Merge pull request #13 from Robbybp/import

More conventional importing of dependencies

Author: Robbybp

src/dulmage_mendelsohn.jl

@@ -17,7 +17,7 @@
 #  This software is distributed under the 3-clause BSD license.
 #  ___________________________________________________________________________
 
-import Graphs as gjl
+import Graphs
 
 import JuMPIn: maximum_matching, _is_valid_bipartition
 
@@ -27,12 +27,12 @@ In this function, matching must contain a key for every matched node.
 I.e., Set(keys(matching)) == Set(values(matching))
 """
 function _get_projected_digraph(
-    graph::gjl.Graph, nodes::Vector, matching::Dict
+    graph::Graphs.Graph, nodes::Vector, matching::Dict
 )
     # Note that we are constructing a graph in a projected space, and must
     # be aware of whether coordinates are in original or projected spaces.
     orig_proj_map = Dict(n => i for (i, n) in enumerate(nodes))
-    n_nodes = gjl.nv(graph)
+    n_nodes = Graphs.nv(graph)
     n_nodes_proj = length(nodes)
     matched_nodes = keys(matching)
     node_set = Set(nodes) # Set of nodes in the original space
@@ -41,7 +41,7 @@ function _get_projected_digraph(
         orig_node = nodes[proj_node]
         if orig_node in matched_nodes
             # In-edges from all (other) neighbors of matched node
-            for nbr in gjl.neighbors(graph, matching[orig_node])
+            for nbr in Graphs.neighbors(graph, matching[orig_node])
                 if nbr != orig_node
                     nbr_proj = orig_proj_map[nbr]
                     push!(edge_set, (nbr_proj, proj_node))
@@ -50,29 +50,29 @@ function _get_projected_digraph(
         end
         # TODO: Out edges?
     end
-    digraph = gjl.DiGraph(n_nodes_proj)
+    digraph = Graphs.DiGraph(n_nodes_proj)
     for (n1, n2) in edge_set
-        gjl.add_edge!(digraph, n1, n2)
+        Graphs.add_edge!(digraph, n1, n2)
     end
     return digraph, orig_proj_map
 end
 
 
-function _get_reachable_from(digraph::gjl.DiGraph, nodes::Vector)
-    n_nodes = gjl.nv(digraph)
+function _get_reachable_from(digraph::Graphs.DiGraph, nodes::Vector)
+    n_nodes = Graphs.nv(digraph)
     source_set = Set(nodes)
-    gjl.add_vertex!(digraph)
+    Graphs.add_vertex!(digraph)
     # Note that root needs to be in this scope so it can be accessed in
     # finally block
     root = n_nodes + 1
     bfs_parents = Vector{Int64}()
     try
         for node in nodes
-            gjl.add_edge!(digraph, root, node)
+            Graphs.add_edge!(digraph, root, node)
         end
-        bfs_parents = gjl.bfs_parents(digraph, root)
+        bfs_parents = Graphs.bfs_parents(digraph, root)
     finally
-        gjl.rem_vertex!(digraph, root)
+        Graphs.rem_vertex!(digraph, root)
     end
     reachable = [
         node for (node, par) in enumerate(bfs_parents)
@@ -82,11 +82,11 @@ function _get_reachable_from(digraph::gjl.DiGraph, nodes::Vector)
 end
 
 
-function dulmage_mendelsohn(graph::gjl.Graph, set1::Set)
+function dulmage_mendelsohn(graph::Graphs.Graph, set1::Set)
     if !_is_valid_bipartition(graph, set1)
         throw(Exception)
     end
-    n_nodes = gjl.nv(graph)
+    n_nodes = Graphs.nv(graph)
     set2 = setdiff(Set(1:n_nodes), set1)
 
     # Compute maximum matching between bipartite sets

src/interface.jl

@@ -26,9 +26,7 @@ import JuMP
 
 import JuMPIn: get_bipartite_incidence_graph, maximum_matching, GraphDataTuple
 
-import Graphs as gjl
-import BipartiteMatching as bpm
-
+import Graphs
 
 """
 Utility function to convert a tuple of nodes and edges into a
@@ -39,17 +37,16 @@ function _tuple_to_graphs_jl(bip_graph)
     # Assumption here is that A and B are disjoint. And also form
     # a partition of 1:nv.
     nv = length(A) + length(B)
-    graph = gjl.Graph(gjl.Edge.(E))
+    graph = Graphs.Graph(Graphs.Edge.(E))
     # If E does not cover all vertices, some vertices may not appear in the
     # graph. Add these missing vertices.
-    n_missing = nv - gjl.nv(graph)
-    gjl.add_vertices!(graph, n_missing)
+    n_missing = nv - Graphs.nv(graph)
+    Graphs.add_vertices!(graph, n_missing)
     # Note that by constructing this graph, we have lost our particular
     # bipartition.
     return graph
 end
 
