implement augmented lagragian method

accosmin-org / libnano

C++ numerical optimization and machine learning utilities using Eigen3

MIT License

4 stars 1 forks source link

Closed accosmin closed 2 years ago

accosmin commented 2 years ago

reason: no need to increase the penalty too much, faster than exterior penalty methods
see: "Structured minimal-memory inexact quasi-Newton method and secant preconditioners for augmented lagrangian optimization", Birgin & Martinez, 2006
see: "Practical augmented lagrangiam methods", Birgin & Martinez, 2007
see: ch 17 in Nocedal & Wright's book