When
Where
Speaker: Abel Palafox Gonzalez, Universidad de Guadalajara
Title: Can numerical representation guide uncertainty? Exploring their impact on geophysical inverse problems
Abstract: The content of this talk touches on numerical representation, inverse problems, geophysics, uncertainty quantification... and a bunch of other interesting stuff. Inverse problems arise in a wide range of practical applications. Typically, inverse problems are ill-posed, making it difficult to obtain precise solutions. On the other hand, Uncertainty Quantification (UQ) is an interdisciplinary field that focuses on understanding how unknown information in the inputs of a system propagates through models and impacts the outputs or predictions. In the context of inverse problems, UQ helps to characterize the sources of uncertainty in observational data and how this uncertainty propagates onto the solutions of the inverse problems. On the other, other hand, in geophysical exploration, the goal is to infer subsurface information from surface measurements of physical properties—these are inverse problems. This talk will provide a brief overview of inverse problems arising in geophysics and their formulation as Bayesian inference problems, a pathway for quantifying and managing uncertainty. In this context, we will discuss how decisions regarding the parameterization of unknowns can enhance the accuracy and reliability of solutions. The talk will also illustrate the role of numerical representations through several examples.