Accession Number:

AD0856001

Title:

Gradient Methods in Mathematical Programming Part 3. Supermemory Gradient Method,

Descriptive Note:

Corporate Author:

RICE UNIV HOUSTON TX AERO-ASTRONAUTICS GROUP

Personal Author(s):

• Cragg, E. E.
• Levy, A. V.

Report Date:

1969-01-01

Pagination or Media Count:

28.0

Abstract:

A generalization of the memory gradient algorithm, called the supermemory gradient algorithm, is investigated. While the memory gradient algorithm is based on remembering one previous iteration, the supermemory gradient algorithm is based on remembering k previous iterations. The added complication is the need for a k 1-dimensional search at each iteration as opposed to the two-dimensional search required by the memory gradient algorithm . Author

Descriptors:

• (*MATHEMATICAL PROGRAMMING
• ITERATIONS)
• COMPUTER PROGRAMMING
• PARTIAL DIFFERENTIAL EQUATIONS
• NUMERICAL ANALYSIS
• DIFFERENCE EQUATIONS
• MATRICES(MATHEMATICS)
• CONVERGENCE
• ALGORITHMS

Subject Categories:

• Operations Research
• Computer Programming and Software

Distribution Statement:

APPROVED FOR PUBLIC RELEASE