examples/color.mod
author Alpar Juttner <alpar@cs.elte.hu>
Mon, 06 Dec 2010 13:09:21 +0100
changeset 1 c445c931472f
permissions -rw-r--r--
Import glpk-4.45

- Generated files and doc/notes are removed
     1 /* COLOR, Graph Coloring Problem */
     2 
     3 /* Written in GNU MathProg by Andrew Makhorin <mao@gnu.org> */
     4 
     5 /* Given an undirected loopless graph G = (V, E), where V is a set of
     6    nodes, E <= V x V is a set of arcs, the Graph Coloring Problem is to
     7    find a mapping (coloring) F: V -> C, where C = {1, 2, ... } is a set
     8    of colors whose cardinality is as small as possible, such that
     9    F(i) != F(j) for every arc (i,j) in E, that is adjacent nodes must
    10    be assigned different colors. */
    11 
    12 param n, integer, >= 2;
    13 /* number of nodes */
    14 
    15 set V := {1..n};
    16 /* set of nodes */
    17 
    18 set E, within V cross V;
    19 /* set of arcs */
    20 
    21 check{(i,j) in E}: i != j;
    22 /* there must be no loops */
    23 
    24 /* We need to estimate an upper bound of the number of colors |C|.
    25    The number of nodes |V| can be used, however, for sparse graphs such
    26    bound is not very good. To obtain a more suitable estimation we use
    27    an easy "greedy" heuristic. Let nodes 1, ..., i-1 are already
    28    assigned some colors. To assign a color to node i we see if there is
    29    an existing color not used for coloring nodes adjacent to node i. If
    30    so, we use this color, otherwise we introduce a new color. */
    31 
    32 set EE := setof{(i,j) in E} (i,j) union setof{(i,j) in E} (j,i);
    33 /* symmetrisized set of arcs */
    34 
    35 param z{i in V, case in 0..1} :=
    36 /* z[i,0] = color index assigned to node i
    37    z[i,1] = maximal color index used for nodes 1, 2, ..., i-1 which are
    38             adjacent to node i */
    39 (  if case = 0 then
    40    (  /* compute z[i,0] */
    41       min{c in 1..z[i,1]}
    42       (  if not exists{j in V: j < i and (i,j) in EE} z[j,0] = c then
    43             c
    44          else
    45             z[i,1] + 1
    46       )
    47    )
    48    else
    49    (  /* compute z[i,1] */
    50       if not exists{j in V: j < i} (i,j) in EE then
    51          1
    52       else
    53          max{j in V: j < i and (i,j) in EE} z[j,0]
    54    )
    55 );
    56 
    57 check{(i,j) in E}: z[i,0] != z[j,0];
    58 /* check that all adjacent nodes are assigned distinct colors */
    59 
    60 param nc := max{i in V} z[i,0];
    61 /* number of colors used by the heuristic; obviously, it is an upper
    62    bound of the optimal solution */
    63 
    64 display nc;
    65 
    66 var x{i in V, c in 1..nc}, binary;
    67 /* x[i,c] = 1 means that node i is assigned color c */
    68 
    69 var u{c in 1..nc}, binary;
    70 /* u[c] = 1 means that color c is used, i.e. assigned to some node */
    71 
    72 s.t. map{i in V}: sum{c in 1..nc} x[i,c] = 1;
    73 /* each node must be assigned exactly one color */
    74 
    75 s.t. arc{(i,j) in E, c in 1..nc}: x[i,c] + x[j,c] <= u[c];
    76 /* adjacent nodes cannot be assigned the same color */
    77 
    78 minimize obj: sum{c in 1..nc} u[c];
    79 /* objective is to minimize the number of colors used */
    80 
    81 data;
    82 
    83 /* These data correspond to the instance myciel3.col from:
    84    http://mat.gsia.cmu.edu/COLOR/instances.html */
    85 
    86 /* The optimal solution is 4 */
    87 
    88 param n := 11;
    89 
    90 set E :=
    91  1 2
    92  1 4
    93  1 7
    94  1 9
    95  2 3
    96  2 6
    97  2 8
    98  3 5
    99  3 7
   100  3 10
   101  4 5
   102  4 6
   103  4 10
   104  5 8
   105  5 9
   106  6 11
   107  7 11
   108  8 11
   109  9 11
   110  10 11
   111 ;
   112 
   113 end;