Comparison of fitting in L-domain and t-domain of simulated noisy exponentials defined by parameters and , superimposed Poisson noise and stationary gaussian noise with different standard deviations .

is the probability that the fitting parameters obtained in the L-domain approximate the true values better than those obtained in the t-domain (i.e., ) under the indicated noise condition. With optimal importance-weighting for Poisson noise (), the comparison is carried out for four different levels of gaussian noise, given as a ().