Semiclassical representation of the Hubbard model

This paper proposes a nonperturbative semiclassical approach for the Hubbard model using an unconventional coherent-state representation with static dynamical variables, which qualitatively reproduces exact results for small systems at finite temperature while offering a framework extendable to multiorbital systems.

The Gist

The Big Picture: The "Too Hard" Puzzle

Imagine you are trying to predict the weather in a tiny, crowded city where every single person (an electron) is constantly bumping into their neighbors, arguing about personal space, and trying to move around.

In physics, this is the Hubbard Model. It's the "standard model" for understanding how electrons behave in materials like copper-oxide superconductors. The problem is that electrons are quantum objects. They don't just sit in one place; they exist in a fuzzy cloud of probabilities, and they can be in two places at once. Calculating exactly what happens when billions of these fuzzy, arguing electrons interact is mathematically impossible for even the most powerful supercomputers.

Physicists usually have to use "approximations" (shortcuts) to get an answer. This paper proposes a new, clever shortcut that is surprisingly accurate and easier to use than the old ones.

---

The Old Way: The "Crowded Dance Floor"

Traditionally, to solve this problem, physicists use a method called the "Path Integral." Imagine the electrons are dancers on a floor.

• The Problem: In the old method, you have to track every single dancer's every possible move, spin, and jump simultaneously. It's like trying to film a dance floor with a million cameras, all recording in slow motion, and then trying to edit the footage into a coherent story. It's computationally exhausting.
• The Old Shortcut: Previous shortcuts tried to freeze the dancers' movements to make the math easier, but this often missed the "chemistry" between the dancers (how they influence each other across the room).

The New Way: The "Two-Team Strategy"

The authors (Yamasaki, Suwa, Batista, and Hoshino) came up with a fresh way to look at the dancers. Instead of treating every electron as a complex, fuzzy quantum object, they split the problem into two distinct teams:

1. The "Static" Team (The Bosses): These represent the average behavior of the electrons. Think of them as the managers of the dance floor. They decide the general mood (is the room hot or cold? are people crowded or sparse?). In the paper's math, these are the "bosonic variables" (represented by angles on a sphere). They are treated as classical (fixed and predictable) for the sake of the calculation.
2. The "Quantum" Team (The Dancers): These are the actual electrons moving around. However, the authors realized they don't need to track all the quantum fuzziness. They only need to track one simple "switch" (a Grassmann variable) that tells us if a spot is empty, has one person, or has two people.

The Analogy:
Imagine a busy restaurant.

• Old Method: You try to calculate the exact path, speed, and conversation of every single waiter and customer in real-time.
• New Method: You assume the waiters (the static team) are standing still, holding their trays in a specific pose. You only track the customers (the quantum team) moving between tables. By simplifying the waiters' movements, you can still predict exactly how many people are eating, how full the tables are, and how fast the food is moving, without needing a supercomputer.

Why is this Special?

The paper highlights three major wins for this new method:

1. It's "Non-Perturbative" (No Guessing):
   Usually, shortcuts work by saying, "Let's assume the interaction is small, then add a little bit more, then a little bit more." This is like trying to build a house by adding one brick at a time and hoping it doesn't fall.
   This new method doesn't guess. It treats the interaction as a whole, solid block. It works even when the electrons are screaming at each other (strong interaction), which is where most other methods fail.

2. It Handles "Spin" and "Charge" Together:
   Electrons have two main personalities: their charge (how much they repel each other) and their spin (like a tiny magnet).
   In this new model, the "managers" (the static variables) naturally split into two spheres: one sphere controls the charge, and the other controls the spin. This makes it incredibly easy to study complex phenomena where charge and spin get tangled together, like superconductivity (where electrons pair up and flow without resistance).

3. It's a "Classical" Approximation of a Quantum World:
   The authors show that if you imagine the system has a huge number of "copies" of itself (a mathematical trick called the "Large-M limit"), this new method becomes exact.

   • Analogy: Think of a single coin flip. It's random (quantum). But if you flip a million coins, the average result is perfectly predictable (classical). This method treats the electrons as if they are part of that "million coin" scenario, allowing us to use simple classical physics to solve a quantum puzzle.

Did it Work?

The authors tested their new method on small, simple systems (one site and two sites) where they knew the exact answer (the "Gold Standard").

• The Result: The new method got the qualitative behavior right. It correctly predicted that electrons would clump together or spread out, and it correctly predicted how magnetic spins would align.
• The Catch: It wasn't perfectly precise numerically. Because the method treats the energy levels as a smooth, continuous slide (like a ramp) rather than distinct steps (like a staircase), there are small errors.
  • Analogy: If you measure a staircase with a ruler that only measures smooth ramps, you'll get the general height right, but you might miss the exact height of the third step. However, for most real-world engineering, knowing the ramp is there is enough to build a bridge.

