Profiler results

C:\Users\Jonas\Desktop> python -m line_profiler gadgetzan.py.lprof
Timer unit: 3.00191e-07 s

Total time: 8.63965 s
File: d:\github\odl\odl\util\phantom.py
Function: _phantom_3d at line 231

Line #      Hits         Time  Per Hit   % Time  Line Contents
==============================================================
   231                                           @profile
   232                                           def _phantom_3d(space, ellipses
):
   233                                               """Create a phantom in 3d s
pace.
   234
   235                                               Parameters
   236                                               ----------
   237                                               space : `DiscreteLp`
   238                                                   The space the phantom s
hould be generated in.
   239                                               ellipses : list of lists
   240                                                   Each row should contain
:
   241                                                   'value', 'axis_1', 'axi
s_2', 'axis_2',
   242                                                   'center_x', 'center_y',
 'center_z',
   243                                                   'rotation_phi', 'rotati
on_theta', 'rotation_psi'
   244                                                   The ellipses should be
contained the he rectangle
   245                                                   [-1, -1, -1] x [1, 1, 1
].
   246
   247                                               Returns
   248                                               -------
   249                                               phantom : `DisceteLpVector`

   250                                                   The phantom
   251                                               """
   252
   253                                               # Blank image
   254         1           35     35.0      0.0      p = np.zeros(space.size)
   255
   256                                               # Create the pixel grid
   257         1      3396097 3396097.0     11.8      points = space.points()
   258         1           41     41.0      0.0      minp = space.grid.min()
   259         1           16     16.0      0.0      maxp = space.grid.max()
   260
   261                                               # move points to [-1, 1]
   262         1       929777 929777.0      3.2      np.subtract(points, (minp +
 maxp)/2.0, out=points)
   263         1      1067001 1067001.0      3.7      np.divide(points, (maxp -
minp)/2.0, out=points)
   264         1           73     73.0      0.0      offset_points = np.empty_li
ke(points)
   265
   266                                               # move to [-1, 1]
   267
   268        11          222     20.2      0.0      for ellip in ellipses:
   269        10           56      5.6      0.0          I = ellip[0]
   270        10          146     14.6      0.0          a2 = ellip[1] ** 2
   271        10           55      5.5      0.0          b2 = ellip[2] ** 2
   272        10           49      4.9      0.0          c2 = ellip[3] ** 2
   273        10           42      4.2      0.0          x0 = ellip[4]
   274        10           42      4.2      0.0          y0 = ellip[5]
   275        10           43      4.3      0.0          z0 = ellip[6]
   276        10          138     13.8      0.0          phi = ellip[7] * np.pi
/ 180
   277        10           67      6.7      0.0          theta = ellip[8] * np.p
i / 180
   278        10           58      5.8      0.0          psi = ellip[9] * np.pi
/ 180
   279
   280                                                   # Create the offset x,y
 and z values for the grid
   281        10     11424022 1142402.2     39.7          np.subtract(points, [x
0, y0, z0], out=offset_points)
   282        10          307     30.7      0.0          scales = [1 / a2, 1 / b
2, 1 / c2]
   283
   284        10          138     13.8      0.0          if any([phi, theta, psi
]):
   285                                                       # Optimization, onl
y rotate if needed.
   286         2          117     58.5      0.0              cphi = np.cos(phi)
   287         2           30     15.0      0.0              sphi = np.sin(phi)
   288         2           15      7.5      0.0              ctheta = np.cos(the
ta)
   289         2           16      8.0      0.0              stheta = np.sin(the
ta)
   290         2           15      7.5      0.0              cpsi = np.cos(psi)
   291         2           14      7.0      0.0              spsi = np.sin(psi)
   292
   293         2           13      6.5      0.0              scales = [1 / a2, 1
 / b2, 1 / c2]
   294         2           25     12.5      0.0              mat = [[cpsi * cphi
 - ctheta * sphi * spsi,
   295         2           22     11.0      0.0                      cpsi * sphi
 + ctheta * cphi * spsi,
   296         2            8      4.0      0.0                      spsi * sthe
ta],
   297         2           14      7.0      0.0                     [-spsi * cph
i - ctheta * sphi * cpsi,
   298         2           10      5.0      0.0                      -spsi * sph
i + ctheta * cphi * cpsi,
   299         2            7      3.5      0.0                      cpsi * sthe
ta],
   300         2            6      3.0      0.0                     [stheta * sp
hi,
   301         2            8      4.0      0.0                      -stheta * c
phi,
   302         2           12      6.0      0.0                      ctheta]]
   303
   304         2      2214451 1107225.5      7.7              rotated = np.dot(m
at, offset_points.T)
   305         2       679724 339862.0      2.4              np.square(rotated,
out=rotated)
   306         2       925947 462973.5      3.2              radius = np.dot(sca
les, rotated)
   307                                                   else:
   308         8      2246663 280832.9      7.8              np.square(offset_po
ints.T, out=offset_points.T)
   309         8      3752264 469033.0     13.0              radius = np.dot(sca
les, offset_points.T)
   310
   311                                                   # Find the pixels withi
n the ellipse
   312        10      1034977 103497.7      3.6          inside = radius <= 1
   313
   314                                                   # Add the ellipse inten
sity to those pixels
   315        10      1100882 110088.2      3.8          p[inside] += I
   316
   317         1         6830   6830.0      0.0      return space.element(p)