Revisiting multi-phase variational problems: A Muckenhoupt weight approach

Technical Summary: Revisiting Multi-phase Variational Problems: A Muckenhoupt Weight Approach

Problem Setting
This paper investigates the regularity theory of local minimizers for multi-phase energy functionals of the form:
$$\mathcal{F}(v, \Omega) = \int_{\Omega} \left( |\nabla v|^p + \sum_{i=1}^N a_i(x)|\nabla v|^{p_i} \right) dx,$$
where $\Omega \subset \mathbb{R}^n$ ($n \geq 2$) is a bounded open set, $1 < p \leq p_1 \leq \dots \leq p_N < \infty$, and $a_i(x)$ are non-negative modulating coefficients. The functional models composite materials with multiple growth phases, where the coefficients $a_i(x)$ dictate the geometric distribution of different material components. The associated Euler-Lagrange equation exhibits highly non-uniform and degenerate ellipticity, particularly where coefficients vanish or where the gradient becomes large.

Methodological Shift and Motivation
Traditionally, regularity results for such functionals (including the double-phase case $N=1$) rely on structural assumptions where the modulating coefficients $a_i(\cdot)$ are Hölder continuous ($C^{0,\alpha}$) and satisfy specific growth gap conditions (e.g., $p_i \leq p + \alpha_i p/n$). These conditions ensure the absence of the Lavrentiev phenomenon and allow for the construction of De Giorgi iteration schemes.

The authors propose a distinct structural framework by replacing the classical Hölder continuity assumption with the condition that the coefficients $a_i$ belong to suitable Muckenhoupt weight classes ($a_i \in \mathcal{A}_{p_i}$). This shift is motivated by the physical reality that coefficients in strongly anisotropic materials may exhibit discontinuities or singularities (e.g., power-type or logarithmic singularities) that violate Hölder continuity but remain admissible within the Muckenhoupt setting. The Muckenhoupt condition provides the necessary and sufficient criteria for the boundedness of the Hardy-Littlewood maximal operator on weighted spaces, which is crucial for density arguments.

Key Contributions and Methodology
The paper develops a regularity theory for these multi-phase functionals under Muckenhoupt assumptions through the following technical steps:

1. Absence of the Lavrentiev Phenomenon: The authors first establish that smooth functions are dense in the natural Musielak-Orlicz-Sobolev space $W^{1,H}(\Omega)$ associated with the functional. This is achieved by proving the boundedness of the Hardy-Littlewood maximal operator on the weighted space defined by the Muckenhoupt coefficients. This density result rules out the Lavrentiev gap, ensuring that the infimum of the energy over the natural space coincides with that over smooth functions.

2. Higher Integrability (Theorem A): Using a refined Gehring-type argument, the paper proves that the energy density $H(x, |\nabla u|)$ of a local minimizer enjoys higher integrability ($L^{1+\delta_0}_{loc}$). This relies on establishing a reverse Hölder inequality for the minimizer, which is derived by combining Caccioppoli-type estimates, Sobolev-Poincaré inequalities adapted to the multi-phase structure, and the self-improving properties of Muckenhoupt weights.

3. Local Boundedness (Theorem B): Building on higher integrability, the authors establish the local boundedness ($L^\infty_{loc}$) of minimizers. This is achieved via a De Giorgi-type iteration scheme. A critical component is the derivation of a specialized level-set inequality that quantifies the oscillation of the minimizer in terms of the multi-phase energy density, accounting for the non-homogeneous growth and the weighted nature of the coefficients.

4. Hölder Continuity (Theorem C): The local Hölder continuity of minimizers is proven by demonstrating the decay of oscillation in smaller balls. This involves a specialized De Giorgi-type level-set inequality that bridges the boundedness and continuity properties, leveraging the previously established $L^\infty$ bounds and the structural properties of the Muckenhoupt weights.

5. Harnack Inequality (Theorem D): The paper proves the Harnack inequality for non-negative local minimizers, stating that the local supremum is controlled by the local infimum. This result is derived by combining a weak Harnack inequality (infimum estimate) and a supremum estimate. The proof is noted as highly non-trivial due to the irregularity of the coefficients, requiring a delicate combination of weighted estimates and scaling arguments to bridge the two one-sided bounds.

Key Results

• Theorem A: Local minimizers satisfy $H(\cdot, |\nabla u|) \in L^{1+\delta_0}_{loc}(\Omega)$ for some $\delta_0 > 0$.
• Theorem B: Local minimizers are locally bounded ($u \in L^\infty_{loc}(\Omega)$), with quantitative estimates relating the $L^\infty$ norm to the local energy.
• Theorem C: Local minimizers are locally Hölder continuous ($u \in C^{0,\alpha}_{loc}(\Omega)$).
• Theorem D: Non-negative local minimizers satisfy the Harnack inequality: $\sup_{B_r} u \leq C \inf_{B_r} u$.

Significance and Claims
The paper claims to provide a new framework for understanding the qualitative behavior of minimizers in non-uniformly elliptic variational problems that complements existing literature. By moving beyond the classical Hölder continuity of modulating coefficients, the authors handle a broader class of irregular coefficients that may vanish or blow up on sets of measure zero.

Most notably, the paper asserts that the proof of the Harnack inequality for non-negative local minimizers in the multi-phase setting with Muckenhoupt modulating coefficients stands as the first result of its kind. This result offers a new quantitative tool to control the local behavior of minimizers in settings where the ellipticity ratio can degenerate or blow up due to the interaction of multiple growth phases and singular weights. The work contributes to the broader understanding of regularity in non-standard growth problems, specifically addressing the challenges posed by the lack of scale-invariance and homogeneity in multi-phase models.