Towards Studying Superconductivity in the Fermi-Hubbard Model on Rydberg Atoms

This paper presents a method using Rydberg atom processors and sample-based quantum diagonalization to calculate the ground state energy of the Fermi-Hubbard model for large U by sampling the Heisenberg model, demonstrating superior convergence and efficiency over random sampling on up to 56 qubits while analyzing the potential for studying emergent superconductivity.

The Gist

Imagine you are trying to solve a massive, incredibly complex puzzle: How do electrons move in a material to create superconductivity? Superconductivity is a magical state where electricity flows with zero resistance, but figuring out exactly how it happens in certain materials (described by the "Fermi-Hubbard model") is so hard that even the world's fastest supercomputers struggle with it.

This paper is like a story about a team of scientists who decided to stop trying to solve the whole puzzle at once. Instead, they built a clever shortcut using a new kind of "quantum microscope" made of Rydberg atoms (super-excited atoms that act like giant magnets).

Here is the breakdown of their adventure, explained with simple analogies:

1. The Problem: The "Impossible" Puzzle

The Fermi-Hubbard model is the rulebook for how electrons dance around in a grid. When the electrons repel each other strongly (a condition called "large U"), the math gets so messy that classical computers crash trying to find the "ground state" (the most stable, lowest-energy arrangement).

• Analogy: Imagine trying to predict the exact path of 56 people running through a crowded maze where everyone is pushing against everyone else. A regular computer tries to calculate every single possible path, gets overwhelmed, and gives up.

2. The Shortcut: The "Shadow" Trick

The scientists realized that while the electron puzzle is hard, there is a simpler, related puzzle called the Heisenberg model (which describes how tiny magnets, or "spins," interact).

• The Insight: In the "large repulsion" limit, the behavior of the electrons is mathematically linked to the behavior of these simpler magnets. If you can figure out how the magnets arrange themselves, you can use that information to solve the harder electron puzzle.
• The Metaphor: It's like trying to figure out the exact shape of a complex 3D sculpture (the electrons). Instead of sculpting it directly, you look at its shadow on the wall (the magnets). The shadow is much easier to capture, and if you know the rules of light, you can reconstruct the 3D shape from the shadow.

3. The Tool: The "Quantum Atom Array"

To capture this "shadow," they used a special quantum computer called Aquila, made by a company called QuEra.

• How it works: Instead of using silicon chips like your laptop, Aquila uses lasers to trap 256 individual atoms in a vacuum. They can turn these atoms into "Rydberg states," making them act like giant, interacting magnets.
• The Process: They used a technique called VQITE (Variational Quantum Imaginary Time Evolution). Think of this as a "cooling" process. They start with the atoms in a hot, chaotic state and slowly "cool" them down using laser pulses until they settle into the most stable, lowest-energy arrangement (the ground state).

4. The Magic Sauce: "Sample-Based Quantum Diagonalization" (SQD)

Once they had the atoms settled into their stable "shadow" state, they didn't just look at the result. They took samples (snapshots) of the atoms' positions.

• The Analogy: Imagine you want to know the average height of a crowd. You could measure everyone (impossible), or you could take a few random snapshots of the crowd and use a smart algorithm to guess the average.
• The Twist: The scientists didn't just take random snapshots. They took snapshots of the atoms after they had been "cooled" by the VQITE process.
• The Result: These "smart samples" were much better at predicting the answer than "random samples." Even when they took 10 times more random samples, the "smart samples" from the cooled atoms still won. It's like a detective who looks at the right clues versus a detective who just guesses randomly.

5. The Victory Lap

The team tested this on a system with 56 qubits (the quantum equivalent of bits).

• The Scale: This is a huge number for quantum computers. It's like solving a puzzle with 56 pieces where the number of possible arrangements is larger than the number of atoms in the universe.
• The Comparison: They also tried the same method on a different type of quantum computer (IBM's gate-based chips) to prove their method works on any hardware, not just the atom-based one.
• The Outcome: Their method successfully calculated the energy and chemical properties of the system, getting very close to the "true" answer that only a perfect computer could find.

Why Does This Matter?

This paper is a major step toward understanding high-temperature superconductivity.

• The Big Picture: If we can understand how electrons behave in these models, we might be able to design new materials that conduct electricity with zero loss at room temperature. This would revolutionize power grids, electric cars, and computers.
• The Takeaway: The scientists showed that we don't need a perfect, error-free quantum computer to make progress. By using a clever "shortcut" (mapping electrons to magnets) and a smart sampling strategy, we can use today's noisy, imperfect quantum computers to solve problems that were previously impossible.

In a nutshell: They built a quantum "shadow puppet" show to solve a math problem that was too big for the world's best supercomputers, proving that with the right tricks, we can start unlocking the secrets of superconductivity today.

Technical Summary

Here is a detailed technical summary of the paper "Towards Studying Superconductivity in the Fermi-Hubbard Model on Rydberg Atoms" by Kubra Yeter-Aydeniz and Nora Bauer.

1. Problem Statement

The Fermi-Hubbard model is the fundamental theoretical framework for understanding high-temperature superconductivity. However, classical simulation methods face significant limitations:

• DMFT (Dynamical Mean Field Theory): Struggles with non-local interactions required for $d$-wave superconductivity.
• DMRG (Density Matrix Renormalization Group): Computationally expensive in dimensions higher than 1D.
• QMC (Quantum Monte Carlo): Suffers from the "sign problem" at low temperatures and struggles to resolve ground state properties.

While quantum computers offer a path forward, current Noisy Intermediate-Scale Quantum (NISQ) devices lack the qubit count and error correction to solve large-scale Hubbard models directly. The challenge is to compute ground state energies and properties for systems with 56 orbitals (exceeding exact classical diagonalization limits) using current hardware, specifically to study regimes relevant to quasi-superconductivity.

