A canonical form for PEPS

David Pérez-García (U Complutense de Madrid)

Oct 04. 2022, 15:30 — 16:40

I will show how geometric invariant theory helps to define a canonical form for arbitrary PEPS, which we call the minimal canonical form. Somehow surprisingly, the construction also gives a new canonical form for MPS. I will comment on some potential applications, in particular to enhance privacy in tensor-network based machine learning. 


Further Information
ESI Boltzmann Lecture Hall
Associated Event:
Tensor Networks: Mathematical Structures and Novel Algorithms (Thematic Programme)
Frank Pollmann (TU Munich)
Norbert Schuch (U of Vienna)
Frank Verstraete (Ghent U)