def test_abslinear(self):
"""
Absolute value function with many iterations
"""
myfunc = simul.abslinear
nvar = 1
vbounds = np.zeros((nvar, 2))
vbounds[0, 0] = -100.
vbounds[0, 1] = 100.
mypb = genop.Genop()
mypb.verbose = False
self.assertEqual(mypb.popsize, 25)
self.assertEqual(mypb.pbcross, 0.7)
self.assertEqual(mypb.pbmut, 0.1)
error = mypb.initialize(myfunc, nvar, vbounds)
self.assertEqual(error, 0)
_, xopt = mypb.optimize(2000)
self.assertTrue(np.abs(xopt[-1] - 2.) < 1.e-3)
def test_eggholder(self):
"""
Eggholder function with a large population size
"""
myfunc = simul.eggholder
nvar = 2
vbounds = np.zeros((nvar, 2))
vbounds[0, 0] = -512.
vbounds[0, 1] = 512.
vbounds[1, 0] = -512.
vbounds[1, 1] = 512.
mypb = genop.Genop(1000)
mypb.verbose = False
self.assertEqual(mypb.popsize, 1000)
self.assertEqual(mypb.pbcross, 0.7)
self.assertEqual(mypb.pbmut, 0.1)
error = mypb.initialize(myfunc, nvar, vbounds)
self.assertEqual(error, 0)
fcost, _ = mypb.optimize(20)
self.assertTrue(fcost[-1] < -900.)
def test_easom(self):
"""
Easom function
"""
myfunc = simul.easom
nvar = 2
vbounds = np.zeros((nvar, 2))
vbounds[0, 0] = -100.
vbounds[0, 1] = 100.
vbounds[1, 0] = -100.
vbounds[1, 1] = 100.
mypb = genop.Genop()
mypb.verbose = False
self.assertEqual(mypb.popsize, 25)
self.assertEqual(mypb.pbcross, 0.7)
self.assertEqual(mypb.pbmut, 0.1)
error = mypb.initialize(myfunc, nvar, vbounds)
self.assertEqual(error, 0)
fcost, _ = mypb.optimize(nbgen=100)
self.assertTrue(np.abs(fcost[-2] - fcost[-1]) < 1.e-12)
def test_maccormick(self):
"""
Mc Cormick function with non default algorithm parameters
"""
myfunc = simul.maccormick
nvar = 2
vbounds = np.zeros((nvar, 2))
vbounds[0, 0] = -1.5
vbounds[0, 1] = 4.
vbounds[1, 0] = -3.
vbounds[1, 1] = 4.
mypb = genop.Genop(90, 0.69, 0.05)
mypb.verbose = False
self.assertEqual(mypb.popsize, 90)
self.assertEqual(mypb.pbcross, 0.69)
self.assertEqual(mypb.pbmut, 0.05)
error = mypb.initialize(myfunc, nvar, vbounds)
self.assertEqual(error, 0)
fcost, _ = mypb.optimize(20)
self.assertTrue(fcost[-1] < -1.91)
def test_rosenbrock(self):
"""
Rosenbrock function with verbose mode and parallelism
"""
myfunc = simul.rosenbrock
nvar = 2
vbounds = np.zeros((nvar, 2))
vbounds[0, 0] = -5.
vbounds[0, 1] = 5.
vbounds[1, 0] = -5.
vbounds[1, 1] = 5.
mypb = genop.Genop(50)
mypb.verbose = True
self.assertEqual(mypb.popsize, 50)
self.assertEqual(mypb.pbcross, 0.7)
self.assertEqual(mypb.pbmut, 0.1)
error = mypb.initialize(myfunc, nvar, vbounds)
self.assertEqual(error, 0)
fcost, _ = mypb.optimize(30, nproc=2)
print(fcost)
self.assertTrue(fcost[-1] < 5.e-2)
def test_linear(self):
"""
Linear function with many iterations
"""
myfunc = simul.linear
nvar = 3
vbounds = np.zeros((nvar, 2))
vbounds[0, 0] = -3.14
vbounds[0, 1] = 3.14
vbounds[1, 0] = 0.
vbounds[1, 1] = 5.
vbounds[2, 0] = -10.
vbounds[2, 1] = 20.
mypb = genop.Genop()
mypb.verbose = False
self.assertEqual(mypb.popsize, 25)
self.assertEqual(mypb.pbcross, 0.7)
self.assertEqual(mypb.pbmut, 0.1)
error = mypb.initialize(myfunc, nvar, vbounds)
self.assertEqual(error, 0)
fcost, _ = mypb.optimize(200)
self.assertTrue(fcost[-1] < -20.)