Sensitivity of linear systems to input orientation and novelty. (July 2018)

• Record Type: Journal Article
• Authors: Kumar, Gautam; Menolascino, Delsin; Ching, ShiNung
• Is Part Of: Automatica. Volume 93(2018)
• Page Start: 462
• Page End: 468
• Publication Date: 2018-07
• Subjects: Input novelty -- Linear systems -- Sensitivity -- Robustness; Automatic control -- Periodicals; Automation -- Periodicals; 629.805
• DOI: 10.1016/j.automatica.2018.03.006
• Languages: English
• ISSNs: 0005-1098

Abstract

Abstract: This paper introduces a framework for quantitative characterization of the sensitivity of time-varying linear systems (or networks) in terms of input orientation. The motivation for such an approach comes from the study of biophysical sensory networks in the brain, wherein responsiveness to both energy and salience (in terms of input orientation and novelty) is presumably critical for mediating behavior and function. Here, we use an inner product to define the angular separation of the current input with respect to past inputs. Then, by constraining input energy, we define an optimal control problem to obtain the minimally novel input –the one that has the smallest relative angle –that effects a given state transfer. We provide analytical conditions for existence and uniqueness for the solution in both continuous and discrete-time. A closed-form expression for the minimally novel input is derived and from this solution, a fundamental relationship between control energy and input orientation sensitivity is highlighted. We provide an example that demonstrates the utility of the developed sensitivity analysis.