betaFeedback

WorksPhysics and Astronomy

Study of the frustrated SU(3) Heisenberg chain with SU(3)-symmetric matrix product states

Abstract

We study the SU(3) Heisenberg chain with nearest- and next-nearest-neighbor antiferromagnetic exchange J1J_1 and J2J_2. Field theory predicts that the trimerized valence-bond solid of this chain has a finite extent in α=J2/J1α=J_2/J_1, followed by a gapless SU(3)1_1 phase, but earlier numerical estimates of the upper boundary range from α≈2.1α\approx2.1 (exact diagonalization) to α≈3.8α\approx3.8 (iDMRG). We use the variational uniform matrix product state (VUMPS) algorithm with the full non-Abelian SU(3) symmetry imposed on the virtual bonds, reaching bond dimensions up to D≈8000D\approx8000, and locate the phase boundaries by finite-entanglement scaling. Resolving the transfer-matrix spectrum and the entanglement spectrum by SU(3) irreducible representation, we find that the trimerized phase extends to αc2≳3.5α_{c2}\gtrsim3.5 and end below α≈5α\approx5, excluding a boundary near α≈2α\approx2. At larger αα our numerical analysis of the effective central charge is consistent with the predicted massless flow from SU(3)2_2 to SU(3)1_1.

Introduction

Lattice models with SU(NN) symmetry describe Mott insulators of alkaline-earth-like atoms with NN populated nuclear-spin states [1, 2], which have been realized with 173Yb{}^{173}\mathrm{Yb} and 87Sr{}^{87}\mathrm{Sr} in optical lattices [3, 4, 5, 6]. For N>2N > 2 two spins in the fundamental representation cannot form a singlet; the smallest singlet involves NN sites. Frustration therefore favors NN-site valence-bond solids rather than dimers.

The simplest frustrated model of this kind is the J1J_1–J2J_2 SU(3) Heisenberg chain [Eq. (1)]. At zero next-nearest-neighbor coupling (J2=0J_2 = 0), it is the integrable Sutherland chain [7], a gapless SU(3)1_1 Wess–Zumino–Witten (WZW) critical point with central charge c=2c = 2. Increasing J2J_2 drives a marginal current–current coupling through zero and opens a gap into a threefold-degenerate trimerized phase at αc1≈0.48\alpha_{\mathrm{c1}} \approx0.48 [8, 9, 10]. The fate of this phase at large α\alpha was clarified by Herviou, Capponi and Lecheminant (HCL) [10]: viewing the chain as a zigzag ladder of two SU(3) chains coupled by J1J_1, they showed that for odd NN, the inter-chain coupling gaps a ZN\mathbb{Z}_N parafermion sector and leaves an SU(NN)2_2 theory perturbed by its adjoint field, which flows without a gap to SU(NN)1_1 [11]. The trimerized phase thus has a finite extent, and a second gapless SU(3)1_1 phase appears at large α\alpha.

The location of the upper boundary αc2\alpha_{\mathrm{c2}} for the trimerized phase and the nature of the large-α\alpha phase are not settled numerically. Exact diagonalization (ED) on rings of up to 27 sites gives αc2=2.1(1)\alpha_{\mathrm{c2}} = 2.1(1) with strong finite-size effects, while iDMRG with U(1) × U(1) color conservation and bond dimension χ≤8192\chi\le8192 shows trimerization up to α≈3.8\alpha\approx3.8, beyond which the data jump at a χ\chi-dependent coupling [10]; an earlier study placed the boundary near 3.5 [8]. At large α\alpha, HCL obtained an effective central charge c≃3.2c \simeq3.2, close to the SU(3)2_2 value 16/516/5, and argued that full SU(3) symmetry would be needed to resolve the flow to c=2c = 2, in analogy with the SU(3) chain in the symmetric representation [12, 13].

Here we address both questions with variational uniform matrix product states (VUMPS) [14, 15] with exact SU(3) symmetry on the virtual bonds, at bond dimensions up to D≈8000D \approx8000 [about 540 SU(3) multiplets]. Our main results are:

  1. An injective SU(3)-symmetric MPS with the fundamental representation on every site cannot be invariant under one-site translation; finite-DD states therefore break translation and generically show some trimerization, so that the phase boundaries have to be located by finite-entanglement scaling (§ 2).

  1. At the lower transition, the extrapolated trimerization becomes positive at αc1≈0.48\alpha_{\mathrm{c1}} \approx0.48, once states up to D=4000D = 4000 are employed, consistent with the earlier ED estimate. By contrast, a fit of the essential-singularity form expected for this SU(3)-symmetric marginal transition cannot locate αc1\alpha_{\mathrm{c1}} at reachable bond dimensions (§ 2.4).

  1. From the bond-dimension scaling of the octet correlation length, the entanglement entropy and the trimerization, the trimerized phase extends to αc2≳3.5\alpha_{\mathrm{c2}} \gtrsim3.5; where it ends is not resolved at the bond dimensions reached (§ 2.5).

  1. Above αc2\alpha_{\mathrm{c2}}, for α∈[4,10]\alpha\in[4, 10], the effective central charge lies below 16/516/5, and where the crossover length is shortest (moderate α≈4\alpha\approx4), it keeps decreasing with DD (Sec. VI).

  1. The jumps of the iDMRG data of Ref. [10] are crossings, at fixed bond dimension, to states outside the SU(3)-symmetric ansatz, which we reproduce with an Abelian iDMRG approach; no SU(3)-symmetric state of either entanglement structure reproduces them (Sec. VII).

After the model and the method (§ 2) we give an overview of the phase diagram (Sec. III), determine the lower and the upper boundary of the trimerized phase (§ 2.4 and § 2.5), discuss the large-α\alpha regime (Sec. VI), and compare with previous work (Sec. VII).

Model and method

Model

We consider

H^=J1∑iP^i,i+1+J2∑iP^i,i+2,α≡J2/J1,(1)\hat{H}=J_{1}\sum_i \hat{P}_{i,i+1}+J_{2}\sum_i \hat{P}_{i,i+2},\qquad\alpha\equiv J_{2}/J_{1}, \tag*{(1)}

where each site carries the fundamental representation 3\mathbf{3} of SU(3) and P^ij\hat{P}_{ij} exchanges the states of sites ii and jj. With SU(3) generators normalized as Tr⁡(S^AS^B)=δAB/2\operatorname{Tr}(\hat{S}^{A}\hat{S}^{B})=\delta^{AB}/2 one has 2∑AS^iAS^jA=P^ij−1/32\sum_A\hat{S}^{A}_{i}\hat{S}^{A}_{j}=\hat{P}_{ij}-1/3, so Eq. (1) differs from the spin form used in Ref. [10] only by a factor of two, and α\alpha coincides with J2/J1J_{2}/J_{1}. We set J1=1J_{1}=1. Drawing the chain as a zigzag ladder (Figure 1), the J2J_{2} bonds form two legs and the J1J_{1} bonds alternate between “rung” and “diagonal” bonds. Two sites cannot form an SU(3) singlet, since 3⊗3=6⊕3ˉ\mathbf{3}\otimes\mathbf{3}=\mathbf{6}\oplus\bar{\mathbf{3}}; the smallest singlet involves three sites, which is why frustration favors period-3 (trimer) rather than bipartite order.

Zigzag chain geometry and the SU(3) permutation operator

Figure 1. Zigzag chain geometry and the SU(3) permutation operator. Sites (circles) carry one of three colors (red = 1, green = 2, blue = 3), illustrating the period-3 pattern of the trimerized phase. Solid lines are nearest-neighbor (J1J_1) bonds; within the zigzag, two geometrically distinct subtypes arise – “rung” bonds (bottom→top, carrying energy ErungE_{\mathrm{rung}}) and “diagonal” bonds (top→next-bottom, EdiagE_{\mathrm{diag}}) – which can have different expectation values whenever translational symmetry is broken. Dashed lines are next-nearest-neighbor (J2=αJ1J_2 = \alpha J_1) bonds running along each row (EJ2E_{J_2}), which introduce geometric frustration. The bracketed region is one unit cell (NUC=6N_{\mathrm{UC}} = 6 sites). Inset: the permutation operator P^ij\hat{P}_{ij} exchanges the SU(3) colors of sites ii and jj. It plays the same role as 2Si⋅Sj2\mathbf{S}_i \cdot\mathbf{S}_j in SU(2) but has no two-site singlet eigenstate, so color-neutral clusters require three sites.

At α=0\alpha=0, Eq. (1) is the Sutherland chain, with exact ground-state energy per site e0=1−ln⁡3−π3/9=−0.703212e_{0}=1-\ln3-\pi\sqrt{3}/9=-0.703212 [7] and an SU(3)1_1 critical point with c=2c=2. At J1=0J_{1}=0 the two legs decouple into two such chains (c=4c=4). For any finite α\alpha, the inter-leg coupling is relevant: it gaps a Z3\mathbb{Z}_{3} parafermion sector (c=4/5c=4/5) and leaves SU(3)2_2 (c=16/5c=16/5) perturbed by its adjoint primary, which flows without a gap to SU(3)1_1 (c=2c=2) [10, 11]. Here c=k(N2−1)/(k+N)c=k(N^{2}-1)/(k+N) for SU(NN)k_k, i.e. 8/4=28/4=2 for k=1k=1 and 16/516/5 for k=2k=2; the parafermion sector accounts for 4−16/5=4/54-16/5=4/5.

We use a unit cell of six sites (three rungs), commensurate with both the period-3 order and the two-leg structure, as in Ref. [10]. All states we obtain have a period of three sites (the bond energies repeat after three sites), thus a three-site cell would describe the same states; the six-site cell, in addition, admits period-2 and period-6 patterns, which do not occur.

SU(3)-symmetric uniform MPS

Ground states are obtained with the variational uniform matrix-product-state algorithm (VUMPS) [14, 15], with the SU(3) symmetry imposed exactly on the virtual spaces [16, 17, 18, 19], as in non-Abelian symmetric DMRG and tensor-network methods [20, 21]: every tensor is stored as reduced blocks between multiplets, with the Clebsch–Gordan structure fixed by the symmetry. Each virtual space is a direct sum of SU(3) irreducible representations (irreps) with multiplicities; the bond dimension DD counts states, so an irrep λ\lambda with multiplicity mλm_{\lambda} contributes mλdim⁡λm_{\lambda}\dim\lambda. At D≈6000D\approx6000 the states carry about 420 multiplets (≈14\approx14 states per multiplet), and the cost is set by the multiplet structure rather than by DD. Because each site adds one box to the Young diagram, the triality (number of boxes mod 3) of the virtual irreps advances by one from bond to bond: every bond carries irreps of a single triality class, t=0t=0 (1,8,10,27,…\mathbf{1},\mathbf{8},\mathbf{10},\mathbf{27},\ldots), t=1t=1 (3,6ˉ,15,…\mathbf{3},\bar{\mathbf{6}},\mathbf{15},\ldots) or t=2t=2 (their conjugates), cycling through the unit cell. The irreps and their multiplicities are not fixed in advance: they are selected by the singular values, and irreps up to dimension 64 occur in our states. The truncation keeps or discards whole multiplets, ranked by their Schmidt values, so the state is exactly SU(3)-symmetric at every step; the triality of the irreps on each bond is fixed by the fusion rules and is not chosen by hand.

The bond dimension is increased in steps of a factor 1.2. Each step starts with a two-site infinite DMRG sweep that truncates to the new DD; this is the step in which the bond spaces are chosen, i.e., which irreps occur on each bond and with which multiplicities, by keeping the multiplets with the largest Schmidt values. VUMPS then optimizes the tensors at fixed bond spaces until the gradient norm

δ=max⁡i∥(I−ALiALi†)A~Ci∥,A~Ci=HACiACi/∥HACiACi∥,(2)\delta=\max_i\left\|(\mathbb{I}-A^{i}_{L}A^{i\dagger}_{L})\widetilde{A}^{i}_{C}\right\|,\qquad\widetilde{A}^{i}_{C}=H^{i}_{A_{C}}A^{i}_{C}/\left\|H^{i}_{A_{C}}A^{i}_{C}\right\|, \tag*{(2)}

falls below 10−910^{-9} (at most 1000 iterations). Here ALiA^{i}_{L} and ACiA^{i}_{C} are the left-orthonormal and center-site tensors of site ii of the cell, and HACiH^{i}_{A_{C}} is the effective Hamiltonian acting on ACiA^{i}_{C}; δ\delta is the norm of the energy gradient projected onto the tangent space of the MPS manifold [14, 15] and vanishes at a variational optimum.

No single-site translation invariance at finite DD

The fundamental representation is a projective representation of PSU(3)==SU(3)/Z3\mathbb{Z}_{3}: the center element z=ωIz=\omega\mathbb{I}, ω=e2πi/3\omega=e^{2\pi i/3}, acts as the phase ω\omega on each site. Consider a single-site uniform MPS AsA^{s} with this symmetry. Each site adds one box to the Young diagram, so AsA^{s} maps virtual irreps of triality tt to triality t+1t+1; with Z=ωZ=\omega on the triality-tt blocks of the virtual space, ZAsZ−1=ωAsZA^{s}Z^{-1}=\omega A^{s}. If ρ\rho is the fixed point of the transfer map E(X)=∑sAsXAs†\mathcal{E}(X)=\sum_s A^{s}XA^{s\dagger}, then E(Zρ)=ω−1Zρ\mathcal{E}(Z\rho)=\omega^{-1}Z\rho; since ZZ is invertible, Zρ≠0Z\rho\ne0, and ω−1≠1\omega^{-1}\ne1 is a second eigenvalue of modulus one. An injective MPS has a unique eigenvalue of modulus one, so an injective SU(3)-symmetric MPS cannot be invariant under one-site translation. This is the matrix-product-state form of the Lieb–Schultz–Mattis theorem [22, 23], shown for SU(2) with half-integer spin in Ref. [24]; a complete derivation is given in Appendix C.

A translation-invariant SU(3)-symmetric MPS exists only in non-injective form, as the equal-weight superposition of the three translates of a period-3 injective state; its correlations do not cluster and it carries no information beyond its components. In the SU(3)-symmetric representation, the translation breaking of an injective state is visible directly in the entanglement spectrum. On cuts with t≠0t \ne0 every irrep has dimension divisible by three, so every Schmidt value is at least threefold degenerate, while the t=0t = 0 cuts contain the singlet sector with non-degenerate values. The spectra of neighboring cuts therefore differ, and every finite-DD state breaks single-site translation: the bond energies are modulated with a period of three sites, with T>0T > 0 in all our states, also in a gapless phase (at α=0\alpha= 0, T=0.006T = 0.006 at D=1034D = 1034). The distinction between the gapless and the trimerized phases is then a matter of DD-scaling: T(D)T(D), ξ8(D)\xi_8(D) and S(D)S(D) saturate in the trimerized phase, while in a gapless phase T→0T \to0 and ξ8\xi_8, SS grow without bound.

Observables

Correlation length. The transfer matrix of the six-site cell, T=∑sAs⊗Aˉs\mathcal{T} = \sum_s A^s \otimes\bar{A}^s, is block diagonal in the SU(3) irrep λ\lambda of its channel (only triality-zero irreps occur). Let tλ,jt_{\lambda,j} be its eigenvalues in sector λ\lambda, ordered by modulus. For a normalized injective MPS the largest eigenvalue is t1,1=1t_{1,1} = 1, in the singlet sector; it corresponds to the norm of the state and carries no correlations. We therefore define, in units of sites,

ξλ=−6ln⁡∣tλ,1∣(λ≠1),ξ1=−6ln⁡∣t1,2∣.(3)\xi_\lambda= -\frac{6}{\ln|t_{\lambda,1}|}\quad(\lambda\ne1),\qquad\xi_1 = -\frac{6}{\ln|t_{1,2}|}. \tag*{(3)}

The adjoint length ξ8\xi_8 is the longest at all couplings, as also found in Ref. [10].

Entanglement. Cutting the infinite chain at one of the six bonds of the unit cell divides it into two half-infinite parts. The reduced density matrix of one part, ρ=Tr⁡∣ψ⟩⟨ψ∣\rho= \operatorname{Tr}_{| \psi\rangle\langle\psi|}, has the squared Schmidt values sk2s_k^2 of that bond as eigenvalues, which follow from the bond matrix of the MPS in mixed canonical form. With SU(3) symmetry they come in multiplets: a Schmidt value in irrep λ\lambda is dim⁡λ\dim\lambda-fold degenerate. For each cell we compute the entanglement entropy S=−∑ksk2ln⁡sk2S = -\sum_k s_k^2 \ln s_k^2 and the entanglement energies ϵk=−ln⁡sk2\epsilon_k = -\ln s_k^2; we report SS averaged over the six cuts, and the largest eigenvalue λmax⁡\lambda_{\max} of ρ\rho over the cuts together with the irrep that carries it.

Trimerization. With Eb=⟨P^i,i+1⟩E_b = \langle\hat{P}_{i,i+1} \rangle on the nearest-neighbor bonds, the trimerization is T=Eweak−13∑k=02Eb+kT = E_{\mathrm{weak}} - \frac{1}{3}\sum_{k=0}^{2} E_{b+k} on the weak bond [10]. In a period-3 state with bond energies {s,s,w}\{s,s,w\}, T=2(w−s)/3=(2/3)σrungT = 2(w-s)/3 = (2/\sqrt{3})\sigma_{\mathrm{rung}}, where

σrung=[12∑k=13(Erung(k)−Eˉrung)2]1/2(4)\sigma_{\mathrm{rung}} = \left[\frac{1}{2}\sum_{k=1}^{3}\left(E_{\mathrm{rung}}^{(k)}-\bar{E}_{\mathrm{rung}}\right)^2\right]^{1/2} \tag*{(4)}

is the sample standard deviation of the three rung energies Erung(k)E_{\mathrm{rung}}^{(k)} of the cell, with mean Eˉrung=13∑kErung(k)\bar{E}_{\mathrm{rung}} = \frac{1}{3}\sum_k E_{\mathrm{rung}}^{(k)}; for {s,s,w}\{s,s,w\} it equals ∣w−s∣/3|w-s|/\sqrt{3}.

