« Index of all namespaces of this project
Basic graph theory algorithms
(->DirectedGraph nodes neighbors)
Positional factory function for class datalog.graph.DirectedGraph.
(component-graph g sccs)
Given a graph, perhaps with cycles, return a reduced graph that is acyclic. Each node in the new graph will be a set of nodes from the old. These sets are the strongly connected components. Each edge will be the union of the corresponding edges of the prior graph.
Similar to a topological sort, this returns a vector of sets. The set of nodes at index 0 are independent. The set at index 1 depend on index 0; those at 2 depend on 0 and 1, and so on. Those withing a set have no mutual dependencies. Assume the input graph (which much be acyclic) has an edge a->b when a depends on b.
(fixed-point data fun max equal)
Repeatedly apply fun to data until (equal old-data new-data) returns true. If max iterations occur, it will throw an exception. Set max to nil for unlimited iterations.
(lazy-walk g n)
(lazy-walk g ns v)
Return a lazy sequence of the nodes of a graph starting a node n. Optionally, provide a set of visited notes (v) and a collection of nodes to visit (ns).
Factory function for class datalog.graph.DirectedGraph, taking a map of keywords to field values.
Return a sequence of indexes of a post-ordered walk of the graph.
(post-ordered-visit g n [visited acc :as state])
Starting at node n, perform a post-ordered walk.
(recursive-component? g ns)
Is the component (recieved from scc) self recursive?
Given a directed graph, return another directed graph with the order of the edges reversed.
Returns, as a sequence of sets, the components of a graph that are self-recursive.
(stratification-list g1 g2)
Similar to dependency-list (see doc), except two graphs are provided. The first is as dependency-list. The second (which may have cycles) provides a partial-dependency relation. If node a depends on node b (meaning an edge a->b exists) in the second graph, node a must be equal or later in the sequence.
Returns the transitive closure of a graph. The neighbors are lazily computed. Note: some version of this algorithm return all edges a->a regardless of whether such loops exist in the original graph. This version does not. Loops will be included only if produced by cycles in the graph. If you have code that depends on such behavior, call (-> g transitive-closure add-loops)