Skip to content

I have created a Gradient Projection Method using Steepest Descent to numerically optimize minimization problems. I also did an investigation into the Damped Spectral Square Root Preconditioning Method, specifically how adding controlled levels of error can actually optimize the search algorithm.

Notifications You must be signed in to change notification settings

kpowers2000/NumericalOptimization1

Repository files navigation

Constrained Least Squares

I have created a Gradient Projection Method using Steepest Descent to numerically optimize minimization problems. Demonstrate this Methods effectiveness with solving Constrained Least Squares, a subclass of the minimization of residuals problem.

I also did an investigation into the Damped Spectral Square Root Preconditioning Method.

Please see the associated research paper, Constrained Least Squares Optimizations with Damped Preconditioning Methods for detailed information.

This paper also shows that controlled levels of error added to the Damped Spectral Square Root reduces the convergence rate of the Gradient Projection Method by a statistically significant factor.

About

I have created a Gradient Projection Method using Steepest Descent to numerically optimize minimization problems. I also did an investigation into the Damped Spectral Square Root Preconditioning Method, specifically how adding controlled levels of error can actually optimize the search algorithm.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages