Fig. 4.

Estimation of parameters by minimization of data assimilation error on the time interval [0, T] for various T. Each red dot is the value of the sum of squares assimilation error for randomly chosen parameters (β′, α′), while the exactis assumed known. The blue dot represents the calculated minimum. For T significantly below 80, the loss function has no well-defined minimum, and the generating parameters (β = 1.1, α = 0.2) are poorly estimated. For larger T, the minimum becomes more pronounced, and the parameters can be well estimated. (Color available online.)