Computation and performance analysis of existing three term conjugate gradient method

• Main Authors: Arizi, N.A.N.M (Author), Buchek, N.F (Author), Fadhilah, N.H (Author), Idalisa, N. (Author), Norizan, N.S (Author)
• Format: Article
• Language: English
• Published: World Academy of Research in Science and Engineering 2019
• Subjects: Conjugate gradient method; Scaling CG; Three term; Unconstrained optimization
• Online Access: View Fulltext in Publisher; View in Scopus

 The three-term Conjugate Gradient (CG) method is employed to solve unconstrained optimization problem (UOP) according to its efficiency and numerical performances. The three-term CG methods used were three-term Hestenes-Stiefel (TTHS) and three term new RMIL (3TNRMIL). The efficiency of the method is shown through comparison with all the three-term CG methods. The original three-term is compared with its modified scaling initial direction. The total test run involved in the numerical experiment was 800 that comprised of 5 test functions. The five test functions that were selected were Extended Himmelblau, Extended Rosenbrock, Extended White and Holst, Extended Beale and Shallow Functions with dimensions of 1200, 3000, 6000, and 9000. The performance profile is analysed based on number of iteration (NOI) and CPU time. Result shows that the modified TTHS performed better than the original TTHS and 3TNRMIL. © 2019, World Academy of Research in Science and Engineering. All rights reserved.