A Shortcut to the Q-Operator

TL;DR

This work provides the first explicit operatorial construction of Baxter's ${\bf Q}$-operator for the compact spin-${\frac{1}{2}}$ XXX chain by introducing a twist and tracing over infinite oscillator states in the auxiliary space. It demonstrates a factorization of the transfer matrices into two independent ${\bf Q}$-operators ${\bf Q}_{\pm}$ and derives the operatorial ${\bf T}{\bf Q}$ relations, enabling Bethe equations to emerge without an explicit Bethe Ansatz. In the zero-twist limit, the authors define SU(2)-invariant ${\bf Q}$-operators via renormalization, as well as a second family ${\bf P}$-operators, linked by generalized Wronskian relations that express the full ${\bf T}$-system. They also present numerical root distributions showing two-cut structures for the Q- and P-polynomials, revealing the dual geometric organization of Bethe roots. The results illuminate the operatorial structure behind AdS/CFT spectral proposals and open pathways to higher-rank and non-compact generalizations.

Abstract

Baxter's Q-operator is generally believed to be the most powerful tool for the exact diagonalization of integrable models. Curiously, it has hitherto not yet been properly constructed in the simplest such system, the compact spin-1/2 Heisenberg-Bethe XXX spin chain. Here we attempt to fill this gap and show how two linearly independent operatorial solutions to Baxter's TQ equation may be constructed as commuting transfer matrices if a twist field is present. The latter are obtained by tracing over infinitely many oscillator states living in the auxiliary channel of an associated monodromy matrix. We furthermore compare and differentiate our approach to earlier articles addressing the problem of the construction of the Q-operator for the XXX chain. Finally we speculate on the importance of Q-operators for the physical interpretation of recent proposals for the Y-system of AdS/CFT.

Figures (2)

• Figure 1: The two-cut root distribution of the polynomials $Q(u)$ (purple squares) and $P(u)$ (blue dots). Here $L=60$, $M=22$ and $|n_k|=1$ for all $k$.
• Figure 2: Root distribution of $Q(u)$ (purple squares) and $P(u)$ (blue dots) with various $\tilde{u}_0$ for $L=60$ and $M=16$. (a) $\tilde{u}_0=0$, (b) $\tilde{u}_0=1$, (c) $\tilde{u}_0=4$. Reading this figure from the bottom one can see that when $|\tilde{u}_0|$ is decreased the two condensates shown in (c) start to approach each other (b) and eventually assemble on one line (a). It is interesting to note that for even length $L$ the two condensates pass through each other without "touching" (like in the figure), while for odd $L$ they "scatter", thereby producing double roots on the imaginary axis for $\tilde{u}_0=0$.