pyssht 1.5.1

Fast spin spherical transforms

SSHT Python Documentation

This guide is intended to explain the python interface of SSHT. For a description of the workings of SSHT see here. The python package also offers an interface to some of the functionality from ducc0, including forward and inverse transforms.

pyssht.forward

flm = pyssht.forward(f, int L, Spin=0, Method='MW', Reality=False)

Performs the forward spherical harmonic transform.

Inputs

• f the signal on the sphere, numpy.ndarray type complex or real, ndim 2. NB different for 'MW_pole' sampling.
• L the band limit of the signal, non-zero positive integer
• Spin the spin of the signal, non-negative integer (default = 0)
• Method the sampling scheme used, string:
  1. 'MW' [McEwen & Wiaux sampling (default)]
  2. 'MW_pole' [McEwen & Wiaux sampling with the south pole as a separate double.]
  3. 'MWSS' [McEwen & Wiaux symmetric sampling]
  4. 'DH' [Driscoll & Healy sampling]
  5. 'GL' [Gauss-Legendre sampling]
• Reality determines if the signal is real or complex, Boolean (default = False)
• backend the backend that runs the transforms:
  1. 'SSHT' this package
  2. 'ducc' interface to ducc0. "MW_pole" is not available in this backend.
• nthreads: number of threads when calling into the 'ducc' backend. Ignored otherwise.

Output

flm the spherical harmonic transform of f, 1D numpy.ndarray type complex

Note on 'MW_pole' sampling

This is the same as the 'MW' sampling however the south pole is expressed as a double only not a vector. Therefore the size of the array is one smaller on the (\theta) direction. f is now a tuple containing either:

• (f_array, f_sp, phi_sp) if complex transform
• (f_array, f_sp) if real transform

pyssht.inverse

f = pyssht.inverse(np.ndarray[ double complex, ndim=1, mode="c"] f_lm not None, L, Spin=0, Method='MW', Reality=False)

Performs the inverse spherical harmonic transform.

Inputs

• flm the spherical harmonic transform of f, numpy.ndarray type complex, ndim 1
• L the band limit of the signal, non-zero positive integer
• Spin the spin of the signal, non-negative integer (default = 0)
• Method the sampling scheme used, string:
  1. 'MW' [McEwen & Wiaux sampling (default)]
  2. 'MW_pole' [McEwen & Wiaux sampling with the south pole as a separate double.]
  3. 'MWSS' [McEwen & Wiaux symmetric sampling]
  4. 'DH' [Driscoll & Healy sampling]
  5. 'GL' [Gauss-Legendre sampling]
• Reality determines if the signal is real or complex, Boolean (default = False)
• backend the backend that runs the transforms:
  1. 'SSHT' this package
  2. 'ducc' interface to ducc0. "MW_pole" is not available in this backend.
• nthreads: number of threads when calling into the 'ducc' backend. Ignored otherwise.

Output

f the signal on the sphere, 2D numpy.ndarray type complex or real. NB different for 'MW_pole' sampling.

Note on 'MW_pole' sampling

This is the same as the 'MW' sampling however the south pole is expressed as a double only not a vector. Therefore the size of the array is one smaller on the (\theta) direction. f is now a tuple containing either:

• (f_array, f_sp, phi_sp) if complex transform
• (f_array, f_sp) if real transform

pyssht.forward_adjoint

f = pyssht.forward_adjoint(np.ndarray[ double complex, ndim=1, mode="c"] f_lm not None, L, Spin=0, Method='MW', Reality=False)

Performs the adjoint of the forward spherical harmonic transform.

Inputs

• flm the spherical harmonic transform of f, numpy.ndarray type complex, ndim 1
• L the band limit of the signal, non-zero positive integer
• Spin the spin of the signal, non-negative integer (default = 0)
• Method the sampling scheme used, string:
  1. 'MW' [McEwen & Wiaux sampling (default)]
  2. 'MWSS' [McEwen & Wiaux symmetric sampling]
• Reality determines if the signal is real or complex, Boolean (default = False)
• backend the backend that runs the transforms:
  1. 'SSHT' this package
  2. 'ducc' interface to ducc0.
• nthreads: number of threads when calling into the 'ducc' backend. Ignored otherwise.

Output

f the signal on the sphere, 2D numpy.ndarray type complex or real.

pyssht.inverse_adjoint

flm = pyssht.inverse_adjoint(f, int L, Spin=0, Method='MW', Reality=False)

Performs the adjoint of the inverse spherical harmonic transform.

