On regularity of solutions to the Navier--Stokes equation with initial data in $\mathrm{BMO}^{-1}$

This paper establishes that mild solutions to the incompressible Navier-Stokes equations with initial data in $\mathrm{BMO}^{-1}$ are weak*-continuous in time and that global solutions vanish in $\mathrm{BMO}^{-1}$ as time approaches infinity.

The Gist

Here is an explanation of the paper "On Regularity of Solutions to the Navier–Stokes Equation with Initial Data in BMO⁻¹" by Hedong Hou, translated into everyday language with creative analogies.

The Big Picture: Predicting the Weather of Fluids

Imagine you are trying to predict how a drop of ink will spread in a glass of water, or how a hurricane will swirl across the ocean. In physics, this is described by the Navier–Stokes equations. These are the "rules of the road" for fluids.

However, these rules are notoriously difficult. If you start with a very messy, chaotic, or "rough" initial state (like a sudden, violent splash), the math often breaks down. It's like trying to predict the path of a leaf in a tornado; the equations might tell you the leaf exists, but they can't tell you exactly where it will be a second later, or if it will suddenly vanish into thin air.

The Problem: The "Rough" Starting Point

Mathematicians have spent decades trying to figure out what happens if you start with a fluid that is extremely rough.

• The Smooth Case: If you start with a calm, smooth fluid, we know the solution behaves nicely. It flows continuously, and we can track it easily.
• The Rough Case (BMO⁻¹): This paper focuses on a specific type of "roughness" called BMO⁻¹. Think of this as a fluid that is full of tiny, chaotic jitters everywhere. It's not smooth at all.

The big question was: If we start with this chaotic fluid, does the solution stay "well-behaved" over time?
Specifically, two things were unknown:

1. Continuity: Does the fluid change smoothly as time passes, or does it jump around wildly?
2. Long-term behavior: If we wait a very long time, does the fluid eventually calm down and disappear, or does it stay chaotic forever?

The Solution: The "Ghostly" Connection

Hedong Hou's paper proves two major things about these chaotic fluids. To understand them, let's use an analogy.

Analogy 1: The "Ghostly" Handshake (Continuity)

Imagine you are shaking hands with a ghost. You can feel the pressure of the hand (the fluid's state), but you can't see the hand clearly.

• The Old View: We knew that if the fluid started smooth, the handshake was firm and continuous.
• The New Discovery: Hou proves that even if the fluid starts as a "ghostly" mess (BMO⁻¹), the handshake is still continuous.
• The Catch: It's a weak handshake. You can't feel every tiny twitch of the ghost's fingers (strong continuity), but you can feel the overall pressure is there and changing smoothly over time.
• Why it matters: This means the fluid doesn't suddenly "teleport" or break the laws of physics. It evolves in a predictable, continuous way, even if it's messy. It's like a chaotic dance that never skips a beat, even if the dancers are stumbling.

Analogy 2: The Fading Echo (Long-Term Behavior)

Imagine you shout into a canyon.

• The Smooth Case: If you shout clearly, the echo fades away completely after a while.
• The Rough Case: We were worried that if you shouted a chaotic, static-filled noise, the echo might get stuck in the canyon forever, bouncing around endlessly.
• The New Discovery: Hou proves that the echo always fades away. Even if you start with the most chaotic noise imaginable (within the BMO⁻¹ rules), as time goes on ($t \to \infty$), the fluid's energy dissipates. The "ghost" eventually fades into silence. The fluid returns to a state of calm (zero).

Why This is a Big Deal

Before this paper, mathematicians knew these results were true for smooth fluids or fluids that were "almost" smooth. But for the specific type of roughness called BMO⁻¹, there was a gap in our knowledge.

Hou filled that gap by using a clever mathematical tool called Tent Spaces.

• The Analogy: Imagine trying to measure the shape of a tent that is being blown by a storm. It's hard to measure the whole thing at once. Instead, Hou broke the tent down into small, manageable poles and fabric sections (mathematical operators). He showed that even though the storm is wild, the structure of the tent (the solution) holds together and eventually settles down.

The "Sharpness" Warning

The paper also includes a fascinating warning. Hou shows that his results are the best possible.

