Optimization plays an important role in the solution of ill-posed inverse problems. Optimization formulations of said problems are often nonlinear, nonconvex, and high-dimensional. Moreover, popular regularization techniques for enforcing the well-posedness of ill-posed inverse problems introduce a further challenge: nonsmoothness. This minisymposium aims to facilitate the exchange of ideas and varied viewpoints on optimisation challenges in the solution of inverse problems.
Mini symposium organizers:
Neil Dizon (University of Helsinky)
Tuomo Valkonen (University of Helsinki & Escuela Politécnica Nacional)

Session 4. Room A5, Thursday 11:30-13:00.
Chair: Neil Dizon (University of Helsinky)
Speakers:
Marcello Carioni (University of Twente) Exact sparse representation recovery in convex optimization
Raphael Kuess (Humboldt-Universität zu Berlin) Modeling, analysis and solution of parameter identification problems in thermo-piezoelectricity
Rossen Nenov (Acoustics Research Institute) Accelerated Griffin-Lim: A fast and provably converging method for phase retrieval

Session 5. Room A5, Thursday 17:30-19:30.
Chair: Tuomo Valkonen (University of Helsinki & Escuela Politécnica Nacional)
Speakers:
Neil Dizon (University of Helsinky) Online optimization for electrical impedance tomography
Felix Schneppe (Technische Universität Braunschweig) The impact of adjoint mismatches in the primal-dual Douglas-Rachford method – Existence of stationary points and convergence.
Kostas Papafitsoros (Queen Mary University of London) Learning data-driven priors for image reconstruction via neural network-based unrolled algorithmic schemes
Tuomo Valkonen (University of Helsinki & Escuela Politécnica Nacional) Interweaved first-order methods for PDE-constrained and bilevel optimisation