Quantitative Hölder estimates for even singular integral operators on patches (Q2165536)

In this article, the authors study the behavior of higher-order Riesz transform operators of even order on characteristic functions \(1_D (x)\) of bounded domains \(D\) of class \(C^{1+\sigma}\) given by \[ T(1_D) (x) = pv \int_{\mathbb{R}^{n}} K(x-y) \, 1_D(y) \, dy, \,\,\,\,\, x \in \mathbb{R}^{n}, \] where \(K(x) = \frac{P_{2l}(x)}{|x|^{n+2l}}\) and \(P_{2l}\) is a homogeneous harmonic polynomial of degree \(2l\), \(l \geq 1\), in \(\mathbb{R}^{n}\). In the classical theory of singular integrals, the function \(T(1_D)\) belongs to the BMO space. The authors study the Hölder regularity of the function \(T(1_D)\). The regularity of this class of functions is studied in terms of the following odd associated operator on the boundary \[ S(f)(x) = pv \int_{\partial D} k(x-y) \, f(y) \, ds(y), \,\,\,\,\, x \in \mathbb{R}^{n}, \] where \(k(x) = \frac{Q_{2l-1}(x)}{|x|^{n+2l-2}}\) and \(Q_{2l-1}\) is a homogeneous harmonic polynomial of degree \(2l-1\). The main results of this work are contained in Theorems 2.2 and 2.3. Theorem 2.2 states that the operator \(S\) maps boundedly \(\dot{C^{\sigma}}(\partial D)\) into \(\dot{C^{\sigma}}(\overline{D}) \cup \dot{C^{\sigma}}(\mathbb{R}^{n} \setminus D)\) and \[ \| S(f) \|_{\dot{C^{\sigma}}(\overline{D}) \cup \dot{C^{\sigma}}(\mathbb{R}^{n} \setminus D)} \leq C (1 + |\partial D|) P(\| D \|_{\ast} + \| D \|_{Lip}) (\| f \|_{C^{\sigma}} + \|f\|_{L^{\infty}} \| D \|_{\dot{C^{1+\sigma}}}), \] holds with \(P\) a polynomial function depending on \(S\), and \(C = C(n, \sigma)\). Theorem 2.3 follows as a corollary of Theorem 2.2. This says that \(T(1_D)\) defines a piecewise \(\dot{C^{\sigma}}\) function and the quantitative bound \[ \| T(1_D) \|_{\dot{C^{\sigma}}(\overline{D}) \cup \dot{C^{\sigma}}(\mathbb{R}^{n} \setminus D)} \leq C (1 + |\partial D|) P(\| D \|_{\ast} + \| D \|_{Lip}) (1 + \| D \|_{\dot{C^{1+\sigma}}}),\tag{1} \] holds with \(P\) a polynomial function depending on \(T\), and \(C = C(n, \sigma)\). Since one of the main motivations of these results is physical, they show the techniques in dimension three. Concretely, they prove Theorem 2.2 for the kernel \(k(x) = \frac{x_1 x_2 x_3}{|x|^{5}}\), \(x=(x_1, x_2, x_3)\). This case is entirely representative for the general case. With this kernel, the authors provide a constructive and direct method, showing the main difficulties and cancellations. The proof is long, and distinguishes three situations: when the two points are on the boundary, near the boundary, and far from the boundary. The motivation for these estimates comes from a classical two-dimensional fluid mechanics problem. Namely, the dynamics of two incompressible immiscible fluids evolving by the inhomogeneous Navier-Stokes equations. In this setting, the estimate \((1)\) is used in the proof of the main result in [\textit{F. Gancedo} and \textit{E. García-Juárez}, `` Global regularity of 2D Navier-Stokes free boundary with small viscosity contrast'', Preprint, \url{arXiv:2109.08764}].