Hi, We recently added ADMM based proximal algorithm in breeze.optimize.proximal.NonlinearMinimizer which uses a combination of BFGS and proximal algorithms (soft thresholding for L1 for example) to solve large scale constrained optimization problem of form f(x) + g(z). Its usage is similar to current lbfgs flow. We also added a library of commonly used proximal algorithms in breeze.
I am experimenting with L1 but I remember there were various admm based PR before. Please feel free to use the solver for your usecases and let us know if it was helpful. The solver is available in mllib master as we already integrated breeze 0.11.2 to mllib. Thanks. Deb
