UCSC Researchers Win Award for Introducing New AI Method for Minimum-effort Materials Engineering

The researchers, including Halder (second from right) and Nodozi (middle), recieve the O. Hugo Schuck Best Application Paper Award at the 2024 American Control Conference. 

Researchers at UC Santa Cruz have developed a groundbreaking method for designing materials with properties unseen in nature. Their work, which combines novel mathematical techniques and artificial intelligence, recently won the prestigious 2024 O. Hugo Schuck Best Application Paper Award from the American Automatic Control Council.

Material Design

Material engineers constantly strive to create materials with properties beyond what nature offers. This often involves manipulating the material's structure using various methods. The key challenge lies in achieving the desired properties with minimal effort, making the process cost-effective.

This research introduces a new approach called "minimum effort stochastic control" that mathematically guarantees the most efficient control algorithm for achieving a specific crystal structure. The researchers successfully applied this method to control the arrangement of colloids (suspended particles) in a solution using an electric field.

A Different Approach

The team's approach is unique because it employs a "distribution-to-distribution" control strategy. Traditional methods focus on point-to-point control, but this new approach considers the inherent randomness within the material's structure. This allows for a more precise manipulation of the material towards the desired ordered state.

Interestingly, this research has implications beyond material engineering. The process led to the creation of a new set of complex mathematical equations that had never been studied before. This advancement contributes to the field of stochastic control, a branch of control theory.

Furthermore, the researchers developed a new type of neural network – a physics-informed neural network – specifically designed to solve these novel equations. Unlike conventional neural networks, this one accounts for the noisy and nonlinear nature of the underlying physics.

This research signifies a significant step towards using AI to accelerate materials design. Traditionally a slow and trial-and-error process, AI offers promising avenues for efficient material discovery.

The researchers plan to explore further applications of their approach and delve deeper into the fundamental mathematics behind their findings. This work holds immense potential for revolutionizing material engineering and beyond.