groups and facilities that can provide the necessary resources for material synthesis and characterization.
Monitor Roadmap Progress
Regularly assess progress against the defined milestones to ensure timely achievement of goals and adjust strategies as needed based on emerging insights.
The Superconductors team is making commendable progress towards the ambitious goal of discovering room-temperature superconductors. By refining machine learning models, conducting in-depth feature analyses, and laying out a strategic roadmap, the team is well-positioned to overcome existing challenges and drive significant advancements in the field.
Crystal structure perturbation analysis on latent space
GHOST Roadmap
Exploring material classes and their latent representations
Superconductivity as a sweet spot of properties
Starting to think about synthesis
Tc prediction model updates