##

Solving Large Convex Minimax Problems using CUDA

This work was presented at SEAL '10. [pdf].

We consider unconstrained minimax problem where the objective function is the
maximum of a finite number of smooth convex functions. We present an iterative
method to compute the optimal solution for the unconstrained convex finite
minimax problem. The algorithm developed estimates the direction of
steepest-descent rapidly and using Armijo’s condition proceeds towards the
solution. Owing to the highly parallel nature of the algorithm, it is highly
suitable for large minimax problems. Algorithm is implemented on Tesla C1060
graphics card using CUDA and numerical comparisons with RGA & CFSQP are
presented.

Back to Home

musically_ut
2014-04-03