Finite-entanglement scaling. At a critical point a finite-DD MPS has a finite ξ(D)∝Dκ\xi(D) \propto D^\kappa, with κ=6/[12/c+1+1]\kappa= 6/[\sqrt{12/c+1}+1] [25], and S=(c/6)ln⁡ξ+constS = (c/6)\ln\xi+ \mathrm{const} [26]. We extract an effective central charge ceff(D)=6 dS/dln⁡ξ8c_{\mathrm{eff}}(D) = 6\,dS/d\ln\xi_8 from the slope of SS versus ln⁡ξ8\ln\xi_8 over the three most recent states up to and including DD (two states whose DD differ by less than 10%, which occur where a run was resumed, are merged); in a gapped phase ceff→0c_{\mathrm{eff}} \to0 as ξ8\xi_8 saturates.

Extrapolation of the trimerization. Every finite-DD SU(3)-symmetric state has T>0T > 0, also in a gapless phase (see § 2 C). Therefore, the value of TT at a given bond dimension does not indicate whether the chain is trimerized; the question is whether T(D)T(D) tends to zero or to a finite value as D→∞D \to\infty. A finite-DD MPS describes the infinite system up to the length scale ξ8(D)\xi_8(D), which plays the role of a finite system size. In a gapless phase an order parameter that vanishes in the thermodynamic limit then decays as T(D)∝ξ8−ΔT(D) \propto\xi_8^{-\Delta} [27, 28], where Δ\Delta is the scaling dimension of the corresponding operator [29, 30]. For the trimerization in the SU(3)1_1 phase this is the operator Tr⁡g\operatorname{Tr} g, built from the primary field gg in (3,3‾)(\mathbf{3},\overline{\mathbf{3}}) with h=hˉ=C2(3)/(k+N)=(4/3)/4=1/3h = \bar{h} = C_2(3)/(k + N) = (4/3)/4 = 1/3, so that Δ=2h=2/3\Delta= 2h = 2/3. In a gapped phase T(D)T(D) instead approaches a finite limit once ξ8(D)\xi_8(D) exceeds the correlation length of the ground state. We therefore fit

T(D)=T∞+Aξ8−Δ(5)T(D) = T_\infty+ A \xi_8^{-\Delta} \tag*{(5)}

to the states in the larger-DD half of each series (at least four), with Δ=2/3\Delta= 2/3 unless stated otherwise. In the gapless phase this form describes the leading finite-DD behavior, with T∞=0T_\infty= 0 up to logarithmic corrections near the Sutherland point (§ 2.4); in the trimerized phase it is an interpolation between the critical behavior at short scales and the finite limit at long scales. Since ξ8\xi_8 stays finite, T∞T_\infty is not a measurement of the order parameter at D=∞D = \infty but a comparative criterion: we compare it with the values obtained in the same way at gapless reference points (α={0,5,6}\alpha= \{0, 5, 6\}, as suggested by field-theory arguments), which measure the bias of the extrapolation and test its dependence on Δ\Delta by varying the exponent between 1/21/2 and 11.

Convergence and bond dimensions

Of the 315 states used in this work, 278 reach δ<10−9\delta< 10^{-9} and 295 reach δ<10−8\delta< 10^{-8}. States at α=2.75\alpha= 2.75–1010 and D≳3500D \gtrsim3500 that reached the iteration limit were continued at fixed DD for up to 1500 further iterations; this changed the energy by less than 10−910^{-9} per site, SS by at most 1.2×10−51.2 \times10^{-5} and ξ8\xi_8 by at most 0.01 sites. We reach D=4000D = 4000 for α=0.45\alpha= 0.45–0.500.50, D=6000D = 6000 for α=0\alpha= 0, 11–2.752.75, 66–1010 and 2020–3232, and D≈8000D \approx8000 for α=3.0\alpha= 3.0–4.04.0 and 55; other couplings are taken to D≈1000D \approx1000–20002000.

Benchmark. At α=0\alpha= 0 and D=5982D = 5982, e0=−0.7032119e_0 = -0.7032119, 3×10−5%3 \times10^{-5}\% above the exact value. ξ8\xi_8 grows as D0.78D^{0.78} (prediction for c=2c = 2: κ=0.87\kappa= 0.87) and ceff=1.92c_{\rm eff} = 1.92, 1.951.95, 1.951.95, 1.981.98 in the windows D∈[200,500)D \in[200, 500), [500,1000)[500, 1000), [1000,2000)[1000, 2000), [2000,4000)[2000, 4000), approaching c=2c = 2 from below. Both the reduced exponent and the slow approach of ceffc_{\rm eff} to c=2c = 2 are expected from the logarithmic corrections generated by the marginally irrelevant coupling of the Sutherland chain [10, 31, 32, 33].

Overview of the phase diagram

Figure 2 summarizes the phase diagram. At small α\alpha the chain is gapless, as at the Sutherland point; above a lower transition αc1\alpha_{c1} it is trimerized, a threefold-degenerate period-3 valence-bond solid whose order parameter is largest near α≈1.25\alpha\approx1.25; above an upper boundary αc2\alpha_{c2} it is gapless again, and at large α\alpha the effective central charge follows the crossover from two decoupled chains towards SU(3)1_1. The two boundaries are determined in § 2.4 and § 2.5, and the large-α\alpha regime is discussed in Sec. VI. Here, we describe the ground state in the three regions; the trimerization, the largest Schmidt weight, and the inverse octet correlation length are shown versus α\alpha for several bond dimensions in Figure 8.

Figure 2. Phase diagram along α=J2/J1\alpha= J_2/J_1 (linear scale at α<1\alpha< 1 and logarithmic above). Shaded: gapless SU(3)1_1 phase (α<αc1=0.48\alpha< \alpha_{c1} = 0.48), trimerized phase, and gapless phase at large α\alpha, where the SU(3)2→_2 \to SU(3)1_1 crossover is expected; grey band: α=3.5\alpha= 3.5–55, where the upper boundary lies but is not resolved (αc2≳3.5\alpha_{c2} \gtrsim3.5, § 2.5). Red: trimerization extrapolated to D→∞D \to\infty. Blue: effective central charge at the largest bond dimension; dashed and dotted red lines mark c=2c = 2 and c=16/5c = 16/5.

Figure 8. SU(3)-symmetric VUMPS data at bond dimensions DD matched to the χ\chi of Ref. [10] (J1=1J_1=1, so J2=αJ_2=\alpha). (a) Trimerization TT (cf. Fig. 3 of Ref. [10]), (b) largest eigenvalue of the half-chain reduced density matrix (cut between trimers; cf. Fig. 4, top), (c) inverse octet correlation length (cf. Fig. 5), (d) entanglement entropy versus ξ8\xi_8 with fits S=(c/6)ln⁡ξ8+bS=(c/6)\ln\xi_8+b over all DD (cf. Fig. 7). The grey band marks the range J2=2.4J_2=2.4–3.83.8 in which the iDMRG data of Ref. [10] jump (χ=256\chi=256 to 81928192). No jump occurs in the SU(3)-symmetric data.

Energy. The ground-state energy per site (Figure 3) is smooth over the whole range. Since e0=EJ1+αEJ2e_0 = E_{J_1} + \alpha E_{J_2} with EJ2=⟨P^i,i+2⟩E_{J_2} = \langle\hat{P}_{i,i+2}\rangle, the Hellmann–Feynman theorem gives de0/dα=EJ2de_0/d\alpha= E_{J_2}; numerical derivatives of e0e_0 agree with the measured EJ2E_{J_2} to 1.5×10−51.5 \times10^{-5} (median deviation at D≈1000D \approx1000; with states up to D=4000D = 4000 near αc1\alpha_{c1}, to 7×10−67 \times10^{-6}). The energy has a shallow maximum at α≈0.22\alpha\approx0.22–0.250.25, where EJ2E_{J_2} changes sign; this maximum is unrelated to the transition at αc1\alpha_{c1}.

Figure 3. Ground-state energy per site. Dotted line: exact value at α=0\alpha= 0 [7]. Inset: de0/dαde_0/d\alpha from finite differences (points) and the measured energy of the J2J_2 bonds (line), equal by the Hellmann–Feynman theorem; the dashed line marks the maximum of e0e_0, where EJ2E_{J_2} changes sign. States at D≈1000D \approx1000 (inset) and the largest DD (main panel).

Bond energies. Figure 4 shows all bond energies of the six-site cell at D≈4000D \approx4000. At α=0\alpha= 0 the J1J_1 bonds are uniform within 0.6%0.6\%, a finite-DD residual of the translation-symmetry breaking discussed in § 2 C. At α=1\alpha= 1 the trimers are visible directly: two strong J1J_1 bonds (⟨P^⟩=−0.69\langle\hat{P}\rangle= -0.69) and one strong J2J_2 bond (−0.60-0.60) form a triangle, separated from the next one by weak bonds (−0.15-0.15 and −0.13-0.13). At α=3.5\alpha= 3.5 and 55 the energy resides in the legs (⟨P^⟩≈−0.66\langle\hat{P} \rangle\approx-0.66 to −0.71-0.71) and the J1J_1 bonds are frustrated (⟨P^⟩>0\langle\hat{P} \rangle> 0); a period-3 modulation remains in both bond types and decreases with α\alpha.

