Coverings and fundamental groups: a new approach

Abstract: The traditional approach of defining the fundamental group first and then constructing universal coverings works well only for the class of Poincar\' e spaces. For general spaces there were several attempts to define generalized coverings (see \cite{BP3}, \cite{BDLM}, and \cite{FisZas}), yet there is no general theory so far that covers all path connected spaces. In this paper we plan to remedy that by changing the order of things: we define the universal covering first and its group of deck transformations is the new fundamental group of the base space. The basic idea is that a non-trivial loop ought to be detected by a covering (not by extension over the unit disk): a loop is non-trivial if there is a covering such that some lift of the loop is a non-loop. So it remains to define coverings: the most natural class is the class of maps that have unique disk lifting property. To make the theory work one needs to add the assumption that path components of pre-images of open sets form a basis of the total space.