Dimerization in 𝒪 ( n ) \mathcal{O}(n) -invariant quantum spin chains
J. E. Björnberg and K. Ryan · 2025
Read Dimerization in 𝒪 ( n ) \mathcal{O}(n) -invariant quantum spin chains by J. E. Björnberg and K. Ryan (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-06103.txt.
Introduction
Björnberg and Ryan prove, in a 2025 math paper, that paired-up states arise in a broad class of quantum spin chains. The chains keep a symmetry named in the title, and the result needs the count in that symmetry to be large enough. Their dimerization claim covers a far wider slice of the phase map than past work. Two endless-volume ground states appear, each shifted by one step from the other. Both are gapped and show fast-fading links.
The states arise as limits of finite chains at low heat, in any order of limits. The method recasts the quantum system as random loops and adapts a trick built for loop models on a honeycomb lattice.
Catalog record
| CREATOR | J. E. Björnberg and K. Ryan |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 136 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-06103
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Dimerization in 𝒪 ( n ) \mathcal{O}(n) -invariant quantum spin chains J. E. Björnberg and K. Ryan Abstract . We establish dimerization in 𝒪 ( n ) \mathcal{O}(n) -invariant quantum spin chains with big enough n n , in a large part of the phase diagram where this result is expected. This includes identifying two distinct ground states which are translations of one unit of eachother, and which both have exponentially decaying correlations. Our method relies on a probabilistic representation of the quantum system in terms of random loops, and an adaptation of a method developed for loop 𝒪 ( n ) \mathcal{O}(n) models on the hexagonal lattice by Duminil-Copin, Peled, Samotij and Spinka. 1. Introduction The most general O(n) -invariant quantum spin system with pair-interactions has Hamiltonian (1.1) H_{\Lambda}=-\sum_{xy\in\mathcal{E}(\Lambda)}\big[uT_{xy}+vQ_{xy}\big],\qquad\text{acting on }(\mathbb{C}^{n})^{\otimes\Lambda}, where u,v\in\mathbb{R} , \Lambda is a finite graph, \mathcal{E}(\Lambda) is its set of nearest-neighbour pairs, and the interaction involves the operators (1.2) T|a,b\rangle=|b,a\rangle,\qquad Q=\frac{1}{n}\sum_{a,b=1}^{n}|b,b\rangle\langle a,a|,\qquad\text{on }(\mathbb{C}^{n})^{\otimes 2}. The behaviour of this model varies widely depending on the graph \Lambda , the parameters u,v , and the value of n (related to the spin S by 2S+1=n ). The model reduces to the spin- \tfrac{1}{2} Heisenberg XXZ model when n=2 , and to the spin-1 bilinear-biquadratic Heisenberg model when n=3 . For u,v\geq 0 , the model has a well-known probabilistic representation as a loop model (related to the interchange process). This paper concerns the model in 1 dimension with u\geq 0,v>0 and n large, where we prove a breaking of translation invariance, known as dimerization, in the ground state. In a separate paper [ 9 ] , we prove that in \mathbb{Z}^{d} for all d\geq 1 , there is exponential decay of correlations at finite low temperature v\geq u and n large, using different techniques. 1.1. Main theorem To state our result on dimerization, we need some definitions. For positive temperature T=\frac{1}{\beta} , the Gibbs state \langle\cdot\rangle_{\Lambda,\beta} in volume \Lambda is the linear map L((\mathbb{C}^{n})^{\otimes\Lambda})\to\mathbb{C} given by (1.3) \langle A\rangle_{\Lambda,\beta}=\frac{1}{Z_{\Lambda,n,\beta}}\mathrm{Tr}\big(A\mathrm{e}^{-\beta H_{\Lambda}}\big),\qquad\text{where }Z_{\Lambda,n,\beta}=\mathrm{Tr}\big(\mathrm{e}^{-\beta H_{\Lambda}}\big). The ground state is given by replacing the Gibbs factor \mathrm{e}^{-\beta H_{\Lambda}} by the projection onto the subspace of the lowest eigenvalue of H_{\Lambda} , equivalently it is the limit \beta\to\infty of the Gibbs state. Infinite-volume Gibbs- or ground-states can be obtained as limits of these as \Lambda\Uparrow\mathbb{Z}^{d} , or characterized using the KMS-condition. For a vector \Psi\in(\mathbb{C}^{n})^{\otimes\Lambda} , write (1.4) \langle A\rangle_{\Lambda,\beta}^{\Psi}=\frac{\langle\Psi|\mathrm{e}^{-\frac{\beta}{2}H_{\Lambda}}A\mathrm{e}^{-\frac{\beta}{2}H_{\Lambda}}|\Psi\rangle}{\langle\Psi|e^{-\beta H_{\Lambda}}|\Psi\rangle}, which we refer to as a seeded state. For \Lambda=\Lambda_{L}:=\{-L+1,\dots,L\}\subset\mathbb{Z} , let (1.5) \Psi_{L}=\sum_{i=1}^{L}\frac{1}{\sqrt{n}}\sum_{a=1}^{n}|a,a\rangle_{-L+2i-1,-L+2i}. Note that Q is the orthogonal projector on \Psi_{1} . By local operator we mean a linear operator A on (\mathbb{C}^{n})^{\otimes\Lambda(A)} for a finite set \Lambda(A) ; we regard A as an operator on (\mathbb{C}^{n})^{\otimes\Lambda} for any \Lambda\supseteq\Lambda(A) by identifying it with A\otimes\hbox{\rm 1\kern-2.70004ptI}_{\Lambda\setminus\Lambda(A)} . The