Data-driven science and engineering : machine learning, dynamical systems, and control /

Unlike other works treating machine learning and control theory, Data-Driven Science and Engineering could not easily serve as a stand-alone resource in either discipline. Brunton and Kutz (both Univ. of Washington) intend this book as a complementary resource for a team-taught course. Readers will need background knowledge of the three domains specified in the subtitle and also experience with linear algebra and differential equations. The first part (of four) deals with signal processing and data transformation, reviewing topics such as principal component analysis (PCA), Fourier transforms, and compressed sensing. In part 2, traditional machine learning topics are discussed but with examples applied toward dynamical systems and their controls. Part 3 provides a review of dynamical systems and linear control theory and introduces the intersection of machine learning with control theory. Part 4 offers a discussion of reduced order models (ROMs), i.e., the mapping of high-dimensionality models to lower-dimensionality models facilitating lower computational complexity, reducing the complexity of simulation and analysis and how these models are interpolated through the sampling of the higher-order system to produce the ROM. Throughout, topics are discussed with theoretical depth and accompanied by a substantial bibliography. The authors also make use of software code snipes (available to readers online). Summing Up: Recommended. Graduate students, researchers, faculty. --Richard S. Stansbury, Embry-Riddle Aeronautical University