Hubbard--Heisenberg Thermodynamic Comparison at Half Filling in a Fixed Staggered Field

arXiv:2605.30663 · math-ph, math.MP · Submitted 2026-05-28 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematical Physics. The day in mathematical physics: rigorous models of physical systems.

Kai: We're looking at the paper: "Hubbard--Heisenberg Thermodynamic Comparison at Half Filling in a Fixed Staggered Field".

Mira: Here's the gist of the paper. We study the repulsive Hubbard model at half filling in the strong-coupling regime, with a staggered magnetic field of strength h.

Kai: The analysis is carried out in the canonical half-filled ensemble, with temperature measured on the Heisenberg scale J0(U) = 4t 2/U.

Mira: Uniformly for h ≤ h0 and βJ0(U) ≥ l0, we prove finite-volume Hubbard–Heisenberg pressure estimates with errors uniform in the system size.

Kai: These estimates pass to thermodynamic limits whenever the limiting pressures exist.

Paper summary: Kai: So we're looking at this paper "Hubbard--Heisenberg Thermodynamic Comparison at Half Filling in a Fixed Staggered Field". What's the main idea here?

Mira: It’s basically studying the repulsive Hubbard model at half filling when there’s a staggered magnetic field, and they are comparing it to the Heisenberg model. The thesis is that they can prove finite-volume estimates for pressure and magnetization errors that stay uniform with respect to system size, provided you're in the strong-coupling regime where temperature is measured on the Heisenberg scale J0(U) equals four t squared over U.

Lev: But for us building hardware, what does this comparison actually tell us about scaling up? Does it just show that things match up in the limit?

Kai: It shows that even with these finite-volume estimates, they pass to the thermodynamic limits whenever those limiting pressures actually exist. That’s a big deal because it means we can trust these comparisons when we try to see what happens on a really large scale.

Mira: The method they use involves this strong-coupling unitary transformation which separates the singly occupied spin sector from all the other sectors, like empty or doubly occupied sites. They compare the effective Hamiltonian on that singly occupied sector with a Heisenberg reference Hamiltonian, and then they control the rest by looking at how that transformed partition function decomposes based on those charge defect sectors.

Lev: So it’s not just a direct comparison of Hamiltonians; they're doing this decomposition of the partition function to manage those other sectors. That sounds like it introduces a lot of complexity for any simulation, doesn't it?

Kai: It does, but the point is that by controlling those remaining sectors through that decomposition, they can derive the magnetization comparison from the pressure comparison using some convexity argument on finite-volume pressures.

Mira: And they also prove charge-sector suppression estimates uniformly for all of h less than or equal to h0 in this strong-coupling regime. They show that the density of those empty or doubly occupied sites and the double occupancy density are small, and the squared staggered charge divided by lambda squared is small.

Lev: That suppression part is key for us because it suggests that at this scale, the results give a quantitative Gibbs-state formulation where we can treat the half-filled repulsive Hubbard model using effective antiferromagnetic spin degrees of freedom while charge fluctuations are suppressed.

Paper summary: Kai: Exactly, so it means the low-energy physics is dominated by those spin degrees of freedom when you look at it on the Heisenberg scale, J0(U) equals four t squared over U.

Mira: The paper then lays out Theorem one point one which gives a Hubbard–Heisenberg pressure comparison on this bounded field window <ref:2605.30663#pg1>. It says there exists a U0 dependent on l0, d, h0 and t such that for all U greater than or equal to U0, all beta greater than zero satisfying beta J0(U) is greater than or equal to l0, all even tori lambda L, and all h less than or equal to h0.

Lev: I need those numbers. What exactly does that pressure comparison look like in practice for running on real hardware?

Kai: The theorem says that the difference between the Hubbard pressure p Hub L beta U (h) and the Heisenberg pressure p Heis L beta U (h) is less than or equal to an error function epsilonFE(U, beta; h0). And this error function is uniform in the system size.

Mira: And they also state that this error function goes to zero as U goes to infinity, provided you keep the condition beta J0(U) greater than or equal to l0. That means for very large U, the pressure estimates converge to the true thermodynamic limit comparison.

Lev: So if we were designing an error correction scheme, this suggests that if we operate above a certain coupling threshold U0, our effective spin picture is really close to the actual Hubbard physics in terms of pressure scaling.

Kai: And they also have Theorem one point two which focuses on the magnetization comparison for fixed positive-field windows I less than or equal to h0 <ref:2605.30663#pg2>. It shows that there exists a U0 dependent on I, l0, d, and t such that for all U greater than or equal to U0, all beta greater than zero satisfying beta J0(U) is greater than or equal to l0, all even tori lambda L, and all h in the window I.

