The Global Convergence of a New Mixed Conjugate Gradient Method for Unconstrained Optimization

Table 3

Numerical results of the NEW4 and NEW5.

Number	Prob.	dim	𝑘/nf/ng/𝑡(NEW4); ‖𝑔𝑘‖^2 (NEW4)	𝑘/nf/ng/𝑡(NEW5); ‖𝑔𝑘‖^2 (NEW5)
1	Powell badly scaled	2	12/25/13/0.0010	9/21/10/0.0010
			0.3511	0.3709
2	Brown badly scaled	2	14/36/15/0.0010	9/26/10/0.0313
			7.6024𝑒−005	1.1840
3	Trigonometric function	10	11/45/27/0.0010	13/51/33/0.0020
			0.0026	0.0017
4	Chebyquad		44/58/47/0.3281	72/82/73/0.3750
			0.0430	0.0251
5	Penalty function 𝐼	100	5/20/12/0.0313	5/20/12/0.0313
			5.4750𝑒−005	5.4750𝑒−005
		500	9/27/16/0.0156	9/27/16/0.0313
			7.0807𝑒−004	7.4979𝑒−004
6	Allgower	500	4/32/8/0.0010	4/32/8/0.0020
			0.0026	0.0017
7	Variable dimension	500	17/53/19/0.0313	19/53/19/0.0313
			3.9292𝑒−008	2.8061𝑒−008
		1000	16/54/16/0.0313	16/54/16/0.0010
			1.1473𝑒−004	1.1473𝑒−004
5	Penalty function 𝐼	1000	11/31/13/0.0313	11/31/13/0.0313
			0.0041	0.0036
8	Integral equation	1000	2/24/2/0.0020	2/24/2/0.0625
			0.0041	0.0936
9	Separable cubic	1000	15/17/16/0.4531	10/12/11/0.2969
			3.8684𝑒−004	7.7732𝑒−004
		3000	16/20/20/4.3750	14/16/15/3.3438
			1.3230𝑒−005	7.4737𝑒−006