On variable Lebesgue spaces

Abstract

The reader will recall that the classical $p$-Lebesgue spaces are those functions defined on a measure space $(X, \mu)$ whose modulus raised to the $p^{\rm th}$ power is integrable. This condition gives many quantitative measurements on the growth of the function, both locally and globally. Results and applications pertaining to such functions are ubiquitous. That said, the constancy of the exponent $p$ when computing $\int_X \abs{f}^p d\mu$ is limiting in the sense that it is intrinsically uniform in scope. Speaking loosely, there are instances in which one is concerned with the $p$ growth of a function in a region $A$ and its $q$ growth in another region $B$. As such, allowing the exponent to vary from region to region (or point to point) is a reasonable course of action.

The task of developing such a theory was first taken up by Wladyslaw Orlicz in the 1930's. The theory he developed, of which variable Lebesgue spaces are a special case, was only intermittently studied and analyzed through the end of the century. However, at the turn of the millennium, several results and their applications sparked a focused and intense interest in variable $L^p$ spaces. It was found that with very few assumptions on the exponent function many of the classical structure and density theorems are valid in the variable-exponent case. Somewhat surprisingly, these results were largely proved using intuitive adaptations of well-established methods. In fact, this methodology set the tone for the first part of the decade, where a multitude of ``affirmative'' results emerged. While the successful adaptation of classical results persists to a large extent today, there are nontrivial situations in which one cannot hope to extend a result known for constant $L^p$.

In this paper, we wish to explore both of the aforementioned directions of research. We will first establish the fundamentals for variable $L^p$. Afterwards, we will apply these fundamentals to some classical $L^p$ results that have been extended to the variable setting. We will conclude by shifting our attention to Littlewood-Paley theory, where we will furnish an example for which it is impossible to extend constant-exponent results to the variable case.