Inputs

• f the signal on the sphere, numpy.ndarray type complex or real, ndim 2.
• L the band limit of the signal, non-zero positive integer
• Spin the spin of the signal, non-negative integer (default = 0)
• Method the sampling scheme used, string:
  1. 'MW' [McEwen & Wiaux sampling (default)]
  2. 'MWSS' [McEwen & Wiaux symmetric sampling]
• Reality determines if the signal is real or complex, Boolean (default = False)
• backend the backend that runs the transforms:
  1. 'SSHT' this package
  2. 'ducc' interface to ducc0.
• nthreads: number of threads when calling into the 'ducc' backend. Ignored otherwise.

Output

flm the spherical harmonic transform of f, 1D numpy.ndarray type complex

pyssht.elm2ind

index = pyssht.elm2ind( int el, int m)

Computes the index in the flm array of a particular harmonic coefficient (\ell ) and (m).

Inputs

• el the scale parameter of the spherical harmonic coefficients, integer from (0) to (L-1), where (L) is the band limit.
• em the azimuthal parameter, integer from -el to el.

Output

Index of the coefficient in flm array, integer

pyssht.ind2elm

(el, em) = pyssht.ind2elm(int ind)

Computes harmonic coefficient (\ell ) and (m) from the index in the flm array.

Inputs

• ind index of the flm array

Output

Tuple containing (el, em)

• el the scale parameter of the spherical harmonic coefficients, integer from (0) to (L-1), where (L) is the band limit.
• em the azimuthal parameter, integer from -el to el.

pyssht.theta_to_index

p = pyssht.theta_to_index(double theta, int L, str Method="MW")

Outputs the (\theta) index (the first) in the 2 dimensional array used to store spherical images. The index returned is that of the closest (\theta) sample smaller then the angle given on input.

Inputs

• theta the angle (\theta)
• L the band limit of the signal, non-zero positive integer
• Method the sampling scheme used, string:
  1. 'MW' [McEwen & Wiaux sampling (default)]
  2. 'MW_pole' [McEwen & Wiaux sampling with the south pole as a separate double.]
  3. 'MWSS' [McEwen & Wiaux symmetric sampling]
  4. 'DH' [Driscoll & Healy sampling]
  5. 'GL' [Gauss-Legendre sampling]

Output

Int p of corresponding to the angle (\theta).

pyssht.phi_to_index

q = pyssht.phi_to_index(double phi, int L, str Method="MW")

Outputs the (\phi) index (the second) in the 2 dimensional array used to store spherical images. The index returned is that of the closest (\phi) sample smaller then the angle given on input.

Inputs

• phi the angle (\phi)
• L the band limit of the signal, non-zero positive integer
• Method the sampling scheme used, string:
  1. 'MW' [McEwen & Wiaux sampling (default)]
  2. 'MW_pole' [McEwen & Wiaux sampling with the south pole as a separate double.]
  3. 'MWSS' [McEwen & Wiaux symmetric sampling]
  4. 'DH' [Driscoll & Healy sampling]
  5. 'GL' [Gauss-Legendre sampling]

Output

Int q of corresponding to the angle (\phi).

pyssht.sample_length

n = pyssht.sample_length(int L, Method = 'MW')

Outputs a size of the array used for storing the data on the sphere for different sampling schemes.

Inputs

• L the band limit of the signal, non-zero positive integer
• Method the sampling scheme used, string:
  1. 'MW' [McEwen & Wiaux sampling (default)]
  2. 'MW_pole' [McEwen & Wiaux sampling with the south pole as a separate double.]
  3. 'MWSS' [McEwen & Wiaux symmetric sampling]
  4. 'DH' [Driscoll & Healy sampling]
  5. 'GL' [Gauss-Legendre sampling]

Output

Int n equal to the collapsed 1 dimensional size of the 2 dimensional array used to store data on the sphere.

pyssht.sample_shape

(n_theta, n_phi) = pyssht.sample_shape(int L, Method='MW')

Outputs a tuple with the shape of the array used for storing the data on the sphere for different sampling schemes.

