Figure 3.

Parameter error. Here is shown the ratio of the fitted parameters to the true parameters after each iteration of optimally chosen experiments according to the three goal functions. ‘Start’ is the scrambled parameter set before the nominal experiment is fitted. Each inset shows a magnified view of the final parameter set error. The black, green and red dashed lines are 10%, 20% and 30% error, respectively. By the final experiment, all parameter errors are less than 30% and most parameter errors are less than 10%. This suggests that the linear approximation of the uncertainties is an appropriate estimate of the underlying parameter error when the data are plentiful and the uncertainties are small.