The Asymptotic State of Decaying Turbulence

Through extensive direct numerical simulations of decaying turbulence with initial Birkhoff-Saffman and Loitsianskii-Kolmogorov-Batchelor spectra, the study confirms non-universal power-law energy decay and highlights the significant influence of boundary effects on the observed universality, while showing strong agreement with Migdal's theoretical predictions for the Birkhoff-Saffman case.

The Gist

Imagine you drop a drop of ink into a glass of still water. At first, it's a tight, chaotic swirl. Over time, it spreads out, gets thinner, and eventually disappears into the water. This is a bit like turbulence in fluids (like air or water) when it's left to decay on its own.

Scientists have been trying to figure out exactly how fast that ink disappears and what rules it follows for over a century. This paper is a massive, high-tech experiment to finally solve that mystery.

Here is the story of the paper, explained simply:

1. The Big Question: Is There One Rule for Everyone?

For a long time, scientists thought there were two main "families" of decaying turbulence, depending on how the chaos started:

• Family A (The "Linear" Start): Imagine the ink drop started with a specific, gentle swirl. Theory says it should fade away at a specific speed (like a car slowing down at a steady rate).
• Family B (The "Angular" Start): Imagine the ink drop started spinning like a top. Theory says it should fade away at a different speed.

The problem? Real-world experiments and computer simulations were messy. Sometimes the ink faded fast, sometimes slow, and the results didn't match the theories. It was like trying to predict how fast a car stops, but the road conditions kept changing.

2. The Super-Computer Experiment

The authors of this paper decided to run the ultimate test using Direct Numerical Simulations (DNS). Think of this as a super-accurate video game where they simulate the physics of water molecules perfectly, without skipping any details.

• The Duration: They ran these simulations for an incredibly long time—up to 200,000 "swirl-turns." To put that in perspective, if a swirl-turn was one second, they watched the ink fade for over 2 days straight without stopping. Previous studies only watched for a few minutes.
• The Precision: They used a "smart grid." As the turbulence got weaker and the swirls got bigger, they automatically zoomed out their camera (reducing the grid size) to keep watching the smallest details without wasting computer power. This ensured they never missed a tiny swirl.

3. The Results: Two Different Stories

When they finally looked at the data, they found that the "Family" theory was mostly right, but with a twist.

• The "Linear" Family (BS): When they started with the gentle swirl, the turbulence faded exactly as the new, fancy theory predicted. It was a perfect match.
• The "Angular" Family (LKB): When they started with the spinning top, the turbulence faded at a different speed than the new theory predicted. It stuck to the old, classic rules.

The Twist: The paper discovered that the speed at which the turbulence fades depends heavily on the very first moment of the chaos. It's like if you drop a ball: if you drop it gently, it bounces one way; if you throw it hard, it bounces another. The "universal" rule that everyone hoped for (that all turbulence fades the same way) doesn't exist for the total energy.

4. The "Boundary Effect" (The Room Size Matters)

One of the biggest discoveries was about "boundary effects."
Imagine you are running a race in a small room. Eventually, you hit the walls, and your running style changes.
In the computer simulations, as the turbulence got older, the big swirls grew so large they started to "feel" the edges of the computer box. This changed the math.

• The Lesson: To see the "true" rules of nature, you have to make sure the turbulence is in a room so big it never hits the walls. The authors found that if you remove these "wall effects" (by looking only at the middle of the flow), the rules become much clearer.

5. The New Theory (Migdal's Theory)

The authors tested a brand-new, very complex theory proposed by a scientist named Migdal. This theory uses ideas from quantum physics (the physics of tiny particles) to explain big, messy fluids.

• The Verdict: The theory was brilliant at predicting the shape of the turbulence and how the tiny swirls behaved inside the big mess. It was like a perfect map of the terrain.
• The Flaw: It failed to predict the speed at which the whole thing faded away for the "spinning top" case. It assumed everyone fades at the same speed, but the simulation proved that the starting conditions matter too much.

6. The "Enstrophy" Surprise

If the total energy (how fast the ink fades) isn't universal, is anything?
The authors looked at Enstrophy (a fancy word for how much "spin" or "twist" is in the fluid).

• The Discovery: While the speed of fading changed based on the start, the decay of the spin was much more consistent. It's like saying: "No matter how you start the race, the way the runners' hearts beat slows down in the same pattern."
• This suggests that while the big picture (energy) is messy and depends on history, the small details (spin) might follow a universal law.

