betaFeedback

WorksPhysics and Astronomy

Finite-volume and thermodynamic-limit lower bounds for the coherently driven Bose–Hubbard model via the Lieb–Yamazaki method

Abstract

We derive rigorous lower bounds on the ground-state energy of the coherently driven Bose–Hubbard model using a Lieb–Yamazaki positive-operator decomposition. The resulting quadratic bounding Hamiltonian is diagonalized by displacement and Bogoliubov transformations, yielding explicit stability conditions and finite-volume bounds in arbitrary dimension. We establish the thermodynamic-limit energy density and characterize the excitation spectrum of the bounding Hamiltonian. We also derive its Gaussian-state observables, including filling, number fluctuations, and second-order correlations, and identify regimes of antibunching and sub-Poissonian statistics. Numerical calculations assess the tightness of the bounds across representative parameter regimes.

Introduction

The Bose–Hubbard model provides a standard setting for the competition between tunneling, local interactions, and particle-number fluctuations in lattice boson systems[1, 2, 3]. Its coherently driven form is relevant to nonlinear bosonic arrays and to protocols in which an external field controls the amplitude and phase of the lattice modes. The drive breaks particle-number conservation and introduces a linear source term, enriching the ground-state structure while removing several simplifications available in the undriven model. These features make the model pertinent to strongly correlated bosons and cold-atom physics, as well as to driven photonic lattices and quantum fluids of light[4, 5, 6, 7]. Experiments on open photonic systems have observed dissipative critical behavior and stabilized Mott states[8, 9], while driven-dissipative theory has addressed critical slowing down, excitation and correlation structure, and chiral or chaotic dynamics[10, 11, 12, 13]. Open photonic realizations motivate the coherent source term, but the present work addresses the closed Hamiltonian sector only; dissipation and steady-state Liouvillian spectra are outside its scope.

Variational trial states readily furnish upper bounds on the ground-state energy, but rigorous lower bounds are considerably less accessible in the presence of both a quartic on-site interaction and a coherent source. A useful lower estimate must control the unbounded local occupation, remain stable under the thermodynamic limit, and retain explicit dependence on the detuning, hopping, and drive. It is also desirable to determine what the bounding construction implies for observables and excitations, rather than for the energy alone. Rigorous Bogoliubov analyses and energy estimates for interacting Bose gases provide complementary control of quadratic approximations and their remainders[14, 15, 16]. Elementary one-site and bond-cluster estimates also follow from operator inequalities and Anderson’s cluster lower-bound construction [17]. We therefore compare the Lieb–Yamazaki result directly with these important baselines rather than claiming uniqueness of the bound.

Herein, the Lieb–Yamazaki positive-operator decomposition is applied[18] to the driven Bose–Hubbard Hamiltonian. Upon the subtraction of a sum of positive operators, a quadratic bounding Hamiltonian is left, and is subsequently reduced by a uniform displacement and a Bogoliubov transformation[19, 20]. An admissible family of decomposition parameters is thereby delimited by the resulting stability condition. By the min–max principle, the quasiparticle gap of the bounding Hamiltonian is then converted into a certified lower threshold for the first excited energy of the full model.

From the construction there is obtained the rigorous lower bound of Theorem II.3 in every dimension d≥1d \ge1 and its explicit thermodynamic-limit energy density. Proposition III.2 identifies the continuous excitation band of the bounding Hamiltonian, whilst from the Bogoliubov vacuum there follow formulas that are exact within the Gaussian ground state ∣0B⟩\lvert0_{B}\rangle of the bounding Hamiltonian h1h_{1} for the order parameter, filling, energy densities, particle-number fluctuations, anomalous averages, and a drive-induced antibunching criterion. Finally, the certified Temple analysis of § 7 clarifies the finite-volume condition under which this Gaussian ansatz can support a Temple correction. The lower-bound property, per se, does not depend upon that correction. Its extensive energy excess makes that condition increasingly restrictive with system size, independently of the quality of the lower energy-density bound.

The Model and the Lieb–Yamazaki Procedure

Model Hamiltonian

We consider an array of coupled single-mode microcavities containing nonlinear media and driven coherently by external radiation. In a frame rotating at the driving frequency, the driven Bose–Hubbard Hamiltonian is

H=∑k=1M[V2ak†2ak2−ℏΔω0ak†ak]−J2∑⟨k,l⟩(ak†al+h.c.)+F∑k=1M(ak†+ak).(1)H = \sum_{k=1}^{M} \left[\frac{V}{2}a_k^{\dagger2}a_k^2-\hbar\Delta\omega_0 a_k^\dagger a_k\right]-\frac{J}{2}\sum_{\langle k,l\rangle}\left(a_k^\dagger a_l+\mathrm{h.c.}\right)+F\sum_{k=1}^{M}\left(a_k^\dagger+a_k\right). \tag*{(1)}

Here, the bosonic operators obey [ak,al†]=δkl[a_k,a_l^\dagger]=\delta_{kl}. The parameter VV is the on-site Kerr nonlinearity, JJ is the nearest-neighbor hopping amplitude, FF is the coherent driving amplitude, and Δω0\Delta\omega_0 is the detuning between the drive and the bare cavity mode. The lattice contains MM sites and is endowed with periodic boundary conditions. The notation ⟨k,l⟩\langle k,l\rangle denotes nearest-neighbor pairs, counted once. Throughout we assume a repulsive interaction V>0V>0. A gauge transformation makes the hopping nonnegative, so we take J≥0J\ge0; a global phase rotation likewise permits real FF.

Rescaling to Dimensionless Variables

For V>0V>0, we measure all energies in units of V/2V/2 and introduce the dimensionless quantities

h=2HV,v0=2ℏΔω0V,j=2JV,f=2FV,e=2EV.(2)h=\frac{2H}{V},\qquad v_0=\frac{2\hbar\Delta\omega_0}{V},\qquad j=\frac{2J}{V},\qquad f=\frac{2F}{V},\qquad e=\frac{2E}{V}. \tag*{(2)}

The stationary Schrödinger equation then takes the form

hψ=eψ.(3)h\psi=e\psi. \tag*{(3)}

In these variables, the on-site interaction has unit coefficient, which makes the algebraic structure used below explicit.

Lieb–Yamazaki Decomposition

Following the positive-operator strategy of Lieb and Yamazaki18, we introduce at every lattice site

Bk=ak†ak−K−g(ak−ak†),(4)B_k = a_k^\dagger a_k - K - g\left(a_k-a_k^\dagger\right), \tag*{(4)}

where KK and gg are real variational parameters. The operator

w2=∑k=1MBk†Bk(5)w^2 = \sum_{k=1}^{M} B_k^\dagger B_k \tag*{(5)}

is positive semidefinite. The remainder in the exact decomposition below is

h1=∑k=1M[(2K−ν0−1)ak†ak+(f−g)(ak†+ak)+g2(ak−ak†)2]−j2∑⟨k,l⟩(ak†al+h.c.)−MK2.(6)\begin{aligned} h_1={}&\sum_{k=1}^{M}\left[(2K-\nu_0-1)a_k^\dagger a_k+(f-g)(a_k^\dagger+a_k)+g^2(a_k-a_k^\dagger)^2\right] \\ &-\frac{j}{2}\sum_{\langle k,l\rangle}(a_k^\dagger a_l+\mathrm{h.c.})-MK^2. \tag*{(6)} \end{aligned}

Proposition II.1 (Positive-Operator Decomposition). Let Bk=ak†ak−K−g(ak−ak†)B_k=a_k^\dagger a_k-K-g(a_k-a_k^\dagger) for real parameters KK and gg. Then the dimensionless Hamiltonian admits the exact decomposition

h=w2+h1,(7)h=w^2+h_1, \tag*{(7)}

where w2=∑k=1MBk†Bk≥0w^2=\sum_{k=1}^{M}B_k^\dagger B_k\geq0 and h1h_1 is given explicitly by Eq. (6). Consequently, h≥h1h\geq h_1 as an operator inequality.

Proof of Proposition II.1. On the common invariant finite-particle core, direct expansion of Bk†BkB_k^\dagger B_k, using [ak,al†]=δkl[a_k,a_l^\dagger]=\delta_{kl}, gives Eq. (7) term by term. Moreover, ⟨ψ,Bk†Bkψ⟩=∥Bkψ∥2≥0\langle\psi,B_k^\dagger B_k\psi\rangle=\lVert B_k\psi\rVert^2\geq0 for every kk, so that summation over kk gives w2≥0w^2\geq0; hence the asserted inequality holds there as a quadratic-form inequality and extends to the corresponding closed forms21. This is the positive-operator step of the Lieb–Yamazaki construction18.

Uniform c-number Shift

The linear terms are eliminated by the real, spatially uniform displacement λ\lambda,

ak=a~k+λ,ak†=a~k†+λ.(8)a_k=\widetilde{a}_k+\lambda,\qquad a_k^\dagger=\widetilde{a}_k^\dagger+\lambda. \tag*{(8)}

While the original polaron application of Lieb and Yamazaki employs a momentum-dependent coupling gμg_\mu, we adopt the uniform ansatz g=constg=\mathrm{const}. This choice preserves the diagonal momentum-space structure of the Bogoliubov transformation. It permits an analytic treatment of the thermodynamic limit and closed-form expressions for the energy density and excitation spectrum. The same framework also permits momentum-dependent couplings.

With the dimensionless detuning ν0\nu_0 of Eq. (2), define

ν~=2K−ν0−1.(9)\widetilde{\nu}=2K-\nu_0-1. \tag*{(9)}

Cancellation of the linear terms fixes the displacement to

λ=−f−gν~−j,ν~≠j.(10)\lambda=-\frac{f-g}{\widetilde{\nu}-j},\qquad\widetilde{\nu}\ne j. \tag*{(10)}

Substitution into Eq. (6) gives

h1=∑k=1M[ν~a~k†a~k+g2(a~k−a~k†)2]−j2∑⟨k,l⟩(a~k†a~l+h.c.)−M(K2+(f−g)2ν~−j).(11)h_1=\sum_{k=1}^{M}\left[\tilde{\nu}\tilde{a}_k^\dagger\tilde{a}_k+g^2\left(\tilde{a}_k-\tilde{a}_k^\dagger\right)^2\right]-\frac{j}{2}\sum_{\langle k,l\rangle}\left(\tilde{a}_k^\dagger\tilde{a}_l+\mathrm{h.c.}\right)-M\left(K^2+\frac{(f-g)^2}{\tilde{\nu}-j}\right). \tag*{(11)}

The last term is the complete c-number contribution, including the −MK2-MK^2 term generated by the Lieb–Yamazaki decomposition.

Fourier Transformation

Translation invariance permits the discrete Fourier representation

a~k=1M∑μ=0M−1bμexp⁡(2πikμM).(12)\tilde{a}_k=\frac{1}{\sqrt{M}}\sum_{\mu=0}^{M-1}b_\mu\exp\left(\frac{2\pi i k\mu}{M}\right). \tag*{(12)}

where [bμ,bμ′†]=δμμ′[b_\mu,b_{\mu'}^\dagger]=\delta_{\mu\mu'} and all momentum indices are understood modulo MM. The quadratic lower-bound Hamiltonian becomes

h1=∑μ=0M−1[ωμbμ†bμ+g2(bμbM−μ+bμ†bM−μ†)]−M(K2+(f−g)2ν~−j+g2).(13)h_1=\sum_{\mu=0}^{M-1}\left[\omega_\mu b_\mu^\dagger b_\mu+g^2\left(b_\mu b_{M-\mu}+b_\mu^\dagger b_{M-\mu}^\dagger\right)\right]-M\left(K^2+\frac{(f-g)^2}{\tilde{\nu}-j}+g^2\right). \tag*{(13)}

where

ωμ=ν~−jcos⁡(2πμM)−2g2.(14)\omega_\mu=\tilde{\nu}-j\cos\left(\frac{2\pi\mu}{M}\right)-2g^2. \tag*{(14)}

A positive normal coefficient requires ωμ>0\omega_\mu>0 for every momentum μ\mu. The stronger stability condition needed by the Bogoliubov transformation is obtained below.

Bogoliubov Transformation

We introduce new canonical bosonic modes cμc_\mu through

bμ=11−Aμ2(cμ+AμcM−μ†).(15)b_\mu=\frac{1}{\sqrt{1-A_\mu^2}}\left(c_\mu+A_\mu c_{M-\mu}^\dagger\right). \tag*{(15)}

where Aμ=AM−μA_\mu=A_{M-\mu} is real and ∣Aμ∣<1|A_\mu|<1. Choosing AμA_\mu to cancel the anomalous terms diagonalizes Eq. (13) and yields

h1=∑μ=0M−111−Aμ2[ωμ(1+Aμ2)+4g2Aμ]cμ†cμ+g2∑μ=0M−1Aμ−M(K2+(f−g)2ν~−j+g2).(16)\begin{aligned} h_1={}&\sum_{\mu=0}^{M-1}\frac{1}{1-A_\mu^2}\left[\omega_\mu(1+A_\mu^2)+4g^2A_\mu\right]c_\mu^\dagger c_\mu\\ &+g^2\sum_{\mu=0}^{M-1}A_\mu-M\left(K^2+\frac{(f-g)^2}{\tilde{\nu}-j}+g^2\right). \tag*{(16)} \end{aligned}

The physical branch, continuously connected to Aμ=0A_\mu=0 as g→0g\to0, is

Aμ=12g2(−ωμ+ωμ2−4g4).(17)A_\mu=\frac{1}{2g^2}\left(-\omega_\mu+\sqrt{\omega_\mu^2-4g^4}\right). \tag*{(17)}

Lemma II.2 (Stability Condition for Bogoliubov Transformation). For g≠0g \ne0, the physical branch of the Bogoliubov transformation defined by Eq. (15) is canonical and diagonalizes a positive quadratic mode Hamiltonian if and only if

ωμ>2g2for all μ.(18)\omega_{\mu} > 2g^{2} \quad\text{for all }\mu. \tag*{(18)}

For j>0j > 0, this condition is equivalent to the constraint on KK given by Eq. (23).

