Webinars

Getting the most out of COMSOL simulations requires efficient use of the available simulation budget as well as validating that the simulation produces results that agree with reality. This webinar series will discuss the role design of experiments, surrogate modeling, and machine learning (ML) tools such as statistical calibration can play in calibrating COMSOL simulations to physical data, validating their accuracy, and maximizing the knowledge gained from their use. With surrogate modeling, first an ML model is trained to predict the results of a particular COMSOL simulation. This ML model is then used in place of the COMSOL simulation to run any desired analyses. The ML model’s rapid prediction of the COMSOL simulation results allows more analyses to be performed and in less time. The success of surrogate modeling requires fast training speed and high prediction accuracy. Without speed, training a model can become infeasible as the scale and complexity of the problem increases. Without high accuracy a ML model’s predictions will have too much uncertainty to be usable. This 2nd of 4 webinars, will cover how SmartUQ addresses the need for speed and accuracy with its best in class ML models. Further discussed will be how SmartUQ’s accuracy and speed advantages are augmented by a flexible approach featuring many unique ML models, specifically designed to handle cases common to engineering simulation.