Selection and Penalty Strategies for Genetic Algorithms Designed to Solve Spatial Forest Planning Problems

Table 3

Best solutions for landscapes A, B, and C. We present the best solution in terms of percent of trivial upper bound (TUB), for each selection/penalty combination. The secondary column heading represents the maximum number of generations. PROP: Proportional selection; BIN DT: Binary, deterministic tournament; 7 DT: A deterministic tournament with 7 competitors; U(2, 10) DT: A deterministic tournament with the number of competitors sampled from a uniform distribution on the range (2, 10); 2–7 DT: A deterministic tournament where the number of competitors increases from 2 to 7 throughout the course of the search.

Penalty	Selection	Landscape A		Landscape B		Landscape C
		750	1500	1500	3000	1500	6000
Static	PROP	88.04	92.39	78.66	87.64	68.90	86.74
	BIN DT	87.72	91.83	82.52	88.03	72.97	83.86
	7 DT	86.94	93.06	84.24	89.85	73.43	84.10
	U(2, 10) DT	87.51	93.07	83.66	89.36	73.70	83.82
	2–7 DT	87.67	92.35	83.03	88.71	72.64	83.94
Dynamic	PROP	89.70	94.82	79.84	90.42	70.11	89.65
	BIN DT	85.44	91.89	82.51	89.66	71.97	85.92
	7 DT	87.77	92.73	85.03	90.91	72.69	84.56
	U(2, 10) DT	87.90	91.80	84.10	90.27	72.75	84.33
	2–7 DT	87.03	91.85	85.35	90.50	74.00	84.34