Proof of Lemma II.2. The Bogoliubov transformation is canonical if and only if its coefficients satisfy ∣Aμ∣<1|A_{\mu}| < 1 and the transformation preserves the commutation relations. Reality of the square root alone permits either ωμ>2g2\omega_{\mu} > 2g^{2} or ωμ<−2g2\omega_{\mu} < -2g^{2}. The latter branch has a negative coefficient of the number operator after diagonalization and is therefore unbounded below. Positivity selects ωμ>2g2\omega_{\mu} > 2g^{2}, for which the physical root in Eq. (17) also obeys ∣Aμ∣<1|A_{\mu}| < 1. For j>0j > 0, the minimum of ωμ\omega_{\mu} occurs at μ=0\mu= 0, which yields the stated constraint on KK. When g=0g = 0, one instead takes Aμ=0A_{\mu} = 0 directly; the positive-mode condition is ωμ>0\omega_{\mu} > 0. This is the standard bosonic Bogoliubov criterion [19, 20].

Extension to dd Dimensions

The construction above does not depend on the dimension of the lattice. Let d≥1d \ge1 and let Λ={1,…,M}d\Lambda= \{1,\ldots,M\}^{d} be the periodic hypercubic lattice of linear size MM, with ∣Λ∣=Md|\Lambda| = M^{d} sites labeled by k=(k1,…,kd)\mathbf{k} = (k_{1},\ldots,k_{d}). We consider

H=∑k∈Λ[V2ak†2ak2−ℏΔω0ak†ak]−J2d∑⟨k,l⟩(ak†al+h.c.)+F∑k∈Λ(ak†+ak),(19)H = \sum_{\mathbf{k}\in\Lambda}\left[\frac{V}{2}a_{\mathbf{k}}^{\dagger2}a_{\mathbf{k}}^{2}-\hbar\Delta\omega_{0}a_{\mathbf{k}}^{\dagger}a_{\mathbf{k}}\right]-\frac{J}{2d}\sum_{\langle\mathbf{k},\mathbf{l}\rangle}\left(a_{\mathbf{k}}^{\dagger}a_{\mathbf{l}}+\text{h.c.}\right)+F\sum_{\mathbf{k}\in\Lambda}\left(a_{\mathbf{k}}^{\dagger}+a_{\mathbf{k}}\right), \tag*{(19)}

which reduces to Eq. (1) for d=1d = 1. The hopping amplitude per spatial direction is thus J/dJ/d. With this normalization, the hopping energy of the zero-momentum mode is independent of dd. The dimensionless quantities are defined as in Eq. (2).

The steps of the one-dimensional derivation carry over as follows.

  1. The operator BkB_{\mathbf{k}} of Eq. (4) acts on a single site, and the hopping term is not affected by the decomposition. Proposition II.1 therefore holds verbatim, with sums over Λ\Lambda and with the constant −MdK2-M^{d}K^{2} in h1h_{1}.

  1. Every site belongs to dd nearest-neighbor pairs, each carrying the coefficient −j/(2d)-j/(2d). The uniform displacement of Eq. (8) thus produces the same linear terms per site as in one dimension, and Eq. (10) for λ\lambda is unchanged. The c-number contribution becomes −Md[K2+(f−g)2/(ν~−j)]-M^{d}\left[K^{2}+(f-g)^{2}/(\tilde{\nu}-j)\right].

  1. The Fourier modes are labeled by μ=(μ1,…,μd)∈{0,…,M−1}d\boldsymbol{\mu} = (\mu_{1},\ldots,\mu_{d}) \in\{0,\ldots,M-1\}^{d},

a~k=1Md/2∑μbμexp⁡(2πik⋅μM),(20)\tilde{a}_{\mathbf{k}} = \frac{1}{M^{d/2}}\sum_{\boldsymbol{\mu}} b_{\boldsymbol{\mu}}\exp\left(\frac{2\pi i\mathbf{k}\cdot\boldsymbol{\mu}}{M}\right), \tag*{(20)}

and Eqs. (13)–(17) hold with μ\mu replaced by μ\boldsymbol{\mu}, with M−μM-\mu replaced by −μ-\boldsymbol{\mu} (modulo MM in each component), with the constant −M(⋯ )-M(\cdots) replaced by −Md(⋯ )-M^{d}(\cdots), and with

ωμ=ν~+ϵμ−2g2,ϵμ=−jd∑α=1dcos⁡(2πμαM).(21)\omega_{\boldsymbol{\mu}} = \tilde{\nu}+\epsilon_{\boldsymbol{\mu}}-2g^{2}, \qquad\epsilon_{\boldsymbol{\mu}}=-\frac{j}{d}\sum_{\alpha=1}^{d}\cos\left(\frac{2\pi\mu_{\alpha}}{M}\right). \tag*{(21)}

Here ϵμ\epsilon_{\boldsymbol{\mu}} is the free-particle dispersion, with ϵ0=−j\epsilon_{\mathbf{0}}=-j for every dd.

  1. The excitation energies and the Bogoliubov coefficients are

Ωμ=ωμ2−4g4,Aμ=12g2(−ωμ+Ωμ),(22)\Omega_{\boldsymbol{\mu}} = \sqrt{\omega_{\boldsymbol{\mu}}^{2}-4g^{4}}, \qquad A_{\boldsymbol{\mu}} = \frac{1}{2g^{2}}\left(-\omega_{\boldsymbol{\mu}}+\Omega_{\boldsymbol{\mu}}\right), \tag*{(22)}

so that g2Aμ=12(Ωμ−ωμ)g^{2}A_{\boldsymbol{\mu}}=\frac{1}{2}(\Omega_{\boldsymbol{\mu}}-\omega_{\boldsymbol{\mu}}); for g=0g = 0 both sides vanish.

  1. For j≥0j \ge0, the minimum of ωμ\omega_\mu is attained at μ=0\mu= 0, where all cosines equal one, and equals ν~−j−2g2\tilde{\nu} - j - 2g^2. The stability condition ωμ>2g2\omega_\mu> 2g^2 of Lemma II.2 is therefore equivalent to the same constraint on KK in every dimension.

Ground-State Energy Lower Bound

Theorem II.3 (Rigorous Lower Bound in dd Dimensions). Let d≥1d \ge1, let HH be the driven Bose–Hubbard Hamiltonian on Λ={1,…,M}d\Lambda= \{1,\ldots,M\}^d defined by Eq. (19), and let h=2H/Vh = 2H/V be the corresponding dimensionless Hamiltonian, with j≥0j \ge0. For any real parameters KK and gg satisfying

K>12(1+ν0+j+4g2),(23)K > \frac{1}{2}\left(1+\nu_0+j+4g^2\right), \tag*{(23)}

the ground-state energy E0E_0 of hh satisfies the rigorous lower bound

E0≥E0,l(d)=g2∑μAμ−Md(K2+(f−g)2ν~−j+g2),(24)E_0 \ge E_{0,l}^{(d)} = g^2 \sum_\mu A_\mu- M^d\left(K^2+\frac{(f-g)^2}{\tilde{\nu}-j}+g^2\right), \tag*{(24)}

where ν~=2K−ν0−1\tilde{\nu}=2K-\nu_0-1, the sum runs over the MdM^d modes μ∈{0,…,M−1}d\boldsymbol{\mu}\in\{0,\ldots,M-1\}^d, and AμA_\mu is given by Eqs. (21) and (22). Equivalently,

E0,l(d)=12∑μ(Ωμ−ωμ)−Md(K2+(f−g)2ν~−j+g2).(25)E_{0,l}^{(d)} = \frac{1}{2}\sum_\mu(\Omega_\mu-\omega_\mu)-M^d\left(K^2+\frac{(f-g)^2}{\tilde{\nu}-j}+g^2\right). \tag*{(25)}

The constraint in Eq. (23) does not depend on dd. For d=1d=1, Eq. (24) reads E0,l(1)=g2∑μ=0M−1Aμ−M[K2+(f−g)2/(ν~−j)+g2]E_{0,l}^{(1)}=g^2\sum_{\mu=0}^{M-1}A_\mu-M\left[K^2+(f-g)^2/(\tilde{\nu}-j)+g^2\right], with AμA_\mu given by Eq. (17).

Proof of Theorem II.3. Proposition II.1 gives h≥h1h\ge h_1 in the quadratic-form sense. The variational (min–max) principle[22] therefore implies inf⁡σ(h)≥inf⁡σ(h1)\inf\sigma(h)\ge\inf\sigma(h_1). The Hamiltonian h1h_1 is quadratic in the bosonic operators and is diagonalized by the Bogoliubov transformation in Eq. (15), which is well-defined by Lemma II.2. The vacuum of the cμc_\mu modes minimizes the diagonalized h1h_1, and its lowest eigenvalue is given by Eq. (24); Eq. (25) follows from Eq. (22). The dimension enters this argument only through the number of sites, MdM^d, and through the sum over modes: the Lieb–Yamazaki decomposition is local, acting on one site at a time, and the remaining steps were adapted to Λ\Lambda in § 2 G. The proof is thus the same for every d≥1d\ge1. □

Summary

It follows that, in every dimension d≥1d\ge1, there is a variational family of rigorous lower bounds for the ground-state energy, parameterized by KK and gg subject to the stability constraint. In the following section, we take the thermodynamic limit M→∞M\to\infty to extract the energy density and excitation spectrum.

THE THERMODYNAMIC LIMIT

Strategy: From Finite Lattice to Thermodynamic Limit

Theorem II.3 provides a rigorous lower bound E0,l(d)E_{0,l}^{(d)} for every finite periodic lattice of MdM^d sites, in any dimension d≥1d\ge1. Its derivation uses the positive-operator inequality of Proposition II.1 and the stability criterion of Lemma II.2. We now prove the existence of the corresponding energy density

e0,l(d)=lim⁡M→∞E0,l(d)Md.(26)e_{0,l}^{(d)}=\lim_{M\to\infty}\frac{E_{0,l}^{(d)}}{M^d}. \tag*{(26)}

It is essential that passage to the thermodynamic limit preserve the finite-volume lower-bound property. Let q=(q1,…,qd)∈[−π,π]d\mathbf{q}=(q_1,\ldots,q_d)\in[-\pi,\pi]^d denote the momentum variable of the dd-dimensional Brillouin zone. If F(q)F(\mathbf{q}) is continuous and 2π2\pi-periodic in each component, the standard convergence theorem for Riemann sums on the grids 2πμ/M2\pi\boldsymbol{\mu}/M gives

lim⁡M→∞1Md∑μF(2πμM)=∫[−π,π]dddq(2π)dF(q),(27)\lim_{M\to\infty}\frac{1}{M^d}\sum_{\boldsymbol{\mu}}F\left(\frac{2\pi\boldsymbol{\mu}}{M}\right)=\int_{[-\pi,\pi]^d}\frac{\mathrm{d}^d q}{(2\pi)^d}F(\mathbf{q}), \tag*{(27)}

see, e.g., Ref.23.

Energy Density in the Thermodynamic Limit

Dividing the finite-volume bound in Eq. (24) by MdM^d yields

E0,l(d)Md=g2Md∑μAμ−(K2+(f−g)2ν~−j+g2).(28)\frac{E_{0,l}^{(d)}}{M^d}=\frac{g^2}{M^d}\sum_{\boldsymbol{\mu}}A_{\boldsymbol{\mu}}-\left(K^2+\frac{(f-g)^2}{\tilde{\nu}-j}+g^2\right). \tag*{(28)}

Introduce the continuous functions

ω(q)=ν~−jd∑α=1dcos⁡qα−2g2,Ω(q)=ω(q)2−4g4,(29)\omega(\mathbf{q})=\tilde{\nu}-\frac{j}{d}\sum_{\alpha=1}^{d}\cos q_\alpha-2g^2,\qquad\Omega(\mathbf{q})=\sqrt{\omega(\mathbf{q})^2-4g^4}, \tag*{(29)}

and

A(q)=12g2(−ω(q)+Ω(q)).(30)A(\mathbf{q})=\frac{1}{2g^2}\left(-\omega(\mathbf{q})+\Omega(\mathbf{q})\right). \tag*{(30)}

The function A(q)A(\mathbf{q}) is the root of g2A2+ω(q)A+g2=0g^2A^2+\omega(\mathbf{q})A+g^2=0 with ∣A∣<1\lvert A\rvert<1, and g2A(q)=12[Ω(q)−ω(q)]g^2A(\mathbf{q})=\frac{1}{2}[\Omega(\mathbf{q})-\omega(\mathbf{q})]. For g=0g=0, the expression g2A(q)g^2A(\mathbf{q}) is understood by its continuous limit as g→0g\to0, which is zero. By Eqs. (21) and (22), ωμ=ω(2πμ/M)\omega_{\boldsymbol{\mu}}=\omega(2\pi\boldsymbol{\mu}/M), Ωμ=Ω(2πμ/M)\Omega_{\boldsymbol{\mu}}=\Omega(2\pi\boldsymbol{\mu}/M), and, for g≠0g\ne0, Aμ=A(2πμ/M)A_{\boldsymbol{\mu}}=A(2\pi\boldsymbol{\mu}/M).

