math.CA — Classical Analysis and ODEs
Representation of Degrees of Convergence by Nonadditive Measures
We characterize the functions that arise as degrees of convergence of sequences in metric spaces with respect to normalized capacities on . More precisely, answering a question of Mi\v{s}\'{\i}k and T{\'o}th in [Fuzzy Sets and Systems \textbf{375} (2019), 179--190], we show that these functions are precisely the -valued upper semicontinuous functions whose positive supports are separable. We obtain further characterizations when the representing capacities are required to be maxitive, subadditive, submodular, superadditive, supermodular, or countably subadditive, as well as for finitely and countably additive probability measures. Remarkably, representation by a normalized submeasure is equivalent to representation by the upper asymptotic density . As a byproduct, if is an uncountable compact metric space, then a map is upper semicontinuous if and only if there exists a sequence in such that $$ f(x)=\inf_{\varepsilon> 0} \mathsf{d}^\star (\{n \in \mathbb{N}: d(x_n,x)<\varepsilon\}) \quad \text{ for all }x \in X. $$ We also characterize the pairs of upper and lower degrees that can be represented simultaneously by a single sequence, with a sharp distinction between countable compact metric spaces and the remaining metric spaces. Lastly, we characterize the existence of universal representing sequences, and investigate the comeagerness of the family of maldistributed sequences.