The Frisch-Parisi conjecture. II: Besov spaces in multifractal environment, and a full solution (Q6163241)

The paper and its companion [\textit{J. Barral} and \textit{S. Seuret}, J. Math. Pures Appl. (9) 175, 76--108 (2023; Zbl 07693670)] are devoted to resolve the Frisch-Parisi conjecture as formulated by Jaffard (cf [\textit{S. Jaffard}, J. Math. Pures Appl. (9) 79, No. 6, 525--552 (2000; Zbl 0963.28009)]). The conjecture asks whether for a given concave continuous function \(\sigma: \mathbb{R} \to [0,d]\cup \{-\infty\}\), one can find a Baire space \(\mathcal{B}\) of functions on \(\mathbb{R}^d\) such that any typical (in the Baire sense) element of \(\mathcal{B}\) \begin{itemize} \item[1.] has \(\sigma\) as its multifractal spectrum, \item[2.] obeys some multifractal formalism. \end{itemize} The construction of spaces satisfying condition 1 is done in [\textit{J. Barral} and \textit{S. Seuret}, J. Math. Pures Appl. (9) 175, 76--108 (2023; Zbl 07693670)]. Developing on those results, the current paper is devoted to a full resolution of the problem. The multifractal (also called singularity) spectrum is understood as follows. Given function \(f:\mathbb{R}^d \to \mathbb{R}\) and index \(q>0\) let \(E(f;q)\) be the set of \(x\in \mathbb{R}^d\), where \(f\) is locally \(q\)-Holder (and no better than that). Then the multifractal spectrum \(\sigma_f\) is a function on \(\mathbb{R}\) defined as \[ \sigma_f (q) = \begin{cases} \dim(E(f;q)) & \text{if } E(f;q)\neq \varnothing \\ -\infty & \text{otherwise}, \\ \end{cases} \] where \(\dim(\cdot)\) is the Hausdorff dimension. The multifractal formalism is defined in the context of a specified \textit{scaling function} \(\zeta_f\), e.g. Frisch and Parisi considered \(\zeta_f\) defined by the relation \[ |h|^{-d} \int |f(x+h)-f(x)|^q \approx |h|^{\zeta_f(q)}, \quad \text{ as } h\to 0. \] Function \(f\) is said to obey the multifractal formalism with scaling \(\zeta_f\) if the multifractal spectrum \(\sigma_f\) equals to the Legendre-Frenchel transform of the scaling function \(\zeta_f\). Valid multifractal formalism makes it possible to calculate/approximate the multifractal spectra using a scaling function which is typically based on ``averaged quantities'', thus allowing stable numerical computations. The methodology of the paper (as well as its companion [\textit{J. Barral} and \textit{S. Seuret}, J. Math. Pures Appl. (9) 175, 76--108 (2023; Zbl 07693670)] and [\textit{S. Jaffard}, J. Math. Pures Appl. (9) 79, No. 6, 525--552 (2000; Zbl 0963.28009)]) is based on methods of wavelet theory. In particular, the scaling functions \(\zeta_f\) are formed using the ``leading'' coefficients in the wavelet decompositions of \(f\).