Corollary III.1 (Energy Density in the Thermodynamic Limit). Let d≥1d\ge1 and j≥0j\ge0. For every admissible pair (K,g)(K,g) satisfying the stability constraint in Eq. (23), the thermodynamic limit of the finite-volume lower-bound density exists and is

e0,l(d)=lim⁡M→∞E0,l(d)Md=g2∫[−π,π]dddq(2π)dA(q)−(K2+(f−g)2ν~−j+g2).(31)e_{0,l}^{(d)}=\lim_{M\to\infty}\frac{E_{0,l}^{(d)}}{M^d}=g^2\int_{[-\pi,\pi]^d}\frac{\mathrm{d}^d q}{(2\pi)^d}A(\mathbf{q})-\left(K^2+\frac{(f-g)^2}{\tilde{\nu}-j}+g^2\right). \tag*{(31)}

Equivalently,

e0,l(d)=12∫[−π,π]dddq(2π)dΩ(q)−(K2+(f−g)2ν~−j+ν~2).(32)e_{0,l}^{(d)}=\frac{1}{2}\int_{[-\pi,\pi]^d}\frac{\mathrm{d}^d q}{(2\pi)^d}\Omega(\mathbf{q})-\left(K^2+\frac{(f-g)^2}{\tilde{\nu}-j}+\frac{\tilde{\nu}}{2}\right). \tag*{(32)}

Without assuming existence of the exact thermodynamic limit, this expression obeys

lim inf⁡M→∞E0(M)Md≥e0,l(d).(33)\liminf_{M\to\infty}\frac{E_0(M)}{M^d}\ge e_{0,l}^{(d)}. \tag*{(33)}

Proof of Corollary III.1. For admissible parameters, ω(q)>2g2\omega(\mathbf{q}) > 2g^2 for every q\mathbf{q}, and hence Ω(q)>0\Omega(\mathbf{q}) > 0 and A(q)A(\mathbf{q}) is continuous and 2π2\pi-periodic in each component. The Riemann-sum limit in Eq. (27) applies with F=AF=A. The remaining terms in Eq. (28) are independent of MM, which proves Eq. (31). Equation (32) follows from g2A=12(Ω−ω)g^2 A = \frac{1}{2}(\Omega-\omega), because the cosines integrate to zero and the Brillouin-zone average of ω(q)\omega(\mathbf{q}) is therefore ν~−2g2\tilde{\nu}-2g^2. Moreover, Theorem II.3 gives E0(M)/Md≥E0,l(d)(M)/MdE_0(M)/M^d \ge E_{0,l}^{(d)}(M)/M^d for every MM. Taking the lower limit proves Eq. (33); if the exact energy density exists, this reduces to the usual thermodynamic-limit inequality.

Continuous Bogoliubov Excitation Spectrum

Proposition III.2 (Continuous Excitation Spectrum). The diagonal coefficient in Eq. (16) is Ωμ=ωμ2−4g4\Omega_\mu= \sqrt{\omega_\mu^2-4g^4}. In the thermodynamic limit these Bogoliubov excitation energies sample the continuous dispersion

Ω(q)=(ν~−jd∑α=1dcos⁡qα−2g2)2−4g4,q∈[−π,π]d.(34)\Omega(\mathbf{q}) = \sqrt{\left(\tilde{\nu}-\frac{j}{d}\sum_{\alpha=1}^{d}\cos q_\alpha-2g^2\right)^2-4g^4}, \qquad\mathbf{q}\in[-\pi,\pi]^d. \tag*{(34)}

The stability condition ωμ>2g2\omega_\mu>2g^2 becomes ω(q)>2g2\omega(\mathbf{q})>2g^2 for every q\mathbf{q}, which for j≥0j\ge0 is equivalent to the constraint on KK in Eq. (23), independently of dd.

Proof of Proposition III.2. Substitution of AμA_\mu into the coefficient of cμ†cμc_\mu^\dagger c_\mu in Eq. (16) gives Ωμ\Omega_\mu above, since AμA_\mu solves g2Aμ2+ωμAμ+g2=0g^2A_\mu^2+\omega_\mu A_\mu+g^2=0. The momentum grids become dense in the Brillouin zone; continuity of Ω(q)\Omega(\mathbf{q}) in the stable region then gives the claimed continuum band. For j≥0j\ge0, the minimum of ω(q)\omega(\mathbf{q}) occurs at q=0\mathbf{q}=0 and equals ν~−j−2g2\tilde{\nu}-j-2g^2 in every dimension, so uniform stability is precisely the constraint in Eq. (23).

Analysis of the Energy Density Integral

For every admissible pair (K,g)(K,g), the strict inequality ω(q)>2g2\omega(\mathbf{q})>2g^2 ensures that the radicand in Eq. (30) is positive throughout the compact Brillouin zone. Consequently, A(q)A(\mathbf{q}) is smooth in q\mathbf{q}, and the integral in Eq. (31) is well-defined. It may be evaluated numerically for general parameters and analytically in special limits.

At the boundary of the admissible region, there may exist a critical momentum qc\mathbf{q}_c for which ω(qc)→2g2\omega(\mathbf{q}_c)\to2g^2. The square root ω(q)2−4g4\sqrt{\omega(\mathbf{q})^2-4g^4} then becomes nonanalytic. Such behavior marks the loss of Bogoliubov stability of the bounding Hamiltonian; it is not evidence of a phase transition of the full driven model. The rigorous bound derived above applies in the open stable region; any extension to its boundary requires a separate limiting argument.

Summary

We have thus rigorously established the thermodynamic limit of the Lieb–Yamazaki lower bound in every dimension d≥1d\ge1. The lower-bound density ϵ0,l(d)\epsilon_{0,l}^{(d)} is given by a closed-form integral expression, and the excitation spectrum is described by the continuous dispersion relation Ω(q)\Omega(\mathbf{q}). In the following section, we derive macroscopic observables from this framework.

Excitation Spectrum and Observables

Order Parameter, Filling Fraction, and Energy Densities

Throughout this subsection, we take d=1d=1; the extension to dd dimensions is given in Remark IV.8. The pair (K,g)(K,g) is admissible in the sense of Eq. (23), and ∣0B⟩|0_B\rangle denotes the normalized vacuum of the Bogoliubov modes of Eq. (15),

cμ∣0B⟩=0for all μ.(35)c_\mu|0_B\rangle=0\quad\text{for all }\mu. \tag*{(35)}

By the proof of Theorem II.3, ∣0B⟩|0_B\rangle is the ground state of the bounding Hamiltonian h1h_1; it depends on KK and gg. All expectation values below are taken in this state, ⟨X⟩=⟨0B∣X∣0B⟩\langle X\rangle=\langle0_B|X|0_B\rangle, and all energies are measured in the dimensionless units of Eq. (2). We write Ωμ=ωμ2−4g4\Omega_\mu=\sqrt{\omega_\mu^2-4g^4} as in Proposition III.2.

Lemma IV.1 (Two-Point Functions). In the state ∣0B⟩|0_B\rangle, ⟨bμ⟩=0\langle b_\mu\rangle=0 and

⟨bμ†bμ′⟩=δμμ′ωμ−Ωμ2Ωμ,⟨bμbμ′⟩=−δμ,M−μ′g2Ωμ,(36)\langle b_\mu^\dagger b_{\mu'}\rangle=\delta_{\mu\mu'}\frac{\omega_\mu-\Omega_\mu}{2\Omega_\mu},\qquad\langle b_\mu b_{\mu'}\rangle=-\delta_{\mu,M-\mu'}\frac{g^2}{\Omega_\mu}, \tag*{(36)}

with momentum indices understood modulo MM. Consequently, the local moments of the displaced operators are independent of the site kk and equal

m≡⟨a~k†a~k⟩=1M∑μ=0M−1ωμ−Ωμ2Ωμ,s≡⟨a~ka~k⟩=−g2M∑μ=0M−11Ωμ,(37)m\equiv\langle\tilde a_k^\dagger\tilde a_k\rangle=\frac{1}{M}\sum_{\mu=0}^{M-1}\frac{\omega_\mu-\Omega_\mu}{2\Omega_\mu},\qquad s\equiv\langle\tilde a_k\tilde a_k\rangle=-\frac{g^2}{M}\sum_{\mu=0}^{M-1}\frac{1}{\Omega_\mu}, \tag*{(37)}

while $\langle\tilde a_k\rangle=0.

Proof of Lemma IV.1. (15) expresses bμb_\mu as a linear combination of cμc_\mu and cM−μ†c^\dagger_{M-\mu}, so that ⟨bμ⟩=0\langle b_\mu\rangle=0 by Eq. (35). Using ⟨cμcμ′†⟩=δμμ′\langle c_\mu c_{\mu'}^\dagger\rangle=\delta_{\mu\mu'} and Aμ=AM−μA_\mu=A_{M-\mu}, we have

⟨bμ†bμ′⟩=δμμ′Aμ21−Aμ2,⟨bμbμ′⟩=δμ,M−μ′Aμ1−Aμ2.(38)\langle b_\mu^\dagger b_{\mu'}\rangle=\delta_{\mu\mu'}\frac{A_\mu^2}{1-A_\mu^2},\qquad\langle b_\mu b_{\mu'}\rangle=\delta_{\mu,M-\mu'}\frac{A_\mu}{1-A_\mu^2}. \tag*{(38)}

Since AμA_\mu in Eq. (17) solves g2Aμ2+ωμAμ+g2=0g^2A_\mu^2+\omega_\mu A_\mu+g^2=0, it follows that 1−Aμ2=−ΩμAμ/g21-A_\mu^2=-\Omega_\mu A_\mu/g^2, which converts Eq. (38) into Eq. (36). For g=0g=0 one has Aμ=0A_\mu=0 and Ωμ=ωμ\Omega_\mu=\omega_\mu, and both sides of Eq. (36) vanish. Inserting the Fourier representation in Eq. (12) yields Eq. (37); the phases cancel in both sums, which proves the independence of kk.

Proposition IV.2 (Order Parameter). In the state ∣0B⟩|0_B\rangle, the expectation value of the annihilation operator is the same at every site and equals the displacement,

⟨ak⟩=λ=−f~−gν~−j,(39)\langle a_k\rangle=\lambda=-\frac{\tilde f-g}{\tilde\nu-j}, \tag*{(39)}

where ν~=2K−ν0−1\tilde\nu=2K-\nu_0-1 as in Eq. (9).

Proof of Proposition IV.2. By Eq. (8), ⟨ak⟩=⟨a~k⟩+λ\langle a_k\rangle=\langle\tilde a_k\rangle+\lambda, and ⟨a~k⟩=0\langle\tilde a_k\rangle=0 by Lemma IV.1. The value of λ\lambda is given by Eq. (10).

Proposition IV.3 (Filling Fraction). In the state ∣0B⟩|0_B\rangle, the mean number of particles per site is

ρ≡⟨ak†ak⟩=λ2+1M∑μ=0M−1ωμ−Ωμ2Ωμ,(40)\rho\equiv\langle a_k^\dagger a_k\rangle=\lambda^2+\frac{1}{M}\sum_{\mu=0}^{M-1}\frac{\omega_\mu-\Omega_\mu}{2\Omega_\mu}, \tag*{(40)}

with λ\lambda given by Eq. (39).

Proof of Proposition IV.3. Since λ\lambda is real, Eq. (8) gives ak†ak=λ2+λ(a~k+a~k†)+a~k†a~ka_k^\dagger a_k=\lambda^2+\lambda(\tilde a_k+\tilde a_k^\dagger)+\tilde a_k^\dagger\tilde a_k. The linear terms have vanishing expectation, and the last term equals mm by Lemma IV.1.

The first term in Eq. (40) is the coherent contribution of the displacement, while the second is the depletion generated by the anomalous terms of h1h_1; the latter vanishes for g=0g=0.

Proposition IV.4 (Kinetic Energy Density). Let

T=−j2∑⟨k,l⟩(ak†al+h.c.)(41)T=-\frac{j}{2}\sum_{\langle k,l\rangle}\left(a_k^\dagger a_l+h.c.\right) \tag*{(41)}

be the hopping part of the dimensionless Hamiltonian hh. In the state ∣0B⟩\lvert0_B\rangle,

⟨T⟩M=ϵ0λ2+1M∑μ=0M−1ϵμωμ−Ωμ2Ωμ,ϵμ=−jcos⁡(2πμM)(42)\frac{\langle T\rangle}{M}=\epsilon_0\lambda^2+\frac{1}{M}\sum_{\mu=0}^{M-1}\epsilon_\mu\frac{\omega_\mu-\Omega_\mu}{2\Omega_\mu},\qquad\epsilon_\mu=-j\cos\left(\frac{2\pi\mu}{M}\right) \tag*{(42)}

where ϵμ\epsilon_\mu is the free-particle dispersion and ϵ0=−j\epsilon_0=-j.

Proof of Proposition IV.4. Let αμ\alpha_\mu denote the Fourier modes of the undisplaced operators aka_k, defined as in Eq. (12). Then T=∑μϵμαμ†αμT=\sum_\mu\epsilon_\mu\alpha_\mu^\dagger\alpha_\mu, and Eq. (8) gives αμ=bμ+λM δμ,0\alpha_\mu=b_\mu+\lambda\sqrt{M}\,\delta_{\mu,0}. By Lemma IV.1, ⟨αμ†αμ⟩=λ2Mδμ,0+(ωμ−Ωμ)/(2Ωμ)\langle\alpha_\mu^\dagger\alpha_\mu\rangle=\lambda^2M\delta_{\mu,0}+(\omega_\mu-\Omega_\mu)/(2\Omega_\mu). Summation over μ\mu and division by MM prove the claim.

Proposition IV.5 (On-Site Energy Density). Let

Vk=ak†2ak2−ν0ak†ak+f(ak†+ak)(43)V_k=a_k^{\dagger2}a_k^2-\nu_0a_k^\dagger a_k+f\left(a_k^\dagger+a_k\right) \tag*{(43)}

be the on-site part of hh, so that h=T+∑k=1MVkh=T+\sum_{k=1}^{M}V_k. In the state ∣0B⟩\lvert0_B\rangle,

⟨Vk⟩=λ4+2λ2(2m+s)+2m2+s2−ν0(λ2+m)+2fλ(44)\langle V_k\rangle=\lambda^4+2\lambda^2(2m+s)+2m^2+s^2-\nu_0(\lambda^2+m)+2f\lambda \tag*{(44)}

where mm and ss are the local moments in Eq. (37). This expression is exact within the Gaussian ground state ∣0B⟩\lvert0_B\rangle of the bounding Hamiltonian h1h_1.

