The scale parameter for the outlier noise is currently fixed and assumes that $\Delta f$ values are (or become) considerably smaller than one. This setting does not work well for multi-objective problems, because for these $\Delta f\not\to 0$ for $x\to x^*$.
The scale parameter for the outlier noise is currently fixed and assumes that$\Delta f$ values are (or become) considerably smaller than one. This setting does not work well for multi-objective problems, because for these $\Delta f\not\to 0$ for $x\to x^*$ .