Compact algorithms for function minimization

This paper presents several function minimization algorithms selected and modified to have short programs and low working-storage requirements. The goal is to provide useful optimization methods for small computers. Such methods would be useful in any situation where storage was at a premium, such as on a parallel computer where each processor had a small local memory, or when large problems were being solved. Comparisons are made with more traditional methods to determine the extent of the sacrifices made in selecting these compact algorithms. Both program and problem costs are examined. Reports of numerical tests are included.