“…In some fields of functional data analysis, noisy random functions have already been studied intensively; a body of literature on this topic encompassing functional linear regression, functional principal components, continuous additive models, etc. can be found in Yao et al (2005), Crambes et al (2009), Paul and Peng (2009), Jiang and Wang (2010), Li and Hsing (2010), Wu et al (2010), Müller et al (2013), Radchenko et al (2015), Hsing and Eubank (2015) and Zhang and Wang (2016), among others. In those works, noisy functional data are typically either (i) pre-smoothed in the first step, and then statistical analysis is performed for the smoothed approximants of the data curves; or (ii) the statistical procedure in question is revisited, and suitably applied directly to the available discrete noisy observations.…”