Proof of Proposition IV.5. The detuning and driving terms follow from Propositions IV.2 and IV.3. For the interaction term, Eq. (8) gives

ak†2ak2=(a~k†2+2λa~k†+λ2)(a~k2+2λa~k+λ2).(45)a_k^{\dagger2}a_k^2=\left(\tilde a_k^{\dagger2}+2\lambda\tilde a_k^\dagger+\lambda^2\right)\left(\tilde a_k^2+2\lambda\tilde a_k+\lambda^2\right). \tag*{(45)}

The state ∣0B⟩\lvert0_B\rangle is the vacuum of operators that are linear in a~k\tilde a_k and a~k†\tilde a_k^\dagger, hence it is quasi-free and Wick’s theorem applies1. Expectation values of odd monomials vanish, and ⟨a~k†2a~k2⟩=2m2+s2\langle\tilde a_k^{\dagger2}\tilde a_k^2\rangle=2m^2+s^2 because ss is real. The remaining even terms of Eq. (45) contribute λ4+4λ2m+2λ2s\lambda^4+4\lambda^2m+2\lambda^2s.

Remark IV.6 (Strong-Condensate Limit). If the displacement dominates the fluctuations, λ2≫max⁡(m,∣s∣)\lambda^2\gg\max(m,|s|), the terms 2m2+s22m^2+s^2 in Eq. (44) may be neglected. To first order in the fluctuation moments one then has the approximation

⟨Vk⟩≈λ4−ν0λ2+2fλ+(4λ2−ν0)m+2λ2s.(46)\langle V_k\rangle\approx\lambda^4-\nu_0\lambda^2+2f\lambda+(4\lambda^2-\nu_0)m+2\lambda^2s. \tag*{(46)}

In contrast with Eq. (44), this is not an identity; it is valid only in the strong-condensate limit.

Corollary IV.7 (Observables in the Thermodynamic Limit). For every admissible pair (K,g)(K,g), the limits M→∞M \to\infty of the quantities above exist. With ω(q)\omega(q) and Ω(q)\Omega(q) as in Eq. (29) for d=1d=1,

m∞=∫−ππdq2πω(q)−Ω(q)2Ω(q),s∞=−g2∫−ππdq2π1Ω(q),ρ∞=λ2+m∞,lim⁡M→∞⟨T⟩M=−jλ2−j∫−ππdq2πcos⁡q ω(q)−Ω(q)2Ω(q).(47)\begin{aligned} m_{\infty} &= \int_{-\pi}^{\pi} \frac{dq}{2\pi}\frac{\omega(q)-\Omega(q)}{2\Omega(q)}, \qquad s_{\infty} = -g^{2}\int_{-\pi}^{\pi}\frac{dq}{2\pi}\frac{1}{\Omega(q)},\\ \rho_{\infty} &= \lambda^{2}+m_{\infty}, \qquad \lim_{M\to\infty}\frac{\langle T\rangle}{M}=-j\lambda^{2}-j\int_{-\pi}^{\pi}\frac{dq}{2\pi}\cos q\,\frac{\omega(q)-\Omega(q)}{2\Omega(q)}. \tag*{(47)} \end{aligned}

and the limit of ⟨Vk⟩\langle V_k\rangle is given by Eq. (44) with mm and ss replaced by m∞m_{\infty} and s∞s_{\infty}. The order parameter λ\lambda is independent of MM.

Proof of Corollary IV.7. In the stable region, Ω(q)>0\Omega(q)>0 for every qq, so the integrands are continuous and 2π2\pi-periodic. The finite-volume expressions are their Riemann sums on the grid q=2πμ/Mq=2\pi\mu/M, and the convergence argument used for Eq. (27) applies23^{23}.

Remark IV.8 (Extension to dd Dimensions). On the lattice Λ={1,…,M}d\Lambda=\{1,\ldots,M\}^{d} of § 2 G, the results of this subsection hold with the following replacements:

  1. the mode averages 1M∑μ=0M−1\frac{1}{M}\sum_{\mu=0}^{M-1} become 1Md∑μ\frac{1}{M^{d}}\sum_{\mu}, with ωμ\omega_{\mu} and Ωμ\Omega_{\mu} replaced by ωμ\omega_{\boldsymbol{\mu}} and Ωμ\Omega_{\boldsymbol{\mu}} of Eqs. (21) and (22), and with δμ,M−μ′\delta_{\mu,M-\mu'} in Eq. (36) replaced by δμ,−μ′\delta_{\mu,-\mu'};

  1. the free-particle dispersion ϵμ=−jcos⁡(2πμ/M)\epsilon_{\mu}=-j\cos(2\pi\mu/M) becomes ϵμ\epsilon_{\mu} of Eq. (21), the hopping operator in Eq. (41) carries the coefficient −j/(2d)-j/(2d), and ⟨T⟩/M\langle T\rangle/M becomes ⟨T⟩/Md\langle T\rangle/M^{d};

  1. the Brillouin-zone integrals and the cosine in Eq. (48) become

∫−ππdq2π⟶∫[−π,π]dddq(2π)d,cos⁡q⟶1d∑α=1dcos⁡qα.(48)\int_{-\pi}^{\pi}\frac{dq}{2\pi}\longrightarrow\int_{[-\pi,\pi]^{d}}\frac{d^{d}q}{(2\pi)^{d}}, \qquad \cos q\longrightarrow\frac{1}{d}\sum_{\alpha=1}^{d}\cos q_{\alpha}. \tag*{(48)}

The order parameter λ\lambda and the expressions of Propositions IV.3 and IV.5 and of Remark IV.6 in terms of λ\lambda, mm, and ss do not change, because they are local and λ\lambda does not depend on dd. All integrands remain continuous, since Ω(q)>0\Omega(q)>0 in the stable region.

Remark IV.9 (Variational Upper Bound). Since h=T+∑kVkh=T+\sum_k V_k and ∣0B⟩|0_B\rangle is a normalized trial state, Propositions IV.4 and IV.5 also give a variational upper bound on the ground-state energy. Combined with Theorem II.3 for d=1d=1,

E0,l(1)M≤E0M≤⟨T⟩M+⟨Vk⟩,(49)\frac{E_{0,l}^{(1)}}{M}\leq\frac{E_0}{M}\leq\frac{\langle T\rangle}{M}+\langle V_k\rangle, \tag*{(49)}

and by Proposition II.1 the difference between the two outer expressions is ⟨Bk†Bk⟩≥0\langle B_k^\dagger B_k\rangle\geq0. In dd dimensions the same statement holds with E0,l(d)E_{0,l}^{(d)} and MdM^{d} in place of E0,l(1)E_{0,l}^{(1)} and MM.

Particle-Number Variance and Correlations {#sec:B}

We now turn from one-body expectation values to local two-body quantities in the state ∣0B⟩|0_B\rangle: the variance of the on-site particle number, the second-order correlation function, and the anomalous averages. Control of these fluctuations is a prerequisite for assessing the quality of the lower bound in § 7. As in § 5 A, we take d=1d=1; since every result below is expressed through λ\lambda, mm, and ss, Remark IV.8 applies without change. Because ∣0B⟩|0_B\rangle is quasi-free, all formulas of this subsection follow from the two-point functions of Lemma IV.1 by Wick’s theorem and are exact, with the exception of Remark IV.14.

We write nk=ak†akn_k=a_k^\dagger a_k and recall from the proof of Proposition IV.5 that

⟨ak†2ak2⟩=λ4+2λ2(2m+s)+2m2+s2.(50)\langle a_k^{\dagger2}a_k^2\rangle=\lambda^4+2\lambda^2(2m+s)+2m^2+s^2. \tag*{(50)}

Lemma IV.10 (Sign of the Local Moments). For g≠0g \ne0, the local moments of Eq. (37) satisfy

m>0,s<0,m+s<0.(51)m > 0,\qquad s < 0,\qquad m+s < 0. \tag*{(51)}

For g=0g=0, m=s=0m=s=0.

Proof of Lemma IV.10. For g≠0g \ne0 and admissible parameters, ωμ>2g2>0\omega_\mu> 2g^2 > 0 and 0<Ωμ<ωμ0 < \Omega_\mu< \omega_\mu, which gives m>0m > 0 and s<0s < 0 term by term. Moreover,

m+s=1M∑μ=0M−1ωμ−2g2−Ωμ2Ωμ,(52)m+s=\frac{1}{M}\sum_{\mu=0}^{M-1}\frac{\omega_\mu-2g^2-\Omega_\mu}{2\Omega_\mu}, \tag*{(52)}

and Ωμ=(ωμ−2g2)(ωμ+2g2)>ωμ−2g2\Omega_\mu=\sqrt{(\omega_\mu-2g^2)(\omega_\mu+2g^2)}>\omega_\mu-2g^2, so every term of the sum is negative. For g=0g=0, Ωμ=ωμ\Omega_\mu=\omega_\mu and both moments vanish.

Proposition IV.11 (Particle-Number Variance). In the state ∣0B⟩\lvert0_B\rangle, the variance of the on-site particle number is

Var⁡(nk)≡⟨nk2⟩−⟨nk⟩2=ρ+2λ2(m+s)+m2+s2,(53)\operatorname{Var}(n_k)\equiv\langle n_k^2\rangle-\langle n_k\rangle^2=\rho+2\lambda^2(m+s)+m^2+s^2, \tag*{(53)}

where ρ=λ2+m\rho=\lambda^2+m is the filling fraction of Eq. (40). This expression is exact within ∣0B⟩\lvert0_B\rangle, the Gaussian ground state of h1h_1.

Proof of Proposition IV.11. The commutation relation [ak,ak†]=1[a_k,a_k^\dagger]=1 gives the operator identity nk=ak†ak+nkn_k=a_k^\dagger a_k+n_k, hence ⟨nk2⟩=⟨ak†2ak2⟩+ρ\langle n_k^2\rangle=\langle a_k^{\dagger2}a_k^2\rangle+\rho. Subtracting ρ2=λ4+2λ2m+m2\rho^2=\lambda^4+2\lambda^2m+m^2 from Eq. (50) yields Eq. (53).

Two limiting cases illustrate Eq. (53). For g=0g=0, m=s=0m=s=0 and Var⁡(nk)=λ2=ρ\operatorname{Var}(n_k)=\lambda^2=\rho, the Poissonian value of a coherent state. For λ=0\lambda=0, Var⁡(nk)=m(1+m)+s2>ρ\operatorname{Var}(n_k)=m(1+m)+s^2>\rho, the super-Poissonian value of a squeezed vacuum. In general, Lemma IV.10 shows that the term 2λ2(m+s)2\lambda^2(m+s) is negative for g≠0g\ne0, so that the number fluctuations are sub-Poissonian, Var⁡(nk)<ρ\operatorname{Var}(n_k)<\rho, if and only if

λ2>m2+s22∣m+s∣.(54)\lambda^2>\frac{m^2+s^2}{2\lvert m+s\rvert}. \tag*{(54)}

Proposition IV.12 (Second-Order Correlation Function). In the state ∣0B⟩\lvert0_B\rangle with ρ>0\rho>0, the local second-order correlation function is

g(2)(0)≡⟨ak†2ak2⟩⟨ak†ak⟩2=λ4+2λ2(2m+s)+2m2+s2(λ2+m)2=1+2λ2(m+s)+m2+s2(λ2+m)2.(55)g^{(2)}(0)\equiv\frac{\langle a_k^{\dagger2}a_k^2\rangle}{\langle a_k^\dagger a_k\rangle^2} =\frac{\lambda^4+2\lambda^2(2m+s)+2m^2+s^2}{(\lambda^2+m)^2} =1+\frac{2\lambda^2(m+s)+m^2+s^2}{(\lambda^2+m)^2}. \tag*{(55)}

This expression is exact within ∣0B⟩\lvert0_B\rangle, the Gaussian ground state of h1h_1, and g(2)(0)−1=[Var⁡(nk)−ρ]/ρ2g^{(2)}(0)-1=[\operatorname{Var}(n_k)-\rho]/\rho^2.

Proof of Proposition IV.12. The first equality is Eq. (50) divided by ρ2\rho^2, with ρ\rho given by Proposition IV.3; the second follows by subtracting ρ2\rho^2 in the numerator. The last statement is a consequence of Proposition IV.11.

Values g(2)(0)>1g^{(2)}(0)>1 correspond to bunching, g(2)(0)=1g^{(2)}(0)=1 to Poissonian statistics, and g(2)(0)<1g^{(2)}(0)<1 to antibunching. For g=0g=0 and λ≠0\lambda\ne0, Eq. (55) gives g(2)(0)=1g^{(2)}(0)=1, as for a coherent state. For λ=0\lambda=0 and g≠0g\ne0, it gives g(2)(0)=2+s2/m2g^{(2)}(0)=2+s^2/m^2, which exceeds 3 because ∣s∣>m\lvert s\rvert>m by Lemma IV.10; this is the strong bunching of a squeezed vacuum. Since m+s<0m+s<0 for g≠0g\ne0, a sufficiently large displacement reverses the sign of g(2)(0)−1g^{(2)}(0)-1: the state ∣0B⟩\lvert0_B\rangle is antibunched precisely when Eq. (54) holds. As λ\lambda is proportional to f−gf-g by Eq. (39), this antibunching is induced by the coherent drive. The use of g(2)(0)g^{(2)}(0) as a blockade and nonclassicality diagnostic parallels recent studies of dynamical, multi-emitter, and squeezing-induced bosonic blockade[24, 25, 26].

Proposition IV.13 (Anomalous Averages). In the state ∣0B⟩|0_{B}\rangle,

⟨ak2⟩=⟨ak†2⟩=λ2+s.(56)\langle a_k^2\rangle=\langle a_k^{\dagger2}\rangle=\lambda^2+s. \tag*{(56)}

Proof of Proposition IV.13. By Eq. (8), ak2=a~k2+2λa~k+λ2a_k^2=\tilde{a}_k^2+2\lambda\tilde{a}_k+\lambda^2. The linear term has vanishing expectation and ⟨a~k2⟩=s\langle\tilde{a}_k^2\rangle=s by Lemma IV.1. The two averages coincide because λ\lambda and ss are real.

