PidIdentifier tends to under-estimate Kp and Ti by 20% on under-sampled datasets.

In cases where the true kp and ti are known, it is is sometimes seen that the PidIdentifier returns a value which is not the global minimum fitscore, but it tends to be slightly lower.(seen in closed-source test on real data)

This means that is should be possible to improve the accuracy of PidIdentifier by possibly playing with the sensitivity of the solver.