-
Notifications
You must be signed in to change notification settings - Fork 12
0. Commented References
Main paper describing the method.
[2] Nocedal, Jorge, and Stephen J. Wright. "Numerical optimization" Second Edition (2006).
Excellent book on optimization that contains most of the theory needed to understand what is being implemented.
Preliminary paper describing the method and its framework. Contains a global analysis of the algorithm.
Describe the local behaviour of the algorithm and gives sufficient conditions to obtain superlinear convergence.
Paper describing techniques to cope with roundoff errors that results from solving equality-constrained programming (EQP) problems using the projected CG method. The projected CG method is an important substep of the method being implemented and this corrections are important to the behaviour of the interior-point method as a hole. I have written a blog post about this (link).
Paper describing the Knitro package that, among other methods, implements the method we intend to implement in Python, internally referred as KNITRO/CG solver.
Paper describing improvements in order to cope with infeasibility.
This paper describe another interior point implementation from Nocedal. This implementation is the one used in KNITRO/DIRECT and also in Matlab interior point method (link). This implementation employs line search method, except when the KKT matrix is singular, and in this case it switches to a trust-region step (similar to the one described on previous papers).
This paper describe a trust-region interior point that have lots of similarities with [1]. I believe this paper had a huge influence in [1].
Some convex QP problems I used for test.
Set of nonlinearly constrained optimization problems.
Preliminary paper describing Byrd-Omojokun SQP method.
Large optimization problem collection.