An Inexact Penalty Decomposition Method for Sparse Optimization

Table 2

Averaged results on 100 instances with size for each sparsity level

Algorithm	s	Time (s)	NNZ	MSE	DF	NS
PD	50	18.2	50			100
	100	21.69	99.95			95
	150	26.36	149.88			89
	200	76.38	297.65			73
	250	102.54	1225.21			0
	270	105.18	1283.92			0
iPD	50	6.72	50			100
	100	8.56	99.99			99
	150	9.78	150			100
	200	10.32	200			100
	250	11.75	656.87			30
	270	12.75	1235.9			1