Development of a deep learning model for predicting crack propagation in rocks

Opportunity Description:

This undergraduate research position focuses on developing data-driven models to predict crack initiation and propagation in geomaterials. The motivation stems from the need to better understand and forecast rock failure processes in energy and hazard applications such as geothermal development, hydrogen storage, and landslide prevention. Ground-truth data will be provided obtained from physical simulations. You will preprocess these simulation outputs, design neural network architectures (e.g., graph or convolutional networks), and evaluate their predictive accuracy for crack growth patterns under varying conditions. Deliverables include a trained deep learning model, visualization of predicted versus simulated crack networks. By the end of the project, you will gain hands-on experience in deep learning, which will prepare you for advanced studies or careers in industry relevant to AI-driven scientific computing.

Faculty Advisor Mentoring Philosophy:

I believe that effective mentoring begins with creating an inclusive and engaging environment where curiosity and confidence can grow together. My approach combines structured guidance and open discussion, so giving students a clear framework while encouraging independent thinking. Through hands-on experimentation, I will help you develop both technical skills and scientific intuition. I view research as a collaborative journey: I will work closely with you to identify your strengths and provide continuous feedback. I am also glad to help to the best of what I can for your future continued study in graduate school or finding a job in industry.