-
 function _maps_to_nodes(con_map, var_map)
     n_nodes = length(con_map) + length(var_map)
     nodes = Vector{Any}([nothing for _ in 1:n_nodes])
@@ -65,7 +62,6 @@ function _maps_to_nodes(con_map, var_map)
     return nodes
 end
 
-
 """
     IncidenceGraphInterface(model; include_inequality = false)
 
@@ -99,7 +95,6 @@ struct IncidenceGraphInterface
     _nodes
 end
 
-
 IncidenceGraphInterface(
     args::GraphDataTuple
 ) = IncidenceGraphInterface(
@@ -109,23 +104,20 @@ IncidenceGraphInterface(
     _maps_to_nodes(args[2], args[3]),
 )
 
-
 IncidenceGraphInterface(
     m::JuMP.Model;
     include_inequality::Bool = false,
 ) = IncidenceGraphInterface(
     get_bipartite_incidence_graph(m, include_inequality = include_inequality)
 )
 
-
 IncidenceGraphInterface(
     constraints::Vector{<:JuMP.ConstraintRef},
     variables::Vector{JuMP.VariableRef},
 ) = IncidenceGraphInterface(
     get_bipartite_incidence_graph(constraints, variables)
 )
 
-
 """
     get_adjacent(
         igraph::IncidenceGraphInterface,
@@ -140,12 +132,11 @@ function get_adjacent(
     constraint::JuMP.ConstraintRef,
 )::Vector{JuMP.VariableRef}
     con_node = igraph._con_node_map[constraint]
-    var_nodes = gjl.neighbors(igraph._graph, con_node)
+    var_nodes = Graphs.neighbors(igraph._graph, con_node)
     variables = [igraph._nodes[n] for n in var_nodes]
     return variables
 end
 
-
 """
     get_adjacent(
         igraph::IncidenceGraphInterface,
@@ -176,12 +167,11 @@ function get_adjacent(
     variable::JuMP.VariableRef,
 )::Vector{JuMP.ConstraintRef}
     var_node = igraph._var_node_map[variable]
-    con_nodes = gjl.neighbors(igraph._graph, var_node)
+    con_nodes = Graphs.neighbors(igraph._graph, var_node)
     constraints = [igraph._nodes[n] for n in con_nodes]
     return constraints
 end
 
-
 """
     maximum_matching(igraph::IncidenceGraphInterface)::Dict
 
@@ -351,7 +341,6 @@ function dulmage_mendelsohn(
     return con_dmp, var_dmp
 end
 
-
 function dulmage_mendelsohn(
     constraints::Vector{<:JuMP.ConstraintRef},
     variables::Vector{JuMP.VariableRef},

src/maximum_matching.jl

@@ -17,46 +17,46 @@
 #  This software is distributed under the 3-clause BSD license.
 #  ___________________________________________________________________________
 
-import Graphs as gjl
-import BipartiteMatching as bpm
+import Graphs
+import BipartiteMatching as BM
 
 
 """
 TODO: This should probably be promoted to Graphs.jl
 """
-function _is_valid_bipartition(graph::gjl.Graph, set1::Set)
-    n_nodes = gjl.nv(graph)
+function _is_valid_bipartition(graph::Graphs.Graph, set1::Set)
+    n_nodes = Graphs.nv(graph)
     all_nodes = Set(1:n_nodes)
     if !issubset(set1, all_nodes)
         throw(Exception)
     end
     set2 = setdiff(all_nodes, set1)
     for node in set1
-        if !issubset(gjl.neighbors(graph, node), set2)
+        if !issubset(Graphs.neighbors(graph, node), set2)
             return false
         end
     end
     for node in set2
-        if !issubset(gjl.neighbors(graph, node), set1)
+        if !issubset(Graphs.neighbors(graph, node), set1)
             return false
         end
     end
     return true
 end
 
 
-function maximum_matching(graph::gjl.Graph, set1::Set)
+function maximum_matching(graph::Graphs.Graph, set1::Set)
     if !_is_valid_bipartition(graph, set1)
         throw(Exception)
     end
-    n_nodes = gjl.nv(graph)
+    n_nodes = Graphs.nv(graph)
     card1 = length(set1)
     nodes1 = sort([node for node in set1])
     set2 = setdiff(Set(1:n_nodes), set1)
     nodes2 = sort([node for node in set2])
-    edge_set = Set((n1, n2) for n1 in nodes1 for n2 in gjl.neighbors(graph, n1))
+    edge_set = Set((n1, n2) for n1 in nodes1 for n2 in Graphs.neighbors(graph, n1))
     amat = BitArray{2}((r, c) in edge_set for r in nodes1, c in nodes2)
-    matching, _ = bpm.findmaxcardinalitybipartitematching(amat)
+    matching, _ = BM.findmaxcardinalitybipartitematching(amat)
     # Translate row/column coordinates back into nodes of the graph
     graph_matching = Dict(nodes1[r] => nodes2[c] for (r, c) in matching)
     return graph_matching