• If you try to demand that the fluid behaves perfectly smoothly (strong continuity) or that it vanishes perfectly (in a strong sense), the math breaks down.
• The Self-Similar Monster: He points out that there are special, self-repeating fluid patterns (like a fractal) that stay the same size forever. If you start with one of these, the fluid never truly "vanishes" in the strictest sense. It just keeps spinning in a perfect loop. This proves that the "weak" continuity and "fading" results Hou found are the absolute limit of what is possible. You can't ask for more.

Summary

In simple terms, this paper says:

"Even if you start the universe of fluids with a chaotic, messy explosion, the laws of physics ensure that the chaos evolves smoothly (in a ghostly, weak sense) and eventually settles down to nothing. The fluid might be wild, but it never goes crazy."

This gives mathematicians and physicists greater confidence in using these equations to model real-world phenomena, from blood flow in veins to the movement of galaxies, even when the starting conditions are far from perfect.

Technical Summary

Here is a detailed technical summary of the paper "On Regularity of Solutions to the Navier–Stokes Equation with Initial Data in BMO$^{-1}$" by Hedong Hou (arXiv:2410.16468v2).

1. Problem Statement

The paper addresses the regularity and long-time behavior of mild solutions to the incompressible Navier–Stokes (NS) equations in $\mathbb{R}^n$:
$$\begin{cases} \partial_t u + (u \cdot \nabla)u = \Delta u - \nabla p, \\ \nabla \cdot u = 0, \\ u(0) = u_0, \end{cases}$$
where the initial data $u_0$ belongs to the critical space BMO$^{-1}$ (the space of distributions $f$ such that $f = \text{div } g$ for some $g \in \text{BMO}$).

Context and Gap:

• Critical Spaces: The NS equation is scale-invariant. The space BMO$^{-1}$ is a critical space (scale-invariant) lying between $L^n$ and $\dot{B}^{-1}_{\infty, \infty}$.
• Existing Results:
  • Koch and Tataru (2001) established global existence of mild solutions in a specific space $X_\infty$ for small initial data in BMO$^{-1}$.
  • Dubois (2002) and Auscher–Dubois–Tchamitchian (2004) proved that for initial data in VMO$^{-1}$ (the closure of Schwartz functions in BMO$^{-1}$), solutions are strongly continuous in time and vanish at infinity.
• The Open Question: For general initial data in BMO$^{-1}$ (which includes singular distributions not in VMO$^{-1}$), the time regularity was unknown. Specifically, it was unclear if mild solutions are continuous in time with values in BMO$^{-1}$, and whether they vanish as $t \to \infty$ in the strong topology.

2. Methodology

The author employs a functional analytic approach based on tent spaces, Hardy–Sobolev spaces, and duality arguments.

Key Tools:

1. Function Spaces:

   • BMO$^{-1}$: Identified as the dual of the homogeneous Hardy–Sobolev space $\dot{H}^{1,1}$. The paper equips BMO$^{-1}$ with the weak-topology* induced by this duality.
   • Tent Spaces ($T^{\infty, q}_\beta$): Used to characterize the solution space $X_T$. The norm involves Carleson measures.
   • Koch–Tataru Space ($X_T$): Defined by $\|u\|_{X_T} = \sup_{t} \|t^{1/2}u(t)\|_{L^\infty} + \|C_T^{(2)}(u)\|_{L^\infty}$.

2. Decomposition of the Integral Operator:
   The mild solution satisfies the integral equation $u(t) = e^{t\Delta}u_0 - B(u,u)(t)$. The core of the proof involves analyzing the bilinear operator $B(u,u)$, which is a linear operator $L$ applied to $f = u \otimes u$.
   The author decomposes the linear operator $L(f)(t) = \int_0^t e^{(t-s)\Delta} P \text{div } f(s) ds$ into two parts using a specific identity involving the heat semigroup:
   $$L(f) = L_0(g) + L_1(f)$$

   • $L_0(g)$: A "maximal regularity" part involving $\Delta e^{(t-s)\Delta}$.
   • $L_1(f)$: A "remainder" part involving $e^{(t+s)\Delta}$.

3. Duality and Density:

   • Since BMO$^{-1}$ is the dual of $\dot{H}^{1,1}$, continuity is proven by showing that the pairing $\langle u(t), \phi \rangle$ is continuous for all test functions $\phi$ in a dense subset of $\dot{H}^{1,1}$ (specifically Schwartz functions with vanishing moments, $S_\infty$).
   • The proof relies heavily on estimates of heat kernels and Gaussian bounds to control the action of the semigroup on these test functions.