Inputs

• L the band limit of the signal, non-zero positive integer
• Method the sampling scheme used, string:
  1. 'MW' [McEwen & Wiaux sampling (default)]
  2. 'MW_pole' [McEwen & Wiaux sampling with the south pole as a separate double.]
  3. 'MWSS' [McEwen & Wiaux symmetric sampling]
  4. 'DH' [Driscoll & Healy sampling]
  5. 'GL' [Gauss-Legendre sampling]

Output

Tuple containing (n_theta, n_phi)

• n_theta the number of samples in the (\theta) direction, integer
• n_phi the number of samples in the (\phi) direction, integer

pyssht.sample_positions

(thetas, phis) = pyssht.sample_positions(int L, Method = 'MW', Grid=False)

Computes the positions on the sphere of the samples.

Inputs

• L the band limit of the signal, non-zero positive integer
• Method the sampling scheme used, string:
  1. 'MW' [McEwen & Wiaux sampling (default)]
  2. 'MW_pole' [McEwen & Wiaux sampling with the south pole as a separate double.]
  3. 'MWSS' [McEwen & Wiaux symmetric sampling]
  4. 'DH' [Driscoll & Healy sampling]
  5. 'GL' [Gauss-Legendre sampling]
• Grid describes if the output is a vector of the sample positions or a 2D array the same shape as the signal on the sphere, default False

Outputs

Tuple containing (thetas, phis)

• thetas positions of the samples in the (\theta) direction
• phis positions of the samples in the (\theta) direction

pyssht.s2_to_cart

(x, y, z) = pyssht.s2_to_cart(theta, phi)

Computes the (x), (y), and (z) coordinates from (\theta) and (\phi) on the sphere.

Inputs

• theta (\theta) values, type numpy.ndarray
• phi (\phi) values, type numpy.ndarray

Output

Tuple containing (x, y, z)

• x the (x) coordinate of each point, type numpy.ndarray
• y the (y) coordinate of each point, type numpy.ndarray
• z the (z) coordinate of each point, type numpy.ndarray

pyssht.cart_to_s2

thetas, phis = cart_to_s2(x, y, z)

Computes the (\theta) and (\phi) on the coordinates on the sphere from (x), (y), and (z) coordinates.

Inputs

• x (x) values, type numpy.ndarray
• y (y) values, type numpy.ndarray
• z (z) values, type numpy.ndarray

Output

Tuple containing (theta, phi)

• theta the (\theta) coordinate of each point, type numpy.ndarray
• phi the (\phi) coordinate of each point, type numpy.ndarray

pyssht.spherical_to_cart

(x, y, z) = pyssht.spherical_to_cart(r, theta, phi)

Computes the (x), (y), and (z) coordinates from the spherical coordinates (r), (\theta) and (\phi).

Inputs

• r (r) values, type numpy.ndarray
• theta (\theta) values, type numpy.ndarray
• phi (\phi) values, type numpy.ndarray

Output

Tuple containing (x, y, z)

• x the (x) coordinate of each point, type numpy.ndarray
• y the (y) coordinate of each point, type numpy.ndarray
• z the (z) coordinate of each point, type numpy.ndarray

pyssht.cart_to_spherical

r, theta, phi = pyssht.cart_to_spherical(x, y, z)

Computes the (\r), (\theta) and (\phi) on the spherical coordinates from (x), (y), and (z) coordinates.

Inputs

• x (x) values, type numpy.ndarray
• y (y) values, type numpy.ndarray
• z (z) values, type numpy.ndarray

Output

Tuple containing (r, theta, phi)

• r the (r) coordinate of each point, type numpy.ndarray
• theta the (\theta) coordinate of each point, type numpy.ndarray
• phi the (\phi) coordinate of each point, type numpy.ndarray

pyssht.theta_phi_to_ra_dec

(dec, ra) = pyssht.theta_phi_to_ra_dec(theta, phi, Degrees=False)

Computes the Right Assension and declination from an array of (\theta) and (\phi) values.

Inputs

• theta (\theta) values, type numpy.ndarray
• phi (\phi) values, type numpy.ndarray
• Degrees defines if the output is in degrees or radians

Output

Tuple containing (dec, ra)

• dec the declination angle, numpy.ndarray
• ra the Right Assension, numpy.ndarray

pyssht.ra_dec_to_theta_phi

(theta, phi) = pyssht.ra_dec_to_theta_phi(ra, dec, Degrees=False)

Computes the (\theta) and (\phi) values from an array of Right Assension and declination values.

