Twisting the Hubbard model into the Momentum-Mixing Hatsugai-Kohmoto Model

The paper introduces the momentum-mixing Hatsugai-Kohmoto (MMHK) model, a continuously deformable framework that systematically restores momentum mixing to the exactly solvable Hatsugai-Kohmoto model, thereby accurately reproducing the Hubbard model's strongly correlated physics with superior convergence rates compared to standard finite-cluster techniques.

The Gist

Imagine you are trying to understand how a crowd of people behaves in a room. In the world of physics, these "people" are electrons, and the "room" is a crystal lattice. The most famous rulebook for how these electrons interact is called the Hubbard Model. It's the gold standard for understanding materials like cuprate superconductors (which can conduct electricity with zero resistance).

However, there's a catch: The Hubbard Model is incredibly difficult to solve. It's like trying to predict the exact path of every person in a mosh pit while they are all bumping into each other. The math gets so messy that even the smartest supercomputers struggle to get a perfect answer, especially for 2D materials (like flat sheets of atoms).

On the other hand, there is a simpler, "cheat code" model called the Hatsugai-Kohmoto (HK) model. It's easy to solve, but it's a bit of a lie. It assumes electrons only care about each other if they are in the exact same "seat" (momentum state), ignoring the fact that in the real world, electrons interact based on their physical location. It's like saying people in a room only bump into each other if they are wearing the exact same hat, ignoring the fact that they might bump into someone standing right next to them.

The Big Idea: Twisting the Cheat Code

The authors of this paper asked a clever question: Can we take this simple "cheat code" model and slowly twist it until it becomes the real, difficult model, without losing our ability to solve it?

They say "Yes." They created a new model called the Momentum-Mixing Hatsugai-Kohmoto (MMHK) model.

Here is the analogy they use:

• The Old Way (HK Model): Imagine you have a room with 100 seats. In the HK model, you group people by their "hat color" (momentum). If two people have the same hat, they repel each other. But people with different hats never interact. This is too simple.
• The New Way (MMHK Model): The authors say, "Let's mix it up." They take a small group of seats (say, 2, 4, or 10 seats) and force the people sitting in them to swap places and interact. They call this "mixing momenta."
  • If you mix 2 seats, you get a slightly better approximation.
  • If you mix 4 seats, it gets even better.
  • If you mix 10 seats, it becomes incredibly accurate.

The Magic Result: Speed and Accuracy

The most surprising part of their discovery is how fast this works.

Usually, when scientists try to approximate a complex system by adding more pieces (like adding more seats to your group), the accuracy improves slowly, like walking up a gentle hill. If you double the number of seats, you only get a little bit closer to the truth.

The authors found that their MMHK model is like a rocket ship.

• When they increased the number of mixed seats from 1 to 10, the model didn't just get a little better; it got 99% accurate to the real Hubbard Model.
• They call this a "square law" improvement. It means that if you double your effort (mixing twice as many momenta), you get four times the accuracy. This is much faster than the standard methods used today.

What Did They Prove?

They tested this new model in two scenarios:

1. One Dimension (A Line of Atoms): They compared their results to the only known perfect solution (the Bethe Ansatz). With just 10 mixed momenta, their model was within 1% of the perfect answer. Standard methods would need thousands of atoms to get that close.
2. Two Dimensions (A Flat Sheet): This is the "hard mode" where the Hubbard Model is usually unsolved. They applied their model to a square grid. Even with a small number of mixed momenta (like 4 or 16), their model successfully reproduced all the known "tricks" of real materials, such as:
   • The Mott Transition: How a material suddenly stops conducting electricity and becomes an insulator.
   • Antiferromagnetism: How electron spins align in a checkerboard pattern.
   • Pseudogaps: A mysterious state where the material acts like it's half-metal and half-insulator.
   • Heat Capacity: How the material stores heat, showing distinct peaks that separate charge and spin behaviors.

Why Does This Matter?

Think of the MMHK model as a high-fidelity simulator.

• Old Simulators: To get a clear picture, you need a massive, expensive supercomputer running for days, and you still might not be sure if the result is perfect.
• The MMHK Simulator: You can get a picture that is 99% clear using a tiny, simple setup. It captures the "soul" of the complex physics (the Mott physics) while remaining mathematically solvable.

The authors conclude that this model offers a new, powerful tool for physicists. It allows them to study strong electron interactions (which are the key to understanding high-temperature superconductors) with a level of speed and precision that was previously impossible, all by simply "mixing" a few momentum states together.

In short: They found a way to turn a simple, solvable toy model into a highly accurate replica of the real, complex world of electrons, and they did it with surprisingly little effort.

Technical Summary

Technical Summary: Twisting the Hubbard Model into the Momentum-Mixing Hatsugai-Kohmoto Model

Problem Statement
The Hubbard model is the foundational theoretical framework for understanding strongly correlated electron systems, such as cuprate superconductors. However, its exact solution remains elusive in dimensions $d=2$ and $d>1$ due to the non-commutativity of kinetic and potential energy terms. This non-commutativity prevents the straightforward ordering of eigenstates by double occupancy, rendering standard analytical techniques insufficient. While numerical methods like Quantum Monte Carlo (QMC), Density Matrix Renormalization Group (DMRG), and Dynamical Mean-Field Theory (DMFT) have provided insights, they face challenges such as the sign problem, finite-size scaling issues, or the need for cluster extensions to capture non-local correlations. Conversely, the exactly solvable Hatsugai-Kohmoto (HK) model captures Mott physics through momentum-space interactions but fails to reproduce key Hubbard features (e.g., antiferromagnetism, specific spectral weight transfer) because it assumes locality in momentum space (long-range in real space) and lacks momentum mixing. The central challenge addressed is finding a systematic way to interpolate between the solvable HK model and the unsolved Hubbard model without sacrificing exact solvability.

