Extreme flows: where physics meets mathematically rigorous bounds

Technical Summary: Extreme Flows: Where Physics Meets Mathematically Rigorous Bounds

Problem Statement
The paper addresses two fundamental open problems in fluid mechanics concerning the "extreme" behavior of solutions to the Navier-Stokes (NS) and Euler equations. The first problem concerns the global existence of smooth solutions versus finite-time blow-up (Question 1.1), specifically whether smooth initial conditions can lead to singularities in finite time in 3D. The second problem concerns the dissipation anomaly (Question 1.2), investigating whether the normalized rate of energy (or enstrophy) dissipation remains non-zero in the inviscid limit ($\nu \to 0$), a phenomenon central to Kolmogorov's statistical theory of turbulence.

The author argues that these questions are intractable via standard analytical methods alone due to the lack of exact solutions and the complexity of the nonlinear dynamics. Instead, the paper proposes a framework to study the limits of possible behavior by combining mathematical analysis, variational optimization, and high-performance scientific computation.

Methodology
The research program follows a three-step iterative process:

1. Derivation of Rigorous Bounds (Step S1): Using energy methods and functional analysis (e.g., Gagliardo-Nirenberg, Poincaré, and Cauchy-Schwarz inequalities), the author derives a priori upper bounds on the growth of quantities of interest (such as enstrophy $E(t)$, $L_q$ norms of velocity, or dissipation rates). These bounds represent fundamental limitations on the behavior of all admissible solutions.
2. Variational Optimization (Step S2): To test the sharpness of these bounds, the author formulates and solve constrained PDE optimization problems. The objective is to find initial conditions (or forcing terms) that maximize the growth of the quantity of interest over a specific time window, subject to constraints on the initial data (e.g., fixed enstrophy or $L_q$ norm).
   • Optimization Formulation: The problems are posed as finding $u_0 = \arg \max \Phi(u(T; u_0))$ over a constraint manifold (e.g., $\{u_0 : E(u_0)=E_0, \nabla \cdot u_0 = 0\}$).
   • Numerical Solution: The author employs a Riemannian gradient method. This involves computing the gradient of the objective functional using the adjoint method (solving the linearized forward system and the backward-in-time adjoint system). The optimization is performed on constraint manifolds with Riemannian structures, utilizing projection and retraction operators to maintain constraints (divergence-free, fixed norm) during the iterative update.
   • Continuation: To explore the parameter space (e.g., large initial enstrophy $E_0$ or long time windows $T$), a continuation approach is used, where solutions for smaller parameters serve as initial guesses for larger ones.
3. Physical Mechanism Analysis (Step S3): If solutions are found that saturate the bounds, their physical structure is analyzed to identify the mechanisms (e.g., vortex stretching, reconnection, filament thinning) responsible for the extreme behavior.

Key Contributions and Results

• 1D Viscous Burgers Flows (Maximum Enstrophy Growth):

  • The paper revisits the problem of maximizing the growth of enstrophy in 1D Burgers flows.
  • It confirms that the instantaneous bound on the growth rate $dE/dt \sim E^{5/3}$ is sharp in terms of the exponent (up to a prefactor) by solving an instantaneous optimization problem.
  • Crucially, it addresses the finite-time growth bound. While a straightforward integration of the instantaneous bound suggests $E_{max} \sim E_0^3$, solving the finite-time optimization problem (Problem 3.2) reveals that the actual maximum growth scales as $E_{max} \sim E_0^{1.531} \approx E_0^{3/2}$.
  • This result was recently confirmed mathematically by Albritton & Nitti (2023), effectively closing the problem. The extreme flows exhibit a self-similar structure where enstrophy initially decreases to conserve energy before rapidly growing.

