A part of my dissertation was an agent-based model (ABM) on probabilistic displacement determined by economic indicators, shock size, and network topography. For this I had to code and incorporate
#Dijkstra algorithm in my
#Netlogo code (around 2009-10). Getting this to work was one of the most exciting things since it is at the heart of the model!
The algorithm itself is inefficient since it does a full graph search either based on distances (path lengths) or number of nodes crossed (hops). It is superceded by the A* algorithm which is a "greedy" heuristic crawl on a network that reduces the search time significantly.
BUT most of the graph network measures, such as betweenness and centrality, implemented in R, Python, C++ etc still use Dijkstra as the main algorithm. It is also the core algorithm for measures that involve distances along weighted links.