Inputs

• dec the declination angle, numpy.ndarray
• ra the Right Assension, numpy.ndarray
• Degrees defines if the input is in degrees or radians, if degrees they are converted

Output

Tuple containing (theta, phi])

• theta (\theta) values, type numpy.ndarray
• phi (\phi) values, type numpy.ndarray

pyssht.make_rotation_matrix

rot_much = pyssht.make_rotation_matrix(list rot)

Computes the 3 by 3 rotation matrix from the Euler angles given on input

Inputs

• rot List of length 3. Each element are the Euler angles [alpha, beta, gamma]

Output

3 by 3 rot_matrix the rotation matrix type ndarray dtype float

pyssht.rot_cart

x_p, y_p, z_p = pyssht.rot_cart(x, y, z, list rot)

Computes the rotations of the cartesian coordinates given a set of Euler angles. The inputs can be any shape ndarrays. For speed if the arrays are 1 or 2 dimensional it is recommended to use pyssht.rot_cart_1D or pyssht.rot_cart_2D.

Inputs

• x (x) values, type numpy.ndarray
• y (y) values, type numpy.ndarray
• z (z) values, type numpy.ndarray
• rot List of length 3. Each element are the Euler angles [alpha, beta, gamma]

Output

Tuple containing (x_p, y_p, z_p) the rotated coordinates the same shape and type as the inputs.

pyssht.rot_cart_1d and pyssht.rot_cart_2d

(x_p, y_p, z_p) = pyssht.rot_cart_1d(np.ndarray[np.float_t, ndim=1] x, np.ndarray[np.float_t, ndim=1] y, np.ndarray[np.float_t, ndim=1] z, list rot)

(x_p, y_p, z_p) = pyssht.rot_cart_2d(np.ndarray[np.float_t, ndim=2] x, np.ndarray[np.float_t, ndim=2] y, np.ndarray[np.float_t, ndim=2] z, list rot)

Computes the rotations of the cartesian coordinates given a set of Euler angles. The inputs can be any shape ndarrays. Same as pyssht.rot_cart except optimised for arrays that are 1 or 2 dimensional.

Inputs

• x (x) values, type numpy.ndarray, dtype float, ndim 1 or 2
• y (y) values, type numpy.ndarray, dtype float, ndim 1 or 2
• z (z) values, type numpy.ndarray, dtype float, ndim 1 or 2
• rot List of length 3. Each element are the Euler angles [alpha, beta, gamma]

Output

Tuple containing (x_p, y_p, z_p) the rotated coordinates the same shape and type as the inputs.

pyssht.plot_sphere

pyssht.plot_sphere(
    f, L, Method='MW', Close=True, Parametric=False,
    Parametric_Scaling=[0.0,0.5], Output_File=None,
    Show=True, Color_Bar=True, Units=None, Color_Range=None,
    Axis=True
)

Plots data on to a sphere. It is really slow and not very good!

Inputs

• f the signal on the sphere, numpy.ndarray type complex or real, ndim 2. NB different for 'MW_pole' sampling.
• L the band limit of the signal, non-zero positive integer
• Method the sampling scheme used, string:
  1. 'MW' [McEwen & Wiaux sampling (default)]
  2. 'MW_pole' [McEwen & Wiaux sampling with the south pole as a separate double.]
  3. 'MWSS' [McEwen & Wiaux symmetric sampling]
  4. 'DH' [Driscoll & Healy sampling]
  5. 'GL' [Gauss-Legendre sampling]
• Close if true the full sphere is plotted (without a gap after the last (\phi) position), default True
• Parametric the radius of the object at a certain point is defined by the function (not just the color), default False
• Parametric_Saling used if Parametric=True, defines the radius of the shape at a particular angle r = norm(f)*Parametric_Saling[1] + Parametric_Saling[0], default [0.0,0.5]
• Output_File if set saves the plot to a file of that name
• Show if True shows you the plot, default False
• Color_Bar if True shows the color bar, default True
• Units is set puts a label on the color bar
• Color_Range if set saturates the color bar in that range, else the function min and max is used
• Axis if True shows the 3d axis, default True

Outputs

None

pyssht.mollweide_projection

f_plot, mask = pyssht.mollweide_projection(
    f, L, resolution=500, rot=None,
    zoom_region=[np.sqrt(2.0)*2,np.sqrt(2.0)],
    Method="MW"
)

Creates an ndarray of the mollweide projection of a spherical image and a mask array. This is useful for plotting results, not to be used for analysis on the plane. Elements in the signal f that are NaNs are marked in the mask. This allows one to plot these regions the color of their choice.

