Editor’s note: This article was initially published in The Daily Gazette, Swarthmore’s online, daily newspaper founded in Fall 1996. As of Fall 2018, the DG has merged with The Phoenix. See the about page to read more about the DG.
This week, the Math and Statistics Department presented its first colloquia of this year’s lecture series: Markov Chain Monte Carlo: Taking Advantage of Chance. Swarthmore stat professor Steve Wang (pronounced Wong) presented his statistical simulation research in modeling paleontological data. Despite the especially specialized nature of the presentation, flyers across campus proclaimed the talk’s only prerequisite to be respiration thus drawing small crowds of inhaling/exhaling students and professors from all different departments.
The study largely aims to uncover the exact primary cause of the end-Permian extinction, an unexplained extinction event occurring 250 million years ago involving the annihilation of nearly 95% of all Earth’s then-living species. Without substantive physical data such as the asteroid craters left behind from the well-known Cretaceous era extinction, paleontologists are hard-pressed for hypotheses explaining such mysterious episodes as the Permian extinction; researchers thus look to statistics for answers. Professor Wang’s research, specifically, uses the currently popular Markov Chain Monte Carlo (MCMC) simulation method in an attempt to recreate an accurate 22-dimension Permian food web and route its collapse thereby providing some basis for modern theories explaining the extinction event.
Complex computer algorithms (comprised of 2000 lines of code!) are used to process various probabilities (the MCMC method) and model approximate extinction levels of the era. Despite the simple nature of the statistical concept behind the research, the program is exceptionally complicated and difficult to execute. Algorithms must incorporate numerous variables, chronicle both indirect and direct effects of the extinction event, and work backwards to match current, fluctuating fossil records and other data.
So far, results have led to many proposed theories postulating oceanic methane shaft shifts, volcanism, and low levels of oxygen as root of the mass extinction. A consensus has yet to be reached among the natural history community, and Professor Wang plans to continue research. Possible future work involves applying these same MCMC methods to other datasets in order to simulate uncertainties and explore how various factors affect extinction models. Watch for more to come in Math/Stat Department’s weekly presentation series.
