def testBijectorOverRange(self):
with self.test_session():
for dtype in (np.float32, np.float64):
skewness = np.array([1.2, 5.], dtype=dtype)
tailweight = np.array([2., 10.], dtype=dtype)
# The inverse will be defined up to where sinh is valid, which is
# arcsinh(np.finfo(dtype).max).
log_boundary = np.log(
np.sinh(
np.arcsinh(np.finfo(dtype).max) / tailweight -
skewness))
x = np.array([
np.logspace(-2, log_boundary[0], base=np.e, num=1000),
np.logspace(-2, log_boundary[1], base=np.e, num=1000)
],
dtype=dtype)
# Ensure broadcasting works.
x = np.swapaxes(x, 0, 1)
y = np.sinh((np.arcsinh(x) + skewness) * tailweight)
bijector = SinhArcsinh(skewness=skewness,
tailweight=tailweight,
validate_args=True)
self.assertAllClose(y,
bijector.forward(x).eval(),
rtol=1e-4,
atol=0.)
self.assertAllClose(x,
bijector.inverse(y).eval(),
rtol=1e-4,
atol=0.)
# On IBM PPC systems, longdouble (np.float128) is same as double except that it can have more precision.
# Type double being of 8 bytes, can't hold square of max of float64 (which is also 8 bytes) and
# below test fails due to overflow error giving inf. So this check avoids that error by skipping square
# calculation and corresponding assert.
if np.amax(y) <= np.sqrt(np.finfo(np.float128).max) and \
np.fabs(np.amin(y)) <= np.sqrt(np.fabs(np.finfo(np.float128).min)):
# Do the numpy calculation in float128 to avoid inf/nan.
y_float128 = np.float128(y)
self.assertAllClose(np.log(
np.cosh(
np.arcsinh(y_float128) / tailweight - skewness) /
np.sqrt(y_float128**2 + 1)) - np.log(tailweight),
bijector.inverse_log_det_jacobian(
y, event_ndims=0).eval(),
rtol=1e-4,
atol=0.)
self.assertAllClose(-bijector.inverse_log_det_jacobian(
y, event_ndims=0).eval(),
bijector.forward_log_det_jacobian(
x, event_ndims=0).eval(),
rtol=1e-4,
atol=0.)
def testSkew(self):
with self.test_session():
# Will broadcast together to shape [3, 2].
x = [-1., 1.]
skewness = [[-1.], [0.], [1.]]
bijector = SinhArcsinh(skewness=skewness, validate_args=True)
y = bijector.forward(x).eval()
# For skew < 0, |forward(-1)| > |forward(1)|
self.assertGreater(np.abs(y[0, 0]), np.abs(y[0, 1]))
# For skew = 0, |forward(-1)| = |forward(1)|
self.assertAllClose(np.abs(y[1, 0]), np.abs(y[1, 1]))
# For skew > 0, |forward(-1)| < |forward(1)|
self.assertLess(np.abs(y[2, 0]), np.abs(y[2, 1]))
def testLargerTailWeightPutsMoreWeightInTails(self):
with self.test_session():
# Will broadcast together to shape [3, 2].
x = [-1., 1.]
tailweight = [[0.5], [1.0], [2.0]]
bijector = SinhArcsinh(tailweight=tailweight, validate_args=True)
y = bijector.forward(x).eval()
# x = -1, 1 should be mapped to points symmetric about 0
self.assertAllClose(y[:, 0], -1. * y[:, 1])
# forward(1) should increase as tailweight increases, since higher
# tailweight should map 1 to a larger number.
forward_1 = y[:, 1] # The positive values of y.
self.assertLess(forward_1[0], forward_1[1])
self.assertLess(forward_1[1], forward_1[2])
def testBijectorOverRange(self):
with self.test_session():
for dtype in (np.float32, np.float64):
skewness = np.array([1.2, 5.], dtype=dtype)
tailweight = np.array([2., 10.], dtype=dtype)
# The inverse will be defined up to where sinh is valid, which is
# arcsinh(np.finfo(dtype).max).
log_boundary = np.log(
np.sinh(
np.arcsinh(np.finfo(dtype).max) / tailweight -
skewness))
x = np.array([
np.logspace(-2, log_boundary[0], base=np.e, num=1000),
np.logspace(-2, log_boundary[1], base=np.e, num=1000)
],
dtype=dtype)
# Ensure broadcasting works.
x = np.swapaxes(x, 0, 1)
y = np.sinh((np.arcsinh(x) + skewness) * tailweight)
bijector = SinhArcsinh(skewness=skewness,
tailweight=tailweight,
validate_args=True)
self.assertAllClose(y,
bijector.forward(x).eval(),
rtol=1e-4,
atol=0.)
self.assertAllClose(x,
bijector.inverse(y).eval(),
rtol=1e-4,
atol=0.)
# Do the numpy calculation in float128 to avoid inf/nan.
y_float128 = np.float128(y)
self.assertAllClose(np.log(
np.cosh(np.arcsinh(y_float128) / tailweight - skewness) /
np.sqrt(y_float128**2 + 1)) - np.log(tailweight),
bijector.inverse_log_det_jacobian(
y).eval(),
rtol=1e-4,
atol=0.)
self.assertAllClose(
-bijector.inverse_log_det_jacobian(y).eval(),
bijector.forward_log_det_jacobian(x).eval(),
rtol=1e-4,
atol=0.)
def testBijectorOverRange(self):
with self.test_session():
for dtype in (np.float32, np.float64):
skewness = np.array([1.2, 5.], dtype=dtype)
tailweight = np.array([2., 10.], dtype=dtype)
# The inverse will be defined up to where sinh is valid, which is
# arcsinh(np.finfo(dtype).max).
log_boundary = np.log(
np.sinh(np.arcsinh(np.finfo(dtype).max) / tailweight - skewness))
x = np.array([
np.logspace(-2, log_boundary[0], base=np.e, num=1000),
np.logspace(-2, log_boundary[1], base=np.e, num=1000)
], dtype=dtype)
# Ensure broadcasting works.
x = np.swapaxes(x, 0, 1)
y = np.sinh((np.arcsinh(x) + skewness) * tailweight)
bijector = SinhArcsinh(
skewness=skewness, tailweight=tailweight, validate_args=True)
self.assertAllClose(y, bijector.forward(x).eval(), rtol=1e-4, atol=0.)
self.assertAllClose(x, bijector.inverse(y).eval(), rtol=1e-4, atol=0.)
