Apply to the School of Engineering

Kevin Yao ME’18 ME’20 Presented Research at the 2020 ASP Division of Fluid Dynamics Conference

POSTED ON: December 21, 2020

Figure 1: Time evolution of a quasi-geostrophic model ocean stream function field comparing the direct numerical simulation of the partial differential equation (PDE) model, top, to the Echo State Network estimation (bottom).

Figure 1: Time evolution of a quasi-geostrophic model ocean stream function field comparing the direct numerical simulation of the partial differential equation (PDE) model, top, to the Echo State Network estimation (bottom).
Figure 2: Trajectories of an object in a quasi-geostrophic ocean flow. Left: intrinsic sensitivity of trajectories, one reference (solid blue) vs one perturbed (dashed blue). Right: Echo State Network estimated trajectory (orange) shows long time accuracy

Figure 2: Trajectories of an object in a quasi-geostrophic ocean flow. Left: intrinsic sensitivity of trajectories, one reference (solid blue) vs one perturbed (dashed blue). Right: Echo State Network estimated trajectory (orange) shows long time accuracy but over very long times, the solution deteriorates and the object ends up in the upper rather than the lower gyre.

Mechanical engineering graduate student, Kevin Yao ME’18 ME’20, presented his research at the American Physical Society's annual Division of Fluid Dynamics Conference, held on November 22-24 in Chicago (virtually). Kevin has also presented this research at the Society for Industrial and Applied Mathematics (SIAM) conference on Applications of Dynamical System in 2019, you can read more here.

With the help of Professor of Physics Philip Yecko and Professor of Applied Mathematics Eric Forgoston of Montclair State University, Kevin's thesis research is on the use of “Echo State Networks” a form of Machine Learning to predict the behavior of time-dependent dynamical systems which originate from a simplified model of ocean flow, known in geophysical fluid
dynamics as the quasi-geostrophic beta plane, or QG. QG models of oceans can be used to study trajectories of objects in the ocean and machine learning models may be very useful in predicting pollutant dispersal or in search and recovery.

More information about the conference can be found here.

More News from The Cooper Union

  • Founded by inventor, industrialist and philanthropist Peter Cooper in 1859, The Cooper Union for the Advancement of Science and Art offers education in art, architecture and engineering, as well as courses in the humanities and social sciences.

  • “My feelings, my desires, my hopes, embrace humanity throughout the world,” Peter Cooper proclaimed in a speech in 1853. He looked forward to a time when, “knowledge shall cover the earth as waters cover the great deep.”

  • From its beginnings, Cooper Union was a unique institution, dedicated to founder Peter Cooper's proposition that education is the key not only to personal prosperity but to civic virtue and harmony.

  • Peter Cooper wanted his graduates to acquire the technical mastery and entrepreneurial skills, enrich their intellects and spark their creativity, and develop a sense of social justice that would translate into action.