Distribution of Attraction Basins in a Family of Simple Glasses
P. Chandra, L.B. Ioffe

TL;DR
This paper investigates how attraction basins are distributed in simple glasses, revealing two distinct phases with different basin size scalings and discussing implications for optimization problems.
Contribution
It identifies two types of glasses with different attraction basin size distributions and explores phase transitions between these regimes.
Findings
Attraction basins are always broadly distributed in simple glasses.
Two phases exist: one with exponentially small largest basins, another with polynomially small basins.
A critical point separates these two regimes when a tuning parameter is varied.
Abstract
We study the distribution of attraction basins as a function of energy in simple glasses. We find that it is always broad. Furthermore we identify two types of glasses, both with an exponentially large number of metastable states. In one type the largest attraction basin is exponentially small, whereas in the other it is polynomially small in the system size N. If there exists a tuning parameter that connects one regime with another, then these two phases are separated by a critical point. We discuss implications for optimization problems.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
