The Network-Enabled Optimization System (NEOS) - a means of solving optimization problems over the Internet

Many optimization methods are available at the present time. The software that implements a particular method may not be available to all users or the software may require a compiler not readily available to all users. The software may require extensive modifications before it can run on a particular site. The Network--Enabled Optimization System (NEOS) is a system that has been designed to reduce these problems and make high quality optimization methods available to a large number of users. One of the goals of NEOS is to eliminate the need for the user to have extensive programming knowledge. Another goal is to make a large number of computer facilities available to the user. This goal is achieved by permitting optimization problems to be submitted to NEOS over the Internet. NEOS then directs one or more computers to solve the problem. NEOS also uses automatic differentiation. This relieves the user from coding expressions for the derivatives or from risking the approximation error inherent in numerical differentiation. A major accomplishment of the NEOS project is to take advantage of function partial separability when possible. Previously this had been considered quite difficult because of the difficulty in computing derivative information. NEOS use of automatic differentiation has eliminated this problem. This research presents an overview of NEOS and shows how it can be used to optimize econometric functions and other functions.