The paper develops a mathematical framework for studying certain types of differential equations where time does not flow continuously but can include sudden jumps. Normally, evolution equations (which describe how physical quantities change over time, like heat spreading through a material) are set up with a standard continuous time variable. Here, the authors replace ordinary time with a more general "Stieltjes" time, meaning the equations are driven by a measure that can assign concentrated weight to specific moments, creating instantaneous jumps. The key modeling choice is to use the "posterior state," meaning the value of the solution just after a jump, rather than just before. This turns out to be the right choice because at jump moments the equation automatically produces a well-posed implicit update rule, similar to what one would use in a stable numerical time-stepping scheme.
The core technical contributions involve proving several foundational results for this generalized setting. These include a product rule for differentiation, a way to reconstruct a strong solution from a weak one, and an energy identity that carefully tracks not just the smooth evolution but also the discrete energy contributed by each jump. The energy identity is particularly subtle because jumps in the solution contribute extra positive terms that must be accounted for. The authors prove that solutions exist using two independent methods: one based on Galerkin approximation (projecting onto finite-dimensional subspaces) and one based on a Rothe scheme (a time-discretization approach adapted to the jump structure of the measure). A worked example with a heat equation driven by Stieltjes time illustrates how the theory applies in practice.
The paper also carefully explains some technical obstacles that arise compared to the classical continuous-time theory. In the standard setting, a classical result called the Aubin-Lions theorem is used to extract convergent subsequences when proving existence. Here, that theorem does not directly apply because the function spaces involved are different when using posterior states. The authors show that in their linear, coercive setting this is not actually a problem, and they work around it without needing that tool. Overall, the paper provides a rigorous and self-contained foundation for parabolic evolution equations driven by general measures, unifying continuous dynamics and discrete jump updates within a single coherent framework.