Comments on: Multivariate functional outlier detection

First of all, we would like to congratulate M. Hubert, P. Rousseeuw and P. Segaert for this very interesting and stimulating work. It is clear that functional data are becoming ubiquitous in many disciplines and the development of appropriate statistical techniques is clearly needed. Moreover, outliers are very likely to occur in this type of data, where many measurements are taken by applying mostly unsupervised procedures. The authors provide several tools that can be successfully applied for detecting outliers when dealing with (even multivariate) functional data. They are very intuitive graphical tools based on suitable depth notions for functional data. We consider that these graphical tools are clearly useful specially in the multivariate setting where it is virtually impossible to visualize directly data curves in order to detect anomalous patterns. In our comment, we will focus on explaining how trimming principles can be also taken into account in the detection of functional outliers.