In measure theory, a sequence of measures that shares a property—such as being absolutely continuous with respect to a reference measure—does not automatically pass that property to its limit. When moving from finite measures to the unbounded ($\sigma$-finite) measures common in physics and stochastic modeling, ensuring this stability becomes significantly more complex.
Work by Lasserre et al. provides a framework for understanding how absolute continuity and density regularity are preserved under topological convergence.
The Problem of Singular Limits
Absolute continuity implies that a measure possesses a density function (a Radon-Nikodym derivative) relative to a reference measure. If a set has zero volume in the reference measure, it must also have zero mass in the absolutely continuous measure.
The classical Vitali-Hahn-Saks (VHS) theorem guarantees that this relationship is preserved in the limit of a sequence of measures, provided the measures are finite and converge "setwise." However, in many applications, measures are defined on infinite domains where they are only $\sigma$-finite. In these cases, the limit can "break" absolute continuity by allowing mass to concentrate on sets that the reference measure considers empty, creating a singularity.
Extending Stability to Unbounded Measures
To address this, Lasserre et al. generalized the VHS theorem to handle unbounded measures by shifting the focus to topological convergence—specifically, convergence against continuous functions that vanish at infinity.
In An extension of the Vitali-Hahn-Saks Theorem, the authors established that the regularity of these density functions is not lost during the limiting process. If a sequence of measures has densities that are bounded within a specific $L^p$ space, the density of the resulting limit measure is guaranteed to remain within that same $L^p$ space. This ensures that the "smoothness" or integrability of the sequence is inherited by the limit.
The Necessity of Density Bounds
The preservation of absolute continuity relies on a specific constraint: the densities must not grow without bound as the sequence progresses. In a Correction to “An extension of the Vitali-Hahn-Saks theorem”, the authors clarified that for the limit measure to inherit absolute continuity, the $L^\infty$ norm of the sequence's densities must remain bounded.
The intuition behind this requirement lies in the relationship between a set and its surrounding open sets. To prove a limit measure is absolutely continuous, one must show that the measure of a null set is zero. By using the regularity of measures, a null set can be enclosed within an open set of arbitrarily small reference measure.
If the $L^\infty$ norm of the densities is bounded, this bound acts as a scaling factor. It prevents the measure of the open set from exploding, even as the sequence of measures evolves. By shrinking the open set toward the null set, the authors use an epsilon-based argument to show that the limit measure of the null set must be zero. Without this $L^\infty$ "safety rail," mass could concentrate so aggressively that the limit measure assigns positive value to a set that the reference measure ignores, destroying absolute continuity.
Key Foundational Work
Correction to “An extension of the Vitali-Hahn-Saks theorem” (1998) — contributed 10 concept elements to this direction.
An extension of the Vitali-Hahn-Saks Theorem (1996) — contributed 7 concept elements to this direction.