Skip to main content

Table 4 Detailed results of SParSE and SParSE++ applied to SIRS disease dynamics on eight initial reaction rates (column 1) that exhibited slowest SParSE convergence

From: SParSE++: improved event-based stochastic parameter search

   SParSE SParSE++  
k 0 \(\hat {p}^{0}\) IE γ OP Interp Tot IE γ OP Leap Interp Tot Gain
[0.079 1.59 2.42] 0.80 20 17 17 1 55 4 3 3 3 3 16 39
[0.081 1.07 2.28] 0.87 18 16 16 0 50 4 3 3 5 0 15 35
[0.067 3.96 2.37] 0.00 17 6 0 5 38 12 11 0 2 0 25 13
[0.064 3.31 1.71] 0.00 13 12 0 5 30 10 9 0 3 0 22 8
[0.035 1.22 2.38] 0.64 11 8 8 2 29 5 4 4 2 0 15 14
[0.040 3.59 1.52] 0.00 13 12 0 3 28 11 10 0 1 0 22 6
[0.080 2.08 2.22] 0.74 11 8 8 1 28 3 2 2 1 0 8 20
[0.031 3.16 1.84] 0.00 11 10 0 6 27 9 8 0 1 0 18 9
  1. Initial probability estimate, number of intermediate event (IE) computations, number of biasing parameter computations, number of over-perturbation (OP) stages, and number of interpolation stages for SParSE simulations are given in columns 2-6, respectively. Column 7 contains the total number of SParSE iterations (row sum) for the initial rate in column 1. SParSE++ statistics are listed in columns 8 to 13, including the number of times the leaping method was used in column 11. The number of iterations saved by employing SParSE++ over SParSE is given in column 14 (difference between columns 7 and 13)