Magnetic structure of antiferromagnetic spin-1 Bose-Hubbard ground states beyond the minimal-total-spin theorem, with continuum contact-interaction and higher-spin extensions

For the spin-1 Bose-Hubbard model of ultracold spinor bosons, Katsura and Tasaki (Phys. Rev. Lett. 110, 130405 (2013)) proved universal, lattice-independent ground-state theorems: for antiferromagnetic spin-dependent coupling $U_2>0$, the unique ground state in each magnetization sector has the smallest total spin compatible with that sector. That constraint, however, does not determine the actual magnetic structure. The problem the source paper explicitly leaves open is: (1) a concrete characterization of the magnetic structures of the $U_2>0$ ground states intrinsic to spinor bosons -- which lattices, fillings, and couplings produce spin-nematic or antiferromagnetic long-range order, dimerization, Haldane-type gapped non-magnetic phases, or criticality -- at the standard of rigor of the total-spin theorems; (2) a rigorous proof of the minimal-total-spin theorem for the continuum gas with repulsive spin-dependent contact interaction $U_2(\mathbf r,\mathbf r')=c_2\delta(\mathbf r-\mathbf r')$, whose lack of mathematical control the source explicitly notes; and (3) the extension of the minimal-spin theorem to the antiferromagnetic sector of the spin-2 model and to spin $S\ge 3$.

Background

Spinor bosons are bosonic atoms carrying hyperfine spin, so that each particle has internal Zeeman states -- for spin-1 atoms the three states $m=+1,0,-1$. Loaded into optical lattices they are described by the spin-1 Bose-Hubbard model: bosons hop between lattice sites $x$ with amplitudes $t_{x,y}\ge 0$, and on each site the interaction is $U_0\,n_x(n_x-1)/2+U_2[(\mathbf S_x)^2/2-n_x]$, where $n_x$ is the site occupation, $\mathbf S_x$ the on-site spin operator, $U_0$ the spin-independent (density-density) coupling, and $U_2$ the spin-dependent spin-exchange coupling. The sign of $U_2$ is a property of the atom: ferromagnetic for $^{87}$Rb, antiferromagnetic for $^{23}$Na. For a finite connected lattice with $N$ bosons and conserved magnetization $M=\sum_x(n_{x,+}-n_{x,-})$, Katsura and Tasaki proved, using only the Perron-Frobenius theorem (a linear-algebra fact: a real symmetric matrix with non-positive off-diagonal entries, connected by non-zero entries, has a unique lowest eigenvector with strictly positive components), results that depend on nothing but the sign of $U_2$ -- not on lattice geometry, hopping, trap, $U_0$, or particle number. For $U_2<0$ the ground state is a saturated ferromagnet with total spin $S_{\rm tot}=N$; for $U_2>0$ the unique ground state in each magnetization sector has the smallest total spin allowed in that sector, with wave-function components of fixed sign $(-1)^{\sum_x n_{x,0}/2}$ in the occupation basis; for $U_2=0$ the ground states have an SU(3)-symmetric degeneracy.

Exactly as with the Lieb-Mattis theorem for the antiferromagnetic Heisenberg model, which pins spin quantum numbers but leaves the physics open, this total-spin constraint does not specify the magnetic structure. A minimal-spin ground state may still exhibit antiferromagnetic long-range order, spin-nematic order (zero average local spin but long-range order of the quadrupole tensor built from $\mathbf S$), dimerization (spontaneous breaking of lattice translation symmetry with doubled period), a Haldane-type gapped topological phase (a gapped phase with no magnetic order, protected by symmetry as in Haldane's conjecture for the spin-1 antiferromagnetic chain), or quantum criticality, depending on the lattice, filling, and couplings. The source paper states as its remaining challenge 'a concrete characterization of nontrivial magnetic structures for $U_2>0$ which are intrinsic to spinor bosons.' At strong coupling with one boson per site, second-order hopping maps the model to the spin-1 bilinear-biquadratic effective spin model (the effective spin-1 chain with nearest-neighbor couplings $a\,\mathbf S_i\cdot\mathbf S_j+b\,(\mathbf S_i\cdot\mathbf S_j)^2$), whose phase diagram contains singlet, nematic, dimerized, and Haldane phases; quantum Monte Carlo finds a dimerized ground state of the one-dimensional model; density-matrix renormalization group and bosonic dynamical mean-field studies map nematic, singlet-Mott, and dimerized phases in higher dimensions. None of these results is a theorem. The only rigorous structural statement beyond the total-spin constraint is the source's Theorem 4, which proves (by rigorous perturbation theory) that on a uniform lattice with zero trap potential and uniform nearest-neighbor hopping $t$, at even integer filling $\nu$ and $U_0/t$, $U_2/t$ above a model-dependent threshold, the unique ground state is a small perturbation of the product state in which each site is occupied by $\nu$ bosons forming an on-site spin singlet, with an energy gap above it -- a phase with trivial, absent magnetic order.

