SParSE++: improved event-based stochastic parameter search

BMC Systems Biology

Table 1 Results of SParSE and SParSE++ applied to the birth-death process

\(\mathcal {P}_{\mathcal {E}}\)	\(\epsilon _{\mathcal {P}_{\mathcal {E}}}\)	Tot. Iter.	Tot. Iter.	Time (s)	Time (s)	No. Gain	No. Loss
		SParSE	SParse++	SParSE	SParSE++
0.40	0.01	293	175	1139.0	684.9	119	1
	0.05	189	139	685.1	517.4	53	3
	0.10	139	121	476.2	457.6	21	3
0.60	0.01	332	171	1128.7	602.2	161	0
	0.05	175	141	539.5	500.1	38	4
	0.10	129	120	378.3	421.6	9	0
0.80	0.01	242	186	728.8	629.8	60	4
	0.05	139	137	449.9	480.8	3	1
	0.10	110	110	342.2	382.7	2	2
0.010	0.001	279	244	3943.5	3656.6	44	9

The first column denotes the target probability, the second column absolute error tolerance, the third column the total number of SParSE samples computed for the 30 initial parameter sets, the fourth column the total number of SParSE++ samples computed for the 30 initial parameter sets, the fifth the total number of iterations gained by running SParSE++ compared to SParSE, and the sixth the total number of iterations lost by running SParSE++ compared to SParSE. N _S=5×10⁴ for all configurations except \(\mathcal {P}_{\mathcal {E}} = 0.01\), where N _S=2×10⁵

ISSN: 1752-0509