The Heisenberg Uncertainty Principle (HUP) is a fundamental result in mathematics and physics stating that a function and its Fourier transform cannot both be sharply concentrated at the same time. Mathematically, this is captured by an inequality involving the spread of a function and the spread of its frequency content. A "sharp" version of such an inequality means we know the best possible constant in it, and we know exactly which functions achieve that best constant. "Stability" means that if a function almost achieves equality, it must be close to one of those optimal functions. The paper's first contribution is a surprisingly short new proof of a sharp stability result for the second-order HUP, achieved by recognizing it as secretly equivalent to a well-known inequality from a different area of analysis, called the Caffarelli-Kohn-Nirenberg inequality, applied to Fourier transforms. What seemed to require a long, technical argument can now be seen as a quick consequence of this connection.
Building on this insight, the authors construct two new families of uncertainty principles where the usual notion of frequency spread (involving a simple derivative) is replaced by a fractional version, meaning something between no derivative and a full derivative. For each member of these families, they find the best possible constant, identify all the functions that achieve it, and prove stability estimates. One family has the especially clean property that its stability constant equals exactly one, regardless of the dimension of the space or the order of the fractional operator. They also handle cases where the fractional power is negative, which corresponds to a classical object called a Riesz potential. These results significantly broaden the landscape of known sharp uncertainty principles.
The paper also goes beyond simple proximity to optimal functions by showing that the "deficit," meaning how far a function is from achieving equality, controls a whole cascade of increasingly refined remainder terms. Each term in this chain is measured against specific explicit functions and tied to gaps in the spectrum of a particular operator, giving a hierarchy of increasingly sharp information about how non-optimal a given function is. For the classical HUP this yields three such remainder terms with optimal constants, and for the second-order HUP it yields four, expressed through special mathematical functions called confluent hypergeometric profiles. As a bonus application, analogous stability chains are derived for vector fields satisfying a curl-free condition, showing the broad reach of the framework.