The paper also formulates a continuum counterpart: $N$ spin-1 bosons in $\mathbb R^3$ with a trap $V(\mathbf r)$, an arbitrary spin-independent interaction $U_0(\mathbf r,\mathbf r')$, and a spin-dependent interaction $U_2(\mathbf r,\mathbf r')$. For the repulsive contact case $U_2(\mathbf r,\mathbf r')=c_2\delta(\mathbf r-\mathbf r')$ with $c_2>0$, the authors state that all conclusions of the minimal-spin theorem should hold, but explicitly decline to claim a proof, since a mathematically rigorous treatment of the delta-function spin-dependent interaction 'may require nontrivial effort'. (The ferromagnetic side is close to the rigorous theorem of Eisenberg and Lieb that interacting bosons with no spin-dependent forces always admit a fully polarized ground state.)

For higher spin the same program meets new structure. Spin-2 bosons have two spin-dependent on-site couplings, $c_1$ and $c_2$, the second tied to an SO(5) singlet-pairing channel; the source notes that its minimal-spin theorem 'seems to have no straightforward extension.' Yang and Katsura later proved the spin-2 analogues of the ferromagnetic and symmetry-enhanced theorems -- saturated ferromagnetism for $c_1<0$ with $c_2>5c_1$, an SO(5)-symmetric ground state with total spin $F_{\rm tot}=0$ or $2$ for $c_1=0,c_2<0$, and SU(5) ferromagnetism for $c_1=c_2=0$ -- but their methods do not reach the antiferromagnetic sector $c_1>0$, and no rigorous minimal-total-spin result exists for spin $S\ge3$.

Finally, as an observable signature, the weak-coupling $S_{\rm tot}=0$ ground state carries a predicted power-law distribution for the number $n$ of spin-0 bosons, $|\eta_n|^2\propto 1/\sqrt n$, derived in the single-mode theory of Law, Pu, and Bigelow and Koashi and Ueda; the source notes it had not been observed experimentally. Lattice experiments have since produced on-site spin singlets in spinor condensates but have not measured this distribution. This experimental challenge is not part of the solving target of this task, which is the theoretical characterization program.

Problem Statement

Solve the three-part ground-state program that the source paper explicitly leaves open for antiferromagnetic spinor bosons.

1. Magnetic structure beyond the total-spin constraint. For the spin-1 Bose-Hubbard Hamiltonian with $U_2>0$ (arbitrary connected lattice, hopping, trap, spin-independent coupling $U_0$, and particle number), give a concrete characterization of the magnetic structure of the ground states going beyond Theorem 2's minimal-total-spin statement. Acceptable forms include: rigorous theorems identifying, for specified lattices, fillings, and coupling regimes, which magnetic structures the ground state realizes -- antiferromagnetic long-range order, spin-nematic (quadrupolar) long-range order, dimerization (spontaneous translation-symmetry breaking), a Haldane-type gapped topological phase, or criticality; or exact ground-state constructions in identified regimes that make the magnetic structure manifest. A rigorous theorem or exact construction restricted to one specified (lattice, filling, coupling) regime is a genuine contribution toward this item, but does not by itself resolve it: the item is resolved only by a characterization covering the stated scope of the model family, or by a rigorous result determining, across that scope, which of the listed structures can and cannot occur.

2. Continuum contact-interaction case. Prove rigorously that the continuum Hamiltonian of $N$ spin-1 bosons in $\mathbb R^3$ with trap $V(\mathbf r)$, arbitrary spin-independent interaction, and spin-dependent contact interaction $U_2(\mathbf r,\mathbf r')=c_2\delta(\mathbf r-\mathbf r')$ with $c_2>0$ -- with the Hamiltonian first properly defined (self-adjoint operator or closed quadratic form) -- has, in each magnetization sector, a unique ground state of minimal total spin with the sign structure and energy-ordering stated in Theorem 2, as the source asserts but does not prove.

3. Higher-spin extension. Prove (or disprove, e.g. by an explicit counterexample) the minimal-total-spin analogue of Theorem 2 for the antiferromagnetic sector of the spin-2 Bose-Hubbard model ($c_1>0$, any $c_2$), and determine, likewise by proof or by explicit counterexample, whether the minimal-total-spin statement of Theorem 2 extends to the antiferromagnetic spin-$S$ Bose-Hubbard model for every $S\ge 3$ -- here the spin-$S$ model is the source's Hamiltonian evaluated with $\mathbf S_x$ the spin-$S$ spin operators and on-site interaction $U_0\,n_x(n_x-1)/2+U_2[(\mathbf S_x)^2/2-n_x]$, so that the antiferromagnetic sector is the single-coupling regime $U_2>0$; more general on-site interactions with the additional invariant channels available at higher spin are a separate question and are not part of this item.