Here is an example of using the function to plot real spherical data.

f_plot, mask = pyssht.mollweide_projection(f, L, Method="MW") # make projection
plt.figure() # start figure
imgplot = plt.imshow(f_real_plot,interpolation='nearest') # plot the projected image
plt.colorbar(imgplot,fraction=0.025, pad=0.04) # plot color bar (these extra keywords make the bar a reasonable size)
plt.imshow(mask_real, interpolation='nearest', cmap=cm.gray, vmin=-1., vmax=1.) # plot the NaN regions in grey
plt.gca().set_aspect("equal") # ensures the region is the correct proportions
plt.axis('off') # removes axis (looks better)
plt.show()

Inputs

• f the signal on the sphere, numpy.ndarray type complex or real, ndim 2.
• L the band limit of the signal, non-zero positive integer
• resolution size of the projected image, default 500
• rot If the image should be rotated before projecting, None. rot should be a list of length 1 or 3. If 1 then the image is rotated around the (z) axis by that amount. If 3 then the image is rotated by the Euler angles given in the list.
• zoom_region the region of the sphere to be plotted, default [np.sqrt(2.0)*2,np.sqrt(2.0)] is the full sphere.
• Method the sampling scheme used, string:
  1. 'MW' [McEwen & Wiaux sampling (default)]
  2. 'MWSS' [McEwen & Wiaux symmetric sampling]
  3. 'DH' [Driscoll & Healy sampling]
  4. 'GL' [Gauss-Legendre sampling]

Outputs

If the input is real:

Tuple containing:

• f_plot the projection of the image as a 2 dimensional ndarray of type float. Masked regions and regions not in the sphere are NaNs to make them clear when plotted
• mask the projection of the masked regions (NaNs in input f) as a 2 dimensional ndarray of type float. Masked regions have a value 0.0 and regions not in the sphere are NaNs to make them clear when plotted.

If the input is complex:

Tuple containing:

• f_plot_real the projection of the real part of the image as a 2 dimensional ndarray of type float. Masked regions and regions not in the sphere are NaNs to make them clear when plotted
• mask_real the projection of the masked regions (NaNs in input f.real) as a 2 dimensional ndarray of type float. Masked regions have a value 0.0 and regions not in the sphere are NaNs to make them clear when plotted.
• f_plot_imag the projection of the imaginary part of the image as a 2 dimensional ndarray of type float. Masked regions and regions not in the sphere are NaNs to make them clear when plotted
• mask_imag the projection of the masked regions (NaNs in input f.imag) as a 2 dimensional ndarray of type float. Masked regions have a value 0.0 and regions not in the sphere are NaNs to make them clear when plotted.

pyssht.equatorial_projection

f_proj_real, mask_real, (f_proj_imag, mask_imag)\
            = pyssht.equatorial_projection(f, int L, int resolution=500,\
             rot=None, list zoom_region=[-1,-1], str Method="MW", \
             str Projection="MERCATOR", int Spin=0)

Creates ndarrays of the projections of a spherical image and a mask array. This is useful for plotting results and performing analysis on the plane. All the spherical samples that fall in one planar pixel is averaged, if no samples fall in a pixel then the pixel is assigned the value of the closest spherical sample. Elements in the signal f that are NaNs are marked in the mask. This allows one to plot these regions the color of their choice.

There are two projections supported.

1. The Mercator projection often used in maps.
2. The Sinusoidal projection a simple equal area projection.

Here is an example of using the function to plot real spherical data using the Mercator projection.

f_proj, mask \
        = pyssht.equatorial_projection(f, L, resolution=500, Method="MW", \
        Projection="MERCATOR")
plt.figure() # start figure
imgplot = plt.imshow(f_proj,interpolation='nearest')# plot the projected image (north part)
plt.colorbar(imgplot) # plot color bar
plt.imshow(mask, interpolation='nearest', cmap=cm.gray, vmin=-1., vmax=1.) # plot the NaN regions in grey
plt.axis('off') # removes axis

plt.show()