# On IBM PPC systems, longdouble (np.float128) is same as double except that it can have more precision.
# Type double being of 8 bytes, can't hold square of max of float64 (which is also 8 bytes) and
# below test fails due to overflow error giving inf. So this check avoids that error by skipping square
# calculation and corresponding assert.
if np.amax(y) <= np.sqrt(np.finfo(np.float128).max) and \
np.fabs(np.amin(y)) <= np.sqrt(np.fabs(np.finfo(np.float128).min)):
# Do the numpy calculation in float128 to avoid inf/nan.
y_float128 = np.float128(y)
self.assertAllClose(
np.log(np.cosh(
np.arcsinh(y_float128) / tailweight - skewness) / np.sqrt(
y_float128**2 + 1)) -
np.log(tailweight),
bijector.inverse_log_det_jacobian(y, event_ndims=0).eval(),
rtol=1e-4,
atol=0.)
self.assertAllClose(
-bijector.inverse_log_det_jacobian(y, event_ndims=0).eval(),
bijector.forward_log_det_jacobian(x, event_ndims=0).eval(),
rtol=1e-4,
atol=0.)
def testBijectorVersusNumpyRewriteOfBasicFunctions(self):
with self.test_session():
skewness = 0.2
tailweight = 2.0
bijector = SinhArcsinh(skewness=skewness,
tailweight=tailweight,
validate_args=True)
self.assertEqual("SinhArcsinh", bijector.name)
x = np.array([[[-2.01], [2.], [1e-4]]]).astype(np.float32)
y = np.sinh((np.arcsinh(x) + skewness) * tailweight)
self.assertAllClose(y, bijector.forward(x).eval())
self.assertAllClose(x, bijector.inverse(y).eval())
self.assertAllClose(
np.sum(np.log(np.cosh(np.arcsinh(y) / tailweight - skewness)) -
np.log(tailweight) - np.log(np.sqrt(y**2 + 1)),
axis=-1),
bijector.inverse_log_det_jacobian(y, event_ndims=1).eval())
self.assertAllClose(
-bijector.inverse_log_det_jacobian(y, event_ndims=1).eval(),
bijector.forward_log_det_jacobian(x, event_ndims=1).eval(),
rtol=1e-4,
atol=0.)
def testBijectorOverRange(self):
with self.test_session():
for dtype in (np.float32, np.float64):
skewness = np.array([1.2, 5.], dtype=dtype)
tailweight = np.array([2., 10.], dtype=dtype)
# The inverse will be defined up to where sinh is valid, which is
# arcsinh(np.finfo(dtype).max).
log_boundary = np.log(
np.sinh(np.arcsinh(np.finfo(dtype).max) / tailweight - skewness))
x = np.array([
np.logspace(-2, log_boundary[0], base=np.e, num=1000),
np.logspace(-2, log_boundary[1], base=np.e, num=1000)
], dtype=dtype)
# Ensure broadcasting works.
x = np.swapaxes(x, 0, 1)
y = np.sinh((np.arcsinh(x) + skewness) * tailweight)
bijector = SinhArcsinh(
skewness=skewness, tailweight=tailweight, validate_args=True)
self.assertAllClose(y, bijector.forward(x).eval(), rtol=1e-4, atol=0.)
self.assertAllClose(x, bijector.inverse(y).eval(), rtol=1e-4, atol=0.)
# Do the numpy calculation in float128 to avoid inf/nan.
y_float128 = np.float128(y)
self.assertAllClose(
np.log(np.cosh(
np.arcsinh(y_float128) / tailweight - skewness) / np.sqrt(
y_float128**2 + 1)) -
np.log(tailweight),
bijector.inverse_log_det_jacobian(y).eval(),
rtol=1e-4,
atol=0.)
self.assertAllClose(
-bijector.inverse_log_det_jacobian(y).eval(),
bijector.forward_log_det_jacobian(x).eval(),
rtol=1e-4,
atol=0.)
def testBijectorVersusNumpyRewriteOfBasicFunctions(self):
with self.test_session():
skewness = 0.2
tailweight = 2.0
bijector = SinhArcsinh(
skewness=skewness,
tailweight=tailweight,
event_ndims=1,
validate_args=True)
self.assertEqual("sinh_arcsinh", bijector.name)
x = np.array([[[-2.01], [2.], [1e-4]]]).astype(np.float32)
y = np.sinh((np.arcsinh(x) + skewness) * tailweight)
self.assertAllClose(y, bijector.forward(x).eval())
self.assertAllClose(x, bijector.inverse(y).eval())
self.assertAllClose(
np.sum(
np.log(np.cosh(np.arcsinh(y) / tailweight - skewness)) -
np.log(tailweight) - np.log(np.sqrt(y**2 + 1)),
axis=-1), bijector.inverse_log_det_jacobian(y).eval())
self.assertAllClose(
-bijector.inverse_log_det_jacobian(y).eval(),
bijector.forward_log_det_jacobian(x).eval(),
rtol=1e-4,
atol=0.)