Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Log-free estimate for the resonant paraproduct in the 3D Navier-Stokes equations

Pylyp Cherevan

TL;DR

This work proves a log-free a priori estimate for the resonant high–high→low paraproduct in the 3D Navier–Stokes equations. It combines phase–geometric integration by parts along an adapted frame, wave-packet discretization at scale \(N^{-1/2}\), and an anisotropic Strichartz estimate on \(N^{-1/2}\)-long time windows, augmented by bilinear decoupling on a rank-3 phase surface. The analysis splits into a wide angular region, where a phase-time gain of \(\lambda^{-3/2}\) is achieved via two angular IBPs and TT* on short windows, and a narrow almost-collinear region where a null-form suppression near the interaction diagonal yields exact \(\lambda^{-1}\) decay in \(\dot H^{-1}\). The results are scale-consistent and hold without smallness assumptions, relying only on divergence-free structure; the method targets a specific resonant component with discussion of possible generalizations. The paper unfolds a modular, global strategy to merge the dyadic pieces into a global bound, paving the way for extending log-free estimates to other nonlinear interactions in the series on the Navier–Stokes problem.

Abstract

We consider the resonant paraproduct (high-high $\to$ low regime) in the nonlinearity $(u\cdot\nabla)u$ for the three-dimensional Navier-Stokes equations. For sufficiently smooth, divergence-free u, we establish the a priori estimate without logarithmic loss $$\|R_N(u)\|{\dot H^{-1}} \lesssim N^{-1}\,\|u\|{\dot H^{1/2}}\,\|u\|_{\dot H^{1}},$$ with a constant independent of the dyadic frequency $N$. The proof combines phase-geometric integration by parts along an adapted frame, wave-packet discretization at scale $N^{-1/2}$, and an anisotropic Strichartz estimate on time windows of length $N^{-1/2}$. In the wide angular region we apply bilinear decoupling on a rank-3 phase surface; in the geometry at hand the minimal curvature yields a gain of order $N^{-1/6+o(1)}$ (with $o(1)\to 0$ as $N\to\infty$), which suffices to remove the logarithmic loss. The contribution from the narrow region is handled separately by an energy argument in $\dot H^{-1}$ using null-form suppression near the interaction diagonal. The resulting bound is scale-consistent and requires no smallness assumptions, only the divergence-free condition on $u$. The analysis is restricted to a single resonant component of the paraproduct; potential extensions are discussed.

Log-free estimate for the resonant paraproduct in the 3D Navier-Stokes equations

TL;DR

This work proves a log-free a priori estimate for the resonant high–high→low paraproduct in the 3D Navier–Stokes equations. It combines phase–geometric integration by parts along an adapted frame, wave-packet discretization at scale , and an anisotropic Strichartz estimate on -long time windows, augmented by bilinear decoupling on a rank-3 phase surface. The analysis splits into a wide angular region, where a phase-time gain of is achieved via two angular IBPs and TT* on short windows, and a narrow almost-collinear region where a null-form suppression near the interaction diagonal yields exact decay in . The results are scale-consistent and hold without smallness assumptions, relying only on divergence-free structure; the method targets a specific resonant component with discussion of possible generalizations. The paper unfolds a modular, global strategy to merge the dyadic pieces into a global bound, paving the way for extending log-free estimates to other nonlinear interactions in the series on the Navier–Stokes problem.

Abstract

We consider the resonant paraproduct (high-high low regime) in the nonlinearity for the three-dimensional Navier-Stokes equations. For sufficiently smooth, divergence-free u, we establish the a priori estimate without logarithmic loss with a constant independent of the dyadic frequency . The proof combines phase-geometric integration by parts along an adapted frame, wave-packet discretization at scale , and an anisotropic Strichartz estimate on time windows of length . In the wide angular region we apply bilinear decoupling on a rank-3 phase surface; in the geometry at hand the minimal curvature yields a gain of order (with as ), which suffices to remove the logarithmic loss. The contribution from the narrow region is handled separately by an energy argument in using null-form suppression near the interaction diagonal. The resulting bound is scale-consistent and requires no smallness assumptions, only the divergence-free condition on . The analysis is restricted to a single resonant component of the paraproduct; potential extensions are discussed.

Paper Structure

This paper contains 153 sections, 25 theorems, 268 equations.

Table of Contents

  1. Setup and notation
  2. Function spaces and filters
  3. Fourier transform and homogeneous norms.
  4. Dyadic projections (Littlewood–Paley).
  5. Angular decomposition.
  6. Conventions.
  7. Remarks on usage.
  8. Initial data regularity
  9. Working hypotheses and functional classes
  10. Class of solutions on $[0,T]$.
  11. Time windows and scale cylinders.
  12. Localized evolution norms.
  13. Narrow region parameter.
  14. Commutator rules and basic references.
  15. Notation and constants.
  16. ...and 138 more sections

Key Result

Theorem 2.1

Let $u:[0,T]\times\mathbb{R}^3\to\mathbb{R}^3$ be a divergence-free solution with $u\in C([0,T];\dot H^{1/2}\cap \dot H^{1})$ (see also §subsec:init, §subsec:hyp). Then for each $N\in 2^{\mathbb{Z}}$ and all $t\in[0,T]$ one has In integral form, where $X_\sigma$ is given in eq:Xsigma. The constant is independent of $N$ and $T$ and depends on $u$ only through the indicated norms.

Theorems & Definitions (66)

  • Theorem 2.1: Log-free estimate for the resonant block
  • Remark 2.2: Periodic case
  • Remark 2.3: Context
  • Remark 3.1: On variables
  • Lemma 3.2: Block structure and radial component
  • proof : Idea of the proof
  • Lemma 3.3: Nondegeneracy of $A_{\mathrm{eff}}$
  • proof
  • Corollary 3.4: Two angular IBPs
  • proof : Idea of the proof
  • ...and 56 more