• 2D Unforced Navier-Stokes Flows (Enstrophy Dissipation Anomaly):

  • The paper investigates whether enstrophy dissipation can remain non-zero in the inviscid limit for 2D unforced flows.
  • Rigorous bounds derived from the convergence of viscous to inviscid flows (Ciampa et al., 2021) suggest the dissipation rate vanishes.
  • By solving an optimization problem to maximize enstrophy dissipation (Problem 3.4), the author constructs families of flows that saturate the upper bound. The numerical data confirms that the dissipation rate scales as $\nu^\alpha$ with $\alpha \approx 1$, consistent with the rigorous upper bound, though the paper notes the bound may allow for a potential logarithmic correction. This confirms the absence of an enstrophy dissipation anomaly in this setting.
  • Six distinct branches of local maximizers were identified, revealing different physical mechanisms (e.g., stretching of thin vorticity filaments) for amplifying palinstrophy.

• 3D Navier-Stokes Flows (Singularity Search):

  • The author applies the framework to search for finite-time singularities in 3D NS flows by maximizing enstrophy and $L_q$ norms.
  • Instantaneous Bounds: The optimization confirms that the instantaneous growth rate bounds (e.g., $dE/dt \sim E^3$) are sharp in terms of the exponent, though the numerical solutions underestimate the theoretical prefactor by approximately seven orders of magnitude.
  • Finite-Time Growth: Solving finite-time optimization problems (Problems 4.3–4.7) for various initial enstrophies and time windows reveals significant transient growth but no evidence of unbounded growth or singularity formation within the computable regime.
  • The maximum growth of enstrophy scales as $E_{max} \sim E_0^{3/2}$, mirroring the 1D Burgers result. The extreme flows feature colliding vortex rings or localized vortex structures, but the growth rates of the norms remain below the critical thresholds required for blow-up (e.g., the exponent for enstrophy growth is $\approx 0.85$, well below the critical value of 2).

• 3D Euler Flows (Singularity Search):

  • For the inviscid Euler system, the author maximizes the $\dot{H}^3$ seminorm over a long time window.
  • Numerical resolution refinement studies show that for short time windows, the objective functional remains finite, indicating well-posedness.
  • However, for a sufficiently long time window ($T=75$), the objective functional diverges as the numerical resolution increases, signaling a potential finite-time singularity.
  • The exponent of growth $\alpha(t)$ in the relation $d\|u\|_{\dot{H}^3}/dt \sim \|u\|_{\dot{H}^3}^\alpha$ remains greater than 1 for the duration of the well-resolved simulation.
  • The resulting flow structure is nearly axisymmetric, featuring two head-on colliding jets and flattened vortex rings, consistent with recent computational findings by Luo & Hou (2014) and Hou (2022), but here emerging spontaneously from the optimization without imposed symmetry.

Significance and Claims
The paper claims that this framework successfully bridges the gap between abstract mathematical analysis and physical reality. By identifying flows that saturate rigorous bounds, the author provides "closed" resolutions to specific model problems (1D Burgers enstrophy growth and 2D unforced enstrophy dissipation), demonstrating that the bounds are sharp in terms of exponents and cannot be fundamentally improved (except perhaps for prefactors or logarithmic corrections).

For the open 3D problems, the paper does not claim to have proven or disproven the existence of singularities. Instead, it provides a systematic, data-driven approach to probe the "edge of regularity." The results suggest that:

1. If singularities exist in 3D Navier-Stokes flows, they likely do not follow trajectories that saturate the known instantaneous bounds.
2. The observed scaling laws ($E_{max} \sim E_0^{3/2}$) appear universal across 1D and 3D models, potentially offering a refined a priori estimate for future theoretical work.
3. The optimization approach successfully identified a candidate scenario for singularity formation in 3D Euler flows, providing a concrete physical structure (axisymmetric colliding jets) for further mathematical investigation.

The author emphasizes that the optimization problems are non-convex, meaning the solutions found are local maximizers, and the search is not exhaustive. However, the consistency of the results across different models and the ability to recover known mathematical bounds lend credibility to the findings. The paper concludes by outlining methodological improvements, such as sum-of-squares optimization, self-similar ansätze, and adaptive mesh refinement, to further advance this research program.