smallest choice of set \Lambda(A) is called the operator’s support . For t\in\mathbb{R} and a local operator A (with support in \Lambda ) we write A(t)=e^{tH_{\Lambda}}Ae^{-tH_{\Lambda}} , and for a state \langle\cdot\rangle we write \langle A;B\rangle=\langle AB\rangle-\langle A\rangle\langle B\rangle . We consider the model ( 1.1 ) with v=1-u and u\in[0,1] , so the Hamiltonian is (1.6) H_{\Lambda}=-\sum_{xy\in\mathcal{E}(\Lambda)}\big[uT_{xy}+(1-u)Q_{xy}\big]. Theorem 1.1 (Dimerization for d=1 ) . Let d=1 and fix u\in[0,1) . There is n_{0}(u) such that for n>n_{0} , the following holds: (1) There are two distinct infinite-volume ground-states \langle\cdot\rangle_{1} , \langle\cdot\rangle_{2} for the model ( 1.6 ), such that for \alpha\in\{1,2\} , (1.7) \langle\cdot\rangle_{\alpha}=\lim_{\begin{subarray}{c}L\to\infty\\ L\in 2\mathbb{Z}+\alpha\end{subarray}}\lim_{\beta\to\infty}\langle\cdot\rangle_{\Lambda_{L}}=\lim_{\begin{subarray}{c}L\to\infty\\ L\in 2\mathbb{Z}+\alpha\end{subarray}}\lim_{\beta\to\infty}\langle\cdot\rangle_{\Lambda_{L}}^{\Psi_{L}}. Moreover, for both \langle\cdot\rangle_{\Lambda_{L}} and \langle\cdot\rangle_{\Lambda_{L}}^{\Psi_{L}} the limits L,\beta\to\infty can also be taken simultaneously, and for \langle\cdot\rangle_{\Lambda_{L}}^{\Psi_{L}} the limits can be taken in any order. The states \langle\cdot\rangle_{\alpha} are 2\mathbb{Z} -invariant and are translations by one unit of each other. (2) Correlations decay exponentially in both limiting states: there exists C>0 such that for any local operators A,B with supports \Lambda(A) , \Lambda(B) respectively, we have for \alpha\in\{1,2\} and for all t\in\mathbb{R} , (1.8) |\langle A;B(t)\rangle_{\alpha}|\leq\|A\|\|B\|\mathrm{e}^{-C(d(\Lambda(A),\Lambda(B))+|t|)}. Here \|A\|=\sup_{\|\psi\|=1}|\langle\psi|A|\psi\rangle| is the operator norm. Remark 1.2 . The decay of correlation implies that the two states \langle\cdot\rangle_{\alpha} , \alpha\in\{1,2\} , are extremal. 1.2. Background The 1-dimensional ground state behaviour of the model ( 1.1 ) is diverse. The expected ground state for the model with n\geq 3 is depicted in Figure 1 . We now summarize the heuristics for that phase diagram, for further details see [ 8 ] or (for n=3 ) [ 17 ] . Figure 1. Expected ground state phase diagram for the spin chain with Hamiltonian ( 1.1 ) for n\geq 3 . Dimerization is expected in the yellow region, and has been established in the darker yellow region, for certain values of n . In this paper we prove dimerization in the range from N to E (E not included), for large enough n . It was previously established at N (for n\geq 3 ) and in a neighbourhood around N (for large n ). The operator T_{x,y} can be thought of as ferromagnetic, and Q_{x,y} anti-ferromagnetic. This leads to the most straightforward part of the phase diagram: the southeast quadrant u\geq 0,v\leq 0 (blue in Figure 1 , including the point E ) is ferromagnetic; there are many ground states, which minimize the Hamiltonian ( 1.1 ) term by term. In the southwest quadrant u<0,v<0 (green in Figure 1 ), the model is “critical”, in the sense that there should be a unique, gapless ground state, with polynomially decaying correlations. Correlations are in fact expected to decay with incommensurate phases, that is, correlations between the origin and x should decay as |x|^{-r}\cos(\omega|x|) , where r,\omega are functions of u,v . See [ 17 ] and [ 21 ] . Between the point W and the Reshetikhin point (red in Figure 1 ), one expects behaviour dependent on the parity of n . For n odd the model is in the Haldane phase: there should be a unique ground state, with a gap and exponentially decaying correlations. For n even, there should be two extremal gapped ground states, which are translations of each other by 1, but which are not dimerized states (as thought in some physics literature). Rather, the O(n) symmetry is broken down to SO(n) , and the two states are related by a site-transformation of determinant -1 . This behaviour for n even was recently discovered in the PhD thesis of Ragone [ 35 ] , and had not been observed in a model before. He studies the point v=-2u , which is a frustration-free point whose ground states are given by matrix product states. For n=3 , this is the AKLT model [ 1 , 2 ] , and for larger n named the SO(n) AKLT model. Ragone studies this point for all n\geq 3 , and proves the behaviour described above. The SO(n) AKLT point should have the same qualitative behaviour as the whole phase between the point W and the Reshetikhin point. General stability results on gapped chains should extend the existence of the gap rigorously to a neighbourhood of the point v=-2u , see [ 32 ] . The point v=-\tfrac{2n}{n-2}u was solved by Reshetikhin [ 37 ] and shown to have no gap. The special case when n=3 is the Takhtajan–Babujian model. Finally, between the point E and