Figure 4. Bond energies ⟨P^ij⟩\langle\hat{P}_{ij} \rangle over two unit cells at D≈4000D \approx4000. Line width is proportional to ∣⟨P^ij⟩∣|\langle\hat{P}_{ij} \rangle|; blue: negative, red: positive. Numbers give the bond energies of the first unit cell; the pattern repeats in the second.

Entanglement spectrum. Resolved by SU(3) irrep (Appendix D), the entanglement spectrum shows the structure of the trimerized phase directly: at α=1\alpha= 1 the t=0t = 0 cut, which separates two trimers, is dominated by a single singlet level (λmax⁡=0.75\lambda_{\max} = 0.75), so that the state is close to a product of SU(3) singlets on three sites. The singlet–octet entanglement gap on this cut varies smoothly with α\alpha and shows no jump at any DD. The singlet carries the largest Schmidt value on the t=0t = 0 cut for all α≤10\alpha\le10.

The lower transition

At αc1\alpha_{c1} a marginal current–current coupling changes sign [10]; the gap opens as exp⁡[−const⁡/(α−αc1)]\exp[-\operatorname{const}/(\alpha-\alpha_{c1})], and just above the transition the correlation length exceeds any length accessible to finite-entanglement methods. Indeed, at D=4000D = 4000 the states at α=0.45\alpha= 0.45–0.500.50 have ξ8≈380\xi_8 \approx380–410410 sites, still growing as D0.75D^{0.75}–D0.79D^{0.79}, and ceff=1.98c_{\mathrm{eff}} = 1.98 for 2000≤D<40002000 \le D < 4000, which cannot be distinguished from a gapless state by ξ8\xi_8 or SS alone.

The extrapolated trimerization T∞T_\infty resolves the transition once the bond dimension is large enough (Figure 5). With D≤1100D \le1100 the extrapolation is biased: T∞T_\infty is negative for all α≤0.50\alpha\le0.50 and at the Sutherland point (−0.0016-0.0016). The bias comes from the logarithmic corrections generated by the marginal coupling, which the power law ξ8−2/3\xi_8^{-2/3} does not capture; it decreases with DD (at α=0\alpha= 0: −0.0016-0.0016, −0.0010-0.0010, −0.0008-0.0008 for D≤1100D \le1100, 2200, 4100), and it is smallest close to αc1\alpha_{c1}, where the marginal coupling vanishes. Using states up to D=4000D = 4000, T∞T_\infty is zero within 3×10−43 \times10^{-4} for α=0.45\alpha= 0.45–0.470.47 and positive and increasing from α=0.48\alpha= 0.48 on: T∞=0.0003(1)T_\infty= 0.0003(1), 0.0006(2), 0.0010(2) and 0.0065(7) at α=0.48\alpha= 0.48, 0.49, 0.50 and 0.55. Here the exponent is not only assumed: fits with a free exponent give 0.63–0.67 at α=0.45\alpha= 0.45–0.50, consistent with Δ=2/3\Delta= 2/3. This gives αc1≤0.48\alpha_{c1} \le0.48, in agreement with the level-crossing analysis of ED spectra, αc1≈0.48\alpha_{c1} \approx0.48 [10], which locates the sign change of the marginal coupling directly. Because the order parameter is exponentially small just above αc1\alpha_{c1}, the data cannot exclude a transition slightly below 0.48; we adopt αc1=0.48\alpha_{c1} = 0.48.

Extrapolated trimerization near the lower transition

Figure 5. Extrapolated trimerization T∞T_\infty near the lower transition, from states with 200≤D≤Dcut200 \le D \le D_{\mathrm{cut}}. Dotted lines: T∞T_\infty at the gapless Sutherland point α=0\alpha= 0 for the same DcutD_{\mathrm{cut}}, a measure of the bias of the extrapolation. Dashed line: αc1=0.48\alpha_{c1} = 0.48 from ED level crossings [10].

A fit of the essential-singularity form expected for this transition does not improve on this (Appendix A). With SU(NN) symmetry the gap opens as exp⁡[−b/(α−αc)]\exp[-b/(\alpha-\alpha_c)], so that T∞=Aexp⁡[−b′/(α−αc)]T_\infty= A\exp[-b^\prime/(\alpha-\alpha_c)]; fitted to T∞(α)T_\infty(\alpha), this form gives αc=0.40\alpha_c = 0.40 (statistical error ±0.007\pm0.007), well below αc1≈0.48\alpha_{c1} \approx0.48, and αc\alpha_c moves to 0.40–0.49 when a constant offset is added to absorb the bias of the extrapolation.

Upper boundary of the trimerized phase

We locate αc2\alpha_{c2} from the bond-dimension dependence of the effective central charge and of the extrapolated trimerization, calibrated on two reference points (Figure 6a). At the Sutherland point ceff=1.96c_{\mathrm{eff}} = 1.96–1.98 for all DD from 2000 to 6000, as for a gapless state with c=2c = 2. Deep in the trimerized phase, at α=1\alpha= 1, ceffc_{\mathrm{eff}} falls to 0.01 and ξ8\xi_8 grows only slowly, to 35 sites at D=6000D = 6000. A gapped state whose correlation exceeds ξ8(D)\xi_8(D) shows

Figure 6. Effective central charge versus bond dimension: slope of SS versus ln⁡ξ8\ln\xi_8 over the three most recent states up to DD, plotted at DD. Where the main run of a coupling starts above D≈1600D \approx1600, the states below its first DD come from a second run started at D≈1400D \approx1400; at the junction both runs give the same state within the difference in DD. (a) Across the upper boundary, with the gapless (α=0\alpha= 0) and gapped (α=1\alpha= 1) references. (b) Large α\alpha, including α=20\alpha= 20, 26 and 32; lines: c=4c = 4 (two decoupled chains), 16/516/5 (SU(3)2_2) and 2 (SU(3)1_1). ceffc_{\mathrm{eff}} probes the renormalization-group trajectory c=4→16/5→2c = 4 \to16/5 \to2 at the length scale ξ8(D)\xi_8(D); since the crossover lengths grow with α\alpha, at fixed DD it increases with α\alpha, and the values in (b) are not a plateau at 16/516/5 (see Sec. VI).

ceffc_{\mathrm{eff}} decreasing with DD, through and below 2; a gapless state in the SU(3)1_1 phase approaches c=2c = 2, and in the crossover regime discussed in Sec. VI stays above it.

For α=2\alpha= 2 to 3.5, ceffc_{\mathrm{eff}} lies below 2 and keeps decreasing up to the largest bond dimensions: ceff=0.63c_{\mathrm{eff}} = 0.63 (α=2\alpha= 2), 1.41 (2.75, both at D=6000D = 6000), 1.45 (3.0), 1.56 (3.25) and 1.68 (3.5, all at D≈8000D \approx8000). At α=3.75\alpha= 3.75 and 4.0 it is 2.24 and 2.35 (D≈8000D \approx8000), decreasing slowly. The curves for α=3.0\alpha= 3.0–3.53.5 have kinks, where ceffc_{\mathrm{eff}} drops by up to 0.5 between consecutive points. They do not come from the entropy: the two leading eigenvalues of the octet sector of the transfer matrix are nearly degenerate (∣t8,2/t8,1∣>0.97|t_{8,2}/t_{8,1}| > 0.97 for D≳1400D \gtrsim1400) and cross at D≈2400D \approx2400, 4000 and 5000–6000 for α=3.0\alpha= 3.0, 3.25 and 3.5, where ξ8\xi_8 switches to the faster-growing branch. With the length ξpair=−6/ln⁡∣t8,1t8,2∣\xi_{\mathrm{pair}} = -6/\ln\sqrt{|t_{8,1}t_{8,2}|}, which is smooth through the crossings, ceffc_{\mathrm{eff}} decreases smoothly with DD and reaches 1.61, 1.66, 1.78 and 1.91 at α=2.75\alpha= 2.75, 3.0, 3.25 and 3.5 at the largest DD, against 1.97 at α=0\alpha= 0; with ξ1\xi_1 the values are lower still. Below 2 for every choice of the length, ceffc_{\mathrm{eff}} thus indicates a gap up to α=3.5\alpha= 3.5. For α≥3.75\alpha\ge3.75 the outcome depends on the length used (2.01 and 2.09 at α=3.75\alpha= 3.75 and 4.0 with ξpair\xi_{\mathrm{pair}}, below 2 with ξ1\xi_1), so that ceffc_{\mathrm{eff}} does not locate the boundary there. Quantities that use no correlation length, the growth rate dS/dln⁡DdS/d\ln D of the entropy and the decay rate −dln⁡T/dln⁡D-d\ln T/d\ln D of the trimerization, decrease with DD at every α=2.75\alpha= 2.75–10, by 15–40% over our range of DD, and change smoothly with α\alpha; only at α=2\alpha= 2 do they approach zero (0.04 and 0.03 at D=6000D = 6000).