The difference ⟨ak2⟩−⟨ak⟩2=s\langle a_k^2\rangle-\langle a_k\rangle^2=s is the anomalous average of the fluctuations. It originates from the pairing of the modes μ\mu and M−μM-\mu in Eq. (36); it is nonzero for g≠0g\ne0 and vanishes for g=0g=0.

Remark IV.14 (Strong-Condensate Limit of the Fluctuations). If λ2≫max⁡(m,∣s∣)\lambda^2\gg\max(m,\lvert s\rvert), the terms m2+s2m^2+s^2 in Eqs. (53) and (55) may be neglected. To first order in the fluctuation moments one then has the approximations

Var⁡(nk)≈λ2(1+2m+2s)+m,g(2)(0)≈1+2(m+s)λ2+O(λ−4).(57)\operatorname{Var}(n_k)\approx\lambda^2(1+2m+2s)+m,\qquad g^{(2)}(0)\approx1+\frac{2(m+s)}{\lambda^2}+O(\lambda^{-4}). \tag*{(57)}

In contrast with Propositions IV.11 and IV.12, these are not identities; they are valid only in the strong-condensate limit. By Lemma IV.10, in this limit g(2)(0)g^{(2)}(0) approaches unity from below whenever g≠0g\ne0.

Corollary IV.15 (Fluctuations in the Thermodynamic Limit). For every admissible pair (K,g)(K,g), the limits M→∞M\to\infty of Var⁡(nk)\operatorname{Var}(n_k), g(2)(0)g^{(2)}(0) (for ρ∞>0)\rho_\infty>0), and ⟨ak2⟩\langle a_k^2\rangle exist. They are given by Eqs. (53), (55), and (56) with mm and ss replaced by the integrals m∞m_\infty and s∞s_\infty of Eq. (47). The inequalities of Lemma IV.10 hold for m∞m_\infty and s∞s_\infty as well.

Proof of Corollary IV.15. The three quantities are continuous functions of (m,s)(m,s) at fixed λ\lambda, and m→m∞m\to m_\infty, s→s∞s\to s_\infty by Corollary IV.7. The integrands of m∞m_\infty, s∞s_\infty, and m∞+s∞m_\infty+s_\infty have the signs established pointwise in the proof of Lemma IV.10.

Boundary Slope of the Bounding Hamiltonian

We finally examine the long-wavelength behavior of the continuous band Ω(q)\Omega(\mathbf{q}) of Proposition III.2, with ω(q)\omega(\mathbf{q}) and Ω(q)\Omega(\mathbf{q}) as in Eqs. (29) and (34). In contrast with § 5 A and IV B, this subsection is formulated directly for arbitrary d≥1d\ge1. We assume j>0j>0; for j=0j=0 the band is flat. Because Ω(q)\Omega(\mathbf{q}) is a function of the continuous momentum q\mathbf{q}, every statement below already refers to the thermodynamic limit, and no separate limiting corollary is needed. We use

ω(0)=ν~−j−2g2,Ω(0)=ω(0)2−4g4=(ν~−j)(ν~−j−4g2).(58)\omega(0)=\tilde{\nu}-j-2g^2,\qquad\Omega(0)=\sqrt{\omega(0)^2-4g^4}=\sqrt{(\tilde{\nu}-j)(\tilde{\nu}-j-4g^2)}. \tag*{(58)}

Proposition IV.16 (Boundary Slope). Let d≥1d\ge1 and j>0j>0.

  1. In the stable region ν~−j>4g2\tilde{\nu}-j>4g^2, the excitation spectrum has a gap, attained at q=0\mathbf{q}=0:

min⁡qΩ(q)=Ω(0)=(ν~−j−2g2)2−4g4>0.(59)\min_{\mathbf{q}}\Omega(\mathbf{q})=\Omega(0)=\sqrt{(\tilde{\nu}-j-2g^2)^2-4g^4}>0. \tag*{(59)}

As q→0\mathbf{q}\to0,

Ω(q)=Ω(0)+∣q∣22m∗+O(∣q∣4),1m∗=(ν~−j−2g2)jd Ω(0),(60)\Omega(\mathbf{q})=\Omega(0)+\frac{|\mathbf{q}|^2}{2m_*}+O(|\mathbf{q}|^4),\qquad\frac{1}{m_*}=\frac{(\tilde{\nu}-j-2g^2)j}{d\,\Omega(0)}, \tag*{(60)}

so that the dispersion is quadratic and the ratio Ω(q)/∣q∣\Omega(\mathbf{q})/|\mathbf{q}| diverges as q→0\mathbf{q}\to0: there is no finite linear boundary slope.

  1. On the boundary of the stable region, ν~−j=4g2\tilde{\nu}-j=4g^2 with g≠0g\ne0, the function Ω(q)\Omega(\mathbf{q}) of Eq. (34) remains well defined and

Ω(q)2=jdS(q)(4g2+jdS(q)),S(q)=∑α=1d(1−cos⁡qα).(61)\Omega(\mathbf{q})^2=\frac{j}{d}S(\mathbf{q})\left(4g^2+\frac{j}{d}S(\mathbf{q})\right),\qquad S(\mathbf{q})=\sum_{\alpha=1}^{d}(1-\cos q_\alpha). \tag*{(61)}

The gap is closed and the dispersion is linear at small momenta,

Ω(q)=c∣q∣+O(∣q∣3),c=∣g∣2jd.(62)\Omega(\mathbf{q})=c|\mathbf{q}|+O(|\mathbf{q}|^3),\qquad c=|g|\sqrt{\frac{2j}{d}}. \tag*{(62)}

Proof of Proposition IV.16. By Eq. (29), ω(q)=ω(0)+jdS(q)\omega(\mathbf{q})=\omega(0)+\frac{j}{d}S(\mathbf{q}) with S(q)≥0S(\mathbf{q})\ge0, and S(q)=12∣q∣2+O(∣q∣4)S(\mathbf{q})=\frac{1}{2}|\mathbf{q}|^2+O(|\mathbf{q}|^4). Hence

Ω(q)2=Ω(0)2+2ω(0)jdS(q)+j2d2S(q)2=Ω(0)2+ω(0)jd∣q∣2+O(∣q∣4).(63)\Omega(\mathbf{q})^2=\Omega(0)^2+\frac{2\omega(0)j}{d}S(\mathbf{q})+\frac{j^2}{d^2}S(\mathbf{q})^2=\Omega(0)^2+\frac{\omega(0)j}{d}|\mathbf{q}|^2+O(|\mathbf{q}|^4). \tag*{(63)}

In the stable region ω(0)>2g2≥0\omega(0)>2g^2\ge0, so the right-hand side of the first equality is increasing in SS and is minimal at q=0\mathbf{q}=0, which gives Eq. (59). Since Ω(0)>0\Omega(0)>0, the square root is smooth near q=0\mathbf{q}=0, and its expansion gives Eq. (60). On the boundary, ω(0)=2g2\omega(0)=2g^2 and Ω(0)=0\Omega(0)=0, and the first equality in Eq. (63) reduces to Eq. (61). Then Ω(q)2=2g2jd∣q∣2+O(∣q∣4)\Omega(\mathbf{q})^2=\frac{2g^2j}{d}|\mathbf{q}|^2+O(|\mathbf{q}|^4), and taking the square root gives Eq. (62).

The two regimes are connected by Eq. (63), which has the relativistic form Ω(q)2≈Ω(0)2+v2∣q∣2\Omega(\mathbf{q})^2\approx\Omega(0)^2+v^2|\mathbf{q}|^2 with v2=ω(0)j/dv^2=\omega(0)j/d, so that m∗v2=Ω(0)m^*v^2=\Omega(0). In the stable region the excitations of the bounding Hamiltonian are massive quasiparticles with effective mass m∗m^*; the dispersion is quadratic for ∣q∣≪Ω(0)/v|\mathbf{q}|\ll\Omega(0)/v and approximately linear with slope vv above this scale. As the boundary is approached, m∗→0m^*\to0, v→cv\to c, and the quadratic window shrinks to the point q=0\mathbf{q}=0; the limits q→0\mathbf{q}\to0 and ν~−j→4g2\tilde{\nu}-j\to4g^2 therefore do not commute. The velocity c=∣g∣2j/dc=|g|\sqrt{2j/d} depends only on the hopping amplitude jj, the parameter gg, and the dimension; it does not depend on the detuning ν0\nu_0 or on the drive amplitude ff, because on the boundary KK is fixed by ν~=j+4g2\tilde{\nu}=j+4g^2.

Two limitations should be kept in mind. First, KK and gg are parameters of the Lieb–Yamazaki decomposition and not of the model, so the gap, m∗m^*, and cc characterize the bounding Hamiltonian h1h_1 for a given choice of (K,g)(K,g); the boundary ν~−j=4g2\tilde{\nu}-j=4g^2 is the edge of the admissible region and not a phase boundary of the driven Bose–Hubbard model. Second, Theorem II.3 and the results of § 5 A and IV B are proved in the open stable region only. On the boundary the integrands of Eq. (47) behave as 1/(c∣q∣)1/(c|\mathbf{q}|) near q=0\mathbf{q}=0, which is integrable for d≥2d\ge2 but not for d=1d=1, where m∞m_\infty and ∣s∞∣|s_\infty| diverge logarithmically as the boundary is approached.

Remark IV.17 (Comparison with the Case g=0g=0). For g=0g=0 the spectrum is Ω(q)=ω(q)\Omega(\mathbf{q})=\omega(\mathbf{q}), the stable region is ν~−j>0\tilde{\nu}-j>0, and Eq. (60) gives the free lattice mass 1/m∗=j/d1/m^*=j/d. On the boundary ν~=j\tilde{\nu}=j one has Ω(q)=jdS(q)=j2d∣q∣2+O(∣q∣4)\Omega(\mathbf{q})=\frac{j}{d}S(\mathbf{q})=\frac{j}{2d}|\mathbf{q}|^2+O(|\mathbf{q}|^4): the gap closes, but the dispersion remains quadratic and the boundary slope vanishes, in agreement with Eq. (62). Moreover, the displacement λ\lambda of Eq. (10) diverges there unless f=0f=0. For g≠0g\ne0, in contrast, λ=−(f−g)/(4g2)\lambda=-(f-g)/(4g^2) is finite on the boundary and the gapless excitations have the linear dispersion of Eq. (62). A finite linear boundary slope at the edge of the stable region is thus a qualitative effect of the pairing terms generated by g≠0g\ne0.

Numerical Illustration of the Spectrum and Local Correlations

Figure 6(a) illustrates the two long-wavelength regimes of Proposition IV.16 for ν0=j=1\nu_0=j=1. The deep-stable curve uses the independently optimized variational parameters at f=0.5f = 0.5, K=1.90722K = 1.90722 and g=−0.42515g = -0.42515; its stability margin is approximately 0.09144 and its gap is Ω(0)≃0.27290\Omega(0) \simeq0.27290, so the dispersion is quadratic at sufficiently small momentum. For comparison, the near-boundary curve uses K=2.0005K = 2.0005 and g=0.5g = 0.5, giving ν−j−4g2=10−3\nu-j-4g^2 = 10^{-3} and a small residual gap Ω(0)≃0.03164\Omega(0) \simeq0.03164. Outside the correspondingly narrow quadratic window it follows the boundary law Ω(q)≃c∣q∣\Omega(q) \simeq c\lvert q\rvert, with

Figure 6. Excitation spectrum and local correlations at ν0=j=1\nu_0=j=1. (a) The solid curve with open-circle markers denotes the gapped, quadratic long-wavelength dispersion at the optimized deep-stable point for f=0.5f=0.5; the dashed curve with square markers denotes a point having stability margin 10−310^{-3}, and the dotted curve is the bounding-Hamiltonian boundary asymptote c∣q∣c|q| for c=∣g∣/2jc=|g|/\sqrt{2j}. (b) The on-site second-order correlation of the stable Gaussian family with fixed (K,g)=(2.0005,0.5)(K,g)=(2.0005,0.5) as the coherent drive is increased. The dashed line marks Poissonian statistics; values above and below it indicate bunching and antibunching, respectively.

c=∣g∣2j=12≃0.70711.(64)c = \frac{\lvert g\rvert}{\sqrt{2j}} = \frac{1}{\sqrt{2}} \simeq0.70711. \tag*{(64)}

This linear dispersion defines a boundary slope, which may be viewed as an effective sound velocity of the excitations of the bounding Hamiltonian h1h_1. It is not a sound velocity established for the full driven model and does not mark a phase transition of that model. Whether the full interacting Hamiltonian has a corresponding low-energy feature must be tested independently, for example by finite-size scaling or tensor-network methods. Collective-mode spectroscopy of Bose–Hubbard systems likewise distinguishes low-energy phase modes from massive amplitude response[27, 28].

Figure 6(b) holds the same stable near-boundary pair (K,g)=(2.0005,0.5)(K,g) = (2.0005,0.5) fixed and varies only the coherent drive. This isolates the displacement-induced change in the Gaussian trial state. The moments mm and ss are evaluated from Eq. (47), the filling and number variance from Eqs. (48) and (53), and g(2)(0)g^{(2)}(0) from Eq. (55). The correlation starts in the bunched regime, reaches its squeezed-vacuum maximum near f=g=0.5f = g = 0.5, and crosses the Poissonian value between f=1.45f = 1.45 and 1.50. At stronger drive it remains below unity (for example, its minimum is approximately 0.9246 near f=2f = 2), demonstrating drive-induced antibunching and sub-Poissonian fluctuations. Thus g(2)(0)<1g^{(2)}(0) < 1 exposes the nonclassical number statistics of this coherently displaced auxiliary Gaussian state; it is not, by itself, a prediction for the ground state of the full interacting Hamiltonian.

