summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorAlex AUVOLAT <alex.auvolat@ens.fr>2013-12-08 18:21:49 +0100
committerAlex AUVOLAT <alex.auvolat@ens.fr>2013-12-08 18:21:49 +0100
commit70bd7ed41a97bd637cc04b61e594d4d17aa93303 (patch)
tree19964ec2f599834b4d8dac4bfcca36668aa73541
parentbf00b49660fee9b13268fac075a955c8485d4b6b (diff)
downloadAlgoProg-Projet-70bd7ed41a97bd637cc04b61e594d4d17aa93303.tar.gz
AlgoProg-Projet-70bd7ed41a97bd637cc04b61e594d4d17aa93303.zip
Ajout de l'heuristique de coloriage.
-rw-r--r--algos.c103
1 files changed, 103 insertions, 0 deletions
diff --git a/algos.c b/algos.c
index 35f1dfd..b329f0c 100644
--- a/algos.c
+++ b/algos.c
@@ -1,6 +1,100 @@
+#include <assert.h>
+#include <stdbool.h>
+
+
#include "algos.h"
+/* Heuristique de coloriage
+ * Trouve un coloriage non-optimal (mais parfois proche) du sous-graphe engendré
+ * par s dans le graphe g.
+ * Ne donne pas le coloriage, mais juste le nombre de couleurs d'un coloriage possible (nombre que l'on cherche à minimiser)
+ * Cet algorithme est assez naif mais il est fait pour tourner sur des graphes de taille petite (de l'ordre de 20), donc un
+ * algorithme même en n^3 ou n^4 est acceptable (ou pas...)
+ */
+
+void color_subgraph_aux(const graph g, int *colors, int *v, const int n) {
+ int i, j, t;
+
+ if (n == 0) return;
+
+ if (n == 1) {
+ colors[0] = 0;
+ return;
+ }
+
+ // Find element with smallest neighbours count
+ int x = -1, m = n + 1;
+ for (i = 0; i < n; i++) {
+ t = 0;
+ for (j = 0; j < n; j++) {
+ if (set_mem(v[j], graph_neighbours(g, v[i]))) t++;
+ }
+ if (t < m) {
+ x = i; m = t;
+ }
+ }
+
+ // Put that element at the beginning of vector v, so that the remaining subgraph is defined by v[1..]
+ if (x != 0) {
+ t = v[x];
+ v[0] = v[x];
+ v[x] = t;
+ }
+
+ // Color remaining subgraph (v[1..])
+ color_subgraph_aux(g, colors + 1, v + 1, n - 1);
+
+ // Find free color
+ colors[0] = -1;
+ bool used[n];
+ for (i = 0; i < n; i++) used[i] = false;
+ for (i = 1; i < n; i++) {
+ if (set_mem(v[i], graph_neighbours(g, v[0])))
+ used[colors[i]] = true;
+ }
+ for (i = 0; i < n; i++) {
+ if (used[i] == false) {
+ colors[0] = i;
+ break;
+ }
+ }
+ assert(colors[0] != -1);
+}
+
+int color_subgraph(const graph g, set s) {
+ int i, m;
+
+ const int n = set_size(s);
+ if (n == 0) return 0;
+
+ int colors[n], vertices[n];
+ for (i = 0; i < n; i++)
+ colors[i] = 0;
+
+ // Put all vertices in vertice vector (order is not important)
+ vertices[0] = elt_of_set(s);
+ set_remove_ip(vertices[0], s);
+ i = 1;
+ while (!is_set_empty(s)) {
+ vertices[i] = elt_of_set_heur(s, vertices[i-1]);
+ set_remove_ip(vertices[i], s);
+ i++;
+ }
+ assert(i == n);
+
+ // Color graph
+ color_subgraph_aux(g, colors, vertices, n);
+
+ // Count colors
+ m = colors[0];
+ for (i = 1; i < n; i++)
+ if (colors[i] > m) m = colors[i];
+
+ return m + 1;
+}
+
+
/*Premier algorithme naïf.
* g : graphe où on cherche les cliques
* k : clique actuelle
@@ -89,6 +183,15 @@ void max_clique_c(const graph g, set k, set c, set a, set *mc) {
// If we have no possibility to explore, return
if (is_set_empty(c)) return;
+ if (set_size(c) == g->N / 2) {
+ // Color graph : we may have no possibility
+ set c_copy = copy_set(c);
+ int color_count = color_subgraph(g, c_copy);
+ delete_set(c_copy);
+ if (set_size(k) + color_count <= set_size(*mc)) return;
+ }
+
+
// Find u that maximises |C inter Gamma(u)|
set c_it = copy_set(c);
int u = elt_of_set(c_it), n = 0;