A Julia version of NUM4LCP.m (fischer_newton.m) in: CutAndDisplace
This code is:
- Memory efficient (allocations in main loop are only due to: Krylov solvers & view statements )
- Relativly easy to call/adapt
- Contains some basic regression tests (comparing directly to MATLAB implementation results).
Original code by Kenny Erleben
This is faster than the MATLAB version of FischerNewton in CutAndDisplace:
NameOfFile | Size of A | MATLAB (seconds) | Julia (seconds) | Relative speedup |
---|---|---|---|---|
FischerNewton\test\Matricies.mat | 1560*1560 | 1.2 | 0.8 | 1.5 |
FischerNewton\test\Matricies_1000.mat | 5225*5225 | 22.3 | 19.5 | 1.14 |
For build here: 24/7/2019 on Windows 64bit 4 core machine vs MATLAB 2018a.