Periodic Lp estimates by R-boundedness: Applications to the Navier-Stokes equations

Thomas Eiter, Mads Kyed, Yoshihiro ShibataarXiv 2022 · DOI · arXiv

TL;DR

This paper introduces a powerful new technique for establishing "maximal regularity" estimates for time-periodic solutions to evolution equations. By leveraging the transference principle and $\mathscr{R}$-boundedness, the authors bypass traditional hurdles that occur when an operator's spectrum includes zero—a common issue in fluid dynamics—and prove the existence of periodic solutions for the Navier-Stokes equations in both moving bounded domains and infinite exterior domains.

What Problem Does This Paper Address?

In the study of oscillations (time-periodic solutions), researchers often want to establish maximal regularity. This means showing that the solution is as "smooth" as the data forcing it, specifically in $L^p$ spaces.

Traditionally, this is done using the Poincaré operator (mapping an initial state to its state after one period) or representation formulas. However, these methods usually require the underlying linear operator $A$ to be invertible at zero frequency (i.e., $0$ must be in the resolvent set). This is a major roadblock for many physical problems, particularly the Navier-Stokes equations in unbounded "exterior" domains, where $0$ is part of the operator's spectrum. In these cases, the steady-state version of the problem isn't well-posed in the same functional framework as the periodic one, making standard techniques fail.

Methods

The authors develop a framework based on Harmonic Analysis rather than the typical "initial value problem" approach:

  1. Operator-Valued Transference Principle: They use a generalization of de Leeuw’s transference principle. This principle states that if a Fourier multiplier is bounded on the real line ($\mathbb{R}$), its restriction to the integers ($\mathbb{Z}$) is a bounded multiplier on the torus ($\mathbb{T}$), which represents periodic time.
  2. $\mathscr{R}$-boundedness: To handle operator-valued multipliers (where the "multiplier" is a family of solution operators), they rely on $\mathscr{R}$-boundedness. This is a stronger form of boundedness required by Weis’s Multiplier Theorem to guarantee $L^p$ estimates in UMD Banach spaces.
  3. Frequency Decomposition: The "aha" method here is splitting the solution into a low-frequency part (a finite number of Fourier modes, including the time-mean) and a high-frequency part.
  4. The high-frequency part is handled via the new transference/$\mathscr{R}$-boundedness technique.
  5. The low-frequency part (especially the zero-mode) is handled separately in bespoke functional spaces that account for the specific physics of the domain (like spatial decay in exterior domains).

Datasets

Not discussed in the paper. This is a purely mathematical research paper focused on analytical proofs and existence theorems; it does not use empirical datasets or numerical simulations.

Key Insights and Ideas

  • Decoupling from Initial Value Problems: By treating time as a torus $\mathbb{T}$, the authors avoid the need to solve an initial value problem and then search for a fixed point. The periodicity is "baked into" the domain.
  • The Power of the Resolvent: The method shows that if you have good estimates for the resolvent problem (the stationary problem with a complex parameter $\lambda$), you can automatically derive periodic estimates, provided those resolvent estimates are $\mathscr{R}$-bounded.
  • Handling the Zero Spectrum: By using a decomposition, they show that the "failure" of the operator at zero frequency doesn't prevent maximal regularity for the oscillating parts. You just need a different strategy for the time-average (the "steady" component).

Key Results

  • Navier-Stokes in Moving Domains: The authors prove the existence and uniqueness of small time-periodic solutions for $N$-dimensional Navier-Stokes equations in a domain that changes shape periodically over time.
  • Navier-Stokes in Exterior Domains: They establish similar existence results for 3D flow around a fixed body in an infinite fluid. They provide specific estimates for how these solutions decay spatially as you move away from the body.
  • General Linear Framework: Theorem 3.1 provides a "recipe" for other researchers: if your evolution equation's resolvent operators satisfy certain $\mathscr{R}$-bounds, you have a path to proving periodic maximal regularity.

Limitations and Future Work

  • Small Data Requirement: Like most existence proofs for the non-linear Navier-Stokes equations, the results only hold for "small" forcing and boundary data. Proving existence for large periodic data remains an open, much harder challenge.
  • Complexity of Moving Exterior Domains: While the paper covers moving bounded domains and fixed exterior domains, the authors explicitly state that the case of a periodically moving body in an infinite fluid (a moving exterior domain) is significantly more complex and is postponed for future research.
  • UMD Space Restriction: The theory relies on the Banach space being a "UMD space" (like $L^q$), which excludes some spaces like $L^1$ or $L^\infty$.

Summary updated

Abstract

General evolution equations in Banach spaces are investigated. Based on an operator-valued version of de Leeuw's transference principle, time-periodic $L^p$ estimates of maximal regularity type are established from $\mathscr{R}$-bounds of the family of solution operators ($\mathscr{R}$-solvers) to the corresponding resolvent problems. With this method, existence of time-periodic solutions to the Navier-Stokes equations is shown for two configurations: in a periodically moving bounded domain and in an exterior domain, subject to prescribed time-periodic forcing and boundary data.
BibTeX
@misc{eiter2022periodic,
  author = {Thomas Eiter and Mads Kyed and Yoshihiro Shibata},
  title = {{Periodic Lp estimates by R-boundedness: Applications to the Navier-Stokes equations}},
  year = {2022},
  eprint = {2204.11290},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2204.11290},
}
Figure