Numerical Optimization and Tightness of Bounds

Let ∣0B⟩\lvert0_B\rangle be the normalized ground state of the bounding Hamiltonian h1h_1, and write the dimensionless Hamiltonian as h=T+∑k∈ΛVkh = T + \sum_{k\in\Lambda} V_k. Its expectation μ=⟨h⟩\mu= \langle h\rangle in this state is a variational upper bound on the ground-state energy E0E_0.

Lemma V.1 (Certified Gap for the Full Hamiltonian). Let hh be the full dimensionless Hamiltonian and h1h_1 the bounding Hamiltonian of Eq. (16). In finite volume and in the stable region, the first excited-state energy of hh satisfies

E1(h)≥E0(h1)+Ω(0).(65)E_1(h) \ge E_0(h_1) + \Omega(0). \tag*{(65)}

Proof of Lemma V.1. By Proposition II.1, h=h1+∑k∈ΛBk†Bk≥h1h = h_1 + \sum_{k\in\Lambda} B_k^\dagger B_k \ge h_1 in the quadratic-form sense. The min–max principle2 therefore gives En(h)≥En(h1)E_n(h) \ge E_n(h_1) for every nn. The diagonal form of h1h_1 has spectrum E0(h1)+∑μnμΩμE_0(h_1) + \sum_\mu n_\mu\Omega_\mu, where nμ∈Z≥0n_\mu\in\mathbb{Z}_{\ge0}. Hence E1(h1)=E0(h1)+min⁡μΩμ=E0(h1)+Ω(0)E_1(h_1) = E_0(h_1) + \min_\mu\Omega_\mu= E_0(h_1) + \Omega(0) by Proposition IV.16, proving the claim. □

We next give an exact and directly implementable expression for the full variance. For self-adjoint operators XX and YY, define

Cov⁡s(X,Y)=12(⟨(X−⟨X⟩)(Y−⟨Y⟩)+(Y−⟨Y⟩)(X−⟨X⟩)⟩).(66)\operatorname{Cov}_{s}(X,Y) = \frac{1}{2}\left(\langle(X-\langle X\rangle)(Y-\langle Y\rangle) + (Y-\langle Y\rangle)(X-\langle X\rangle)\rangle\right). \tag*{(66)}

With W=∑k∈ΛVkW = \sum_{k\in\Lambda} V_k, the exact variance decomposition is

Var⁡(h)=Var⁡(T)+2Cov⁡s(T,W)+∑k,l∈ΛCov⁡s(Vk,Vl).(67)\operatorname{Var}(h) = \operatorname{Var}(T) + 2\operatorname{Cov}_{s}(T,W) + \sum_{k,l\in\Lambda}\operatorname{Cov}_{s}(V_k,V_l). \tag*{(67)}

In a translation-invariant state on the periodic lattice, the last term is

Md∑r∈ΛCov⁡(V0,Vr).(68)M^d \sum_{r\in\Lambda} \operatorname{Cov}(V_0,V_r). \tag*{(68)}

Consequently it cannot in general be replaced by MdVar⁡(V0)M^d\operatorname{Var}(V_0): that replacement would require all inter-site covariances to vanish.

For completeness, define xk=a~kx_k=\tilde a_k and the translation-invariant contractions

Nr=⟨x0†xr⟩,Sr=⟨x0xr⟩,(69)N_r=\langle x_0^\dagger x_r\rangle,\qquad S_r=\langle x_0x_r\rangle, \tag*{(69)}

which are the inverse discrete Fourier transforms of, respectively, (ωμ−Ωμ)/(2Ωμ)(\omega_\mu-\Omega_\mu)/(2\Omega_\mu) and −g2/Ωμ-g^2/\Omega_\mu. Thus N0=mN_0=m and S0=sS_0=s. If u=o1⋯o2pu=o_1\cdots o_{2p} is any even word in the xk,xk†x_k,x_k^\dagger, put

W(u)=∑pairings P∏(i,j)∈P⟨oioj⟩,W(u)=0for odd words,(70)\mathcal{W}(u)=\sum_{\text{pairings }P}\prod_{(i,j)\in P}\langle o_io_j\rangle,\qquad\mathcal{W}(u)=0\quad\text{for odd words}, \tag*{(70)}

where the four ordered contractions follow from Nr,SrN_r,S_r and ⟨xkxl†⟩=δkl+Nk−l\langle x_kx_l^\dagger\rangle=\delta_{kl}+N_{k-l}. Expanding

T=−j2d∑⟨k,l⟩[(xk†+λ)(xl+λ)+(k↔l)],(71)T=-\frac{j}{2d}\sum_{\langle k,l\rangle}\left[(x_k^\dagger+\lambda)(x_l+\lambda)+(k\leftrightarrow l)\right], \tag*{(71)}
Vk=(xk†+λ)2(xk+λ)2−ν0(xk†+λ)(xk+λ)+f(xk†+xk+2λ)(72)V_k=(x_k^\dagger+\lambda)^2(x_k+\lambda)^2-\nu_0(x_k^\dagger+\lambda)(x_k+\lambda)+f(x_k^\dagger+x_k+2\lambda) \tag*{(72)}

as X=∑ucuuX=\sum_u c_u u, define the completely explicit bilinear Wick sum

C(X,Y)=12∑u,vcudv{W(uv)+W(vu)−2W(u)W(v)}.(73)\mathcal{C}(X,Y)=\frac{1}{2}\sum_{u,v}c_ud_v\{\mathcal{W}(uv)+\mathcal{W}(vu)-2\mathcal{W}(u)\mathcal{W}(v)\}. \tag*{(73)}

All words in Eq. (73) have length at most eight.

Proposition V.2 (Energy Variance in the Bogoliubov Vacuum). In the state ∣0B⟩|0_B\rangle, the exact full-Hamiltonian variance is

Var⁡(h)=C(T,T)+2C(T,W)+C(W,W),W=∑kVk.(74)\operatorname{Var}(h)=\mathcal{C}(T,T)+2\mathcal{C}(T,W)+\mathcal{C}(W,W),\qquad W=\sum_k V_k. \tag*{(74)}

Equivalently, with Q=∑kBk†BkQ=\sum_k B_k^\dagger B_k,

Var⁡(h)=Var⁡(Q)=C(Q,Q).(75)\operatorname{Var}(h)=\operatorname{Var}(Q)=\mathcal{C}(Q,Q). \tag*{(75)}

Equations (70)–(74) are finite lattice sums in λ\lambda, mm, ss, NrN_r, SrS_r and give the requested exact expressions for Var⁡(T)\operatorname{Var}(T), Cov⁡s(T,W)\operatorname{Cov}_s(T,W), and Var⁡(W)\operatorname{Var}(W).

Proof of Proposition V.2. The first identity is the algebraic variance decomposition. The state is quasi-free, so Wick’s theorem evaluates every term exactly as Eq. (73). Finally h=h1+Qh=h_1+Q and h1∣0B⟩=E0(h1)∣0B⟩h_1|0_B\rangle=E_0(h_1)|0_B\rangle; consequently constants and all mixed covariances with h1h_1 drop out, giving Eq. (75).

Theorem V.3 (Temple’s Inequality with Certified Gap). Let μ=⟨h⟩\mu=\langle h\rangle be the variational upper bound of Remark IV.9. If μ<E0(h1)+Ω(0)\mu<E_0(h_1)+\Omega(0), then

E0≥μ−Var⁡(h)E0(h1)+Ω(0)−μ.(76)E_0\geq\mu-\frac{\operatorname{Var}(h)}{E_0(h_1)+\Omega(0)-\mu}. \tag*{(76)}

Together with Theorem II.3, this gives

max⁡{E0,l(d),μ−Var⁡(h)E0(h1)+Ω(0)−μ}≤E0≤μ.(77)\max\left\{E_{0,l}^{(d)},\mu-\frac{\operatorname{Var}(h)}{E_0(h_1)+\Omega(0)-\mu}\right\}\leq E_0\leq\mu. \tag*{(77)}

Proof of Theorem V.3. Lemma V.1 supplies the lower spectral threshold E1∗=E0(h1)+Ω(0)E_1^*=E_0(h_1)+\Omega(0). The stated condition implies μ<E1∗≤E1(h)\mu<E_1^*\leq E_1(h), so Temple’s inequality gives Eq. (76). Taking the stronger of this result and Theorem II.3, and using the variational principle for the upper bound, proves Eq. (77).

Elementary Site and Bond Baselines

The Lieb–Yamazaki estimate should be judged against two elementary rigorous lower bounds. Define the one-site ground energy

ε1(η,f)=inf⁡spec⁡[a†a2−ηa†a+f(a†+a)](78)\varepsilon_1(\eta,f)=\inf\operatorname{spec}\left[a^\dagger a^2-\eta a^\dagger a+f(a^\dagger+a)\right] \tag*{(78)}

and the two-site bond Hamiltonian

hb=Vk+Vl2d−j2d(ak†al+al†ak),Vk=ak†2ak2−ν0nk+f(ak†+ak).(79)h_b=\frac{V_k+V_l}{2d}-\frac{j}{2d}(a_k^\dagger a_l+a_l^\dagger a_k),\qquad V_k=a_k^{\dagger2}a_k^2-\nu_0n_k+f(a_k^\dagger+a_k). \tag*{(79)}

Then the exact energy density satisfies

E0(M)Md≥ε1(ν0+j,f),E0(M)Md≥dinf⁡spec⁡hb.(80)\frac{E_0(M)}{M^d}\ge\varepsilon_1(\nu_0+j,f),\qquad\frac{E_0(M)}{M^d}\ge d\inf\operatorname{spec}h_b. \tag*{(80)}

Indeed, −(ak†al+al†ak)≥−(nk+nl)-(a_k^\dagger a_l+a_l^\dagger a_k)\ge-(n_k+n_l) follows from (ak−al)†(ak−al)≥0(a_k-a_l)^\dagger(a_k-a_l)\ge0; summing bonds gives the site estimate. For the second estimate, h=∑⟨k,l⟩hbh=\sum_{\langle k,l\rangle}h_b and there are dMddM^d bonds, so the Anderson cluster argument [17] applies.

Table 1 reports a cutoff-converged numerical comparison in one dimension. The bond bound is substantially sharper than the Lieb–Yamazaki bound at every listed point, while the site bound can also be competitive. Thus the value of the present construction lies primarily in its analytic parameter dependence and its explicitly solvable auxiliary Gaussian state, not in uniformly optimal energy bounds.

(ν0,j,f)(\nu_0,j,f)LYsitebondM=4M=4 ED
(1,1,0)(1,1,0)−2.2500-2.2500−2.0000-2.0000−1.7321-1.7321−1.5988-1.5988
(1,1,0.5)(1,1,0.5)−2.9622-2.9622−2.8930-2.8930−2.6415-2.6415−2.5987-2.5987
(1,1,2)(1,1,2)−6.5354-6.5354−6.5359-6.5359−6.3781-6.3781−6.3616-6.3616
(1,0.2,5)(1,0.2,5)−13.4519-13.4519−13.4032-13.4032−13.3787-13.3787−13.3783-13.3783
(1,5,5)(1,5,5)−30.0301-30.0301−30.1787-30.1787−29.5555-29.5555−29.4296-29.4296

Table 1. Lower-bound energy densities in d=1d=1; larger values are sharper. The last column is the converged M=4M=4 truncated-Hamiltonian reference.

Optimization of the Variational Parameters

We now optimize the thermodynamic lower-bound energy density of Corollary III.1 over the admissible decomposition parameters. For fixed model parameters (ν0,j,f)(\nu_0,j,f), the formal optimum is the supremum over the open stable set,

εl∗=sup⁡K>12(1+ν0+j+4g2)∣g∣≤2ε0,l(1)(K,g).(81)\varepsilon_l^*=\sup_{\substack{K>\frac{1}{2}(1+\nu_0+j+4g^2)\\ |g|\le2}}\varepsilon_{0,l}^{(1)}(K,g). \tag*{(81)}

For numerical work we keep a distance ϵ\epsilon from the boundary and define

εl∗(ϵ)=max⁡(K,g)∈Sϵ∣g∣≤2ε0,l(1)(K,g),Sϵ={(K,g):K≥12(1+ν0+j+4g2)+ϵ}.(82)\varepsilon_l^*(\epsilon)=\max_{\substack{(K,g)\in S_\epsilon\\ |g|\le2}}\varepsilon_{0,l}^{(1)}(K,g),\qquad S_\epsilon=\left\{(K,g):K\ge\frac{1}{2}(1+\nu_0+j+4g^2)+\epsilon\right\}. \tag*{(82)}

and denote an optimizing pair by (Kl∗,gl∗)(K_l^*,g_l^*). The parameters KK and gg belong to the Lieb–Yamazaki decomposition; this optimization is over those variational parameters, not a limit such as g→0g\to0. We use ϵ=10−4\epsilon=10^{-4} in the numerical calculations. At regular boundary

points the energy density has a finite continuous limit, so this regulator is not part of the formal supremum; it stabilizes the numerical quadrature. In one dimension, however, individual Bogoliubov moments and the same-pair upper estimate can diverge or depend logarithmically on ϵ\epsilon near a gapless optimizer. We distinguish two upper-bound energy-density protocols. The upper-bound energy density evaluated at the lower-bound optimizer is

euat-lower≡eu(Kl∗,gl∗),(83)e_u^{\mathrm{at\text{-}lower}} \equiv e_u(K_l^*,g_l^*), \tag*{(83)}

whereas independent upper-bound energy-density optimization gives

eumin⁡≡min⁡(K,g)∈Sϵeu(K,g).(84)e_u^{\min} \equiv\min_{(K,g)\in S_\epsilon} e_u(K,g). \tag*{(84)}

For brevity, figures label el∗(ϵ)e_l^*(\epsilon) as el∗e_l^*. The associated euat-lowere_u^{\mathrm{at\text{-}lower}} is explicitly regulator dependent, whereas eumin⁡e_u^{\min} is the primary upper estimate used in the regime maps. All numerical multistart searches, including the independent minimization in Eq. (84), are restricted to ∣g∣≤2|g|\le2; their reported best points are not proofs of global extrema. Only the same-pair gap obeys the decomposition identity

Δat-lower=euat-lower−el∗=⟨B0†B0⟩(Kl∗,gl∗)≥0,(85)\Delta_{\mathrm{at\text{-}lower}} = e_u^{\mathrm{at\text{-}lower}} - e_l^* = \left\langle B_0^\dagger B_0 \right\rangle_{(K_l^*,g_l^*)} \ge0, \tag*{(85)}

where the equality follows from the positive-operator decomposition and the fact that ∣0B⟩|0_B\rangle is the ground state of h1h_1. For the independently optimized bracket one has only eumin⁡−el∗≥0e_u^{\min}-e_l^*\ge0 by the variational and lower-bound principles.

