Last Saturday, I attended SIAM GLS 2015, hosted at Grand Valley State University in Grand Rapids, MI. SIAM GLS 2015 was the first math conference I attended, and I had a really great experience. The conference spanned a variety of math modeling projects involving differential equations.
Professor J. T. Oden presented the opening keynote. Oden reminded us of the larger context of the philosophy and challenges in using models for prediction. Oden reviewed the importance of model selection, calibration, and validation. Oden then argued that Bayes’ theorem is a natural way to deal with uncertainty and demonstrated how Occum’s Razor and Bayes’ theorem could provide a comprehensive framework for model selection, calibration, and validation.
Professor Douglas Arnold presented on the “fundamental theorem of numerical analysis.” In particular, Arnold reviewed the importance of consistency and stability as prerequisites for convergence with examples drawn from his work in finite elements.
Professor Erika Camacho gave a particularly moving and inspiring presentation on her experience as a minority and through STEM, escaping poverty. Camacho emphasized the importance of the pipeline of mentors in STEM that transform students’ lives and challenged us to get to know our students and take a personal interest. Camacho gave several wonderful examples of how she helps students turn their ideas into feasible projects, getting the students hooked on math in the process.
In the breakout sessions, I enjoyed hearing from faculty, postdocs, and students about their work on various modeling projects. Professor Mahmood presented on models for geophysical mass flows resulting from volcano eruptions. These models are used to create maps detailing safe and unsafe zones as well as evacuation routes. Professor Mahmood presented on using the Ornstein-Uhlenbeck SDE to develop a computationally-cheaper model than the standard PDE models. (The OU SDE is the generalized form of the Langevin equation, which I’ve written about previously.) Other interesting presentations covered models of bacteria dynamics, vortex dipole flows, and using Markov Chain Monte Carlo algorithms to parameterize multi-physics PDE models of oil production from oil reservoirs.
On my end, I presented a poster on the BigTop Bazaar model. The poster was received with interest. I had a number of questions about the model, stochastic differential equations in general, and my plans for validating the model.
Overall, I had a really good time at the conference, enjoying learning about the work of other researchers, and found folks to be really nice and curious. I look forward to going again next year!