The Bernstein-Sato polynomial is a special algebraic object attached to any polynomial or analytic function. It encodes deep information about the singularities of that function, and its roots are closely connected to important invariants in geometry and algebra, such as the monodromy of the function and quantities called "poles of the zeta function." Computing or even estimating these roots is generally very hard, and a central goal in singularity theory is to find geometric ways to bound or describe them.
One established approach uses "log resolutions," which are geometric procedures that untangle the complicated structure of a singularity by passing to a smoother space where the singular behavior becomes easier to analyze. Classical results let you estimate the roots of the Bernstein-Sato polynomial using data read off from an ordinary log resolution. This paper extends that framework by allowing the resolving space to be an "orbifold" rather than a smooth manifold. Orbifolds are spaces that look smooth almost everywhere but have mild, well-controlled quotient singularities at some points. Because orbifold resolutions can be simpler and more efficient than fully smooth ones, they carry different numerical data that can lead to sharper or more accessible estimates.
The main contribution is a set of new bounds on where the roots of the Bernstein-Sato polynomial can sit, expressed in terms of the geometric data coming from an orbifold log resolution. This is useful because in many practical situations a natural resolution of a singularity is an orbifold rather than a smooth space, and the classical machinery did not directly apply. The results here fill that gap and provide tools that can give better estimates in cases where the smoothest available resolution still retains some orbifold structure.