Summary: What Does This Mean?

This paper is a giant step forward in understanding chaos.

1. There is no single "Universal Decay Rate": How fast turbulence dies depends on how it was born.
2. The "Room" Matters: You have to be careful about the size of your experiment (or computer simulation) because hitting the walls changes the results.
3. New Theory is Half-Right: The new quantum-inspired theory is amazing at describing the structure of the chaos, but it needs to be tweaked to account for the different ways turbulence can start.
4. The Future: Instead of asking "How fast does turbulence die?", scientists should ask "How does the spin die?" because that seems to be the true universal rule.

In short: Chaos is messy, but if you look at the right parts of the mess, you can find order.

Technical Summary

1. Problem Statement

The long-time evolution of decaying homogeneous isotropic turbulence (HIT) is a fundamental problem in fluid mechanics. While theoretical frameworks exist, there is no consensus on the asymptotic decay rate of turbulent kinetic energy ($E_n \sim t^{-n}$).

• Classical Theories: Two primary regimes are predicted based on the conservation of different invariants:
  • Birkhoff-Saffman (BS): Conservation of linear momentum implies a low-wavenumber spectrum $E(k) \sim k^2$, predicting a decay exponent $n = 6/5$ (1.2).
  • Loitsianskii-Kolmogorov-Batchelor (LKB): Conservation of angular momentum implies $E(k) \sim k^4$, predicting $n = 10/7$ ($\approx 1.43$).
• The Gap: Previous experimental and numerical studies have yielded inconsistent decay exponents ($n \approx 1.15$ to $1.45$). These discrepancies are attributed to:
  • Lack of control over initial low-wavenumber spectra.
  • Insufficient simulation durations (often failing to reach the true asymptotic state).
  • Finite-domain effects (boundary interactions) that distort the decay as the integral length scale grows.
• New Context: The paper aims to rigorously test Migdal's recent theory (2023–2025), which utilizes loop-space calculus and quantum field theory concepts to propose a universal asymptotic decay law ($n = 5/4 = 1.25$) independent of initial conditions.

2. Methodology

The authors employed Direct Numerical Simulations (DNS) with unprecedented duration and resolution to isolate the asymptotic state.

• Simulation Setup:

  • Domain: 3D cubic periodic box ($L_{box} = 2\pi$).
  • Solver: Pseudospectral method with phase-shift de-aliasing (up to $k_{max} \approx \sqrt{2}N/3$).
  • Initial Conditions: Two distinct spectral forms were strictly controlled at low wavenumbers:
    • BS Case: $E(k) \sim k^2$ for small $k$.
    • LKB Case: $E(k) \sim k^4$ for small $k$.
  • Reynolds Numbers: Initial Taylor-microscale Reynolds numbers ($Re_\lambda$) ranging from 30 to 145.
  • Duration: Simulations ran for up to 200,000 initial eddy-turnover times ($T_{eddy,0}$), far exceeding previous studies (typically $\sim 1,000$).

• Dynamic Grid Modification:

  • To manage computational cost over such long durations, the authors implemented a dynamic regridding strategy.
  • As the Kolmogov scale ($\eta$) grows and the flow becomes less resolved, the grid resolution ($N$) is halved when the resolution parameter $k_{max}\eta$ exceeds a threshold.
  • Threshold Validation: Sensitivity analysis determined a threshold of $k_{max}\eta = 6.0$ to be safe. At this point, truncating high-wavenumber modes results in negligible energy loss ($\sim 5 \times 10^{-9}$) and does not alter the decay exponent compared to unmodified simulations.

• Statistical Robustness: Multiple independent realizations (using different random seeds) were performed for each $Re_\lambda$ to ensure statistical convergence.

3. Key Contributions

1. Longest Duration DNS: The study extends simulation times to $\sim 2 \times 10^5$ eddy turnover times, allowing the flow to reach a truly asymptotic state and minimizing transient effects.
2. Strict Spectral Control: Unlike previous studies that often used forced turbulence spectra, this work explicitly initializes the flow with pure $k^2$ and $k^4$ low-wavenumber scalings to isolate the effects of initial large-scale structures.
3. Systematic Testing of Migdal's Theory: The paper provides the first comprehensive DNS validation of Migdal's loop-space theory, comparing predictions for decay exponents, length scales, and structure functions against high-fidelity data.
4. Boundary Effect Analysis: The authors introduce a "bulk parameter" analysis (filtering out the lowest wavenumbers) to demonstrate how "boundary effects" (finite domain size and initial low-$k$ modes) influence the apparent universality of energy decay.

