Spectral functions of the strongly interacting 3D Fermi gas

This paper presents an efficient real-time method combining the Keldysh path integral with the self-consistent T-matrix approximation to directly compute spectral functions of the strongly interacting 3D Fermi gas, demonstrating improved dynamical accuracy over traditional analytic continuation and resolving debates regarding the existence of a pseudogap regime above the critical temperature.

The Gist

Imagine you are trying to understand how a crowd of people behaves in a packed concert hall. In physics, this "crowd" is a gas made of atoms (specifically, fermions) that are so cold they behave like a single quantum wave. When these atoms interact strongly, it's like everyone in the crowd is holding hands and dancing in a complex, synchronized routine.

Physicists have been very good at predicting the average behavior of this crowd (like how hot the room is or how much pressure is building up). However, predicting the individual movements of the dancers (how they respond to a sudden beat or a specific note) has been a nightmare.

Here is a simple breakdown of what this paper achieves, using everyday analogies:

1. The Problem: The "Blurry Photo" vs. The "Live Video"

For decades, scientists tried to figure out the individual movements of these atoms using a mathematical trick called Imaginary Time.

• The Analogy: Imagine trying to understand a fast-moving car by taking a photo of it through a foggy window. You can see the general shape and where it is, but the details are blurry. To get the real picture, you have to use a computer program to "clean up" the photo and guess what the car looked like in real-time. This is called Numerical Analytic Continuation (NAC).
• The Issue: This "cleaning up" process is mathematically unstable. It's like trying to guess the exact lyrics of a song by listening to a muffled recording; you might get the general tune right, but the specific notes will be wrong, or you might invent notes that weren't there. This led to disagreements between theory and experiment.

2. The Solution: The "Direct Live Stream"

The authors of this paper developed a new method to calculate these movements directly in real-time, skipping the "foggy photo" step entirely.

• The Analogy: Instead of taking a blurry photo and trying to fix it later, they set up a high-definition camera that records the concert live. They didn't have to guess; they just watched what happened.
• How they did it: They used a mathematical framework called the Keldysh Path Integral. Think of this as a special type of camera that can record both the "forward" motion of time and the "backward" motion simultaneously, allowing them to capture the full story without losing information.

3. The Technical Hurdle: The "Oscillating Guitar String"

Calculating these live movements is incredibly hard because the math involves waves that vibrate incredibly fast (like a guitar string being plucked thousands of times a second).

• The Problem: If you try to measure these fast vibrations with a standard ruler (a computer grid), you need a ruler with billions of tiny marks to catch every wiggle. This would require a supercomputer that doesn't exist yet.
• The Trick: The authors invented a clever "interpolation" technique.
  • The Analogy: Imagine trying to draw a perfect circle. Instead of plotting millions of dots, you realize the circle follows a simple rule. You draw a few key points and then use a smooth curve to connect them.
  • They realized that while the vibrations are fast, they follow a predictable pattern. They mathematically "peeled off" the fast vibrations, leaving behind a smooth, slow-moving shape that is easy for a computer to handle. This allowed them to get high-precision results without needing a super-dense grid.

4. The Discovery: The "Ghost Gap"

One of the main goals of this research was to settle a debate about something called the Pseudogap.

• The Debate: In some materials (like high-temperature superconductors), scientists see a "gap" in the energy levels where particles should be, even before the material becomes a superconductor. It's like a dance floor where people are starting to pair up and hold hands before the music officially starts.
• The 2D vs. 3D Mystery: In 2D (flat) systems, this "pre-pairing" (pseudogap) is very clear. But in 3D (our real world), previous blurry-photo methods suggested a weak gap existed.
• The New Finding: Using their "live stream" method, the authors found that in 3D, the gap is extremely weak and almost invisible right above the critical temperature. The "ghost" isn't really there in a significant way; the thermal noise (the crowd's jostling) washes it out. This settles a long-standing argument in the field.

5. Why This Matters

• Accuracy: Their method matches experimental data much better than the old "blurry photo" methods.
• Versatility: Because their method is based on general rules (convolutions), it can be applied to other systems, like:
  • Systems where the atoms have different masses (like a mix of heavy and light dancers).
  • Systems that are out of balance (like a crowd rushing to an exit).
  • 2D materials (flat sheets of atoms).
• Future Tech: Understanding these "live" dynamics is crucial for developing new quantum technologies, better superconductors, and understanding how materials behave under extreme conditions.

In Summary:
The authors built a new mathematical "HD camera" that films the quantum dance of atoms directly, avoiding the blurry, error-prone "photo editing" of the past. This allows them to see the true nature of these interactions, proving that in 3D, the mysterious "pre-pairing" gap is much fainter than previously thought.

Technical Summary

1. Problem Statement

Theoretical computation of dynamical properties (specifically spectral functions) in strongly interacting quantum many-body systems, such as the unitary Fermi gas, faces a significant bottleneck.

• The Challenge: Standard theoretical approaches often rely on imaginary-time (Matsubara) formalism because it is numerically stable for thermodynamic quantities. However, extracting real-frequency dynamical properties from imaginary-time data requires Numerical Analytic Continuation (NAC).
• The Limitation: NAC is a mathematically ill-posed inverse problem. It is highly sensitive to numerical noise and often yields uncontrolled errors, resulting in broadened spectral features and inaccurate dynamical quantities. This prevents a rigorous comparison between theory and high-precision experimental radio-frequency (rf) spectroscopy data.
• The Specific Context: The self-consistent T-matrix approximation is known to accurately predict thermodynamic properties (e.g., critical temperature, Tan contact) but has historically struggled to produce reliable spectral functions due to the reliance on NAC.