Figure 1 compares the optimized thermodynamic lower-bound energy density el∗e_l^* with the upper-bound energy density euat-lowere_u^{\mathrm{at\text{-}lower}} throughout 0≤f≤50\le f\le5 at fixed ν0=1\nu_0=1 for the reference hopping j=1j=1, weak hopping j=0.2j=0.2, and strong hopping j=5j=5. Numerically converged eigenvalues of the nmax⁡n_{\max}-truncated periodic M=4M=4 chain are included in every panel. We retain nmax⁡=10n_{\max}=10 outside the mixed strong regime; every marker with f≥4f\ge4 and j≥4j\ge4 was recomputed adaptively at nmax⁡=12,14,16,18n_{\max}=12,14,16,18 until consecutive energy densities differed by less than 10−410^{-4}. These points are exact for their finite-dimensional truncated Hamiltonians and serve as numerical reference values, rather than exact thermodynamic energy densities. Exact diagonalization is restricted by this rapid Hilbert-space growth; matrix-product-state and DMRG methods provide the standard scalable alternative in one dimension [29, 30, 31, 32, 33]. In panel (a), at f=0.5f=0.5, the optimization gives Kl∗≃1.6268K_l^*\simeq1.6268, gl∗≃0.2517g_l^*\simeq0.2517, el∗≃−2.9623e_l^*\simeq-2.9623, and euat-lower≃−2.0378e_u^{\mathrm{at\text{-}lower}}\simeq-2.0378, so the relative same-pair gap Δat-lower/∣el∗∣\Delta_{\mathrm{at\text{-}lower}}/|e_l^*| is about 31.2%31.2\%. This number is regulator dependent near the stability boundary. Unless independent minimization is stated explicitly, all gaps discussed for Figure 1–Figure 3 use this same-pair protocol. Comparing the panels shows that weak hopping produces the narrowest relative envelope at the upper end of the scan: at f=5f=5 the relative gaps are approximately 3.0%3.0\%, 2.3%2.3\%, and 6.4%6.4\% for j=1j=1, 0.20.2, and 55, respectively. Thus all three envelopes tighten with increasing drive, although the approach is slowest in the strong-hopping case. The color map in Figure 5 shows the lower-bound landscape at f=0.5f=0.5 and marks its numerically found maximum in the stable region.

Figure 1. Optimized thermodynamic lower-bound energy density el∗e_l^* and variational upper-bound energy density $e_u^{\mathrm{at\mathchar-lower}}$ evaluated at the same lower-bound optimizer versus drive amplitude over $0 \le f \le 5$ for (a) $j = 1.0$, (b) $j = 0.2$, and (c) $j = 5.0$, at fixed $\nu_0 = 1$. Small black dots are converged eigenvalues of the $n_{\max} = 10$ truncated periodic chain with $M = 4$ outside the mixed strong regime; black crosses for $f \ge 4$, $j \ge 4$ use adaptive $n_{\max} = 14$–$18$ with successive energy-density changes below $10^{-4}$ (see {xref}Fig. 8`). These finite-chain markers are numerical references, not TDL energy densities. The numerical step is Δf=0.05\Delta f = 0.05.

Figure 3. Optimized thermodynamic lower-bound energy density el∗e_l^* and variational upper-bound energy density euat-lowere_u^{\mathrm{at\text{-}lower}} evaluated at the same lower-bound optimizer versus detuning over 0≤ν0≤50 \le\nu_0 \le5, beginning at exact resonance, for (a) j=0.2j = 0.2, (b) j=1.0j = 1.0, and (c) j=5.0j = 5.0, at fixed strong drive f=5f = 5. Small black dots are converged eigenvalues of the nmax⁡=10n_{\max} = 10 truncated periodic chain with M=4M = 4 for j=0.2j = 0.2 and 1.01.0; black crosses for j=5j = 5 use adaptive nmax⁡=14n_{\max} = 14–1818 with successive energy-density changes below 10−410^{-4} (see Figure 8). These finite-chain markers are numerical references, not TDL energy densities. The numerical step is Δν0=0.1\Delta\nu_0 = 0.1.

Figure 5. Thermodynamic lower-bound energy-density landscape e0,l(1)(K,g)e_{0,l}^{(1)}(K,g) for ν0=j=1\nu_0=j=1 and f=0.5f=0.5. The hatched region is outside the stability constraint regularized by ϵ=10−4\epsilon=10^{-4}, and the star indicates the numerical maximum. The optimization uses a nine-start L-BFGS-B search over ∣g∣≤2|g|\leq2 inside the stable region.

The larger cutoff is essential in the mixed strong regime. At j=f=5j=f=5, increasing nmax⁡n_{\max} from 6 to 10 lowers the truncated-chain energy density from approximately −27.4099-27.4099 to −29.4181-29.4181; the adaptive calculation converges at nmax⁡=16n_{\max}=16 to approximately −29.4296-29.4296. The converged value lies between the thermodynamic lower and upper curves. Thus the earlier points above the upper curve were a local-Hilbert-space truncation artifact. The reduced cutoff error makes placement inside the thermodynamic envelope a reliable numerical diagnostic at the plotted resolution.

Figure 2 gives the complementary scan in the hopping parameter at fixed drive. For weak drive, f=0.2f=0.2, the absolute gap has a shallow minimum of approximately 0.780.78 near j=0.2j=0.2 and then broadens substantially to about 5.455.45 at j=5j=5. At strong drive, f=5f=5, it grows much more gently, from approximately 0.310.31 at j=0j=0 to 1.931.93 at j=5j=5; relative to ∣el∗∣|e_l^*|, the corresponding endpoint gaps are about 2.4%2.4\% and 6.4%6.4\%. At every sampled point the finite-chain ED marker lies between the two thermodynamic curves.

Figure 2. Optimized thermodynamic lower-bound energy density el∗e_l^* and variational upper-bound energy density euat−lowere_u^{\mathrm{at-lower}} evaluated at the same lower-bound optimizer versus hopping parameter over 0≤j≤50 \le j \le5 for (a) f=0.2f = 0.2, (b) f=1.0f = 1.0, and (c) f=5.0f = 5.0, at fixed ν0=1\nu_0 = 1. Small black dots are converged eigenvalues of the nmax⁡=10n_{\max} = 10 truncated periodic chain with M=4M = 4 outside the mixed strong regime; black crosses for f≥4f \ge4, j≥4j \ge4 use adaptive nmax⁡=14n_{\max} = 14–1616 with successive energy-density changes below 10−410^{-4} (see Figure 8). These finite-chain markers are numerical references, not TDL energy densities. The numerical step is Δj=0.05\Delta j = 0.05.

Figure 3 starts at exact resonance, ν0=0\nu_0 = 0, and tracks the bounds with increasing detuning at fixed strong drive f=5f = 5. At resonance the absolute gaps are approximately 0.228, 0.320, and 1.800 for j=0.2j = 0.2, 1, and 5, respectively. The corresponding stability margins min⁡μ(ωμ−2q2)\min_\mu(\omega_\mu- 2q^2) are approximately 0.7251, 0.3108, and 2×10−42 \times10^{-4}, so every optimum is stable; the strong-hopping solution lies close to the regularized stability boundary. The absolute gaps increase monotonically over the plotted detuning interval, reaching approximately 0.786, 1.423, and 2.739 at ν0=5\nu_0 = 5. All finite-chain ED points again lie between the two thermodynamic curves throughout the scan.

Figure 4 separates finite-size effects from the thermodynamic approximation at ν0=j=1\nu_0 = j = 1. The finite-volume Lieb–Yamazaki lower bound uses the four discrete momenta of the M=4M = 4 ring and rigorously bounds its exact ground-state energy from below. Evaluating the finite-volume variational energy and minimizing it independently over KK and gg in the stable region supplies the corresponding optimized upper bound. The thermodynamic upper curve is minimized in the same way. The separation of each lower–upper pair is therefore a direct measure of the best bracket obtained from the Bogoliubov ansatz for that system size, independently of the thermodynamic limit. Numerically, every recomputed nmax⁡=12n_{\max} = 12 ED point lies inside this finite-volume bracket over 0≤f≤30 \leq f \leq3; the smallest observed distances from the lower and upper curves are approximately 0.144 and 0.179, respectively. The cutoff ED values approximate the untruncated finite-volume energy to which the rigorous bounds apply. The finite-volume and thermodynamic upper bounds differ by at most approximately 0.0100 over the scan and their curves are nearly superposed. The same is true of the lower bounds, whose maximum absolute difference is approximately 0.0120. Thus finite-size corrections to both optimized bounds are small for M=4M = 4 in this parameter range, even though the finite-volume curves are the ones that rigorously bracket the finite system.

Figure 4. Finite-size comparison at ν0=1\nu_0 = 1, j=1j = 1, and 0≤f≤30 \le f \le3. Blue solid and dashed curves are the optimized thermodynamic Lieb–Yamazaki lower-bound energy density el∗e_l^* and independently minimized variational upper-bound energy density eumin⁡e_u^{\min}. Red dotted and dash-dotted curves are their finite-volume M=4M = 4 counterparts. Each upper curve is independently minimized over KK and gg inside the stable region, while each lower curve is independently maximized there. Black points are converged eigenvalues of the nmax⁡=12n_{\max} = 12 truncated Hamiltonian for M=4M = 4; their changes from nmax⁡=10n_{\max} = 10 are below 1.1×10−81.1 \times10^{-8}, well within the 10−410^{-4} criterion of Fig. 8. These finite-chain markers are numerical references, not TDL energy densities. The finite-volume pair isolates the ansatz error at fixed size, while its displacement from the blue curves displays finite-size effects.

The four curves use four separate optimizing pairs (K,g)(K, g) at every drive. For example, Table II gives the values at f=0.2f = 0.2. The lower-bound pairs maximize the corresponding Lieb–Yamazaki functional, whereas the upper-bound pairs independently minimize the variational energy; all four optimizations are restricted to the stable region.

BoundK∗K^*g∗g^*
Thermodynamic lower1.5401230.141461
Thermodynamic upper1.753143−0.341058
Finite-volume lower, M=4M = 41.5399110.141042
Finite-volume upper, M=4M = 41.745619−0.323582

Table II. The four independently optimized stable parameter pairs at f=0.2f = 0.2, ν0=1\nu_0 = 1, and j=1j = 1.

Numerical Assessment of the Temple Estimate

For a translation-invariant Bogoliubov vacuum, the decomposition gives

μ−E0(h1)=⟨Q⟩=Md⟨B0†B0⟩=MdΔM.(86)\mu- E_0(\mathbf{h}_1) = \langle Q\rangle= M^d\langle B_0^\dagger B_0\rangle= M^d\Delta_M. \tag*{(86)}

The Temple applicability condition is therefore

MdΔM<Ω(0).(87)M^d\Delta_M < \Omega(0). \tag*{(87)}

Unless ΔM\Delta_M vanishes with volume, its left-hand side is extensive while the certified one-particle gap is intensive. Hence a fixed trial family can satisfy the test only below a parameter-dependent size threshold; closeness to the stability boundary makes the condition still more restrictive because Ω(0)\Omega(0) closes. At a given point, the auxiliary gap would have to exceed MdΔMM^d\Delta_M; equivalently, if the present test fails, it must increase by a factor greater than MdΔM/Ω(0)M^d\Delta_M/\Omega(0) while the excess is held fixed. This is a necessary threshold, not evidence that any particular extension of the ansatz will achieve it. A finite cutoff imposed directly on the Gaussian state can also spoil this identity, so numerical tests must evaluate the Gaussian moments consistently, for example through the Wick formulas above.

This scaling observation replaces a universal negative verdict based on a small cutoff scan. At a particular finite size the condition may pass or fail, and a different admissible decomposition may improve the threshold, but neither outcome by itself diagnoses the spectrum of the full model. Equation (76) remains a valid conditional finite-volume refinement; it does not yield a thermodynamic energy-density correction for a nonzero limiting ΔM\Delta_M.

Regime Maps and Limitations of the Ansatz

The four regime maps in Figure 7 summarize where the optimized Bogoliubov construction is quantitatively tight. Panels (a) and (c) compare the (j,f)(j,f) plane at detunings ν0=1\nu_0=1 and 00, while panels (b) and (d) compare the (ν0,f)(\nu_0,f) plane over −4≤ν0≤4-4\leq\nu_0\leq4 at hoppings j=1j=1 and 0.10.1. The relative gap is below 10%10\% throughout all four sampled planes by f=2f=2; the numerical contour is already reached between approximately f=1.1f=1.1 and 1.61.6, depending on the remaining parameter. Consequently the especially conservative statement that the bounds are narrow for strong drive f≳3f\gtrsim3 and moderate hopping j≲2j\lesssim2 is satisfied, as is the corresponding statement for ν0≥1\nu_0\geq1 and f≥2f\geq2. The map in fact shows that the latter precision extends over the full sampled hopping and detuning ranges at f≥2f\geq2.