The extrapolated trimerization (Figure 7) gives the same picture: T∞=0.030T_\infty= 0.030, 0.022 and 0.019 at α=2.75\alpha= 2.75, 3.0 and 3.25, 0.006 at both α=3.5\alpha= 3.5 and 3.75, and T∞=0.0013T_\infty= 0.0013 at α=4.0\alpha= 4.0, compatible with zero given the bias of the extrapolation (the same fit gives T∞=−0.002T_\infty= -0.002 at α=5\alpha= 5, in the gapless phase). Varying the number of states in the slope from three to five changes ceffc_{\mathrm{eff}} by at most 0.14 (0.27 at α=3.5\alpha= 3.5), without moving any value across c=2c = 2.

Figure 7. Trimerization extrapolated to D→∞D \to\infty, T∞T_\infty (points, exponent 2/3), the range of T∞T_\infty for exponents between 1/2 and 1 (band), and TT at the largest DD (line); inset: the region near αc2\alpha_{c2}. Open points: couplings with Dmax⁡<3000D_{\max} < 3000 near the upper boundary (α=3.4\alpha= 3.4, 3.6), not included in the band. Dashed: αc1=0.48\alpha_{c1} = 0.48; grey band: α=3.5\alpha= 3.5–5, where the upper boundary lies but is not resolved.

Because Eq. (5) with Δ=2/3\Delta= 2/3 describes only the gapless SU(3)1_1 phase (§ 2), Figure 7 also shows T∞T_\infty for exponents between 1/2 and 1 (band). Deep in the phase T∞T_\infty hardly depends on the exponent (0.351–0.354 at α=1\alpha= 1), and fits with a free exponent give 1.0–1.6 for α=2\alpha= 2–3.5 and larger values of T∞T_\infty, so that Δ=2/3\Delta= 2/3 is the conservative choice there. At the gapless reference points α=0\alpha= 0, 5 and 6 the extrapolation gives values between −0.011 and +0.006 over this range of exponents, which sets the resolution of the criterion T∞>0T_\infty> 0. For every exponent in the range, T∞T_\infty at α=3.5\alpha= 3.5 lies above the values at α=5\alpha= 5 and 6 (e.g. 0.016 against 0.006 and 0.003 for exponent 1), whereas at α=4.0\alpha= 4.0 it lies only slightly above them. Using the singlet correlation length instead of ξ8\xi_8 in the extrapolation shifts T∞T_\infty upward by 0.003–0.007 at all α=2.75\alpha= 2.75–6, including the gapless reference points, and leaves this comparison unchanged.

We conclude that the trimerized phase extends at least to α=3.5\alpha= 3.5: T∞T_\infty lies clearly above the gapless reference values up to α=3.25\alpha= 3.25 (0.02–0.03) and above them for every exponent at α=3.5\alpha= 3.5, and ceffc_{\mathrm{eff}} is below 2 for every choice of the correlation length up to α=3.5\alpha= 3.5. At α≥5\alpha\ge5, the data behave as at the gapless reference points; where the trimerized phase ends between α=3.5\alpha= 3.5 and 5 is not resolved at the bond dimensions reached here.

Large : the SU(3)2_2→SU(3)1_1 crossover

For SU(NN), c=k(N2−1)/(k+N)c = k(N^2 - 1)/(k + N): c=4c = 4 for two decoupled SU(3)2_2 chains, 16/5 for SU(3)2_2 and 2 for SU(3)1_1. Beyond αc2\alpha_{c2} the field theory predicts that the inter-leg coupling first gaps the Z3\mathbb{Z}_3 sector, leaving SU(3)2_2, which then flows without a gap to SU(3)1_1 [10, 11].

With SU(3) symmetry the effective central charge decreases from the two-chain value towards the SU(3)1_1 value (Figure 6b). For all α=4\alpha= 4–10 it lies below 16/5 at D≳3000D \gtrsim3000: 2.35 (α=4\alpha= 4) and 2.62 (5) at D≈8000D \approx8000, and 2.89–3.16 for α=6\alpha= 6–10 at D=6000D = 6000. At α=4\alpha= 4 and 5 it keeps decreasing with DD (at α=5\alpha= 5 from 2.81 at D≈4200D \approx4200 to 2.62 at D≈8000D \approx8000), and at every α=4\alpha= 4–10 the values from states with D≥3000D \ge3000 are lower than from states with D≤2000D \le2000.

The departure from 16/5 is seen first at the smallest α\alpha, as expected for the massless flow, whose crossover length grows as the inter-leg coupling J1/J2J_1/J_2 decreases.

The values in Figure 6b are therefore set by the accessible bond dimension rather than forming a plateau: ceffc_{\mathrm{eff}} measures the central charge at the length scale ξ8(D)\xi_8(D), and the crossover lengths grow with α\alpha, the Z3\mathbb{Z}_3 gap length as (J2/J1)3/2(J_2/J_1)^{3/2} [10]. At α=20\alpha= 20–32, where ξ8\xi_8 reaches 86–94 sites at D≈6000D \approx6000, ceffc_{\mathrm{eff}} stays at 3.4–3.6, between 16/5 and 4, and the crossover is not resolved. Reaching c=2c = 2 itself would require bond dimensions well above D≈8000D \approx8000 even at α=4\alpha= 4 and 5.

Comparison with previous work

Exact diagonalization

Our bound αc2≳3.5\alpha_{c2} \gtrsim3.5 excludes the ED estimate αc2=2.1(1)\alpha_{c2} = 2.1(1) [10]: at α=2\alpha= 2–2.5 all our diagnostics show a gapped, trimerized state, with ξ8=59\xi_8 = 59–71 sites at D=6000D = 6000 and T∞=0.12T_\infty= 0.12–0.05. Since the correlation length near the upper boundary exceeds 70 sites, rings of up to 27 sites used in ED are far from the thermodynamic limit there; moreover, level spectroscopy assumes that the transition is described by SU(3)1_1 with a marginal perturbation, which holds only beyond the crossover length of the SU(3)2_2 →SU(3)1_1 flow (Sec. VI). It is compatible with the position α≈3.8\alpha\approx3.8 of the jump in the largest-χ\chi iDMRG data [10], but is obtained independently of it.

iDMRG with Abelian symmetry

Figure 8 shows our data in the same form as the iDMRG data of Ref. [10], at bond dimensions matched to their χ\chi (both count states on a bond). For α≲2.4\alpha\lesssim2.4 the two calculations agree well: the inverse octet correlation length at α=0\alpha= 0 is 0.016, 0.0095, 0.0055 for D≈256D \approx256, 512, 1024 (Ref. [10]: 0.019, 0.010, 0.006), the trimerization peaks at T=0.37T = 0.37–0.385 near α=1.25\alpha= 1.25 (0.38), the largest reduced-density-matrix eigenvalue at α=1\alpha=1 is 0.770.77 (0.7450.745 at J2=1.2J_2=1.2), and the central charge at α={0.2,0.4}\alpha=\{0.2,0.4\} is {1.97,1.99}\{1.97,1.99\} (Ref. [10]: {1.99,1.98}\{1.99,1.98\}).

In the iDMRG data of Ref. [10], all these quantities jump at a coupling that depends on χ\chi, from J2≈2.4J_2\approx2.4 at χ=256\chi=256 to ≈3.8\approx3.8 at χ=8192\chi=8192, after which TT is zero. Our data show no such jump at any DD up to $8000 [Figure 8(a)–(c) show D≤6000D\leq6000; grey band: range of the jumps]. Because an injective SU(3)-symmetric MPS cannot become translation invariant, the absence of a jump in our data is not by itself evidence against a transition; the origin of the jumps is examined below and in Appendix B.

At large α\alpha, Ref. [10] found c≃3.26c\simeq3.26–3.333.33 for J2=4.6J_2=4.6–88 with U(1)×\timesU(1) symmetry and χ≤8192\chi\leq8192, close to the SU(3)2_2 value. With SU(3) symmetry and D≳3000D\gtrsim3000 we find ceffc_{\mathrm{eff}} below 16/516/5 for all α=4\alpha=4–1010 and decreasing with DD at α=4\alpha=4 and 55 (Sec. VI).

Origin of the iDMRG jumps

Could the jumps of Ref. [10] be a switch between two different SU(3)-symmetric states? Such states exist: on the t=0t=0 cuts the largest Schmidt value can be carried by the singlet (“singlet type”) or by the octet (“octet type”), and in our scans the dominant irrep changes from the singlet to the octet near α≈22\alpha\approx22. We therefore optimized both types at the same α\alpha and DD (α=2.5\alpha=2.5–3030, D≈225D\approx225, 490490 and 970970), starting from a state of each type, and compared their energies. Where the iDMRG data jump, the singlet type is clearly the ground state: for α≤10\alpha\le10 it is lower at every DD, by 10−310^{-3}–10−210^{-2} per site, far above the convergence error. The two types become nearly degenerate only for α≥15\alpha\ge15, where their energy difference is ≲10−3\lesssim10^{-3} per site at D≈490D \approx490 and 970 and changes sign with DD. A switch between SU(3)-symmetric states therefore cannot produce the jumps, and the change of the dominant irrep near α≈22\alpha\approx22 is a crossing of two nearly degenerate approximations, not a phase transition.

