Overview

System Level Synthesis (SLS) is a very new topic in control theory, but is expanding rapidly in controls and controls-adjacent fields. It allows scalable, systematic design of optimal controllers for systems with sparsity, locality, delay, and other constraints on communication and computation internal to the controller - constraints which made pre-SLS methods intractable. SLS-related papers have already won an Axelby prize and student paper prizes at both ACC and CDC. However, the pace of progress and breadth of the current frontier can make it appear difficult to contribute, when almost the opposite is true; we aim to correct this in the workshop.

While the mathematical tools behind SLS are no harder than pre-SLS methods, the concepts and broad applications are new and can be challenging to the uninitiated. This is exacerbated by the fact that pre-SLS methods handled sparsity, locality, delay, and other communication/computational constraints in a very confusing way - it takes some effort and guidance to switch to the SLS way of thinking, in which such constraints are easily accommodated. Broadly speaking, SLS is a new controller parametrization. Many branches of control theory (including model predictive control, nonlinear control, robust control) benefit from the application of the SLS parametrization, which results in scalable, local algorithms that accommodate communication and computational limits. Such limits are omnipresent in large cyberphysical systems, in biology and neuroscience, and even social systems - thus, SLS greatly expands the domains where control theory can be rigorously applied. Additionally, while SLS was originally motivated by communication constraints, it also appears to greatly facilitate the use of learning for control, which will also be a featured topic of the workshop. Overall, any CDC attendee can benefit from a deeper understanding of SLS theory.

Goal and expected outcomes

• Teach the basics of SLS theory

• Enable participants can leverage SLS theory for the benefit of their existing research

• Enable participants to leverage their existing expertise to contribute to the frontiers of SLS research

This is an unusual and exciting opportunity; for experts in both theory and applications, the frontier of SLS theory is broad and accessible. Furthermore, SLS theory opens up many new avenues of potential research, especially in areas where communication constraints were previously an obstacle.

Additional video content will be made available to participants in order to supplement and complement the core workshop material. Potential topics include

• SLS-related applications and examples from biology, neuroscience, and other disciplines not covered in the workshop

• The role of SLS in a nascent theory of engineering and biological architectures

• Expansion content related to workshop talks

Target audience

We invite all interested theorists and practitioners, especially those with an interest in incorporating scalability, communication constraints, and/or closed-loop performance guarantees into their algorithms. Knowledge of standard control theory and linear algebra is sufficient. If you can attend CDC, you can attend and benefit from this workshop.

Organizers: Jing Shuang (Lisa) Li, Jing Yu, Carmen Amo Alonso, John Doyle