precomputing_example.sagews

load("gt.sage")
load("gt_precomputed_database.sage")
load("conjecturing.py")

# Invariants
conjecture(graph_objects, all_invariants, 0, precomputed = precomputed_invariants_for_conjecture())

# Properties
propertyBasedConjecture(graph_objects, properties, 0, precomputed = precomputed_properties_for_conjecture())

"""
    The method precomputed_invariants_for_conjecture() is loaded by load("gt_precomputed_database.sage")
    As long as you have the .db file in your working directory, this method will take care of everything.

    Nico is working on a "maximum precomputed subset" method, to be released soon.
"""

loaded utilities loaded invariants loaded properties loaded theorems loaded graphs Remember to load DIMACS and Sloane graphs if you want them Remember to load DIMACS and Sloane graphs if you want them

load("gt.sage")
load("gt_precomputed_database.sage")
load("conjecturing.py")

loaded utilities loaded invariants loaded properties loaded theorems loaded graphs Remember to load DIMACS and Sloane graphs if you want them

sparselist = []

for g in graph_objects:
    if g.order()<60:
        sparselist.append(g)
print len(sparselist)

532

ecp = efficiently_computable_properties
ecp.append(Graph.is_hamiltonian)
property_of_interest = ecp.index(Graph.is_hamiltonian)

theorem_s1 = lambda g: g.is_bipartite() and g.is_strongly_regular()
theorem_s2 = lambda g: g.is_circular_planar() and g.is_cartesian_product()
theorem_s3 = lambda g: g.is_circular_planar() and g.is_two_connected()

theorems = [Graph.is_cycle, Graph.is_clique, theorem_s1, is_ore, is_dirac, is_chvatal_erdos, theorem_s2,
is_haggkvist_nicoghossian, is_genghua_fan, is_planar_transitive, is_generalized_dirac, is_lindquester, theorem_s3]

precomp = precomputed_properties_for_conjecture()

conjs = propertyBasedConjecture(sparselist, ecp, property_of_interest, theory = theorems, time=5, precomputed = precomp, verbose=False, debug=False)
for c in conjs:
    print c