Mira: The magnetization comparison m Hub L beta U (h) minus m Heis L beta U (h) is less than or equal to an error function epsilonmag(U, beta; I). Again, this error function goes to zero as U goes to infinity under the same condition on beta.

Lev: That’s helpful for the spin sector because it gives us a quantitative way to bound how far off our effective spin picture is when we fix a magnetic field in that range. What about charge? Does this paper address charge fluctuations at all?

Paper summary: Kai: Yes, they have Theorem one point nine which deals with charge-sector suppression and the absence of macroscopic Charge Density Waves, or CDW <ref:2605.30663#pg1>. They show there exists a U0 dependent on l0, d, h0 and t such that for all U greater than or equal to U0, all beta greater than zero satisfying beta J0(U) is greater than or equal to l0, all even tori lambda L, and all h less than or equal to h0.

Mira: This theorem gives an error function epsilonch(U, beta; h0) that bounds the average charge density at a point qx in the system. They also show that for staggered charge density waves Cch L, the average is bounded by epsilonch(U, beta; h0), and twice the square of this average is less than or equal to epsilonch(U, beta; h0).

Lev: That suppression of charge fluctuations at a quantitative level means we can probably focus our error correction overhead on the spin sector because the charge part isn't causing macroscopic problems in this regime.

Kai: So, putting it together, these results give us a quantitative Gibbs-state formulation of the strong-coupling picture where we see that the half-filled repulsive Hubbard model is described at the Heisenberg scale J0(U) by effective antiferromagnetic spin degrees of freedom while charge fluctuations are suppressed. This paper is really showing us how to isolate those relevant degrees of freedom.

Mira: The authors are essentially using this strong-coupling unitary transformation to separate the physics into these distinct sectors, and they use that separation to control the remaining terms in a way that allows for these pressure, magnetization, and charge estimates uniformly across system sizes in finite volumes.

Lev: So what I see is that if you build an error correction protocol based on this picture, you’d focus your resources on accurately mimicking the Heisenberg spin dynamics because those are what they've shown are robust against the charge noise at this scale.

Kai: And the main point of these theorems is that even though we have these finite-volume estimates, they reliably pass to thermodynamic limits when you meet those conditions on U and beta J0(U).

Mira: The main implication for us is that it provides a rigorous way to argue that the low-energy thermodynamics of this model at the Heisenberg scale is governed by effective antiferromagnetic spin dynamics.

Lev: It’s a solid foundation for realizing how we might approach simulating these strongly correlated systems with error correction, focusing on the spin sector where things are controlled and predictable.

Conclusion: Mira: So, we're wrapping up this session on "Hubbard--Heisenberg Thermodynamic Comparison at Half Filling in a Fixed Staggered Field".

Kai: It boils down to them taking the Hubbard model and seeing how it behaves when you add that staggered magnetic field, comparing it directly to the Heisenberg model.

Mira: And they prove that at half filling, if you operate in this strong-coupling regime where temperature is set by J0 of U equals four t squared over U, you can reliably estimate the pressure and magnetization errors across finite volumes.

Lev: So they’re saying the effective spin picture really holds up when you look at the thermodynamics on that specific energy scale, J0 of U.

Kai: Exactly. They use a strong-coupling transformation to separate the physics into spins and charge sectors, showing that the charge fluctuations are suppressed in this regime.

Mira: That’s significant because it suggests that for these strongly correlated systems at half filling, you can treat them primarily as antiferromagnetic spins without worrying about big charge density waves messing things up.

Lev: For our error correction work, this means we can focus our efforts on the spin degrees of freedom because the charge noise is effectively dampened by this suppression.

Kai: The authors provide quantitative bounds for pressure, magnetization, and even for charge density waves that go to zero as U goes to infinity under certain conditions.

Mira: So in simple terms, they’ve shown that at half filling and strong coupling, you can use the Heisenberg model as a very good approximation for the Hubbard model when you are interested in its low-energy thermodynamics.

Lev: It gives us a concrete target for what we should be measuring on real hardware when trying to understand these materials.

Kai: And it opens up new avenues for how we might design error correction schemes that prioritize preserving those spin correlations over charge ones.

Department of Mathematics, Hokkaido University

math-ph, math.MP

Submitted: 2026-05-28

Updated: 2026-10-28

Comments: Substantially revised version with corrected proofs and improved bounds on pressure, magnetization, and charge-defect density

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 90/100

The gist: We study the repulsive Hubbard model at half filling in the strong-coupling regime, with a staggered magnetic field of strength h.

