Link:https://cp-algorithms.com/graph/mst_prim.html#sparse-graphs-om-log-n

Edge in all the other places across document contains the weight and the already-selected node to which it is directed towards. For example: Thei element invector<Edge> min_e(N) implies that the smallest cost with which it can be connected to an already-selected node is 'w' and that already-selected node is 'to'.

However, in the implementation for sparse graphs, theset<Edge> contains all edges which connect selected to not-selected nodes, where 'to' now contains a node which is not already selected.

Implying at one place theto inEdge refers to selected nodes, whereas in other places it refers to non-selected nodes. While the algorithm still works best, but I think this bit can be clarified.