The three parts share one objective: complete the rigorous ground-state theory of antiferromagnetic spinor bosons by replacing the total-spin constraint with an actual determination of the magnetic structure, across the lattice, continuum, and higher-spin settings of the same theorem program. The source's fourth open item -- experimental observation of the $1/\sqrt n$ spin-0 number statistics of the singlet ground state -- is an experimental question, not a theoretical one, and is therefore not part of the solving target.

References

1. H. Katsura and H. Tasaki, 'Ground States of the Spin-1 Bose-Hubbard Model', Phys. Rev. Lett. 110, 130405 (2013). DOI: 10.1103/PhysRevLett.110.130405; arXiv:1212.6035, https://arxiv.org/abs/1212.6035
2. H. Yang and H. Katsura, 'Rigorous Results for the Ground States of the Spin-2 Bose-Hubbard Model', Phys. Rev. Lett. 122, 053401 (2019). DOI: 10.1103/PhysRevLett.122.053401; arXiv:1803.09441, https://arxiv.org/abs/1803.09441
3. E. Eisenberg and E. H. Lieb, 'Polarization of Interacting Bosons with Spin', Phys. Rev. Lett. 89, 220403 (2002). DOI: 10.1103/PhysRevLett.89.220403; arXiv:cond-mat/0207042, https://arxiv.org/abs/cond-mat/0207042
4. A. Imambekov, M. Lukin and E. Demler, 'Spin-exchange interactions of spin-one bosons in optical lattices: Singlet, nematic, and dimerized phases', Phys. Rev. A 68, 063602 (2003). DOI: 10.1103/PhysRevA.68.063602; arXiv:cond-mat/0306204, https://arxiv.org/abs/cond-mat/0306204
5. V. Apaja and O. F. Syljuasen, 'Dimerized ground state in the one-dimensional spin-1 boson Hubbard model', Phys. Rev. A 74, 035601 (2006). DOI: 10.1103/PhysRevA.74.035601, https://doi.org/10.1103/PhysRevA.74.035601
6. A. F. Hincapie-F, R. Franco and J. Silva-Valencia, 'Spin-1 Bose-Hubbard model with two- and three-body interactions', Phys. Lett. A 382, 1760-1765 (2018). DOI: 10.1016/j.physleta.2018.04.029; arXiv:1707.08195, https://arxiv.org/abs/1707.08195
7. X. Zan, J. Liu, J. Han, J. Wu and Y. Li, 'Phase diagrams and multistep condensations of spin-1 bosonic gases in optical lattices', Sci. Rep. 8, 9143 (2018). DOI: 10.1038/s41598-018-27503-0; arXiv:1711.05470, https://arxiv.org/abs/1711.05470
8. L. de Forges de Parny and V. G. Rousseau, 'Phase diagrams of antiferromagnetic spin-1 bosons on a square optical lattice with the quadratic Zeeman effect', Phys. Rev. A 97, 023628 (2018). DOI: 10.1103/PhysRevA.97.023628; arXiv:1801.00449, https://arxiv.org/abs/1801.00449
9. H. Yang, H. Nakano and H. Katsura, 'Symmetry-protected topological phases in spinful bosons with a flat band', Phys. Rev. Res. 3, 023210 (2021). DOI: 10.1103/PhysRevResearch.3.023210; arXiv:2003.01705, https://arxiv.org/abs/2003.01705
10. L. Zhao, T. Tang, Z. Chen and Y. Liu, 'Lattice-induced rapid formation of spin singlets in spin-1 spinor condensates', arXiv:1801.00773 (2018), https://arxiv.org/abs/1801.00773
11. C. K. Law, H. Pu and N. P. Bigelow, 'Quantum Spins Mixing in Spinor Bose-Einstein Condensates', Phys. Rev. Lett. 81, 5257-5261 (1998). DOI: 10.1103/PhysRevLett.81.5257, https://doi.org/10.1103/PhysRevLett.81.5257
12. M. Koashi and M. Ueda, 'Exact Eigenstates and Magnetic Response of Spin-1 and Spin-2 Bose-Einstein Condensates', Phys. Rev. Lett. 84, 1066-1069 (2000). DOI: 10.1103/PhysRevLett.84.1066; arXiv:cond-mat/9906313, https://arxiv.org/abs/cond-mat/9906313
13. E. H. Lieb and D. Mattis, 'Ordering Energy Levels of Interacting Spin Systems', J. Math. Phys. 3, 749-751 (1962). DOI: 10.1063/1.1724276, https://doi.org/10.1063/1.1724276
14. C. Binegar, J. O. Austin-Harris, S. E. Begg, P. Sigdel, T. Bilitewski and Y. Liu, 'Lattice-enabled detection of spin-dependent three-body interactions', arXiv:2510.01158 (2025), https://arxiv.org/abs/2510.01158