Key concepts

Hubbard–Heisenberg pressure comparison
This compares the pressure calculated from the full Hubbard model with the pressure from a simpler Heisenberg reference model. The paper proves that this difference is small (bounded by an error term) in finite volumes and vanishes in the thermodynamic limit, confirming that at high U, the Hubbard model behaves like a Heisenberg system regarding its energy scale.
Strong-coupling unitary transformation
This mathematical tool is used to simplify the complex Hubbard Hamiltonian. It separates the system into sectors: one where spins are singly occupied and others involving empty or doubly occupied sites. This separation allows researchers to compare the effective physics on the spin sector with a simpler Heisenberg reference.
Heisenberg scale J0(U)
This is a specific energy scale defined as 4t^2/U, where t is the hopping parameter and U is the on-site repulsion. The analysis focuses on the strong-coupling regime where this scale dictates the low-energy behavior of the system, suggesting that spin interactions dominate over charge fluctuations.

Terminology

Summary

We study the repulsive Hubbard model at half filling in the strong-coupling regime, with a staggered magnetic field of strength h. The analysis is carried out in the canonical half-filled ensemble, with temperature measured on the Heisenberg scale J0(U) = 4t 2/U. Uniformly for h ≤ h0 and βJ0(U) ≥ l0, we prove finite-volume Hubbard–Heisenberg pressure estimates with errors uniform in the system size. These estimates pass to thermodynamic limits whenever the limiting pressures exist.

The proof uses a strong-coupling unitary transformation which separates the singly occupied spin sector from sectors containing empty or doubly occupied sites. On the singly occupied sector, the effective Hamiltonian is compared with the Heisenberg reference Hamiltonian; the remaining sectors are controlled through a decomposition of the transformed partition function according to the set of empty or doubly occupied sites. For fixed positive staggered-field windows I ⋐ (0, h0], the magnetisation comparison is then derived from the pressure comparison by convexity of finite-volume pressures.

We also prove charge-sector suppression estimates, uniformly for h ≤ h0: in the large positive-U Heisenberg-scale regime, the density of empty or doubly occupied sites and the double-occupancy density are small, and the squared staggered charge divided by Λ2 is small. Thus the results give a quantitative Gibbs-state formulation of the strong-coupling picture in which the half-filled repulsive Hubbard model is described, at the Heisenberg scale, by effective antiferromagnetic spin degrees of freedom, while charge fluctuations are suppressed.

Theorem 1.1 (Hubbard–Heisenberg pressure comparison on the bounded field window). Fix l0 > 0. There exist U0 = U0(l0, d, h0, t) and an error function εFE(U, β; h0) ≥ 0 such that, for all U ≥ U0, all β > 0 satisfying βJ0(U) ≥ l0, all even tori ΛL, and all h ≤ h0,

p Hub LβU (h) − p Heis LβU (h)

≤ εFE(U, β; h0). Moreover,

lim U→∞ supβJ0(U)≥l0

εFE(U, β; h0) = 0.

Theorem 1.2 (Fixed positive-field magnetisation comparison). Fix l0 > 0 and I ⋐ (0, h0]. There exist U0 = U0(I, l0, d, t) and an error function εmag(U, β; I) ≥ 0 such that, for all U ≥ U0, all β > 0 satisfying βJ0(U) ≥ l0, all even tori ΛL, and all h ∈ I,

m Hub LβU (h) − m Heis LβU (h)

≤ εmag(U, β; I). Moreover,

lim U→∞ supβJ0(U)≥l0

εmag(U, β; I) = 0.

Theorem 1.9 (Charge-sector suppression and absence of macroscopic CDW). There exist U0 = U0(l0, d, h0, t) and an error function εch(U, β; h0) ≥ 0 such that, for all U ≥ U0, all β > 0 satisfying βJ0(U) ≥ l0, all even tori ΛL, and all h ≤ h0, one has

1/ΛL X x∈ΛL ω Hub LβU (qx) ≤ εch(U, β; h0). Consequently,

1/ΛL X x∈ΛL ω Hub LβU (nx↑nx↓) ≤ 1/2 εch(U, β; h0).

Moreover,

1/ΛL X x∈ΛL ω Hub LβU (Cch L) ≤ εch(U, β; h0), and

1/ΛL 2 ω Hub LβU (Cch L) squared ≤ εch(U, β; h0).

Finally,

lim U→∞ supβJ0(U)≥l0

εch(U, β; h0) = 0.

Corollary 1.5 (Infinite-volume pressure and fixed-field magnetisation comparison). Assume Assumption 1.3. Under the assumptions of Theorem 1.1, one has, for all h ≤ h0,

p Hub βU (h) − p Heis βU (h)

≤ εFE(U, β; h0).

Let I ⋐ (0, h0]. Under the assumptions of Theorem 1.2, one has, for every h ∈ int I ∩ D Hub βU ∩ D Heis βU,

the magnetisation comparison

m Hub βU (h) − m Heis βU (h)

≤ εmag(U, β; I).

