H.B. Keller Colloquium

Inverse problems are ubiquitous. Usually, we use measurements to infer an unknown vector or function. What if the unknown is a probability measure? Seeking a measure that generates data consistent with given observations leads to an optimization problem over probability space. However, the complex geometric nature of this space prevents the direct use of standard arguments and solvers. We unravel some of the surprises that emerge in this setting and discuss potential solutions.