For what it's worth, statistical machine learning and other CS people have recently "discovered" classical convex analysis - for the last few years papers have

For what it's worth, statistical machine learning and other CS people have recently "discovered" classical convex analysis - for the last few years papers have been full of stochastic subgradients and the like. The next big thing has got to be the material on decompositions from the 50s and 60s.