Inputs

• f the signal on the sphere, numpy.ndarray type complex or real, ndim 2.
• L the band limit of the signal, non-zero positive integer
• resolution size of the projected image, default 500
• rot If the image should be rotated before projecting, None. rot should be a list of length 1 or 3. If 1 then the image is rotated around the (z) axis by that amount. If 3 then the image is rotated by the Euler angles given in the list.
• zoom_region the region of the sphere to be plotted in radians. The first element is the angle left and right of the centre, default is np.pi for both projections. The second element is up and down of the equator, default is np.pi/2 for the Sinusoidal projection and 7*np.pi/16 for the Mercator projection.
• Method the sampling scheme used, string:
  1. 'MW' [McEwen & Wiaux sampling (default)]
  2. 'MWSS' [McEwen & Wiaux symmetric sampling]
  3. 'DH' [Driscoll & Healy sampling]
  4. 'GL' [Gauss-Legendre sampling]
• Projection string describing which of the projections to use. Use "MERCATOR" for the Mercator projection and "SINE" for the Sinusoidal projection, default is "MERCATOR"
• Spin the spin of the signal. If the signal has non-zero spin then on projection the signal must be rotated to account for the changing direction of the definition of the signal. By setting this to a non-zero integer will ensure this rotation is performed.

Outputs

If the input is real:

Tuple containing:

• f_plot the projection of the image as a 2 dimensional ndarray of type float. Masked regions and regions not in the sphere are NaNs to make them clear when plotted
• mask the projection of the masked regions (NaNs in input f) as a 2 dimensional ndarray of type float. Masked regions have a value 0.0 and regions not in the sphere are NaNs to make them clear when plotted.

If the input is complex:

Tuple containing:

• f_plot_real the projection of the real part of the image as a 2 dimensional ndarray of type float. Masked regions and regions not in the sphere are NaNs to make them clear when plotted
• mask_real the projection of the real part of the masked regions (NaNs in input f) as a 2 dimensional ndarray of type float. Masked regions have a value 0.0 and regions not in the sphere are NaNs to make them clear when plotted.
• f_plot_imag the projection of the imaginary part of the image as a 2 dimensional ndarray of type float. Masked regions and regions not in the sphere are NaNs to make them clear when plotted
• mask_imag the projection of the imaginary part of the masked regions (NaNs in input f) as a 2 dimensional ndarray of type float. Masked regions have a value 0.0 and regions not in the sphere are NaNs to make them clear when plotted.

pyssht.polar_projection

f_proj_north_real, mask_north_real, f_proj_south_real, mask_south_real,\
(f_proj_north_imag, mask_north_imag, f_proj_south_imag, mask_south_imag\ )
        = pyssht.polar_projection(f, int L, int resolution=500, rot=None,\
        float zoom_region=-1, str Method="MW", str Projection="OP",int Spin=0):

Creates an two ndarrays of the polar projection of a spherical image and a mask array. This is useful for plotting results and performing analysis on the plane. All the spherical samples that fall in one planar pixel is averaged, if no samples fall in a pixel then the pixel is assigned the value of the closest spherical sample. Elements in the signal f that are NaNs are marked in the mask. This allows one to plot these regions the color of their choice.

All the projections are centred around a pole. There are three projections supported.

1. The Gnomic projection, defined by drawing a line from the centre of the circle trough the sphere on to the plane.
2. The Stereographic projection, defined by drawing a line starting at the opposite pole through the sphere to the plane.
3. The Orthographic, defined by a vertical projection on the plane.

Here is an example of using the function to plot real spherical data.

f_proj_north, mask_north, f_proj_south, mask_south \
        = pyssht.polar_projection(f, L, resolution=500, Method="MW", \
        Projection="OP")
plt.figure() # start figure
imgplot = plt.imshow(f_proj_north,interpolation='nearest')# plot the projected image (north part)
plt.colorbar(imgplot) # plot color bar
plt.imshow(mask_north, interpolation='nearest', cmap=cm.gray, vmin=-1., vmax=1.) # plot the NaN regions in grey
plt.axis('off') # removes axis (looks better)

plt.figure()
imgplot = plt.imshow(f_proj_south,interpolation='nearest')# plot the projected image (south part)
plt.colorbar(imgplot)
plt.imshow(mask_south, interpolation='nearest', cmap=cm.gray, vmin=-1., vmax=1.)
plt.title("orthographic projection south")
plt.axis('off')
plt.show()

