[MERGE] Multiprocessing max clique solver

alexgiving · Oct 1, 2023 · 5e68ace · 5e68ace
1 parent ac0aa74
commit 5e68ace
Show file tree

Hide file tree

Showing 3 changed files with 36 additions and 17 deletions.
diff --git a/README.md b/README.md
@@ -1,9 +1,9 @@
 # HeuroGraph Library
 
-![GitHub stars](https://img.shields.io/github/stars/alexgiving/HeuroGraph-Library?style=social)
+<!-- ![GitHub stars](https://img.shields.io/github/stars/alexgiving/HeuroGraph-Library?style=social)
 ![GitHub forks](https://img.shields.io/github/forks/alexgiving/HeuroGraph-Library?style=social)
 ![GitHub issues](https://img.shields.io/github/issues/alexgiving/HeuroGraph-Library)
-![GitHub contributors](https://img.shields.io/github/contributors/alexgiving/HeuroGraph-Library)
+![GitHub contributors](https://img.shields.io/github/contributors/alexgiving/HeuroGraph-Library) -->
 
 Welcome to the HeuroGraph-Library! This is a Python library designed for solving graph-related problems using heuristic algorithms. Whether you're working on graph coloring, maximum clique, or other graph optimization problems, this library provides a set of powerful heuristic algorithms to assist you in finding efficient solutions.
 

diff --git a/benchmark_coloring.py b/benchmark_coloring.py
@@ -8,8 +8,8 @@
 
 from heurograph import Graph
 from heurograph.coloring import (color_graph_greedy_randomized_sorted,
-                                color_graph_greedy_sorted,
-                                color_graph_greedy_sorted_shuffled)
+                                 color_graph_greedy_sorted,
+                                 color_graph_greedy_sorted_shuffled)
 
 
 class ResultColumns(Enum):

diff --git a/heurograph/clique/greedy_randomized_clique.py b/heurograph/clique/greedy_randomized_clique.py
@@ -1,4 +1,6 @@
 import random
+from functools import partial
+from multiprocessing import Pool, cpu_count
 
 import networkx as nx
 import numpy as np
@@ -15,27 +17,44 @@ def nx_clique(graph: Graph) -> list[int]:
     return best_clique
 
 
+def _get_clique(start_vertex: int, graph: Graph) -> list[int]:
+     clique = [start_vertex]
+     vertices_candidates = graph.neighbors(start_vertex)
+     np.random.shuffle(vertices_candidates)
+
+     for vertex in vertices_candidates:
+         is_connected = True
+         for clique_vertex in clique:
+             if clique_vertex not in graph.neighbors(vertex):
+                 is_connected = False
+                 break
+         if is_connected:
+             clique.append(vertex)
+     return clique
+
+
 def greedy_randomized_clique(graph: Graph, iterations: int = 10) -> list[int]:
     best_clique = []
     vertices = graph.vertexes
 
     for _ in range(iterations):
         clique = []
         random_vertex = random.choice(vertices)
+        clique = _get_clique(random_vertex, graph)
+        if len(clique) > len(best_clique):
+            best_clique = clique
+    return best_clique
 
-        clique.append(random_vertex)
 
-        vertices_candidates = graph.neighbors(random_vertex)
-        np.random.shuffle(vertices_candidates)
+def greedy_randomized_mp_clique(graph: Graph, iterations: int = 10, num_proc: int = -1) -> list[int]:
+    best_clique = []
+    vertices = graph.vertexes
 
-        for vertex in vertices_candidates:
-            is_connected = True
-            for clique_vertex in clique:
-                if clique_vertex not in graph.neighbors(vertex):
-                    is_connected = False
-                    break
-            if is_connected:
-                clique.append(vertex)
-        if len(clique) > len(best_clique):
-            best_clique = clique
+    random_vertices = np.random.choice(vertices, iterations)
+    num_process = num_proc if num_proc > 1 else cpu_count()
+
+    partial_get_clique =  partial(_get_clique, graph=graph)
+    with Pool(num_process) as pool:
+        cliques = pool.map(partial_get_clique, random_vertices)
+    best_clique = max(cliques, key=len)
     return best_clique