Figure 7. Regime maps of the optimized thermodynamic bounds on (a) the (j,f)(j,f) plane at ν0=1\nu_0=1, (b) the (ν0,f)(\nu_0,f) plane at j=1j=1, (c) the (j,f)(j,f) plane at exact resonance ν0=0\nu_0=0, and (d) the (ν0,f)(\nu_0,f) plane at weak hopping j=0.1j=0.1. Both detuning panels cover −4≤ν0≤4-4 \le\nu_0 \le4. Color, with logarithmic color normalization, represents the relative bracket width (eumin⁡−e∗)/∣e∗∣(e_u^{\min}-e^*)/\lvert e^*\rvert, where eumin⁡e_u^{\min} is independently minimized, clipped at 50%. Hatched regions have a relative bracket exceeding 50%. Thick dashed and dotted contours denote, respectively, the stability margin min⁡μ(ωμ−2g2)=10−3\min_\mu(\omega_\mu-2g^2)=10^{-3} and a relative gap of 10%. The haloed solid contour denotes g(2)(0)=1g^{(2)}(0)=1; the white dash-dotted lines in the detuning panels mark exact resonance. Every displayed optimum is stable, so no hatched unstable region is present.

The red dashed contour marks a stability margin of 10−310^{-3}. All optimized points shown are inside the stable region, as required by the construction; the contour identifies the narrow strip in which the solution approaches its boundary. At large hopping, in particular near j=5j=5, the lower-bound optimizer remains close to this contour over the displayed drive interval, which helps explain the comparatively wider bracket there. Crossing the actual boundary would make the Bogoliubov frequencies imaginary and the transformation physically inadmissible.

The black contours in Figure 7 mark g(2)(0)=1g^{(2)}(0)=1 for the Gaussian state selected by the lower-bound optimum. In panels (a) and (c) they separate a bunched region at weak hopping and weak drive from the antibunched region generated by coherent driving; exact resonance enlarges the bunched region. At weak hopping j=0.1j=0.1, panel (d) similarly exhibits a pronounced bunched tongue at small drive. Extending the detuning range to negative values also reveals a bunched pocket in panel (b), centered on weak drive and negative detuning, which was absent from the previously displayed nonnegative-detuning window. These are correlation crossovers, not phase transitions.

As a reading guide, the upper portions of all four panels (f≳2f\gtrsim2) are the narrow-bracket regime, below the dotted 10%10\% contour. The low-drive parts contain the widest brackets and all crossings of the haloed g(2)(0)=1g^{(2)}(0)=1 contour, while the dashed stability-margin contour identifies where the optimized auxiliary Hamiltonian approaches its gapless boundary.

The map should not be read as a conventional equilibrium phase diagram. In particular, this method does not reproduce the Mott lobes of the undriven Bose–Hubbard model. The coherent term f(ak†+ak)f(a_k^\dagger+ a_k) explicitly breaks the number-conserving U(1)U(1) symmetry, while the uniform displacement λ\lambda builds the ansatz around a coherent state. Thus every f>0f > 0 induces some coherence and removes the sharp symmetry-based Mott transition, although smooth remnants and crossovers may remain.

The Gaussian Bogoliubov vacuum also cannot capture all strong non-Gaussian correlations that may become important at very large hopping or close to the stability boundary. The widening lower–upper bracket in these regions is therefore a useful diagnostic that more elaborate correlated states or methods such as DMRG may be needed[30, 33]. It is not a failure of rigor: the Lieb–Yamazaki lower bound remains a strict lower estimate throughout the stable region, while the independently minimized Gaussian expectation is a variational upper estimate. Figure 1–Figure 3 likewise place the M=4M = 4 ED data inside the reported brackets. For example, evaluating the upper energy density at the lower-bound optimum gives, at f=0.5f = 0.5, j=ν0=1j = \nu_0 = 1, Δat-lower/∣el∗∣≃31.2%\Delta_{\mathrm{at\text{-}lower}}/|e_l^*| \simeq31.2\% at ϵ=10−4\epsilon= 10^{-4}. This same-pair value is regulator dependent near the stability boundary. Independent upper minimization used in the present map gives (eumin⁡−el∗)/∣el∗∣≃21.2%(e_u^{\min} - e_l^*)/|e_l^*| \simeq21.2\%. The reduction measures how rigidly the Gaussian state tied to the lower-bound optimum brackets the energy versus what the same Gaussian family can achieve under a separate variational minimization; the residual width diagnoses limitations of that ansatz rather than a numerical or mathematical inconsistency.

Convergence with the Local Occupation Cutoff

We finally tested the local Fock-space truncation used in the finite-chain ED data. Figure 8 follows the M=4M = 4 ground-state energy density through nmax⁡=20n_{\max} = 20 in four representative regimes. Using e(20)e(20) as the reference, the absolute error remains below 10−410^{-4} at all subsequent sampled cutoffs from nmax⁡=6n_{\max} = 6, 10, 10, and 12 in cases A–D, respectively. The corresponding cutoffs for the stricter 10−510^{-5} test are 6, 10, 12, and 14. Thus the default nmax⁡=10n_{\max} = 10 is amply converged for the weak, strong-drive, and strong-hopping examples. It is not uniformly adequate in the mixed strong regime: for case D its error at nmax⁡=10n_{\max} = 10 is 3.13×10−33.13 \times10^{-3}, and nmax⁡≥12n_{\max} \ge12 (or 14 for 10−510^{-5} accuracy) should be used there. The criterion used throughout is therefore an absolute energy-density change below 10−410^{-4} between consecutive adaptive cutoffs, supplemented by the explicit convergence study against e(20)e(20); no unverified relative 10−610^{-6} criterion is assumed.

Figure 8. Convergence of the M=4M = 4 ED ground-state energy density with the local occupation cutoff. The panels show (a) the weak case (f,j,ν0)=(0.5,0.5,1)(f,j,\nu_0) = (0.5, 0.5, 1), (b) strong drive (5,1,1)(5,1,1), (c) strong hopping (1,5,1)(1,5,1), and (d) the mixed strong case (4,4,2)(4,4,2). Insets display ∣e(nmax⁡)−e(20)∣|e(n_{\max}) - e(20)| on a logarithmic scale; dashed and dotted horizontal lines mark 10−410^{-4} and 10−510^{-5}, respectively.

Conclusions

A family of rigorous lower bounds has been derived for the ground-state energy of the coherently driven Bose–Hubbard model on periodic hypercubic lattices. By the positive-operator decomposition, followed by displacement and Bogoliubov diagonalization, the lower-bound energy density e0,l(d)e^{(d)}_{0,l} is obtained and its thermodynamic limit is established for every d≥1d \ge1. By the same construction there is produced the continuous band Ω(q)\Omega(\mathbf{q}) of Proposition III.2, including its gap, effective mass, and boundary slope (an effective sound velocity of the excitations of the bounding Hamiltonian h1h_1, not of the full driven model). Within the ground state of the bounding Hamiltonian, the order parameter, filling, energy densities, number variance, g(2)(0)g^{(2)}(0), and anomalous correlators are all explicit; in particular, the formulas identify the regime of drive-induced antibunching and sub-Poissonian fluctuations.

Finite-volume optimization for the M=4M = 4 ring changes either optimized bound from its thermodynamic-limit counterpart by at most approximately 1.2×10−21.2 \times10^{-2} over the reported scan. The dominant width of the energy bracket therefore reflects the limitations of the Gaussian ansatz rather than finite-size effects. The independently optimized upper bound and the regime maps further locate where this bracket widens and trace the g(2)(0)=1g^{(2)}(0) = 1 crossover curves separating antibunched and bunched correlations. Convergence of the ED markers is parameter dependent: nmax⁡=10n_{\max} = 10 meets the 10−410^{-4} criterion in the weak and axis-strong regimes, whereas mixed strong regimes require adaptive local cutoffs starting at nmax⁡=12n_{\max} = 12 and extending through nmax⁡=18n_{\max} = 18 when necessary.

All reported observables, including g(2)(0)g^{(2)}(0), the excitation spectrum Ω(q)\Omega(\mathbf{q}), and the boundary slope cc, are rigorous properties of the ground state ∣0B⟩\lvert0_B\rangle of the auxiliary bounding Hamiltonian h1h_1. They are not established observables or ground-state properties of the full driven Hamiltonian hh. Establishing whether the auxiliary gapless boundary has a counterpart in the full model requires independent finite-size scaling or tensor-network calculations such as DMRG or iTEBD.

The accuracy attainable with this quadratic reference state is also delimited by the numerical results. For the representative parameters ν0=j=1\nu_0 = j = 1 and f=0.5f = 0.5, the same-pair separation euat−lower−el∗e_u^{\mathrm{at-lower}} - e_l^* is about 31.2% of the lower-bound magnitude at ϵ=10−4\epsilon= 10^{-4}, although this same-pair number is regulator dependent near a gapless stability boundary; the independently optimized upper bound is the more robust numerical comparator. Lemma V.1 makes the Temple test rigorous in finite volume, but its applicability condition reads MdΔM<Ω(0)M^{d}\Delta_M < \Omega(0). Thus an extensive trial-state energy excess eventually overwhelms the intensive certified gap. Temple’s inequality remains a conditional finite-size refinement, not a general thermodynamic correction or a test of whether the ansatz controls the full spectrum.

The elementary site and Anderson bond bounds added in § 7 A are sharper at several representative points, which sets a clear benchmark for the present analytic construction. Natural extensions include finite-temperature bounds, non-hypercubic or inhomogeneous geometries, and bounding states that incorporate non-Gaussian or higher-order correlations. The exactly soluble h1h_1 is thus retained as a Gaussian reference par excellence, rather than as a surrogate for the full interacting theory. Enlarging the positive-operator family or combining it with systematically improved variational states may increase the certified excitation threshold and reduce the energy variance, the two quantities that govern whether the finite-volume Temple condition can hold.

Acknowledgments

OpenAI Codex CLI and Qwen were used as language and coding assistance tools during the preparation and revision of this manuscript. All suggestions and generated material were critically reviewed by the authors, who bear full responsibility for the content of the work.

M.J. would like to dedicate this work to the memory of his Mother.

References

  1. [1]M. P. A. Fisher, P. B. Weichman, G. Grinstein, and D. S. Fisher, Physical Review B 40, 546 (1989).
  2. [2]D. Jaksch, C. Bruder, J. I. Cirac, C. W. Gardiner, and P. Zoller, Physical Review Letters 81, 3108 (1998).
  3. [3]M. Greiner, O. Mandel, T. Esslinger, T. W. Hänsch, and I. Bloch, Nature 415, 39 (2002).
  4. [4]I. Carusotto and C. Ciuti, Reviews of Modern Physics 85, 299 (2013).
  5. [5]M. J. Hartmann, Journal of Optics 18, 104005 (2016).
  6. [6]C. Noh and D. G. Angelakis, Reports on Progress in Physics 80, 016401 (2017).
  7. [7]J. Bloch, I. Carusotto, and M. Wouters, Nature Reviews Physics 4, 470 (2022).
  8. [8]T. Fink, A. Schade, S. Höfling, C. Schneider, and A. Imamoglu, Nature Physics 14, 365 (2018).
  9. [9]R. Ma, B. Saxberg, C. Owens, N. Leung, Y. Lu, J. Simon, and D. I. Schuster, Nature 566, 51 (2019).
  10. [10]F. Vicentini, F. Minganti, R. Rota, G. Orso, and C. Ciuti, Physical Review A 97, 013853 (2018).
  11. [11]T. Grab, Physical Review A 99, 043607 (2019).
  12. [12]D. Dahan, G. Arwas, and E. Grosfeld, npj Quantum Information 8, 14 (2022).
  13. [13]L. Rassaert, T. Ramos, T. Roscilde, and D. Porras, Physical Review Letters 135, 203603 (2025).
  14. [14]M. Lewin, P. T. Nam, S. Serfaty, and J. P. Solovej, Communications on Pure and Applied Mathematics 68, 413 (2015).
  15. [15]M. Napiórkowski, R. Reuvers, and J. P. Solovej, Archive for Rational Mechanics and Analysis 229, 1037 (2018).
  16. [16]B. Brietzke, S. r. Fournais, and J. P. Solovej, Communications in Mathematical Physics 376, 323 (2020).
  17. [17]P. W. Anderson, Physical Review 83, 1260 (1951).
  18. [18]E. H. Lieb and K. Yamazaki, Physical Review 111, 728 (1958).
  19. [19]N. N. Bogoliubov, Journal of Physics (USSR) 11, 23 (1947).
  20. [20]J.-P. Blaizot and G. Ripka, Quantum Theory of Finite Systems (MIT Press, 1986).
  21. [22]M. Reed and B. Simon, Methods of Modern Mathematical Physics. IV: Analysis of Operators (Academic Press, 1978).
  22. [24]S. Ghosh and T. C. H. Liew, Physical Review Letters 123, 013602 (2019).
  23. [25]R. Trivedi, M. Radulaski, K. A. Fischer, S. Fan, and J. Vučković, Physical Review Letters 122, 243602 (2019).
  24. [26]A. Kowalewska-Kudłaszyk, S. I. Abo, G. Chimczak, J. Perina, Jan, F. Nori, and A. Miranowicz, Physical Review A 100, 053857 (2019).
  25. [27]C. Menotti and N. Trivedi, Physical Review B 77, 235120 (2008).
  26. [28]M. Endres, M. Cheneau, T. Fukuhara, C. Weitenberg, P. Schauß, C. Gross, L. Mazza, M.-C. Baňuls, L. Pollet, I. Bloch, and S. Kuhr, Nature 487, 454 (2012).
  27. [29]S. R. White, Physical Review Letters 69, 2863 (1992).
  28. [30]U. Schollwöck, Annals of Physics 326, 96 (2011).
  29. [31]R. Roth, Annals of Physics 349, 117 (2014).
  30. [32]J. Hauschild and F. Pollmann, SciPost Physics Lecture Notes , 5 (2018).
  31. [33]S. Paeckel, T. Köhler, A. Swoboda, S. R. Mamnana, U. Schollwöck, and C. Hubig, Annals of Physics 411, 167998 (2019).

Paper details

vlnc:W_pwdnvbz5
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

downloads · 2

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

  • Physics and Astronomy

Codex, Qwen (multiple versions)

View statement
Looking for related papers…
Contents