Corollary 1.7 (Positive fixed-field Hubbard magnetisation). Fix l0 > 0 and I ⋐ (0, h0], and set hI:= inf I > 0. Let εmag(U, β; I) be the error function in Theorem 1.2. Let Cd < ∞ be the constant in Lemma D.5, and define

mlb Hub(U, β; I):= 1/2 tanh βhI 4 − 4Cd hI J0(U) − εmag(U, β; I).

Then, after increasing the strong-coupling threshold if necessary, for all U above this threshold, all β > 0 satisfying βJ0(U) ≥ l0, all even tori ΛL, and all h ∈ I, one has

m Hub LβU (h) ≥ mlb Hub(U, β; I).

Moreover,

lim U→∞ infβJ0(U)≥l0 mlb Hub(U, β; I) = 1/2. The final lower bound by 1/4 follows by increasing the strong-coupling threshold.

The main strategy involves using a strong-coupling unitary transformation to separate the singly occupied spin sector from sectors containing empty or doubly occupied sites, comparing the effective Hamiltonian on the singly occupied sector with the Heisenberg reference model, and controlling higher-order remainders and charge defects. The comparison is uniform in system size for finite volumes and passes to thermodynamic limits under specified conditions. The results show that at half filling in the strong-coupling regime, charge fluctuations are suppressed, and low-energy thermodynamics are governed by effective antiferromagnetic spin degrees of freedom at the Heisenberg scale J0(U) = 4t 2/U.

Relevant parts quoted:

"We study the repulsive Hubbard model at half filling in the strong-coupling regime, with a staggered magnetic field of strength h. The analysis is carried out in the canonical half-filled ensemble, with temperature measured on the Heisenberg scale J0(U) = 4t 2/U. Uniformly for h ≤ h0 and βJ0(U) ≥ l0, we prove finite-volume Hubbard–Heisenberg pressure estimates with errors uniform in the system size."

"The proof uses a strong-coupling unitary transformation which separates the singly occupied spin sector from sectors containing empty or doubly occupied sites. On the singly occupied sector, the effective Hamiltonian is compared with the Heisenberg reference Hamiltonian; the remaining sectors are controlled through a decomposition of the transformed partition function according to the set of empty or doubly occupied sites."

"Thus results give a quantitative Gibbs-state formulation of the strong-coupling picture in which the half-filled repulsive Hubbard model is described, at the Heisenberg scale, by effective antiferromagnetic spin degrees of freedom, while charge fluctuations are suppressed."

"Theorem 1.1 (Hubbard–Heisenberg pressure comparison on the bounded field window). Fix l0 > 0. There exist U0 = U0(l0, d, h0, t) and an error function εFE(U, β; h0) ≥ 0 such that, for all U ≥ U0, all β > 0 satisfying βJ0(U) ≥ l0, all even tori ΛL, and all h ≤ h0,"

p Hub LβU (h) − p Heis LβU (h)"

≤ εFE(U, β; h0)."

"Theorem 1.2 (Fixed positive-field magnetisation comparison). Fix l0 > 0 and I ⋐ (0, h0]. There exist U0 = U0(I, l0, d, t) and an error function εmag(U, β; I) ≥ 0 such that,"

m Hub LβU (h) − m Heis LβU (h)"

≤ εmag(U, β; I)."

Theorem 1.9 (Charge-sector suppression and absence of macroscopic CDW). There exist U0 = U0(l0, d, h0, t) and an error function εch(U, β; h0) ≥ 0 such that,

1/ΛL X x∈ΛL ω Hub LβU (qx) ≤ εch(U, β; h0)."

Improvements for AI systems

  1. Robust thermodynamic limit estimation for strong coupling regimes allows AI systems to predict macroscopic physical properties of half-filled Hubbard models by comparing them rigorously to effective antiferromagnetic Heisenberg spin models on the Heisenberg scale J0(U) = 4t2/U.

  2. Quantitative error bounds for pressure and magnetisation comparison allow AI systems to determine the accuracy of effective low-energy descriptions, specifically providing a quantitative Gibbs-state formulation where charge fluctuations are suppressed in the large positive U limit.

  3. The ability to bound the density of empty or doubly occupied sites and double occupancy density allows AI systems to rigorously confirm that charge defects are dilute and macroscopic staggered charge order is excluded in the thermodynamic normalization used by the model.

  4. The fixed-field magnetisation comparison theorem enables AI systems to establish explicit lower bounds for Hubbard magnetisation, such as mHub(U, beta; I) >= mlbHub(U, beta; I), providing a quantitative transfer statement from the Heisenberg reference model to the full Hubbard model under positive external fields.

  5. The LS/SW diagonalisation scheme and its convergence properties allow AI systems to compute effective Hamiltonian parameters J(h) and heff(h) which capture renormalized interactions, enabling the comparison of these intermediate effective spin Hamiltonians with the fixed-field Heisenberg reference model.

Related papers