search

SciML / NonlinearSolve.jl

High-performance and differentiation-enabled nonlinear solvers (Newton methods), bracketed rootfinding (bisection, Falsi), with sparsity and Newton-Krylov support.
https://docs.sciml.ai/NonlinearSolve/stable/
MIT License
221 stars 39 forks source link

Normal Form Linear Solve in Trust Region Methods #312

Closed avik-pal closed 6 months ago

avik-pal commented 7 months ago

Currently we cannot use Normal Form equations in TR. This can be added similar to how Gauss Newton and Levenberg work. Allows us to use Iterative Solvers like GMRES.

In the long run, we want to support LSLQ and such solvers but that requires a minor rewrite of JacVec and VecJac to provide the correct adjoint operators