Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This study investigates the dynamics of the Random Quadratic Form (RQF) on a sphere under the influence of random Brownian forcing, revealing that while the overall statistical properties remain unchanged, the synchronization characteristics are significantly altered. Specifically, the introduction of even minimal forcing leads to a transition from partial to full synchronization, driven by symmetry breaking phenomena. The findings elucidate the multiscale behavior of the system, highlighting the importance of initialization bias in continuous-time machine learning models like Neural ODEs and transformers.
Even a tiny amount of random forcing can transform partial synchronization into complete synchronization in systems governed by Random Quadratic Forms.
We study the Random Quadratic Form (RQF) on a sphere in the presence of random Brownian forcing. We show that the forcing does not effectively change the law of the process but affects the synchronization properties of the system. While the RQF without forcing exhibits partial synchronization due to the intrinsic symmetries, the introduction of an arbitrarily small forcing results in long-term symmetry breaking and leads to full synchronization. In this work we focus on the small forcing regime and recover the multiscale behavior of the two-point process. We show that in the first stage the model converges to an anti-polar configuration due to the symmetries of the RQF and in the second stage the two clusters meet due to the symmetry breaking phenomenon. The model is motivated by continuous-time machine learning models such as Neural ODEs and continuous-time formulations of transformers. In particular, the results of this work explain the role of the bias and the scale of its initialization.