3. Key Contributions and Results

Main Theorem 1: Time Regularity (Weak*-Continuity)

Theorem 1.1: Let $u_0 \in \text{BMO}^{-1}$ be divergence-free, and let $u \in X_T$ be a mild solution. Then:

• $u \in C([0, T); \text{BMO}^{-1})$ with respect to the weak-topology*.
• $u(0) = u_0$.
• The solution satisfies the bound: $\sup_{0 \le t < T} \|u(t)\|_{\text{BMO}^{-1}} \lesssim \|u_0\|_{\text{BMO}^{-1}} + \|u\|_{X_T}^2$.

Significance: This extends the strong continuity result of Dubois (which required $u_0 \in \text{VMO}^{-1}$) to the full space BMO$^{-1}$. The author proves that while strong continuity fails for general BMO$^{-1}$ data (due to the presence of self-similar solutions), weak-continuity* holds.

Main Theorem 2: Long-Time Behavior

Theorem 1.2: Let $u_0 \in \text{BMO}^{-1}$ and $u \in X_\infty$ be a global mild solution. Then:

• $u(t) \to 0$ in BMO$^{-1}$ (weak*-topology) as $t \to \infty$.

Significance: This establishes that even for large initial data (provided a global solution exists in $X_\infty$), the solution decays to zero in the critical space.

Sharpness of Results (Remark 1.3)

The author demonstrates that these results are optimal:

• Weak-Continuity is Best Possible:* The heat semigroup is only weak*-continuous on BMO$^{-1}$, not strongly continuous.
• Strong Decay Fails: If one considers the strong topology of BMO$^{-1}$, the solution does not necessarily vanish at infinity. The paper cites the existence of non-zero self-similar solutions (homogeneous of degree $-1$) for small initial data. For such solutions, $\|u(t)\|_{\text{BMO}^{-1}} = \text{const} \neq 0$. This contrasts with the VMO$^{-1}$ case where strong decay holds.

4. Technical Proof Outline

1. Reduction to Linear Problems:
   The proof reduces the nonlinear problem to proving that the linear operator $L(f)$ maps specific tent space intersections ($L^\infty_{-1} \cap T^{\infty, 2}_{-1/2} \cap T^{\infty, 1}$) into $C_0([0, \infty); \text{BMO}^{-1})$.

2. Analysis of $L_0(g)$ (Section 4):

   • Continuity at $t=0$: Proven via duality. The author estimates the integral $\int_0^t \int |g| |e^{(t-s)\Delta}\Delta \phi|$ by splitting the time domain and using properties of tent spaces and the heat kernel.
   • Continuity at $t>0$: Uses the strong continuity of the heat semigroup on $\dot{H}^{1,1}$ and the dominated convergence theorem.
   • Limit at $t \to \infty$: Relies on the decay of the heat semigroup acting on test functions in $\dot{H}^{1,1}$.

3. Analysis of $L_1(f)$ (Section 5):

   • This term involves $e^{(t+s)\Delta}$, which provides better decay properties.
   • The proof uses the duality between $T^{\infty, 1}$ and $T^{1, \infty}$ to bound the integral against test functions in $\dot{H}^{0,1}$.
   • Strong continuity of the heat semigroup on $\dot{H}^{0,1}$ is utilized to show convergence to zero.

5. Significance and Impact

• Completes the Regularity Picture: The paper resolves the long-standing question of time regularity for NS solutions with BMO$^{-1}$ initial data, bridging the gap between the VMO$^{-1}$ (strong continuity) and $\dot{B}^{-1}_{\infty, \infty}$ (ill-posedness) regimes.
• Clarifies Topology Dependence: It rigorously distinguishes between strong and weak*-topologies in critical spaces, showing that while self-similar solutions prevent strong decay, they still vanish in the weak*-sense.
• Methodological Advancement: The decomposition technique for the bilinear operator and the precise use of tent space duality provide a robust framework for analyzing critical regularity problems in fluid dynamics.
• Uniqueness Context: While the paper does not solve the uniqueness problem for large BMO$^{-1}$ data (which remains open and is suggested to be false by recent numerical evidence), it establishes the necessary regularity properties for any existing mild solution.

In summary, Hou proves that mild solutions to the Navier–Stokes equations with BMO$^{-1}$ initial data are weak-continuous* in time and decay to zero in the weak*-topology as $t \to \infty$, while demonstrating that strong continuity and strong decay are generally impossible due to the existence of self-similar solutions.