def test_nonneighboring_distance_cost_close():
g, node_indices, neighbors = make_one_component_engine_data()
settings = top.OptimizerSettings(minimum_dist_others=10,
others_weight=1)
x = np.array([
[0, 0],
[0, 100],
[100, 100],
[3, 4]
]).reshape((-1, 1))
expected_cost = 50 # 2 nodes * (D - sqrt( (x2-x1)^2 + (y2-y1)^2 ))^2 per node
common = 2 * -2 * (10 - 5) / 5 # 2 nodes * (-2 * (D - norm) / norm) per node
expected_grad = np.array([
[common * -3, common * -4],
[0, 0],
[0, 0],
[common * 3, common * 4]
]).flatten()
ct = top.NonNeighboringDistance(g, node_indices, neighbors, settings)
cost, grad = ct.evaluate(top.make_costargs(x))
assert cost == expected_cost
np.testing.assert_equal(expected_grad, grad)
def make_cost_terms(g, node_indices, neighbors, settings):
c1 = top.NeighboringDistance(g,
node_indices,
neighbors,
settings,
internal=True)
c2 = top.NeighboringDistance(g,
node_indices,
neighbors,
settings,
internal=False)
c3 = top.NonNeighboringDistance(g, node_indices, neighbors, settings)
c4 = top.CentroidDeviation(g, node_indices, neighbors, settings)
c5 = top.RightAngleDeviation(g, node_indices, neighbors, settings)
c6 = top.HorizontalDeviation(g, node_indices, neighbors, settings)
return [c1, c2, c3, c4, c5, c6]
def test_nonneighboring_distance_cost_far():
g, node_indices, neighbors = make_one_component_engine_data()
settings = top.OptimizerSettings(minimum_dist_others=10,
others_weight=1)
x = np.array([
[0, 0],
[0, 100],
[100, 100],
[100, 200]
]).reshape((-1, 1))
expected_cost = 0
expected_grad = np.zeros(8)
ct = top.NonNeighboringDistance(g, node_indices, neighbors, settings)
cost, grad = ct.evaluate(top.make_costargs(x))
assert cost == expected_cost
np.testing.assert_equal(expected_grad, grad)
def test_nonneighboring_distance_cost_multiple_nodes_close():
g, node_indices, neighbors = make_two_component_engine_data()
settings = top.OptimizerSettings(minimum_dist_others=20,
others_weight=1)
x = np.array([
[0, 0], # 1
[100, 0], # c1.1
[103, 4], # c2.1
[106, 1], # c1.2
[109, 5], # c2.2
[200, 0] # 2
]).reshape((-1, 1))
cost_11_21 = 2 * (20 - 5) ** 2
cost_11_22 = 2 * (20 - np.sqrt(106)) ** 2
cost_12_21 = 2 * (20 - np.sqrt(18)) ** 2
cost_12_22 = 2 * (20 - 5) ** 2
expected_cost = cost_11_21 + cost_11_22 + cost_12_21 + cost_12_22
# 2 nodes * (-2 * (D - norm) / norm) per node
common_11_21 = 2 * -2 * (20 - 5) / 5
common_11_22 = 2 * -2 * (20 - np.sqrt(106)) / np.sqrt(106)
common_12_21 = 2 * -2 * (20 - np.sqrt(18)) / np.sqrt(18)
common_12_22 = 2 * -2 * (20 - 5) / 5
expected_grad = np.array([
[0, 0],
[common_11_21 * -3 + common_11_22 * -9, common_11_21 * -4 + common_11_22 * -5],
[common_11_21 * 3 + common_12_21 * -3, common_11_21 * 4 + common_12_21 * 3],
[common_12_21 * 3 + common_12_22 * -3, common_12_21 * -3 + common_12_22 * -4],
[common_11_22 * 9 + common_12_22 * 3, common_11_22 * 5 + common_12_22 * 4],
[0, 0]
]).flatten()
ct = top.NonNeighboringDistance(g, node_indices, neighbors, settings)
cost, grad = ct.evaluate(top.make_costargs(x))
assert cost == expected_cost
np.testing.assert_equal(expected_grad, grad)