Since the position of the iDMRG jumps moves with χ\chi, it is not a measurement of a phase boundary; the boundaries determined in § 2.4 and § 2.5 from the DD-scaling of ξ8\xi_8, SS and TT do not rely on it. Finally, we checked directly that the jumps are reproduced when only Abelian color conservation is imposed, using an independent infinite-DMRG code (Appendix B). At χ=128\chi= 128 its energy and entanglement entropy agree with those of the SU(3)-symmetric state with the same number of states up to J2=3.15J_2 = 3.15 (within 1.1×10−41.1 \times10^{-4} per site and 0.01), and the entropy drops between J2=3.15J_2 = 3.15 and 3.3; at J2=3.5J_2 = 3.5 the Abelian state is less entangled (S=2.074S = 2.074 against 2.167) and lower in energy by 2.5×10−32.5 \times10^{-3} per site.

We also test the symmetry from the Schmidt values and the color charges of the Schmidt states: in an SU(3)-symmetric state the Schmidt values form degenerate multiplets, and the summed Schmidt weight at each color weight is invariant under permutations of the three colors. Before the jump, the Abelian states are close in energy but already partially breaking SU(3)-symmetry—the effect decreases with χ\chi: at J2=2.0J_2 = 2.0 the eight Schmidt weights following the dominant singlet, which carry the color charges of an octet, have a relative spread (max−min)/mean of 0.41, 0.14 and 0.02 at χ=32\chi= 32, 64 and 128, and at χ=128\chi= 128 the color-permutation asymmetry of the Schmidt weights is at most 0.022 for J2≤3.15J_2 \le3.15. After the jump, at J2=3.5J_2 = 3.5 and χ=128\chi= 128, the asymmetry is 0.37 and the octet weights spread by twice their mean (Table I): the state breaks SU(3). The same comparison at χ=64\chi= 64 shows the motion of the jump: at J2=3.0J_2 = 3.0 and 3.15 the χ=64\chi= 64 states lie (1.9(1.9–2.3)×10−32.3) \times10^{-3} below the SU(3)-symmetric ones, with strongly split octet weights, whereas at χ=128\chi= 128 the same couplings are again close in energy to the SU(3)-symmetric branch. The broken state is not the translation-invariant superposition of three shifted SU(3)-symmetric states either: each component would carry only about χ/3\chi/3 states, giving an energy close to that of an SU(3)-symmetric state with D≈χ/3D \approx\chi/3, well above the SU(3)-symmetric state with D≈χD \approx\chi, and an entropy larger by ln⁡3\ln3. A continuous symmetry with a non-conserved order parameter cannot be broken spontaneously in the ground state of a one-dimensional chain [34], and the advantage of the broken state is indeed confined to small bond dimension: at J2=3.5J_2 = 3.5 its energy at χ=128\chi= 128, −2.2092-2.2092, lies above that of the SU(3)-symmetric state at D=7998D = 7998, −2.2167-2.2167. The jump is thus a crossing, at fixed bond dimension, to an SU(3)-breaking state that is favorable only at small χ\chi, and its position moves to larger J2J_2 with increasing χ\chi, as in Ref. [10].

Conclusions

Using uniform matrix product states with exact SU(3) symmetry at bond dimensions up to D≈8000D \approx8000, we have determined the phase diagram of the SU(3) J1J_1–J2J_2 Heisenberg chain (Figure 2). A gapless SU(3)1_1 phase at small α\alpha is followed, from αc1≈0.48\alpha_{c1} \approx0.48, by a threefold-degenerate trimerized phase, whose order parameter and correlation length are largest and shortest, respectively, near α≈1.25\alpha\approx1.25 (T∞=0.36T_\infty= 0.36, ξ8≈29\xi_8 \approx29 sites). The trimerized gapless state extends at least to αc2≳3.5\alpha_{c2} \gtrsim3.5, and at α≥5\alpha\ge5 the chain is gapless; where between the two it ends is not resolved at the bond dimensions reached here. This resolves the disagreement between exact diagonalization (αc2=2.1\alpha_{c2} = 2.1) and iDMRG (α≈3.8\alpha\approx3.8) in favor of the larger value, which we obtain from the bond-dimension scaling of the entanglement and of the trimerization rather than from the position of a jump.

The jumps found in iDMRG with U(1)×\timesU(1) symmetry are crossings, at fixed bond dimension, to SU(3)-breaking states that are favorable only at small bond dimension. An injective SU(3)-symmetric MPS cannot be invariant under one-site translation (a translation-invariant one exists only as a non-clustering superposition of three translates), so finite-DD SU(3)-symmetric states always carry some trimerization and show no such jumps. Phase boundaries in SU(NN) chains with the fundamental representation on each site should therefore be located by finite-entanglement scaling, as done here, or by spectroscopy that tracks the sign change of the marginal coupling directly.

At large α\alpha the effective central charge lies below the SU(3)2_2 value 16/516/5 for all α=4\alpha= 4–10 and decreases with bond dimension where the crossover length is shortest, as expected for the massless flow to SU(3)1_1.

Three questions remain open. Reaching the SU(3)1_1 value c=2c = 2 requires correlation lengths beyond the crossover length of the SU(3)2→_2 \toSU(3)1_1 flow, i.e. bond dimensions well above those used here, the more so at larger α\alpha. The position of αc2\alpha_{c2} above 3.5 and a possible transition slightly below αc1=0.48\alpha_{c1} = 0.48 are limited by the exponentially slow opening of the gap at both boundaries; a level-spectroscopy analysis of the transfer matrix or of excitations resolved by irrep would determine both boundaries independently of finite-entanglement scaling.

Acknowledgments

The authors acknowledge support from the Simons Foundation in the framework of the call “Centers of Excellence in Ukraine”, grant No. SFI-PD-COE-00032628.

Essential-singularity fits

Figure 9 shows fits of T∞(α)T_{\infty}(\alpha) on the window α∈[0.47,1.0]\alpha\in[0.47, 1.0]. Here T∞T_{\infty} is extrapolated from all available states of each coupling with D≥282D \ge282: up to D≈4000D \approx4000 at α=0.47\alpha= 0.47–0.500.50, D≈1000D \approx1000–14001400 at α=0.475\alpha= 0.475 and 0.550.55–0.750.75, and D≈6000D \approx6000 at α=1\alpha= 1. The window is fitted with the essential-singularity form expected for the lower transition. At αc1\alpha_{c1}, a marginal current–current coupling g∝α−αcg \propto\alpha- \alpha_c changes sign [10]. With SU(NN) symmetry, it is the only coupling that flows dg/dl∝g2dg/dl \propto g^2, so that strong coupling is reached after l∗∝1/(α−αc)l^* \propto1/(\alpha-\alpha_c) and ξ∝el∗=exp⁡[b/(α−αc)]\xi\propto e^{l^*} = \exp[b/(\alpha-\alpha_c)]. Since T∞∝ξ−2/3T_{\infty} \propto\xi^{-2/3},

Figure 9. Fit of the SU(NN) essential-singularity form, Eq. (6), to the extrapolated order parameter, using the singlet (left) and octet (right) correlation length in the extrapolation. The vertical line marks α=0.48\alpha= 0.48. The plotted quantity is σ∞=(3/2)T∞\sigma_{\infty} = (\sqrt{3}/2)T_{\infty}.

T∞=Ae−b′/(α−αc)(6)T_{\infty} = A e^{-b'/(\alpha-\alpha_c)} \tag*{(6)}

for α>αc\alpha> \alpha_c. With T∞T_{\infty} extrapolated using the singlet or the octet correlation length, the fit gives αc=0.400\alpha_c = 0.400 and 0.3960.396 (±0.007\pm0.007 and ±0.005\pm0.005), with χ2/ν=15\chi^2/\nu= 15 and 2828; on the narrower windows α∈[0.50,1.0]\alpha\in[0.50, 1.0] and [0.55,1.0][0.55, 1.0] it gives αc=0.399\alpha_c = 0.399–0.4060.406, with χ2/ν\chi^2/\nu up to 58. Adding a constant offset as a free parameter (to absorb the bias discussed in § 2.4) moves αc\alpha_c to 0.400.40–0.490.49. The location of the transition cannot be obtained in this way at reachable bond dimensions.