The "Magic Transformation"

One of the coolest parts of the paper is that the authors didn't just invent a shortcut; they proved that their shortcut is actually a different language for the same problem.

They showed that their new way of writing the equations is mathematically identical to a famous transformation involving Majorana fermions (a type of particle that is its own antiparticle) and Kondo lattices (a model where moving electrons interact with fixed magnetic atoms).

The Takeaway:
They took a messy, hard-to-solve quantum problem and translated it into a language where the "messy" parts are handled by simple, static managers, and the "moving" parts are handled by simple quantum switches.

Summary for the General Audience

This paper is like inventing a new GPS navigation system for electrons.

• Old GPS: Tries to calculate the traffic of every single car, pedestrian, and bird in the city. It crashes the computer.
• New GPS: Assumes the traffic lights and road signs (the "managers") are fixed, and only tracks the cars (the "electrons"). It uses a clever map that simplifies the road network but still gets you to the destination accurately.

This new approach allows physicists to study complex materials (like high-temperature superconductors) much faster and with less computing power, opening the door to designing better electronics and energy technologies in the future.

Technical Summary

1. Problem Statement

The Hubbard model is a fundamental framework for understanding strongly correlated electron systems, exhibiting phenomena such as Mott insulation, magnetism, and high-temperature superconductivity. However, obtaining quantitatively reliable solutions for this genuine quantum many-body problem is notoriously difficult.

• Limitations of Existing Methods:
  • Perturbation Theory: Expands in either interaction strength ($U$) or hopping ($t$), failing in intermediate coupling regimes.
  • Variational Methods: Depend heavily on the choice of trial wavefunctions.
  • Standard Semiclassical Approximations: Typically formulated via path integrals using Hubbard-Stratonovich (HS) transformations. These often introduce auxiliary bosonic fields locally, neglecting imaginary-time fluctuations. A major drawback is the arbitrariness in choosing the HS field, which becomes problematic in multi-orbital systems and often fails to capture non-local correlations (e.g., $d$-wave superconductivity) or intertwined spin-charge orders.

The authors aim to develop a non-perturbative semiclassical approximation that is applicable at finite temperatures, naturally incorporates intersite correlations, and is readily extendable to multi-orbital systems without imposing arbitrary physical constraints.

2. Methodology

The authors propose a novel theoretical framework based on an unconventional coherent-state representation of the local Hilbert space, guided by the principle of minimizing the use of Grassmann variables.

A. Graded Coherent-State Construction

Instead of the standard fermionic coherent state (which uses a Grassmann variable for every spin-orbital), the authors construct a generalized coherent state for a single site $i$ that respects the $\mathbb{Z}_2$ grading of fermion parity:
$$|\Omega_i, \psi_i\rangle = |\alpha_i\rangle + \psi_i |\beta_i\rangle$$

• Structure:
  • $|\alpha_i\rangle$: Represents the even-parity sector (empty state $|0\rangle$ and doubly occupied state $|\uparrow\downarrow\rangle$). This sector is parametrized by bosonic coordinates $\Omega^c_i$ (charge sector).
  • $|\beta_i\rangle$: Represents the odd-parity sector (singly occupied states $|\uparrow\rangle, |\downarrow\rangle$). This sector is parametrized by bosonic coordinates $\Omega^s_i$ (spin sector).
  • $\psi_i$: A single Grassmann variable (spin-independent) that mediates transitions between the even and odd sectors.
• Symmetry: This construction naturally realizes an $SU(2)_\eta \times SU(2)_{\text{spin}}$ symmetry, where the bosonic manifold is the product of two Bloch spheres ($S^2_c \times S^2_s$), rather than the full complex projective space $CP^3$.
• Path Integral: The partition function is formulated by inserting a resolution of unity based on these states. The resulting effective action is bilinear in the Grassmann fields, a significant simplification compared to the quartic interactions in standard formulations.

B. Semiclassical Approximation

The approximation is achieved by treating the bosonic variables $\Omega_i(\tau)$ as static (neglecting imaginary-time dependence), while retaining the full quantum dynamics of the Grassmann variables.

• Resulting Hamiltonian: For a fixed configuration of static bosonic fields $\Omega$, the system reduces to an effective one-body Hamiltonian for spinless fermions ($d_i$):
  $$H_{\text{eff}}[\Omega] = -\sum_i [\mu + (U-2\mu)|C_4(\Omega_i)|^2] d^\dagger_i d_i + \sum_{\langle ij \rangle} (T_{ij} d^\dagger_i d_j + \Delta_{ij} d^\dagger_i d^\dagger_j + \text{H.c.})$$
  Here, $T_{ij}$ and $\Delta_{ij}$ are effective hopping and pairing potentials determined by the bosonic parameters.
• Evaluation: The partition function is evaluated by integrating over the bosonic configurations weighted by the trace of the Boltzmann factor of $H_{\text{eff}}$.