2. Methodology

The authors propose a hybrid quantum-classical approach leveraging Sample-Based Quantum Diagonalization (SQD) combined with a perturbative mapping between the Fermi-Hubbard and Heisenberg models.

A. Perturbative Mapping (Large-$U$ Limit)

The core theoretical insight relies on the relationship between the Fermi-Hubbard model and the Heisenberg model in the limit where the on-site repulsion $U$ is much larger than the hopping term $t$ ($U \gg t$).

• Mapping: The authors map the Fermi-Hubbard model (with spin-up and spin-down orbitals) to two coupled Heisenberg spin chains.
• Constraint: In the large-$U$ limit, double occupancy is forbidden. This allows the system to be simulated by preparing the anisotropic Heisenberg (XXZ) model on the quantum hardware.
• Spin-Fermion Conversion: During sampling, a bitstring measured on the "spin-up" chain is used to construct the full Hubbard configuration by assigning the complementary pattern to the "spin-down" chain (ensuring exactly one electron per site).
• Anisotropy: To accommodate the native capabilities of the QuEra Aquila processor, the authors utilize the anisotropic Heisenberg model ($J_{xy} \neq J_z$), which corresponds to a spin-asymmetric Hubbard model ($t_\uparrow \neq t_\downarrow$).

B. State Preparation: VQITE

To prepare the ground state of the target anisotropic Heisenberg model on the QuEra Aquila Rydberg atom processor:

• Algorithm: Variational Quantum Imaginary Time Evolution (VQITE) is used. This avoids the non-unitary operations of standard QITE by using a variational ansatz.
• Hardware Native: The ansatz is implemented using the native Rydberg Hamiltonian of the Aquila processor, controlled by time-dependent Rabi frequencies ($\Omega$), detunings ($\Delta$), and phases ($\phi$).
• Optimization: Hyperparameters were optimized on simulated hardware and found to converge for system sizes up to $L=10$, allowing extrapolation for larger systems.

C. Sample-Based Quantum Diagonalization (SQD)

Once the approximate ground state is prepared on the quantum hardware:

1. Sampling: The quantum processor is measured to generate a set of bitstrings (configurations).
2. Projection: These samples define a subspace. The full Hubbard Hamiltonian is projected onto this subspace.
3. Diagonalization: A classical computer diagonalizes the projected Hamiltonian (using the iterative Davidson method) to find the ground state energy and observables within that subspace.
4. Iteration: The process is self-consistent, updating occupation numbers until convergence.

D. Benchmarking

The authors compared their method against:

• Random Sampling: Using uniform superposition states instead of VQITE-prepared states.
• Gate-Based Implementation: Running the SQD workflow on IBM Quantum (ibm-pittsburgh) using a gate-based VQE ansatz (Efficient SU2) to demonstrate hardware agnosticism.

3. Key Contributions

1. Largest Ground State Calculation: Achieved ground state energy calculations for the Fermi-Hubbard model on quantum hardware with up to 56 orbitals, the largest reported to date.
2. First Cloud Analog VQITE: Demonstrated the first implementation of VQITE on cloud-accessed analog quantum hardware (QuEra Aquila).
3. Superiority of VQITE Sampling: Proved that sampling from a VQITE-prepared state yields significantly better accuracy than random sampling, even when the random sampling uses 10$\times$ more shots (10,000 vs. 1,000).
4. Hardware Agnostic Framework: Validated the SQD workflow on both analog (Rydberg) and gate-based (Superconducting) architectures, showing comparable performance between the two for the 56-qubit case.
5. Chemical Potential Analysis: Successfully computed the chemical potential and its derivative, critical for identifying phase transitions and superconducting regimes.

4. Results

• Energy Convergence: For the 56-orbital system, the VQITE-sampled SQD method converged to a ground state energy significantly closer to the exact value (calculated via DMRG) than random sampling.
  • At 56 orbitals, the energy gap between VQITE-sampled and random-sampled results was nearly 10 units, with VQITE finding a much lower (better) energy estimate.
• Scalability: The method remained effective up to 56 qubits, whereas random sampling failed to converge to the ground state even with increased shot counts.
• Chemical Potential: While calculating the chemical potential ($\mu$) proved challenging due to error cancellation (difference of two noisy values), the VQITE-sampled results generally tracked the exact values better than random sampling.
• Hardware Comparison: The gate-based IBM Quantum results were comparable to the analog Aquila results, confirming the robustness of the SQD framework across different quantum architectures.

5. Significance and Future Directions

• Path to Superconductivity: This work establishes a viable pathway to study emergent superconductivity in the Fermi-Hubbard model on NISQ devices. By leveraging the perturbative link to the Heisenberg model, the authors bypass the need to directly simulate the complex Hubbard dynamics on current noisy hardware.
• Regime Expansion: The current study is limited to the anisotropic (spin-asymmetric) regime due to hardware constraints. Future work aims to move toward the isotropic regime (spin-symmetric) and introduce doping (away from half-filling) to access the true quasi-superconducting phase.
• Dimensionality: The current results are 1D. The framework is designed to be extended to 2D systems, which are necessary for studying $d$-wave superconductivity.
• Methodological Impact: The demonstration that quantum sampling (VQITE) outperforms random sampling by an order of magnitude in shot efficiency validates the necessity of high-quality state preparation for near-term quantum advantage in condensed matter physics.

In summary, this paper presents a successful proof-of-concept for using Rydberg atom processors and SQD to solve large-scale Fermi-Hubbard models, overcoming classical computational limits and demonstrating a clear advantage of quantum sampling strategies over brute-force random sampling.