This article reviews the range of delivery platforms that have been developed for the PySAL open source Python library for spatial analysis. This includes traditional desktop software (with a graphical user interface, command line or embedded in a computational notebook), open spatial analytics middleware, and web, cloud and distributed open geospatial analytics for decision support. A common thread throughout the discussion is the emphasis on openness, interoperability, and provenance management in a scientific workflow. The code base of the PySAL library provides the common computing framework underlying all delivery mechanisms.
Adjacency and neighbor structures play an essential role in many spatial analytical tasks. The computation of adjacenecy structures is non-trivial and can form a significant processing bottleneck as the total number of observations increases. We quantify the performance of synthetic and real world binary, first-order, adjacency algorithms and offer a solution that leverages Python's high performance containers. A comparison of this algorithm with a traditional spatial decomposition shows that the former outperforms the latter as a function of the geometric complexity, i.e the number of vertices and edges.