def test_min_max():
# min: num.Fit1D.get_min()
x = np.linspace(-10,10,100)
y = (x-5)**2.0 + 1.0
spl = Spline(x,y)
xmin = spl.get_min()
ymin = spl(xmin)
np.testing.assert_almost_equal(xmin, 5.0)
np.testing.assert_almost_equal(ymin, 1.0)
# max: num.Fit1D.get_max()
y = -(x-5)**2.0 + 1.0
spl = Spline(x,y)
xmax = spl.get_max()
ymax = spl(xmin)
np.testing.assert_almost_equal(xmax, 5.0)
np.testing.assert_almost_equal(ymax, 1.0)
def test_api():
x = np.linspace(-10, 10, 100)
y = (x - 5)**2.0 + 1.0
spl = Spline(x, y, k=2, s=0.1, eps=0.11)
for kw in ['k', 's']:
assert kw in list(spl.splrep_kwargs.keys())
assert spl.splrep_kwargs['k'] == 2
assert spl.splrep_kwargs['s'] == 0.1
# scalar
assert type(spl(1.0)) != type(np.array(1.0))
def test_min_max():
# min: num.Fit1D.get_min()
x = np.linspace(-10, 10, 100)
y = (x - 5)**2.0 + 1.0
spl = Spline(x, y)
xmin = spl.get_min()
ymin = spl(xmin)
np.testing.assert_almost_equal(xmin, 5.0)
np.testing.assert_almost_equal(ymin, 1.0)
# max: num.Fit1D.get_max()
y = -(x - 5)**2.0 + 1.0
spl = Spline(x, y)
xmax = spl.get_max()
ymax = spl(xmin)
np.testing.assert_almost_equal(xmax, 5.0)
np.testing.assert_almost_equal(ymax, 1.0)
def test_spline():
x = np.linspace(0,10,100)
y = np.sin(x)
spl = Spline(x,y)
assert (y - spl(x) < spl.eps).all()
assert (y - splev(x, spl.tck) < spl.eps).all()
assert not spl.is_mono()
np.testing.assert_array_almost_equal(spl(x,der=1), np.cos(x), decimal=4)
x = np.linspace(0,10,100)
y = x**2.0 - 5
spl = Spline(x,y)
assert spl.is_mono()
y = -(x**2.0 - 5)
spl = Spline(x,y)
assert spl.is_mono()
y0s = [5,0,-40]
xabs = [[0,2], [1,3], [6,8]]
x0s = [1,2,7]
# use bracket [x[0], x[-1]] for brentq()
for y0 in y0s:
np.testing.assert_almost_equal(y0, spl(spl.invsplev(y0)))
# use smaller bracket
for y0,xab in zip(y0s, xabs):
np.testing.assert_almost_equal(y0, spl(spl.invsplev(y0, xab=xab)))
# use start guess for newton()
for y0,x0 in zip(y0s, x0s):
np.testing.assert_almost_equal(y0, spl(spl.invsplev(y0, x0=x0)))
# root
np.testing.assert_almost_equal(spl.invsplev(0.0), spl.get_root())
def test_spline():
x = np.linspace(0, 10, 100)
y = np.sin(x)
spl = Spline(x, y)
assert (y - spl(x) < spl.eps).all()
assert (y - splev(x, spl.tck) < spl.eps).all()
assert not spl.is_mono()
np.testing.assert_array_almost_equal(spl(x, der=1), np.cos(x), decimal=4)
x = np.linspace(0, 10, 100)
y = x**2.0 - 5
spl = Spline(x, y)
assert spl.is_mono()
y = -(x**2.0 - 5)
spl = Spline(x, y)
assert spl.is_mono()
y0s = [5, 0, -40]
xabs = [[0, 2], [1, 3], [6, 8]]
x0s = [1, 2, 7]
# use bracket [x[0], x[-1]] for brentq()
for y0 in y0s:
np.testing.assert_almost_equal(y0, spl(spl.invsplev(y0)))
# use smaller bracket
for y0, xab in zip(y0s, xabs):
np.testing.assert_almost_equal(y0, spl(spl.invsplev(y0, xab=xab)))
# use start guess for newton()
for y0, x0 in zip(y0s, x0s):
np.testing.assert_almost_equal(y0, spl(spl.invsplev(y0, x0=x0)))
# root
np.testing.assert_almost_equal(spl.invsplev(0.0), spl.get_root())
