Optimization and Control
In both engineered and evolved systems, a theme that often comes up is that of optimization – either through deliberate design or through natural selection operating on phenotypic variations. We have worked on a range of problems involving optimization and control in static and dynamic settings in both deterministic and stochastic contexts, e.g. strategies for the control of of posture in humans in sports such as slack-lining and tight-rope walking, possible control mechanisms involved in the evolution of gliding flight in animals, learning how to move (e.g. crawl, walk, swim), the dynamics and control of intermittent flight and swimming, strategies for controlling protein misaggregation, optimal transport in active matter, the control of aging in complex networks and multi-cellular organisms etc. An interesting philosophical question that arises in these ventures is the difference between causality and purpose: optimization presumes a purpose but can yield causal dynamical laws (that can hide purpose); conversely causal dynamical laws can sometimes be derived from an optimization principle (that might take time to glean). Is there a primacy of one approach over another? How can one tell? When does it matter? We are beginning to think about these questions particularly in the context of behavior – see the section on Ethology and Cognition for more.
C. Weber, T.C. Michaels and L. Mahadevan, eLife 8:e42315, 2019. [DOI] [View PDF] [Download PDF]