Inputs

• f the signal on the sphere, numpy.ndarray type complex or real, ndim 2.
• L the band limit of the signal, non-zero positive integer
• resolution size of the projected image, default 500
• rot If the image should be rotated before projecting, default None. rot should be a list of length 1 or 3. If 1 then the image is rotated around the (z) axis by that amount. If 3 then the image is rotated by the Euler angles given in the list.
• zoom_region the region of the sphere to be plotted in radians, default np.pi/2 is the full half sphere for the orthographic and stereographic projections and np.pi/4 for the gnomic projection as the equator is at infinity in this projection.
• Method the sampling scheme used, string:
  1. 'MW' [McEwen & Wiaux sampling (default)]
  2. 'MWSS' [McEwen & Wiaux symmetric sampling]
  3. 'DH' [Driscoll & Healy sampling]
  4. 'GL' [Gauss-Legendre sampling]
• Projection string describing which of the projections to use. Use "GP" for the Gnomic projection, "SP" for the Stereographic projection and "OP" for the Orthographic projection, default is "OP"
• Spin the spin of the signal. If the signal has non-zero spin then on projection the signal must be rotated to account for the changing direction of the definition of the signal. By setting this to a non-zero integer will ensure this rotation is performed.

Outputs

If the input is real:

Tuple containing:

• f_plot_north the projection of the north part of the image as a 2 dimensional ndarray of type float. Masked regions and regions not in the sphere are NaNs to make them clear when plotted
• mask_north the projection of the north part of the masked regions (NaNs in input f) as a 2 dimensional ndarray of type float. Masked regions have a value 0.0 and regions not in the sphere are NaNs to make them clear when plotted.
• f_plot_south the projection of the south part of the image as a 2 dimensional ndarray of type float. Masked regions and regions not in the sphere are NaNs to make them clear when plotted
• mask_south the projection of the south part of the masked regions (NaNs in input f) as a 2 dimensional ndarray of type float. Masked regions have a value 0.0 and regions not in the sphere are NaNs to make them clear when plotted.

If the input is complex:

Tuple containing:

• f_plot_north_real the projection of the north part of the real part of the image as a 2 dimensional ndarray of type float. Masked regions and regions not in the sphere are NaNs to make them clear when plotted
• mask_north_real the projection of the north part of the real part of the masked regions (NaNs in input f) as a 2 dimensional ndarray of type float. Masked regions have a value 0.0 and regions not in the sphere are NaNs to make them clear when plotted.
• f_plot_south_real the projection of the south part of the real part of the image as a 2 dimensional ndarray of type float. Masked regions and regions not in the sphere are NaNs to make them clear when plotted
• mask_south_real the projection of the south part of the real part of the masked regions (NaNs in input f) as a 2 dimensional ndarray of type float. Masked regions have a value 0.0 and regions not in the sphere are NaNs to make them clear when plotted.
• f_plot_north_imag the projection of the north part of the imaginary part of the image as a 2 dimensional ndarray of type float. Masked regions and regions not in the sphere are NaNs to make them clear when plotted
• mask_north_imag the projection of the north part of the imaginary part of the masked regions (NaNs in input f) as a 2 dimensional ndarray of type float. Masked regions have a value 0.0 and regions not in the sphere are NaNs to make them clear when plotted.
• f_plot_south_imag the projection of the south part of the imaginary part of the image as a 2 dimensional ndarray of type float. Masked regions and regions not in the sphere are NaNs to make them clear when plotted
• mask_south_imag the projection of the south part of the imaginary part of the masked regions (NaNs in input f) as a 2 dimensional ndarray of type float. Masked regions have a value 0.0 and regions not in the sphere are NaNs to make them clear when plotted.

pyssht.dl_beta_recurse

dl = pyssht.dl_beta_recurse(np.ndarray[ double, ndim=2, mode="c"] dl not None,\
            double beta, int L, int el, \
            np.ndarray[ double, ndim=1, mode="c"] sqrt_tbl not None,\
            np.ndarray[ double, ndim=1, mode="c"] signs not None)

Compute the el-th plane of the Wigner small-d functions (from the (el-1)-th plane) using Risbo's method.

Inputs

• dl the Wigner plane for all m and n, indexed dl[m][n] of size (2L-1)(2*L-1)
• beta angle to calculate Wigner D matrix at, type double
• L the band limit of the signal, non-zero positive integer
• el is the current harmonic degree (i.e. dl input should already be computed for el-1, and dl output will be computed for el)
• sqrt_tbl precomputed square-roots from (0) to (2*(L-1)+1)
• signs precomputed ((-1)^m) signs from (m=0) to (L)

Outputs

Numpy ndarray dl, type float_ of the Wigner plane for all m and n

pyssht.dln_beta_recurse

dl = pyssht.dln_beta_recurse(np.ndarray[ double, ndim=1, mode="c"] dl not None,\
            np.ndarray[ double, ndim=1, mode="c"] dlm1 not None, double beta,\
            int L, int el, int n, np.ndarray[ double, ndim=1, mode="c"] sqrt_tbl not None,\
            np.ndarray[ double, ndim=1, mode="c"] signs not None)

Compute the el-th line of the Wigner small-d functions for given n (from the (el-1)-th and (el-2)-th lines) using 3-term recursion of Kostelec.

