# Graduate Student / Postdoc Seminar

#### Capturing Intermittent and Low-Frequency Spatiotemporal Patterns in High-Dimensional Data

**Speaker:**
Dimitris Giannakis

**Location:**
Warren Weaver Hall 1302

**Date:**
Friday, November 30, 2012, 1 p.m.

**Synopsis:**

In this talk we present work done in collaboration with faculty and students at the Center for Atmosphere Ocean Science (CAOS) on methods for extracting spatial patterns of variability from high-dimensional time series. This problem arises in many data rich areas, including geosciences, fluid dynamics, and molecular dynamics. Here, we discuss an algorithm called Nonlinear Laplacian Spectral Analysis (NLSA), which represents spatial and temporal patterns through singular value decomposition of a family of maps acting on scalar functions on the nonlinear data manifold. The use of orthogonal basis functions (determined by means of graph Laplace-Beltrami eigenfunction algorithms) and time-lagged embedding are other key ingredients of NLSA designed to capture intermittency, rare events, and other nonlinear dynamical features which are not accessible through classical approaches such as principal components analysis. We present applications of NLSA to detection of decadal and intermittent patterns of variability in the North Pacific sector of comprehensive climate models, and multiscale physical modes of the Madden-Julian Oscillation in infrared brightness temperature satellite data.