I am so happy to share some highlights from the WCCM 2024 conference, which was an incredible experience filled with groundbreaking research and inspiring discussions.
I had the privilege of assisting Dr. Gianmarco Mengaldo in his short course titled "Scientific Machine Learning: Application to Computational Mechanics." Dr. Mengaldo explored the various applications of interpretable AI in engineering and weather, providing invaluable insights.
Picture 1:Â Dr. Mengaldo delivering his short course, with my assistance.
I was particularly delighted to attend several impactful plenaries, including: Mathematical and Computational Foundations for Predictive Digital Twin by Dr.Karen Willcox.
Buckling Phenomena: From Computational Aspects to the Design of Aerospace Composite Structures by Dr. Chiara Bisagni.
Picture 2-4: Dr Wilcox delivering her plenary session.
Attending the mini-symposiums was a highlight of the conference. Here are a few standout talks:
Convolutional Variational Physics-Informed Neural Networks to Solve a Finite Element Formulation by Mohammed Abda from Polytechnique Montreal. Mohammad proposed an extension to Physics Informed Neural Networks (PINN) to approximate weak-form residuals of time-dependent PDEs using convolution operations.
Picture 5: Mohammad Abda delivering his session.
Coupling Variational Data Assimilation and Operator Learning for Effective State Estimation in Complex Systems by Stiven Briand MASSALA. Stiven introduced a hybrid twin approach combining variational data assimilation (PBDW) with operator learning (DeepONets) to reconstruct the state of an imperfect physical system.
Picture 6: Stiven Briand Masssala delivering his session.
Deep Generative Modeling for Data-Driven Identification of Noisy, Non-Stationary Dynamical Systems by Dr Doris Voina from the University of Washington. Doris combined a framework for data-driven discovery of stochastic differential equations with a neural network module to identify switches in parametric time dependencies, resulting in the novel Dynamic HyperSINDy method.