2026年7月8日 (水) 15:00〜16:00
L5D23 and zoom
Eigenvalues/vectors of tensors are a modern concept extending those of matrices. They appear in various non-linear contexts, such as spin glasses, general relativity, artificial intelligence, quantum information theory, string theory, and so on, while the matrix counterpart in linear-contexts. Understanding universal properties across various tensor eigen problems is important, since this assures their wide applicability. I talk about our recent finding of a universality of tensor eigenvalue/vector distributions, which has been achieved by systematically computing them using quantum field theoretical techniques.