def from_tree_transform(cls, node, ctx):
return functional_models.Gaussian2D(amplitude=node['amplitude'],
x_mean=node['x_mean'],
y_mean=node['y_mean'],
x_stddev=node['x_stddev'],
y_stddev=node['y_stddev'],
theta=node['theta'])
def give_gauss(image_in, x_peak, y_peak):
gauss_init = functional_models.Gaussian2D(amplitude=image_in.max(),\
x_mean=x_peak, \
y_mean=y_peak, \
bounds={'amplitude':(np.max(image_in)/1.5,\
2*np.max(image_in) )} )
#If the amplitude is not bounded, it sometimes finds "negative" amplitude fitting...
return fitter(gauss_init, fit_x, fit_y, image_in)
def set_xy_weights(self, centroid=[520, 460]):
# We can weight the events based on their distance from the source.
# It's fine to just use region filtering instead, but in the case of really high count rates this might be useful.
H, xcenters, ycenters = self.make_image(plot=False)
g_init = functional_models.Gaussian2D(amplitude=np.max(H), x_mean=centroid[0], y_mean=centroid[1],\
x_stddev=10.0, y_stddev=10.0)
fit_g = fitting.LevMarLSQFitter()
g = fit_g(g_init, xcenters, ycenters, H)
self.xy_weights = g(self.x, self.y) / g.amplitude
return self.xy_weights
def test__identical_to_astropy_gaussian_model__include_different_ellipticity_from_x_and_y_stddev(
self,
):
pixel_scales = 0.1
x_stddev = (
3.0e-5
* (units.deg).to(units.arcsec)
/ pixel_scales
/ (2.0 * np.sqrt(2.0 * np.log(2.0)))
)
y_stddev = (
2.0e-5
* (units.deg).to(units.arcsec)
/ pixel_scales
/ (2.0 * np.sqrt(2.0 * np.log(2.0)))
)
theta_deg = 0.0
theta = Angle(theta_deg, "deg").radian
gaussian_astropy = functional_models.Gaussian2D(
amplitude=1.0,
x_mean=2.0,
y_mean=2.0,
x_stddev=x_stddev,
y_stddev=y_stddev,
theta=theta,
)
shape = (5, 5)
y, x = np.mgrid[0 : shape[1], 0 : shape[0]]
kernel_astropy = gaussian_astropy(x, y)
kernel_astropy /= np.sum(kernel_astropy)
kernel = aa.Kernel2D.from_as_gaussian_via_alma_fits_header_parameters(
shape_native=shape,
pixel_scales=pixel_scales,
y_stddev=2.0e-5,
x_stddev=3.0e-5,
theta=theta_deg,
normalize=True,
)
assert kernel_astropy == pytest.approx(kernel.native, 1e-4)
def test__identical_to_astropy_gaussian_model__circular_no_rotation_different_pixel_scale(
self,
):
pixel_scales = 0.02
x_stddev = (
2.0e-5
* (units.deg).to(units.arcsec)
/ pixel_scales
/ (2.0 * np.sqrt(2.0 * np.log(2.0)))
)
y_stddev = (
2.0e-5
* (units.deg).to(units.arcsec)
/ pixel_scales
/ (2.0 * np.sqrt(2.0 * np.log(2.0)))
)
gaussian_astropy = functional_models.Gaussian2D(
amplitude=1.0,
x_mean=2.0,
y_mean=2.0,
x_stddev=x_stddev,
y_stddev=y_stddev,
theta=0.0,
)
shape = (5, 5)
y, x = np.mgrid[0 : shape[1], 0 : shape[0]]
kernel_astropy = gaussian_astropy(x, y)
kernel_astropy /= np.sum(kernel_astropy)
kernel = aa.Kernel2D.from_as_gaussian_via_alma_fits_header_parameters(
shape_native=shape,
pixel_scales=pixel_scales,
y_stddev=2.0e-5,
x_stddev=2.0e-5,
theta=0.0,
normalize=True,
)
assert kernel_astropy == pytest.approx(kernel.native, 1e-4)
Max_x = 0
Max_y = 0
Val_sum = 0
for i in range(0, m - 1):
for j in range(0, n - 1):
Val_sum += z[i][j] #Find Sum to calculate Mean
if (z[i][j] > Max):
Max = z[i][j]
Max_x = i
Max_y = j
Mean = Val_sum / (float)(m * n)
print Mean
# Fit the data using astropy.modeling
p_init = M.Gaussian2D(Max, Max_x, Max_y, 1, 10)
#p_init = M.Ring2D(210,Max_x,Max_y,0,30)
f = fitting.LinearLSQFitter()
p = f(p_init, x, y, z)
# Plot the data with the best-fit model
plt.figure(figsize=(8, 2.5))
plt.subplot(1, 4, 1)
plt.imshow(z, interpolation='nearest', vmin=0, vmax=Max)
plt.title("Data")
plt.subplot(1, 4, 2)
plt.imshow(p(x, y), interpolation='nearest', vmin=0, vmax=Max)
plt.title("Model")
plt.subplot(1, 4, 3)