The paper tackles a classic problem in probability theory: figuring out how long it takes for a randomly wandering particle (Brownian motion) to first reach a moving boundary. Specifically, the boundary moves according to a power law in time, like a parabola or a cubic curve. These "first passage" problems matter in finance, physics, and biology, but they are notoriously hard to solve when the boundary is not a simple straight line. The authors focus on boundaries defined by whole-number powers of two or greater.
The main contribution is a new mathematical construction called a "generalized image measure." The classical approach to solving heat flow problems near boundaries uses the method of images, where you place a fictional mirror source of heat on the other side of a wall to enforce the right boundary condition. Here, the boundary is moving and curved, so the authors generalize this idea into something more flexible: a signed measure (meaning it can have both positive and negative parts) that is fixed in time but spread across a spatial coordinate. They show this object satisfies an equation of a specific type (a Volterra equation of the second kind) whose building block is a special mathematical function called a Wright function. Once they find this image measure, they can extract the actual first-passage probability density by tracking the flux of probability at the boundary.
The paper validates the approach in two ways. First, the authors prove rigorous mathematical bounds confirming that the relevant quantities converge properly. Second, they connect their new framework to earlier work on moving zeros and differential equations, showing the two perspectives are consistent. To make the results concrete, they run a numerical experiment using a cubic boundary, comparing the formula against large-scale computer simulations of random walks. The two agree well, giving confidence that the method works in practice and could be extended to other curved boundaries.