Final solution to the problem of relating a true copula to an imprecise copula

Abstract

In this paper we solve in the negative the problem proposed in this journal (I. Montes et al., Sklar's theorem in an imprecise setting, Fuzzy Sets and Systems, 278 (2015), 48-66) whether an order interval defined by an imprecise copula contains a copula. Namely, if $\mathcal{C}$ is a nonempty set of copulas, then $\underline{C}=\inf \{C{\}}_{C\in \mathcal{C}}$ and $\overline{C}=\sup \{C{\}}_{C\in \mathcal{C}}$ are quasi-copulas and the pair $(\underline{C},\overline{C})$ is an imprecise copula according to the definition introduced in the cited paper, following the ideas of $p$-boxes. We show that there is an imprecise copula $(A,B)$ in this sense such that there is no copula $C$ whatsoever satisfying $A⩽C⩽B$. So, it is questionable whether the proposed definition of the imprecise copula is in accordance with the intentions of the initiators. Our methods may be of independent interest: We upgrade the ideas of Dibala et al. (Defects and transformations of quasi-copulas, Kybernetika, 52 (2016), 848-865) where possibly negative volumes of quasi-copulas as defects from being copulas were studied.

20 pages; added Conclusion, added some clarifications in proofs, added some explanations at the beginning of each section, corrected typos, results remain the same

Cited by in corpus (11)

• Some multivariate imprecise shock model copulas
• Exact upper bound for copulas with a given diagonal section
• Quasi-copulas as linear combinations of copulas
• Normal cones corresponding to credal sets of lower probabilities
• Coherence and avoidance of sure loss for standardized functions and semicopulas
• Constructing copulas from shock models with imprecise distributions
• Extending multivariate sub-quasi-copulas
• A complete characterization of normal cones and extreme points for $p$-boxes
• Relation between Blomqvist's beta and other measures of concordance of copulas
• A full scale Sklar's theorem in the imprecise setting
• Extreme mass distributions for quasi-copulas