def generate_random_MILPInstance(numVars=40,
numCons=20,
density=0.2,
maxObjCoeff=10,
maxConsCoeff=10,
tightness=2,
rand_seed=2):
cs, vs, objective, A, b = GenerateRandomMIP(numVars=numVars,
numCons=numCons,
density=density,
maxObjCoeff=maxObjCoeff,
maxConsCoeff=maxConsCoeff,
tightness=tightness,
rand_seed=rand_seed)
A = np.asmatrix(pd.DataFrame.from_dict(A).to_numpy())
objective = CyLPArray(list(objective.values()))
b = CyLPArray(b)
l = CyLPArray([0] * len(vs))
u = CyLPArray([maxObjCoeff] * len(vs))
return MILPInstance(A=A,
b=b,
c=objective,
l=l,
u=u,
sense=['Max', '<='],
integerIndices=list(range(len(vs))),
numVars=len(vs))
def generate_random_value_functions(num_probs=40, num_evals=40):
for prob_num in range(num_probs):
num_vars = 4
num_constrs = 2
density = np.random.uniform(.2, .8)
max_obj_coef = np.random.randint(10, 100)
max_const_coef = np.random.randint(10, 100)
tightness = 15
# get a random A and objective
cs, vs, objective, A, b = GenerateRandomMIP(
numVars=num_vars,
numCons=num_constrs,
density=density,
maxObjCoeff=max_obj_coef,
maxConsCoeff=max_const_coef,
tightness=15)
A = np.asmatrix(pd.DataFrame.from_dict(A).to_numpy())
objective = CyLPArray(list(objective.values()))
l = CyLPArray([0] * len(vs))
u = CyLPArray([max_obj_coef] * len(vs))
# make a new directory to save all these instances
fldr = f'example_value_functions/instance_{prob_num}'
os.mkdir(fldr)
for eval_num in range(num_evals):
# make a random b for each evaluation of this instance
b = CyLPArray([
np.random.randint(
int(num_vars * density * max_const_coef / tightness),
int(num_vars * density * max_const_coef / 1.5))
for i in range(num_constrs)
])
instance = MILPInstance(A=A,
b=b,
c=objective,
l=l,
u=u,
sense=['Max', '<='],
integerIndices=list(range(len(vs))),
numVars=len(vs))
for i in instance.integerIndices:
instance.lp.setInteger(i)
file = f'evaluation_{eval_num}'
instance.lp.writeMps(f'{os.path.join(fldr, file)}.mps')