Methodology: The Momentum-Mixing HK (MMHK) Model
The authors propose the Momentum-Mixing Hatsugai-Kohmoto (MMHK) model, a continuous deformation of the standard HK model that reintroduces momentum mixing.

• Construction: The method involves grouping $n$ distinct momenta into a "cell" and hybridizing them. In the $n$-MMHK model, each momentum state $k$ in the reduced Brillouin zone (rBZ) is decorated by $n$ sites (or orbitals).
• Hamiltonian: The kinetic energy term becomes a matrix $g_{\alpha\alpha'}(k)$ describing hopping between the $n$ sites within the cell, while the interaction term remains diagonal in the mixed basis but acts on the $n$ sites per momentum state.
• Limit: As $n \to N$ (where $N$ is the total number of lattice sites), the reduced Brillouin zone shrinks to a single point, and the interaction term becomes purely local in real space, recovering the standard on-site Hubbard model.
• Solvability: Despite the non-commutativity introduced by mixing, the Hamiltonian remains exactly solvable for moderate $n$ because it reduces to finding the eigenstates of an $n$-site cluster for each $k$-point.

Key Contributions and Results

1. Rapid Convergence in 1D:

   • In one dimension, the ground-state energy of the $n$-MMHK model converges to the exact Bethe ansatz result of the Hubbard model with remarkable speed.
   • At $n=10$, the deviation is less than 1%.
   • The convergence scales as $1/n^{\gamma}$ with $\gamma \approx 2$ (specifically $1.83$ to $2.07$ depending on $U$), significantly outperforming standard finite-cluster DMRG calculations which scale inversely linearly ($1/L$).
   • The model correctly captures the vanishing of the Mott gap as $U \to 0$, a feature often missed by finite-size calculations with open boundary conditions.

2. Reproduction of 2D Physics:

   • Applying the $n$-MMHK model to a square lattice (using $n=4$ and $n=16$ clusters) reproduces state-of-the-art simulation results.
   • Mott Transition: The model correctly predicts the transition to an insulating state for any non-zero $U$ at half-filling (due to Fermi surface nesting) and captures the correct behavior of double occupancy, which starts at $1/4$ (matching the non-interacting Hubbard limit) rather than $1/2$ (the band HK limit).
   • Magnetic Correlations: Unlike the ferromagnetic instability of the band HK model, the $n$-MMHK model exhibits antiferromagnetic (AF) correlations as the leading spin response, consistent with Hubbard physics.
   • Spectral Functions: Using $n=16$ (a $4 \times 4$ cluster per $k$), the spectral function $A(k, \omega)$ quantitatively matches QMC and cluster perturbation theory results, capturing the narrowing of Hubbard bands and particle-hole asymmetry for $t' \neq 0$.
   • Pseudogap: The model successfully generates a pseudogap in the density of states (DOS) for underdoped regions when next-nearest-neighbor hopping ($t'$) is included, consistent with recent studies on twisted boundary condition clusters.
   • Thermodynamics: The heat capacity exhibits a two-peak structure separating charge and spin degrees of freedom, and low-temperature behavior follows a power law consistent with charge-neutral excitations.

3. Dynamical Spectral Weight Transfer (DSWT):

   • The $n$-MMHK model captures the non-trivial dynamical mixing between upper and lower Mott sub-bands.
   • It reproduces the phenomenon where the lower Hubbard band weight exceeds the simple electron count ($1+x$), a feature previously only captured by complex numerical methods, not analytically solvable models.

Scaling and Theoretical Justification
The paper argues that the efficiency of the MMHK model stems from the introduction of quantum fluctuations via momentum mixing. The convergence rate scales as $1/n^2$ in 1D and potentially faster ($1/n^3$) in quasi-2D ladders, suggesting that in higher dimensions, fewer mixed momenta are required to achieve convergence. Theoretically, the authors posit that both the HK and Hubbard models lie in the same universality class governed by the breaking of a discrete $Z_2$ symmetry at the Mott fixed point. Since the momentum-mixing construction constitutes a local deformation that preserves the Luttinger surface (the zero surface of the Green's function), the charge dynamics remain robust and converge rapidly to the Hubbard limit.

Significance and Claims
The authors claim that the MMHK model offers a powerful, alternative tool for studying strongly correlated quantum matter. Its primary significance lies in providing a systematic, controllable, and exactly solvable framework that interpolates between the solvable HK limit and the complex Hubbard model.

• Efficiency: It achieves quantitative accuracy with a small number of mixed momenta ($n=4$ to $16$), offering a computational advantage over large-cluster numerical methods.
• Insight: It bridges the gap between analytical solvability and numerical complexity, allowing for analytical insights into phenomena like DSWT and the pseudogap that are typically inaccessible in solvable models.
• Universality: The work suggests that the essential charge physics of the Mott transition is robust against local deformations, reinforcing the idea that the charge gap is the universal feature of Mott insulators, while spin sector differences (e.g., AF vs. FM) arise from specific details of the interaction.

The paper concludes that the MMHK model serves as an efficient simulator for Mott/Hubbard physics, enabling the study of the interplay between correlations, topology, and non-equilibrium dynamics with a simplified yet faithful platform.