Table 4.

FVA results (barrier solver) for ME76664 before and after lifting

		Iterations		Infeasibilities
Flux	Before	After	Before	After
1	76669	23084	9.4×10−2	9.0×10−2
5001	34721	66731	4.7×100	8.8×10−2
10001	58672	85819	9.3×10−2	9.3×10−4
15001	28032	47901	9.7×10−2	1.6×10−2
20001	18715	30433	3.2×10−2	3.0×10−2
25001	12224	17973	3.5×102	8.8×10−2
30001	19621	35111	8.8×10−2	3.1×10−2
35001	6743	7630	1.0×10−2	6.8×10−2
40001	47609	4111	8.8×10−2	9.4×10−2
45001	9117	9980	1.0×10−1	3.2×10−2
50001	9567	49350	9.5×10−2	9.5×10−2
55001	23985	13362	9.6×10−2	1.1×10−1
60001	99067	27075	1.1×10−1	9.1×10−2
65001	44796	11509	3.1×10−1	8.9×10−2
70001	17045	14393	4.0×10−2	6.5×10−2
75001	20790	14908	9.0×10−2	5.6×10−6

FVA results for the E. coli Metabolic-Expression model ME76664 using the CPLEX barrier solver. Iterations and sum of infeasibilities before and after lifting. The first column lists which variable is being maximized. Columns 2 and 3 list the total iterations for the barrier solver and the simplex crossover (with barrier requiring only about 100 iterations in all cases). CPLEX barrier appears less reliable than the CPLEX simplex solvers on this model.