Abelian iDMRG check

The Abelian calculations use an independent infinite-DMRG code (ITensors [35]) with the same six-site cell and Hamiltonian convention as the VUMPS calculations, and final bond dimensions χ=64\chi= 64 and 128128. The code conserves the numbers of particles of two of the three colors modulo 6, i.e. a Z6×Z6\mathbb{Z}_6 \times\mathbb{Z}_6 subgroup of the U(1)×U(1)\mathrm{U}(1) \times\mathrm{U}(1) color symmetry used in Ref. [10], which keeps the number of charge sectors on the bonds of the infinite chain bounded; like Ref. [10], it does not impose SU(3). Its local bond

J2J_2χ\chiΔE\Delta EAAσ8\sigma_8
2.032—0.0090.41
2.064−3.5×10−4-3.5 \times10^{-4}0.0150.14
2.0128+2.8×10−5+2.8 \times10^{-5}0.0020.02
3.032—0.2862.43
3.064−1.9×10−3-1.9 \times10^{-3}0.0340.80
3.0128−3.0×10−5-3.0 \times10^{-5}0.0180.48
3.1532—0.2852.43
3.1564−2.3×10−3-2.3 \times10^{-3}0.0390.87
3.15128−1.1×10−4-1.1 \times10^{-4}0.0220.60
3.532—0.2812.44
3.564−3.6×10−3-3.6 \times10^{-3}0.0470.99
3.5128−2.5×10−3-2.5 \times10^{-3}0.3711.98

Table 1. Abelian iDMRG states compared with SU(3)-symmetric VUMPS states with the same number of states: energy difference per site ΔE=EAb−ESU(3)\Delta E = E_{\mathrm{Ab}} - E_{\mathrm{SU(3)}} (no SU(3)-symmetric state with D<46D < 46, so none at χ=32\chi= 32), color-permutation asymmetry AA of the Schmidt weights, and spread σ8\sigma_8 of the octet weights. Both AA and σ8\sigma_8 vanish for an SU(3)-symmetric state. The jump at χ=128\chi= 128 lies between J2=3.15J_2 = 3.15 and 3.3.

energies are evaluated on a finite block with approximate environments and do not show the period-3 structure of the infinite state; we therefore compare only the energy per site and the entanglement entropy, which the code reports for one cut. The SU(3)-symmetric reference states for this comparison were computed at small bond dimension (D=46,63,90,123,160D = 46, 63, 90, 123, 160) by two-site growth and VUMPS, as in the main calculations; their energies and least-entangled-cut entropies are interpolated linearly in ln⁡D\ln D to D=χD = \chi. At χ=128\chi= 128 the Abelian and the SU(3)-symmetric energies agree within 1.1×10−41.1 \times10^{-4} per site for J2=2.0J_2 = 2.0–3.153.15, and the entropies within 0.01. The entropy of the Abelian state rises smoothly up to J2=3.15J_2 = 3.15 (S=2.116S = 2.116) and drops to 2.077 at J2=3.3J_2 = 3.3 and 2.074 at 3.5, where the SU(3)-symmetric state has $S = 2.167 (Figure 10).

Figure 10. (a) Entanglement entropy versus J2J_2 (J1=1J_1 = 1, so J2=αJ_2 = \alpha): SU(3)-symmetric VUMPS at D≈128D \approx128 (least-entangled cut) and Abelian iDMRG (color numbers conserved modulo 6) at χ=64\chi= 64 and 128. (b) Energy difference per site between the Abelian state at χ=128\chi= 128 and the SU(3)-symmetric state with the nearest bond dimension (D=129D = 129–141).

Figure 11. Entanglement spectra resolved by SU(3) irrep on the two inequivalent cuts (triality t=0,1t = 0, 1; the t=2t = 2 cut is the conjugate of t=1t = 1), at the Sutherland point (α=0\alpha= 0), in the trimerized phase (α=1\alpha= 1), near the upper boundary (α=3.5\alpha= 3.5) and at large α\alpha (α=5\alpha= 5). Each level of irrep λ\lambda is dim⁡λ\dim\lambda-fold degenerate; 15′15' denotes the irrep with Dynkin labels (4,0)(4, 0).

To test the symmetry of the Abelian states we record, at the end of each χ\chi stage, the Schmidt values sks_k of the central cut together with the color charges (N1,N2)(N_1, N_2) mod 6 of the Schmidt states. Taking the charges relative to the largest Schmidt state, with ΔN3=−ΔN1−ΔN2\Delta N_3 = -\Delta N_1-\Delta N_2, we form the summed weight P(w)P(w) of the Schmidt states with color weight w=(ΔN1,ΔN2,ΔN3)w = (\Delta N_1, \Delta N_2, \Delta N_3) and the asymmetry

A=max⁡π∈S312∑w∣P(w)−P(πw)∣,(7)A = \max_{\pi\in S_3} \frac{1}{2}\sum_w \lvert P(w) - P(\pi w)\rvert, \tag*{(7)}

which vanishes for an SU(3)-symmetric state. The test is necessary but not sufficient, since a splitting inside a multiplet that keeps the color-permutation symmetry leaves A=0A = 0; we therefore also record the relative spread σ8=(max⁡−min⁡)/mean\sigma_8 = (\max- \min)/\mathrm{mean} of the eight weights s22,…,s92s_2^2,\ldots,s_9^2, which in an SU(3)-symmetric state form the octet following the singlet. Table I lists both together with the energy difference. Before the jump both measures decrease with χ\chi; at J2=3.5J_2 = 3.5 and χ=128\chi= 128 the state is strongly SU(3)-breaking. Repeated runs of the code at χ=64\chi= 64 reach different broken states, with energies differing by up to $5.5 \times 10^{-3}persite,whiletheper site, while the\chi = 128 results are reproduced exactly; the table uses one set of runs throughout.

No one-site translation invariance of injective SU(3)-symmetric MPS

Figure 12. Singlet–octet entanglement gap ϵ8−ϵ1\epsilon_8-\epsilon_1 on the t=0t=0 cut, at bond dimensions matched to Ref. [10] and at the largest DD. Dashed line: αc1=0.48\alpha_{c1}=0.48; grey band: α=3.5\alpha=3.5–5, where the upper boundary lies (§ 2.5).

Consider a uniform MPS with tensors AsA^s, s=1,2,3s=1,2,3, acting on a virtual space VV of dimension DD, and its transfer map E(X)=∑sAsXAs†\mathcal{E}(X)=\sum_s A^s X A^{s\dagger}. The MPS is injective (normal) if, after normalization, 11 is the only eigenvalue of E\mathcal{E} of modulus one, it is non-degenerate, and its eigenvector ρ\rho is positive definite; equivalently, connected correlations decay exponentially and the state is the unique ground state of a local parent Hamiltonian. We show that such an MPS cannot be invariant under SU(3) acting in the fundamental representation u(g)u(g) on every site.

(i) If an injective MPS is invariant under u(g)⊗Nu(g)^{\otimes N} for all NN, there are a unitary representation VgV_g on VV, defined up to phases, and a one-dimensional representation eiθge^{i\theta_g} such that ∑s′u(g)ss′As′=eiθgVgAsVg†\sum_{s'}u(g)_{ss'}A^{s'}=e^{i\theta_g}V_gA^sV_g^\dagger [22, 23, 24, 36].

(ii) SU(3) has no non-trivial one-dimensional representations, so eiθg=1e^{i\theta_g}=1; it is simply connected, so VgV_g can be chosen as a linear representation and decomposes into irreps. By Schur’s lemma the center element z=ωIz=\omega I, ω=e2πi/3\omega=e^{2\pi i/3}, acts on an irrep of triality tt as ωt\omega^t, so that Vz=Z≡⨁tωtIVtV_z=Z\equiv\bigoplus_t\omega^t I_{V_t}, where VtV_t collects the irreps of triality tt.

(iii) In the fundamental representation u(z)=ωu(z)=\omega. Hence ZAsZ−1=ωAsZA^sZ^{-1}=\omega A^s: AsA^s maps VtV_t to Vt+1V_{t+1}.

(iv) Using AsZ=ω−1ZAsA^sZ=\omega^{-1}ZA^s, E(Zρ)=∑sAsZρAs†=ω−1ZE(ρ)=ω−1Zρ\mathcal{E}(Z\rho)=\sum_s A^sZ\rho A^{s\dagger}=\omega^{-1}Z\mathcal{E}(\rho)=\omega^{-1}Z\rho. Since ZZ is invertible and ρ\rho is positive definite, Zρ≠0Z\rho\ne0; thus ω−1≠1\omega^{-1}\ne1 is a second eigenvalue of E\mathcal{E} of modulus one, which contradicts injectivity.

The same argument applied to the tensor of an LL-site cell, on which the center acts as ωL\omega^L, shows that an injective SU(3)-symmetric MPS requires L≡0(mod3)L\equiv0\pmod3; the six-site cell used here satisfies this. If the injectivity assumption is dropped, a one-site invariant SU(3)-symmetric MPS exists: V=V0⊕V1⊕V2V=V_0\oplus V_1\oplus V_2 with AsA^s cyclic between the blocks, for which E\mathcal{E} has the eigenvalues 11, ω\omega and ω2\omega^2 on the unit circle. In the thermodynamic limit this state is the equal-weight superposition of the three translates of a period-3 injective state; its correlations do not cluster. The argument is the SU(3) counterpart of the one for SU(2) with half-integer spin, where the center Z2\mathbb{Z}_2 plays the role of Z3\mathbb{Z}_3 [22, 23, 24].