Inputs

• dl the Wigner line for el for non-negative m and given n of size L
• dlm1 is the line for el-1 and dlp1 is the line computed for el+1
• beta angle to calculate Wigner D matrix at, type double
• L the band limit of the signal, non-zero positive integer
• el el is the current harmonic degree
• n the third index in Wigner D matrices
• sqrt_tbl precomputed square-roots from (0) to (2*(L-1)+1)
• signs precomputed ((-1)^m) signs from (m=0) to (L)

Outputs

Numpy ndarray dl, type float_ the Wigner line for el for non-negative m and given n of size

pyssht.generate_dl

dl_array = pyssht.generate_dl(double beta, int L)

Generates the small Wigner D matrices up to a given band limit for a given (\beta)

Inputs

• beta angle to calculate Wigner D matrix at, type double
• L the band limit of the signal, non-zero positive integer

Outputs

Numpy ndarray dl_array, type float_ of the small Wigner D matrices

pyssht.rotate_flms

flm_rotated = pyssht.rotate_flms(
                np.ndarray[ double complex, ndim=1, mode="c"] f_lm not None,\
                double alpha, double beta, double gamma, int L, dl_array=None,\
                M=None, Axisymmetric=False, Keep_dl=False)

Function to rotate a set of spherical harmonic coefficients by the set of Euler angles (\alpha, \beta, \gamma ) using the (z,y,z) convention.

Inputs

• flm the spherical harmonic transform of f, numpy.ndarray type complex, ndim 1
• alpha rotation angle (\alpha), type double
• beta rotation angle (\beta), type double
• gamma rotation angle (\gamma), type double
• L the band limit of the signal, non-zero positive integer
• dl_array if set should be the precomputed small Wigner D matrix for angle (\beta) and harmonic band limit L. If not set this is calculated in the function. (This parameter is ignored when using the ducc backend.)
• M if set is the azimuthal band limit of the function to be rotated, default M=L.
• Axisymmetric set if the function is axisymmetric and axisymmetric harmonic coefficients are parsed.
• Keep_dl if set the output is changed to allow one to keep the computed dl_array. (This parameter is ignored when using the ducc backend.)
• backend the backend that runs the transforms:
  1. 'SSHT' this package
  2. 'ducc' interface to ducc0

Output

If Keep_dl is not set the output is the rotated set of spherical harmonic coefficients. If it is the output is a tuple (flm_rotated, dl_array), ie the rotated harmonic coefficients and the small Wigner D matrix computed for that band limit and (\alpha) value.

pyssht.guassian_smoothing

fs_lm = pyssht.guassian_smoothing(np.ndarray[ double complex, ndim=1, mode="c"] f_lm not None, int L, sigma_in=None, bl_in = None)

Smooths a set of harmonic coefficients either with a precomputed smoothing kernel bl or with a Gaussian given on input.

Inputs

• f_lm the spherical harmonic transform of f, numpy.ndarray type complex, ndim 1
• L the band limit of the signal, non-zero positive integer
• sigma_in the input sigma of the Gaussian to smooth the signal with, default None
• bl_in the smoothing kernel to smooth the signal with, default None

Output

fs_lm the smoothed harmonic coefficients.

pyssht.create_ylm

ylm = pyssht.create_ylm(thetas, phis, int L, int Spin=0, str recursion='Kostelec')

Computes spherical harmonic functions for all el and all 0<=|m|<= el using various recursions.

Inputs

• thetas positions of the samples in the (\theta) direction
• phis positions of the samples in the (\phi) direction
• L the band limit of the signal, non-zero positive integer
• Spin the spin of the signal, non-negative integer (default = 0)
• recursion the recursion scheme used, string:
  1. 'Kostelec' [3-term recursion, e.g. Kostelec (default)]
  2. 'Risbo' [Risbo recursion]
  3. 'NumericalRecipes' [Numerical Recipes]

Output

ylm the spherical harmonics indexed ylm[ind][theta][phi], where ind = pyssht.elm2ind(el, m)