The paper tackles a specific estimation problem in point-process theory: recovering the background intensity of a Hawkes process when the triggering mechanism is already known. This setup appears in applications where the self-exciting component is well understood but the exogenous baseline is not, making the background a nuisance parameter that must be inferred from event data.

The authors treat the background not as a simple parametric function but as an unknown finite positive measure with a regular density. To estimate it, they propose a method based on projected Wasserstein-Fisher-Rao flows, which evolve a measure along a geometry that combines optimal transport with intensity adjustments. This choice of flow is designed to handle the nonnegative and mass-preserving structure of the background measure.

Because the source is only an abstract, the full algorithmic details and experimental results are not yet available. The contribution, as stated, is a principled flow-based estimator for a well-defined inverse problem in Hawkes-process inference.