On quantum separation of variables

TL;DR

This paper introduces a novel quantum separation of variables method that constructs a basis from the transfer matrix to fully characterize the spectrum of quantum integrable models, extending classical integrability concepts.

Contribution

It develops a new approach to construct the separate variables basis in quantum integrable models using transfer matrix actions, generalizing classical action-angle variable construction.

Findings

Constructed a basis for Y(gln) models and their trigonometric/elliptic versions.

Applied the scheme to Y(gl2) and Y(gl3) models, reproducing and extending known results.

Characterized the spectrum of various quantum integrable lattice models.

Abstract

We present a new approach to construct the separate variables basis leading to the full characterization of the transfer matrix spectrum of quantum integrable lattice models. The basis is generated by the repeated action of the transfer matrix itself on a generically chosen state of the Hilbert space. The fusion relations for the transfer matrix, stemming from the Yang-Baxter algebra properties, provide the necessary closure relations to define the action of the transfer matrix on such a basis in terms of elementary local shifts, leading to a separate transfer matrix spectral problem. Hence our scheme extends to the quantum case a key feature of Liouville-Arnold classical integrability framework where the complete set of conserved charges defines both the level manifold and the flows on it leading to the construction of action-angle variables. We work in the framework of the quantum inverse scattering method. As a first example of our approach, we give the construction of such a basis for models associated to Y(gln) and argue how it extends to their trigonometric and elliptic versions. Then we show how our general scheme applies concretely to fundamental models associated to the Y(gl2) and Y(gl3) R-matrices leading to the full characterization of their spectrum. For Y(gl2) and its trigonometric deformation a particular case of our method reproduces Sklyanin's construction of separate variables. For Y(gl3) it gives new results, in particular through the proper identification of the shifts acting on the separate basis. We stress that our method also leads to the full characterization of the spectrum of other known quantum integrable lattice models, including in particular trigonometric and elliptic spin chains, open chains with general integrable boundaries, and further higher rank cases that we will describe in forthcoming publications.