Numerical solution of huge-scale quasiseparable optimization problems

Аннотация

The paper studies approaches to numerical solving huge-scale quasiseparable optimization problems. The main idea is based on using gradient methods with simple iteration structure instead more intelligent techniques, which is widely used for solving traditional, small-sized problems. The results of numerical experiments for a number of test quasiseparable optimization problems with dimensions up to 10¹⁰ variables are presented.