In this menu, one makes the hypothesis that B is an additive white Gaussian noise. Then, we seek Z that minimizes the risk subject to the constraint that its regularity is that of Y plus a shift. One can show that this yields an asymptotically minimax estimator.
First specify your Analyzed signal, and the desired Hölder exponent shift. The algorithm needs to know the standard deviation of the noise. It is often a good idea to experiment with different values. To do this, check the Specify button, and enter a value of your choice for Standard Deviation. Alternatively, you may let the system estimate the power of the noise for you: Uncheck Specify so that Automatic appears instead. A new box called Estimated Standard Deviation pops up, in which the estimated value will be displayed when you hit Compute. The output signal is called mden_signal#. Typical values for the shift in this case are around 2.