BEGIN:VCALENDAR VERSION:2.0 PRODID:-//Talks.cam//talks.cam.ac.uk// X-WR-CALNAME:Talks.cam BEGIN:VEVENT SUMMARY:hIPPYlib-MUQ: A Bayesian Inference Software Framework for Integrat ion of Data with Complex Predictive Models under Uncertainty - Noemi Petra (University of California\, Merced) DTSTART:20230425T133000Z DTEND:20230425T143000Z UID:TALK198418@talks.cam.ac.uk DESCRIPTION:Bayesian inference provides a systematic framework for integra tion of data with mathematical models to quantify the uncertainty in the s olution of the inverse problem. However\, the solution of Bayesian inverse problems governed by complex forward models described by partial differen tial equations (PDEs) remains prohibitive with black-box Markov chain Mont e Carlo (MCMC) methods. We present hIPPYlib-MUQ\, an extensible and scalab le software framework that contains implementations of state-of-the art al gorithms aimed to overcome the challenges of high-dimensional\, PDE-constr ained Bayesian inverse problems. These algorithms accelerate MCMC sampling by exploiting the geometry and intrinsic low-dimensionality of parameter space via derivative information and low rank approximation. The software integrates two complementary open-source software packages\, hIPPYlib and MUQ. hIPPYlib solves PDE-constrained inverse problems using automatically- generated adjoint-based derivatives\, but it lacks full Bayesian capabilit ies. MUQ provides a spectrum of powerful Bayesian inversion models and alg orithms\, but expects forward models to come equipped with gradients and H essians to permit large-scale solution. By combining these two complementa ry libraries\, we created a robust\, scalable\, and efficient software fra mework that realizes the benefits of each and allows us to tackle complex large-scale Bayesian inverse problems across a broad spectrum of scientifi c and engineering disciplines. To illustrate the capabilities of hIPPYlib- MUQ\, we present a comparison of a number of MCMC methods available in the integrated software on several high-dimensional Bayesian inverse problems . These include problems characterized by both linear and nonlinear PDEs\, low and high levels of data noise\, and different parameter dimensions. T he results demonstrate that large (~ 50 x) speedups over conventional blac k box and gradient-based MCMC algorithms can be obtained by exploiting Hes sian information (from the log-posterior)\, underscoring the power of the integrated hIPPYlib-MUQ framework.\nThis work is joint work with: Ki-Tae K im\, Umberto Villa\, Matthew Parno\, Youssef Marzouk and Omar Ghattas\nRea ding material: This work is to appear in Transaction of Mathematical Softw are (TOMS) and the manuscript can be found at:  \;https://arxiv.org/ab s/2112.00713 LOCATION:Seminar Room 1\, Newton Institute END:VEVENT END:VCALENDAR