PerformanceEstimation / PEPit

PEPit is a package enabling computer-assisted worst-case analyses of first-order optimization methods.
https://pepit.readthedocs.io/en/latest/
MIT License
76 stars 10 forks source link

Add parameterized frugal resolvent splitting example #99

Open peterbarkley opened 1 month ago

peterbarkley commented 1 month ago

This adds an example which builds a parameterized frugal resolvent splitting for a monotone inclusion problem. The Malitsky-Tam n operator splitting, Ryu's 3 operator splitting, Douglas-Rachford, and a new block splitting algorithm can be written in this parameterized form. It also adds a PEPit use case comparing these algorithms with fixed step size and with optimized step sizes and optimized matrices built using the dual of the PEP as in Ryu, Taylor, Bergeling, et al (2020) Operator splitting performance estimation: Tight contraction factors and optimal parameter selection.