C. Theoretical Justification (Large-$M$ Limit)

The authors demonstrate that this semiclassical limit is controlled and systematic. By embedding the Hubbard model into a sequence of theories with $M$ identical replicas and projecting onto the completely symmetric sector, they show that the semiclassical approximation corresponds to the large-$M$ limit ($M \to \infty$). In this limit, quantum fluctuations are suppressed by $1/M$, making the saddle-point (static) approximation asymptotically exact.

D. Operator Mapping

By deriving the operator form of the path-integral representation, the authors establish an exact nonlinear transformation mapping the original Hubbard model onto a Majorana-Kondo lattice model. This maps the system to itinerant Majorana fermions coupled to localized pseudospins, providing a new perspective on the model's structure.

3. Key Contributions

1. Novel Coherent-State Formalism: Introduction of a graded coherent state using only one Grassmann variable per site, separating spin and charge degrees of freedom into independent bosonic manifolds ($S^2 \times S^2$).
2. Non-Perturbative Scheme: A method applicable at finite temperature that does not rely on expansions in $U$ or $t$.
3. Systematic Control: Proof that the approximation is the leading order of a controlled $1/M$ expansion, providing a rigorous foundation for the semiclassical limit.
4. Exact Transformation: Derivation of a nonlinear fermionic transformation linking the Hubbard model to a Majorana-Kondo lattice, offering new insights into spin-charge separation and effective interactions.
5. Multi-Orbital Generalization: A natural extension of the formalism to multi-orbital systems using the graded algebra structure ($SO(2N)/U(N)$), avoiding the complexity of standard multi-orbital coherent states.

4. Results and Benchmarks

The validity of the approach was tested against exact solutions for one-site (Hubbard atom) and two-site systems.

• One-Site System (Hubbard Atom):

  • Qualitative Agreement: The method correctly reproduces the qualitative behavior of particle number and double occupancy as functions of chemical potential and temperature.
  • Quantitative Deviation: The semiclassical density of states is continuous (uniform distribution), whereas the exact spectrum is discrete (delta functions). This leads to algebraic asymptotic behavior in the semiclassical limit versus exponential behavior in the exact solution.
  • Half-Filling: At half-filling ($U' = U - 2\mu = 0$), the semiclassical and exact densities of states coincide perfectly because the quantum nature of the remaining fermion is retained explicitly.

• Two-Site System:

  • Spin Correlations: The method captures the temperature dependence of spin correlations $\langle S_1 \cdot S_2 \rangle$. However, in the large-$U$ limit, the correlation magnitude is reduced (approaching $-1/4$ instead of the exact quantum singlet value of $-3/4$). This is attributed to the classical treatment of spins (classical vectors of length $S=1/2$ vs. quantum spin $1/2$).
  • Energy Scale: The characteristic energy scale of the semiclassical results is lower than the exact quantum results (by a factor related to the classical limit), but the functional dependence on temperature and filling is qualitatively correct.
  • Hopping Amplitude: The method reproduces the suppression of hopping at half-filling for large $U$ and the enhancement upon doping. In the non-interacting limit ($U \to 0$), the hopping amplitude approaches 1 (semiclassical) vs. 2 (exact), reflecting the reduction of internal spin degrees of freedom in the effective spinless fermion picture.

5. Significance and Outlook

• Intertwined Orders: The framework is particularly well-suited for studying systems with simultaneous spin and charge symmetry breaking (e.g., coexistence of magnetism and superconductivity) because it naturally treats both sectors as classical variables.
• Computational Efficiency: By reducing the problem to a static bosonic integral over a manageable manifold ($S^2 \times S^2$) coupled to a solvable one-body fermion problem, the method offers a computationally inexpensive alternative to Quantum Monte Carlo (which suffers from sign problems) or Dynamical Mean-Field Theory (DMFT).
• Future Applications: The authors suggest extending this approach to:
  • Multi-site and 2D systems: To explore Kosterlitz-Thouless physics and vortex excitations.
  • Non-equilibrium dynamics: Leveraging the success of semiclassical Langevin dynamics in localized systems.
  • First-principles calculations: Combining with electronic structure methods to study real materials.

In summary, this paper provides a robust, non-perturbative semiclassical framework for the Hubbard model that bridges the gap between exact quantum solutions and classical approximations, offering a new tool for investigating complex correlated electron phenomena.