The trace-free Beurling--Ahlfors transform and the Bourgain--Brezis problem for Hodge systems

Technical Summary: The Trace-Free Beurling–Ahlfors Transform and the Bourgain–Brezis Problem for Hodge Systems

Problem Statement
The paper addresses the Bourgain–Brezis problem concerning critical underdetermined partial differential equations, specifically Hodge systems on $\mathbb{R}^n$. The central question is whether, for a given exact differential form $dv$ in a critical Sobolev space $\dot{W}^{n/p,p}\Lambda^l$ (or more generally, Triebel–Lizorkin and Besov spaces), there exists a primitive $u$ such that $du = dv$ and $u$ belongs to the intersection of the critical space and $L^\infty$.

While the canonical solution $u = -|D|^{-2}\nabla f$ (in the divergence case) or its Hodge analogues often fail to be bounded due to the failure of the critical Sobolev embedding $\dot{W}^{n/p,p} \not\subset L^\infty$, Bourgain and Brezis previously established that a bounded solution exists for specific cases (e.g., $p=n$ in the divergence equation, or $p=2$ in the Hilbertian setting for Hodge systems). The general conjecture posits that this phenomenon holds for all $1 < p < \infty$ and all form degrees $1 \le l \le n-1$. Prior to this work, progress was limited: the dual method was confined to the Hilbertian case ($p=2$), and approximation methods were effective only for large $p$ (specifically where $n < p+l$), leaving the range $1 < p \le 2$ largely inaccessible for general form degrees.

Methodology
The paper introduces a novel approach within the dual formulation of the problem. The core methodology relies on the identification and exploitation of a trace-free Beurling–Ahlfors transform ($S$) acting on differential forms.

1. Definition of $S$: The author defines $S$ as a canonical normalization of the generalized Beurling–Ahlfors transform $\tilde{S} = P - P^\perp$, where $P$ and $P^\perp$ are the Hodge projections onto exact and coexact forms, respectively. The transform is defined as:
   $$S = \frac{n-l}{n}P - \frac{l}{n}P^\perp = \frac{1}{2}\left(\tilde{S} - \frac{2l-n}{n}\text{Id}\right).$$
   This normalization ensures that the trace of the symbol $S(\xi)$ vanishes for all $\xi \neq 0$.

2. Reflection Cancellation: A key technical insight is that the matrix symbol of $S$ decomposes into scalar multipliers of the form $\frac{\xi_i^2 - \xi_j^2}{|\xi|^2}$ and $\frac{\xi_i \xi_j}{|\xi|^2}$. These multipliers are odd under specific orthogonal reflections. This symmetry allows the author to prove an endpoint cancellation estimate: the operator $|D|^{-n}S$ maps finite Radon measures (and specifically $L^1_0$ functions) boundedly into $L^\infty$. This is a significant improvement over the identity operator, which does not possess this boundedness property from measures to $L^\infty$.

3. Duality Principle: The paper utilizes a duality principle (Proposition 3.2) which equates the existence of bounded selections for the primal Hodge system to an endpoint estimate for the coexact component of integrable forms in the dual space. Specifically, proving that $\|\omega\|_X \lesssim \|\omega\|_{L^1_0} + \|P\omega\|_X$ for coexact forms allows one to recover the Bourgain–Brezis selection estimate.

4. Multilinear Extensions: To move beyond the Hilbertian case ($p=2$), the author extends the linear reflection cancellation to multilinear settings. By expanding powers of dyadic blocks (Littlewood–Paley theory) and utilizing the reflection symmetry of the multipliers, the paper derives estimates for non-Hilbertian exponents. This involves controlling multilinear expressions of the form $\int (\Delta_j S\omega) (\Delta_j u)^{k}$ using the oddness of the multipliers.

Key Contributions and Results

• Completion of the Hilbertian Case: The paper proves the Bourgain–Brezis conjecture for the Hilbertian case ($p=2$) in every dimension $n \ge 2$ and for every form degree $1 \le l \le n-1$. This resolves a gap where previous dual methods were limited to specific degrees (e.g., $l=1$ or $l=n-1$).

  • Result: $d[\dot{H}^{n/2}\Lambda^l] = d[(\dot{H}^{n/2} \cap L^\infty)\Lambda^l]$.

• New Non-Hilbertian Ranges: The paper establishes Bourgain–Brezis estimates for a new family of exponents arbitrarily close to $p=1$.

  • Triebel–Lizorkin Scale: For $1 < p \le q \le 2$, if $q'$ is even and $p'/q' \in \mathbb{N}$, the conjecture holds.
  • Besov Scale: For $1 < p \le 2$ and $1 \le q \le p$, if $p'$ is an even integer, the conjecture holds.
  • Sobolev Consequence: By setting $q=2$ (identifying $\dot{W}^{n/p,p} = \dot{F}^{n/p}_{p,2}$), the conjecture is proven for $p = \frac{2k}{2k-1}$ for integers $k \ge 1$. This provides a sequence of exponents converging to 1 where the conjecture holds for all dimensions and form degrees.

• Endpoint Hodge Decompositions: The results yield new Hodge decompositions where both the exact and coexact potentials can be chosen to be bounded. Specifically, for the identified critical spaces, the space of exact forms $d[\dot{X}\Lambda^{l-1}]$ admits a decomposition where the potential lies in $(\dot{X} \cap L^\infty)\Lambda^{l-1}$.

• Intrinsic Nonlinearity: The paper proves that the bounded selections guaranteed by these estimates cannot be realized by bounded linear operators (except in the trivial case where the critical space already embeds into $L^\infty$, i.e., Besov spaces with $q=1$). This confirms that the "correction" of the singular canonical solution to a bounded one is inherently nonlinear.

Significance
The paper claims that the dual approach to the Bourgain–Brezis problem is "substantially more flexible than previously understood." By isolating the trace-free Beurling–Ahlfors structure, the author demonstrates that the dual method is not restricted to the Hilbertian setting ($p=2$) but can be extended via multilinear reflection estimates to cover critical exponents near $p=1$. This complements existing approximation methods (which work well for large $p$) and provides a unified framework for the conjecture across the critical parameter range. The work also clarifies the structural reasons (reflection symmetry of multipliers) behind the endpoint cancellations that make these estimates possible, offering a deeper geometric understanding of the Hodge system's behavior at critical regularity.