The project consists of implementing in Scipy an interior-point method for nonlinear problems. The main goal is to implement a constrained optimization algorithm able to deal with a large (and possibly sparse) problems for which the constrained optimization methods currently implemented in Scipy (namely SLSQP and COBYLA) are largely inappro- priate to deal with. Implement benchmark problems and integrate quasi-Newton (namely BFGS, SR1 and L-BFGS) and finite differences (using graph coloring schemes to deal with sparse structures) approximations to the method are also part of the project.


Antônio Horta Ribeiro


  • Ralf Gommers
  • Nikolay Mayorov
  • Matt Haberland