A New Algorithm for Solving Large-Scale Generalized Eigenvalue Problem Based on Projection Methods

Table 1

The minimum relative residual of Algorithms 1–4 in Example 1.

N.A	m = 16,N = 32		M = 18, N = 32 m = 16		M = 32, m = 4, l = 4		M = 4, L = 4, N = 32
					Threshold =
	SS	NE	CIRR	NE	BSS	NE	BSSRR	NE
n = 200	0.82114	16	8.8577E − 008	16	1.9902E − 008	16	7.9051E − 016	16
n = 400	0.7919	16	3.2359E − 06	16	5.1353E − 09	16	9.8608E − 16	16
n = 600	0.8393	16	3.2571E − 05	16	1.1753E − 07	16	1.3489E − 15	16
n = 800	0.8665	16	8.8640E − 05	16	2.6448E − 05	16	1.6268E − 15	16
n = 1000	0.8508	16	7.7542EE − -05	16	2.7825E − 05	16	2.2567E − 15	16
n = 1500	0.8671	16	5.6523E − 05	16	8.7373E − 005	16	1.9151E − 15	16