Koichiro Yawata (Institute of Science, Tokyo)

Title: Reduced-Order Control of Coupled Phase Oscillator Populations and Hypergraph-Based Approaches to Combinatorial Optimization Problems

Abstract: 

In this seminar, I will present two recent pieces of work from my research: one on reduced-order control of coupled oscillator populations and another on hypergraph-based approaches to combinatorial optimization.

1. Reduced-Order Control of Coupled Oscillator Populations via Frequency-Dependent Forcing
We propose a frequency-dependent forcing framework for heterogeneous populations of coupled phase oscillators. Using the Ott-Antonsen ansatz, we derive a low-dimensional description of the collective dynamics and study control methods for the macroscopic phase and synchronization level.

2. Hypergraph Neural Networks Accelerate MUS Enumeration
We propose a hypergraph-neural-network-based method for accelerating MUS enumeration in constraint satisfaction problems. By combining hypergraph modeling with reinforcement learning, the method reduces satisfiability checks and improves enumeration efficiency.

This seminar will take place in Room S08 at the Faculty of Sciences.