Entanglement spectra

Figure 11 shows the entanglement spectra, ϵ=−ln⁡λ\epsilon=-\ln\lambda with λ\lambda the eigenvalues of the half-chain reduced density matrix, resolved by SU(3) irrep. At α=1\alpha=1 the t=0t=0 cut is dominated by a single singlet level (λmax⁡=0.75\lambda_{\max}=0.75), separated by Δϵ=3.7\Delta\epsilon=3.7 from the first octet, and the neighboring cuts by a single 33 or 3ˉ\bar{3}, as for a product of SU(3) singlets on three sites. At α=0\alpha=0, 3.53.5 and 55 the spectra are dense, with many irreps at low entanglement energy. Figure 12 shows the gap ϵ8−ϵ1\epsilon_8-\epsilon_1 between the lowest octet and the lowest singlet level on the t=0t=0 cut (cf. Figure 4, bottom, of Ref. [10]). It peaks at α≈1.25\alpha\approx1.25, where it is independent of DD, and decreases slowly with DD outside the trimerized phase. It shows no jump at any DD.

Code availability

The code used to obtain and analyze the data (based on MPSKit and TensorKit [16, 17]) will be made publicly available upon publication.

References

References

  1. [1]A. V. Gorshkov, M. Hermele, V. Gurarie, C. Xu, P. S. Julienne, J. Ye, P. Zoller, E. Demler, M. D. Lukin, and A. M. Rey, Two-orbital SU(N) magnetism with ultracold alkaline-earth atoms, Nat. Phys. 6, 289 (2010).DOI
  2. [2]M. A. Cazalilla and A. M. Rey, Ultracold Fermi gases with emergent SU(N) symmetry, Rep. Prog. Phys. 77, 124401 (2014).DOI
  3. [3]S. Taie, R. Yamazaki, S. Sugawa, and Y. Takahashi, An SU(6) Mott insulator of an atomic Fermi gas realized by large-spin Pomeranchuk cooling, Nat. Phys. 8, 825 (2012).DOI
  4. [4]G. Pagano, M. Mancini, G. Cappellini, P. Lombardi, F. Schäfer, H. Hu, X.-J. Liu, J. C. Catanio, C. Sias, M. Inguscio, and L. Fallani, A one-dimensional liquid of fermions with tunable spin, Nat. Phys. 10, 198 (2014).
  5. [5]X. Zhang, M. Bishof, S. L. Bromley, C. V. Kraus, M. S. Safronova, P. Zoller, A. M. Rey, and J. Ye, Spectroscopic observation of SU(N)-symmetric interactions in Sr orbital magnetism, Science 345, 1467 (2014).DOI
  6. [6]C. Hofrichter, L. Riegger, F. Scazza, M. Höfer, D. R. Fernandes, I. Bloch, and S. Fölling, Direct probing of the Mott crossover in the SU(N) Fermi-Hubbard model, Phys. Rev. X 6, 021030 (2016).DOI
  7. [7]B. Sutherland, Model for a multicomponent quantum system, Phys. Rev. B 12, 3795 (1975).
  8. [8]P. Corboz, A. M. Läuchli, K. Totsuka, and H. Tsunetsugu, Spontaneous dimerization in a bilinear-biquadratic S=1 zig-zag chain, Phys. Rev. B 76, 220404 (2007).DOI
  9. [9]S. Rachel, R. Thomale, M. Führinger, P. Schmitteckert, and M. Greiter, Spinon confinement and the Haldane gap in SU(n) spin chains, Phys. Rev. B 80, 180420(R) (2009), arXiv:0904.3882.DOI
  10. [10]L. Herviou, S. Capponi, and P. Lecheminant, Even-odd effects in the J1-J2 SU(N) Heisenberg spin chain, Phys. Rev. B 107, 205135 (2023), arXiv:2302.14090.DOI
  11. [11]P. Lecheminant, Massless renormalization group flow in SU(N)k perturbed conformal field theory, Nucl. Phys. B 901, 510 (2015), arXiv:1509.01680.DOI
  12. [12]P. Nataf and F. Mila, Exact diagonalization of Heisenberg SU(N) chains in the fully symmetric and antisymmetric representations, Phys. Rev. B 93, 155134 (2016), 1601.00922.DOI
  13. [13]P. Nataf, S. Gozel, and F. Mila, Edge states and universality class of the critical two-box symmetric SU(3) chain, Phys. Rev. B 104, L180411 (2021), arXiv:2107.09306.DOI
  14. [14]V. Zauner-Stauber, L. Vanderstraeten, M. T. Fishman, F. Verstraete, and J. Haegeman, Variational optimization algorithms for uniform matrix product states, Phys. Rev. B 97, 045145 (2018), 1701.07035.
  15. [15]L. Vanderstraeten, J. Haegeman, and F. Verstraete, Tangent-space methods for uniform matrix product states, SciPost Phys. Lect. Notes , 7 (2019), 1810.07006.DOI
  16. [16]L. Devos, M. Van Damme, and J. Haegeman, MPSKit.jl (2026), version 0.13.13.
  17. [17]L. Devos and J. Haegeman, TensorKit.jl: A Julia package for large-scale tensor computations, with a hint of category theory (2025), arXiv:2508.10076.arxiv.org/abs/2508.10076
  18. [18]SUNRepresentations.jl, https://github.com/QuantumKitHub/SUNRepresentations.jl (2026), accessed October 2026.
  19. [19]A. Alex, M. Kalus, A. Huckleberry, and J. von Delft, A numerical algorithm for the explicit calculation of SU(N) and SL(N,C) Clebsch-Gordan coefficients, J. Math. Phys. 52, 023507 (2011), arXiv:1009.0437.DOI
  20. [20]I. P. McCulloch and M. Gulácsi, The non-Abelian density matrix renormalization group algorithm, Europhys. Lett. 57, 852 (2002), cond-mat/0012319.
  21. [21]A. Weichselbaum, Non-abelian symmetries in tensor networks: A quantum symmetry space approach, Ann. Phys. 327, 2972 (2012), 1202.5664.DOI
  22. [22]E. Lieb, T. Schultz, and D. Mattis, Two soluble models of an antiferromagnetic chain, Ann. Phys. 16, 407 (1961).
  23. [23]I. Affleck and E. H. Lieb, A proof of part of Haldane’s conjecture on spin chains, Lett. Math. Phys. 12, 57 (1986).
  24. [24]M. Sanz, M. M. Wolf, D. Pérez-García, and J. I. Cirac, Matrix product states: Symmetries and two-body Hamiltonians, Phys. Rev. A 79, 042308 (2009), arXiv:0901.2223.DOI
  25. [25]F. Pollmann, S. Mukerjee, A. M. Turner, and J. E. Moore, Theory of finite-entanglement scaling at one-dimensional quantum critical points, Phys. Rev. Lett. 102, 255701 (2009).pubmed.ncbi.nlm.nih.gov/19659096
  26. [26]P. Calabrese and J. Cardy, Entanglement entropy and quantum field theory, J. Stat. Mech. , P06002 (2004).
  27. [27]L. Tagliacozzo, T. R. de Oliveira, S. Iblisdir, and J. I. Latorre, Scaling of entanglement support for matrix product states, Phys. Rev. B 78, 024410 (2008).DOI
  28. [28]B. Pirvu, G. Vidal, F. Verstraete, and L. Tagliacozzo, Matrix product states for critical spin chains: Finite-size versus finite-entanglement scaling, Phys. Rev. B 86, 075117 (2012).DOI
  29. [29]I. Affleck, Exact critical exponents for quantum spin chains, non-linear σ-models at θ = π and the quantum Hall effect, Nucl. Phys. B 265, 409 (1986).DOI
  30. [30]I. Affleck, Critical behaviour of SU(n) quantum chains and topological non-linear σ-models, Nucl. Phys. B 305, 582 (1988).DOI
  31. [31]I. Affleck, D. Gepner, H. J. Schulz, and T. Ziman, Critical behaviour of spin-s Heisenberg antiferromagnetic chains: analytic and numerical results, J. Phys. A: Math. Gen. 22, 511 (1989).
  32. [32]S. Eggert, Numerical evidence for multiplicative logarithmic corrections from marginal operators, Phys. Rev. B 54, R9612 (1996), arXiv:cond-mat/9602026.DOI
  33. [33]C. Itoi and M.-H. Kato, Extended massless phase and the Haldane phase in a spin-1 isotropic antiferromagnetic chain, Phys. Rev. B 55, 8295 (1997), arXiv:cond-mat/9605105.DOI
  34. [34]S. Coleman, There are no Goldstone bosons in two dimensions, Commun. Math. Phys. 31, 259 (1973).DOI
  35. [35]M. Fishman, S. R. White, and E. M. Stoudenmire, The ITensor software library for tensor network calculations, SciPost Phys. Codebases , 4 (2022).DOI
  36. [36]D. Pérez-García, M. M. Wolf, M. Sanz, F. Verstraete, and J. I. Cirac, String order and symmetries in quantum spin lattices, Phys. Rev. Lett. 100, 167202 (2008), 0802.0447.

Paper details

vlnc:W_mns5wk8j
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 1

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents