Towards Analog Computing: An Electronically Reconfigurable Nanoelectromechanical Oscillator Network

Author: Habermehl, Scott Thomas Philpott

Year: 2026

Degree: Dissertation (Ph.D.)

Advisor: Roukes, Michael L.

Committee Members: Marandi, Alireza; Alicea, Jason F.; Crutchfield, James P.; Roukes, Michael L.

Option: Applied Physics

Abstract

Networks of nonlinear limit-cycle oscillators exhibit a broad range of collective dynamical behaviors, from simple synchronization to exotic, partially phase-locked chimera states. Such dynamics are universal to many natural and technological systems and may prove useful as building blocks for special-purpose analog computers. In this thesis, we present the design, construction, and analysis of a network of eight nonlinear nanoelectromechanical oscillators with electronic feedback and coupling. Precise control over the system parameters, including the nodal frequencies, nodal nonlinearities, inter-nodal coupling strengths, and network topology, enables the systematic study of synchronization, pattern formation, and computation in complex networks with real-world dynamics, parameter disorder, and noise. While investigating the network's dynamics on different graph structures --- namely, all 8-node circulant graphs, a wheel graph, and Erdos--Renyi random graphs --- we observed a rich set of synchronized and unsynchronized states, arising even in the weak coupling limit. Comparison with an accompanying theoretical model reveals mechanisms that cause physical networks to deviate from idealized behavior. Finally, we report recent efforts to scale the network to a larger number of nodes by integrating an array of nanoelectromechanical resonators with an application-specific integrated circuit.