2. Methodology

The authors propose a novel method to compute the self-consistent T-matrix approximation directly in real frequencies using the Keldysh path integral formalism, thereby bypassing the need for analytic continuation.

• Keldysh Formalism: They utilize the real-time Keldysh contour to define the partition function and action. This allows for the direct calculation of retarded ($G^R$), advanced ($G^A$), and Keldysh ($G^K$) propagators.
• Self-Consistent T-Matrix: The approximation is derived from the 2-particle-irreducible (2PI) effective action (or Luttinger-Ward functional). This ensures the conservation of particle number, energy, and momentum. The self-energies for fermions and the pairing field are calculated self-consistently.
• Convolution Structure & Interpolation: The core technical innovation lies in handling the convolution integrals required to compute self-energies (Eqs. 30–33).
  • The Difficulty: Direct Fourier transforms in real time involve "chirped oscillations" (oscillations with frequency increasing linearly with momentum, e.g., $e^{-ik^2t}$). Sampling these on a standard equidistant grid requires prohibitively large grid sizes to satisfy the Nyquist criterion.
  • The Solution: The authors developed a piecewise interpolation scheme.
    1. They transform frequency coordinates to the bare dispersion ($\omega_\alpha = \omega + k^2/2m_\alpha + \mu_\alpha$), making the functions slowly varying in momentum space.
    2. They use spline interpolation for the slowly varying parts of the integrand.
    3. They analytically integrate the rapidly oscillating factors (like $\sin(kr)$ and $e^{-ivk^2t}$) using Hermite interpolation and exact integration of the resulting polynomial terms.
  • Result: This avoids sampling the fast oscillations directly, allowing the use of sparse grids (approx. 500 points) while maintaining high accuracy. This reduces computational cost by orders of magnitude compared to a naive Fast Fourier Transform (FFT) approach.

3. Key Contributions

1. Direct Real-Frequency Solver: The development of an efficient algorithm to solve the self-consistent T-matrix equations directly in real time, eliminating the need for numerical analytic continuation.
2. Validation against Imaginary Time: The method is validated by transforming the real-time results back to imaginary time and comparing them with established Matsubara calculations. The agreement is within 1% for thermodynamic quantities and propagators, demonstrating that the real-time method preserves thermodynamic consistency.
3. Benchmarking NAC: The authors compare their direct real-frequency results with state-of-the-art NAC (using the Maximum Entropy Method) applied to the same imaginary-time data. They demonstrate that NAC systematically overestimates spectral broadening and shifts peaks, failing to capture sharp quasiparticle features.
4. Resolution of the Pseudogap Debate: The method is applied to investigate the existence of a pseudogap in the 3D unitary Fermi gas above the critical temperature ($T_c$).

4. Key Results

• Spectral Accuracy: The direct real-frequency method reproduces the correct asymptotic behavior of spectral functions (e.g., $A(k, \omega) \sim \omega^{-5/2}$ at high frequencies) and the Tan contact density with high fidelity.
• Failure of NAC: Comparison with NAC results (Fig. 8) shows that NAC produces spectra that are too broad and shifted to higher frequencies. The "pseudogap" features observed in NAC are artifacts of the broadening inherent in the Maximum Entropy method.
• Pseudogap in 3D:
  • The study finds that a weak pseudogap exists only very close to the critical temperature ($T \lesssim 0.19 T_F$, where $T_c \approx 0.156 T_F$).
  • At $T_c$, the pseudogap signature is not pronounced due to thermal fluctuations broadening the spectrum.
  • Crucially, the authors conclude that no significant pseudogap regime exists in the self-consistent T-matrix approximation well above $T_c$. This contradicts some previous claims based on NAC or non-self-consistent approximations.
• Quasiparticle Lifetime: The method correctly captures the long lifetime of particle states near the Fermi surface at low temperatures, a feature often washed out in NAC.

5. Significance and Outlook

• Theoretical Reliability: This work establishes a reliable, controlled framework for calculating dynamical quantities in strongly correlated systems without the uncontrolled errors of NAC.
• Experimental Benchmarking: The ability to compute precise spectral functions allows for direct, quantitative comparison with rf-spectroscopy experiments in ultracold atoms, bridging the gap between theory and experiment for dynamical properties.
• Versatility: The method relies on the convolution theorem and is not limited to the unitary Fermi gas. The authors highlight its applicability to:
  • Spin- or mass-imbalanced Fermi gases (relevant for FFLO phases).
  • 2D systems.
  • Bose-Fermi mixtures.
  • Non-equilibrium dynamics (e.g., quench dynamics, pump-probe experiments).
• Future Directions: The approach opens the door to studying non-Fermi liquid behaviors and scaling regimes in other quantum critical systems where spectral functions are difficult to compute.

In summary, this paper presents a major methodological breakthrough in quantum many-body theory, replacing the ill-defined step of analytic continuation with a robust, direct real-time calculation, thereby resolving long-standing discrepancies regarding the spectral properties of the unitary Fermi gas.