4. Key Results

A. Energy Decay Exponents

• BS Regime ($E(k) \sim k^2$): After an initial transient, the energy decays as $E_n \sim t^{-n}$ with $n \approx 1.25$ ($5/4$). This aligns perfectly with Migdal's universal prediction and is slightly higher than the classical Saffman prediction ($n=1.2$).
• LKB Regime ($E(k) \sim k^4$): The energy decays with $n \approx 10/7$ ($\approx 1.43$), consistent with the classical Loitsianskii/Kolmogorov prediction.
• Conclusion on Universality: The decay exponent is not universal; it depends heavily on the initial low-wavenumber spectrum. Migdal's theory holds for the BS case but fails to predict the $10/7$ exponent for the LKB case.

B. Length Scales and Self-Similarity

• Migdal Length ($L_M$): Defined via spectral moments, $L_M$ grows as $L_M \sim t^{0.53}$ for both BS and LKB cases. This is close to the theoretical $t^{1/2}$ prediction.
• Integral Length ($L$): $L$ grows as $t^{2/5}$ (BS) and $t^{0.36}$ (LKB).
• Breakdown Mechanism: The power-law decay persists until the integral length scale $L$ reaches $\sim 10-15\%$ of the domain size. At this point, finite-domain effects cause deviations from the power laws. The LKB case maintains self-similarity longer because its integral scale grows more slowly.

C. Energy Spectra and Structure Functions

• Spectral Evolution: The initial $k^{-5/3}$ inertial range disappears quickly. It is replaced by an intermediate region with a $k^{-1}$ slope (observed in both cases, though more distinct in BS).
• High-Wavenumber Tail: The theory predicts a $k^{-7/2}$ tail. This was not observed in the DNS, likely obscured by the exponential viscous rolloff at the moderate Reynolds numbers ($Re_\lambda \le 145$).
• Second-Order Structure Function ($\zeta_2$): The local scaling exponent of the velocity structure function, $\zeta_2(r, t)$, collapses onto a universal curve predicted by Migdal's theory for both BS and LKB cases. This suggests that while the global decay rate is non-universal, the internal spectral shape and structure function are universal.

D. Enstrophy Decay

• Enstrophy ($\Omega$) decays as $\Omega \sim t^{-(n+1)}$.
• When plotted against the Migdal length ($L_M$), the relationship $\Omega \sim L_M^{-4.5}$ is observed.
• Interestingly, the LKB case (which failed the energy decay test) shows a closer agreement with the theoretical $\Omega$ vs. $L_M$ scaling ($\alpha \approx -4.55$) than the BS case. This suggests enstrophy decay may be a more robust indicator of universality than energy decay.

E. Sensitivity to Low-Wavenumber Truncation

• By filtering out the first few wavenumbers ($k_0 > 0$) from the energy calculation, the authors showed that the "bulk" energy decay exponent changes.
• As $k_0$ increases, the measured decay exponent for the BS case shifts away from $1.25$ toward higher values, and the LKB case shifts away from $10/7$.
• This confirms that the "universality" of the decay rate is highly sensitive to the lowest wavenumbers (boundary effects), implying that the classical invariants (linear vs. angular momentum) dictate the decay rate.

5. Significance and Conclusion

The paper fundamentally challenges the notion of a single universal decay exponent for decaying turbulence.

• Refinement of Theory: It validates Migdal's theory for the internal structure of turbulence (structure functions, length scale growth, and enstrophy scaling) but highlights its limitation regarding the global energy decay rate, which remains dependent on initial large-scale conditions.
• Role of Boundary Effects: The study demonstrates that "boundary effects" (finite domain size and specific low-$k$ initial conditions) are the primary source of non-universality in energy decay.
• Future Outlook: The authors suggest that the quest for a universal energy decay exponent may be ill-posed. Instead, universality might be better defined through enstrophy decay or internal spectral properties. The work calls for higher Reynolds number simulations to resolve the predicted $k^{-7/2}$ spectral tail and further reconcile the coexistence of universal internal structures with non-universal decay rates.

In summary, the research provides the most rigorous numerical evidence to date that while the mechanism of turbulence decay (internal structure) may be universal, the rate of decay is dictated by the specific large-scale invariants present at the start of the flow.