def test_condition(self):
N = 10
date = Date('2019-01-01')
engine = DummyEngine('dummy', np.array([date]), N)
# test basic indicator
ind = cf.Indicator(cf.Vector(np.arange(5)) - 2, 0.1)
val = ind.evaluate(engine, date)
assert_equal([0, 0, 0.5, 1, 1], val)
ind = cf.Indicator(cf.Vector([1.95, 1.97, 2, 2.03, 2.05]) - 2, 0.1)
val = ind.evaluate(engine, date)
assert_almost_equal([0, 0.2, 0.5, 0.8, 1], val, decimal=10)
# test max
val = cf.Max(cf.Vector(np.arange(5)),
cf.Constant(2)).evaluate(engine, date)
assert_almost_equal([2, 2, 2, 3, 4], val)
val = cf.Max(cf.Vector([3, 2, 1, 4, 5]),
cf.Vector(np.arange(5))).evaluate(engine, date)
assert_almost_equal([3, 2, 2, 4, 5], val)
val = cf.Max(cf.Vector([1.95, 1.975, 2, 2.025, 2.05]),
cf.Constant(2)).evaluate(engine, date)
assert_almost_equal([2, 2, 2, 2.025, 2.05], val)
def test_vector(self):
N = 10
date = Date('2019-01-01')
engine = DummyEngine('dummy', np.array([date]), N)
x = cf.Vector(np.ones(N))
y = cf.Constant(10)
# arithmetic
assert_equal(np.ones(N) * -9, (x - y).evaluate(engine, date))
assert_equal(np.ones(N) * -1, -x.evaluate(engine, date))
assert_equal(np.ones(N) * 10, (x * 10).evaluate(engine, date))
def test_cache(self):
N = 10
dates = [
Date('2019-01-01'),
Date('2019-02-01'),
Date('2019-03-01'),
Date('2019-04-01'),
Date('2019-05-01')
]
engine = DummyEngine('dummy', np.array(dates), N)
class MockCashflow(cf.Cashflow):
def __init__(self, map):
super(MockCashflow, self).__init__()
self.spot = map
def _evaluate(self, eng, dt: Date) -> np.ndarray:
return np.array(self.spot[dt])
# single variable corner case
index = MockCashflow(
{k: v
for k, v in zip(dates, [101, 100, 100, 99, 101])})
alive = cf.Cache(1)
alive.assign(cf.Indicator(index - cf.Constant(100)) * alive.state)
vals = [alive.evaluate(engine, d) for d in dates]
assert_almost_equal([[1], [0.5], [0.25], [0], [0]], vals)
# multivariate normal case
index = MockCashflow({
k: v
for k, v in zip(dates, [[101, 100], [100, 100], [100, 100],
[99, 100], [101, 99]])
})
alive = cf.Cache([1, 1])
alive.assign(cf.Indicator(index - cf.Constant(100)) * alive.state)
vals = [alive.evaluate(engine, d) for d in dates]
assert_almost_equal(
[[1, 0.5], [0.5, 0.25], [0.25, 0.125], [0, 0.0625], [0, 0]], vals)
def test_caching(self):
dates = np.array(
[Date('2019-01-01'),
Date('2019-02-01'),
Date('2019-03-01')])
engine = DummyEngine('dummy', dates, 1000)
x = cf.Constant(1)
y = cf.Constant(2)
self.assertEqual(0, len(x.cache))
self.assertEqual(0, len(y.cache))
x.evaluate(engine, dates[0])
self.assertEqual(1, len(x.cache))
# cached
x.value = 10
val = x.evaluate(engine, dates[0])
self.assertEqual(1, val)
# not cached
y.value = 10
val = y.evaluate(engine, dates[0])
self.assertEqual(10, val)
def test_arithmetic(self):
dates = np.array(
[Date('2019-01-01'),
Date('2019-02-01'),
Date('2019-03-01')])
engine = DummyEngine('dummy', dates, 1000)
x = cf.Constant(1)
y = cf.Constant(3)
# addition
self.assertEqual(4, (x + y).evaluate(engine, dates[0]))
self.assertEqual(6, (x + 5).evaluate(engine, dates[0]))
# subtraction
self.assertEqual(-2, (x - y).evaluate(engine, dates[0]))
self.assertEqual(0, (x - 1).evaluate(engine, dates[0]))
# multiplication
self.assertEqual(3, (x * y).evaluate(engine, dates[0]))
self.assertEqual(4, (x * 4).evaluate(engine, dates[0]))
# negation
val = -x.evaluate(engine, dates[0])
self.assertEqual(-1, val)