Research Article

Global Convergence of a Modified Spectral Conjugate Gradient Method

Table 2

Comparison of efficiency with the other methods.

Function Algorithm Dim GV NINF CT(s)

Kowalik and Osborne mfr 4
prp 4 8 . 9 5 2 1 𝑒 0 0 7 833 26191 1.2970
mprp 4 9 . 9 6 9 8 𝑒 0 0 7 6235 35425 3.5940
msprp 4 9 . 9 5 6 0 𝑒 0 0 7 7059 37976 4.9850

Broyden banded mfr 6 8 . 9 4 6 9 𝑒 0 0 7 40 505 0.0780
prp 6 8 . 4 6 8 4 𝑒 0 0 7 268 9640 0.4840
mprp 6 8 . 9 0 2 9 𝑒 0 0 7 102 1319 0.0940
msprp 6 9 . 3 2 7 6 𝑒 0 0 7 44 556 0.0940

Discrete boundary mfr 6 9 . 1 5 3 1 𝑒 0 0 7 107 509 0.0780
prp 6 7 . 8 9 7 0 𝑒 0 0 7 269 11449 0.4690
mprp 6 8 . 2 8 0 7 9 𝑒 0 0 7 157 1473 0.0930
msprp 6 9 . 9 4 3 6 𝑒 0 0 7 165 1471 0.1410

Variably dimensioned mfr 8 7 . 3 4 1 1 𝑒 0 0 7 57 1233 0.1250
prp 8 7 . 3 4 1 1 𝑒 0 0 7 113 7403 0.3290
mprp 8 9 . 0 9 0 0 𝑒 0 0 7 69 1544 0.0780
msprp 8 7 . 3 4 1 1 𝑒 0 0 7 57 1233 0.1100

Broyden tridiagonal mfr 9 9 . 1 8 1 5 𝑒 0 0 7 129 2173 0.1250
prp 9 6 . 4 5 8 4 𝑒 0 0 7 113 5915 0.2500
mprp 9 7 . 3 5 2 9 𝑒 0 0 7 187 2967 0.1250
msprp 9 9 . 2 3 6 3 𝑒 0 0 7 82 1304 0.1100

Linear-rank1 mfr 10 9 . 7 4 6 2 𝑒 0 0 7 84 3762 0.1720
prp 10 4 . 5 6 4 7 𝑒 0 0 7 98 6765 0.2810
mprp 10 6 . 9 1 4 0 𝑒 0 0 7 51 2216 0.0780
msprp 10 6 . 6 6 3 0 𝑒 0 0 7 50 2162 0.1250

Linear-full rank mfr 12 7 . 6 9 1 9 𝑒 0 0 7 9 36 0.0160
prp 12 8 . 2 5 0 7 𝑒 0 0 7 47 1904 0.1090
mprp 12 7 . 6 9 1 9 𝑒 0 0 7 9 36 0.0630
msprp 12 7 . 6 9 1 9 𝑒 0 0 7 9 36 0.0150