Matthias Ehrhardt is a prize fellow at Bath university and I first met him when I visited Cambridge in 2017-2018. Back then I was just a first-year PhD student and it was hard for me to fully understand his work on stochastic primal-dual hybrid gradient (https://arxiv.org/abs/1706.04957). However, I have more experience now and just in two weeks I managed to do a nice extension of my recent work (https://arxiv.org/abs/1905.11535) to the same setting as in the work by Matthias. The work is motivated by imaging application and I hope it will find important applications in practice.
