Experiments to Cycle Decomposition

$ make
$ make run

Exp 1: Layered Graphs

  • 10 layers and repeat 1000 times (default seed).
vector<int> layers = {1, 99, 100, 100, 100, 100, 100, 100, 100, 100, 99, 1};
config.repeat = 1000;

Result:

not-inf-ratio = 0.01700, finite-avg = 27.00000
  • 20 layers; each layer 80 vertices; repeat 1000 times (default seed).
not-inf-ratio = 0.00100, finite-avg = 55.00000

Exp 2: GNP Graph

  • n = 1000, p = 0.5
not-inf-ratio = 0.05800, finite-avg = 17.36207
  • n = 1000, p = 0.3
not-inf-ratio = 0.05400, finite-avg = 18.40741
  • n = 1000, p = 0.1
not-inf-ratio = 0.05300, finite-avg = 17.32075
  • n = 1000, p = 0.9
not-inf-ratio = 0.06400, finite-avg = 16.23438

Exp 3: Random Eliminating Edges with a Fixed Permutation

  • add edges according to each layer -> always disconnected.

  • shuffle edges randomly then random eliminate w.p. 0.5 ->

n = 1000, L = 10, 10 layers.
not-inf-ratio = 0.44600, finite-avg = 43.53812
  • random layered constant degree graph