def test_unique_columns(self):
"""Make sure unique_columns gives the expected answers"""
a = numpy.array([[1, 1, 1, 0, 0, 0],
[0, 1, 1, 1, 0, 0],
[0, 1, 1, 1, 0, 0],
[1, 1, 1, 0, 0, 0],
[1, 1, 1, 1, 1, 0]])
ans1 = numpy.array([[0, 1, 1, 1, 0, 0],
[1, 1, 1, 0, 0, 0],
[1, 1, 1, 1, 1, 0]])
ans0 = numpy.array([[0, 0, 0, 0, 0],
[0, 0, 0, 0, 1],
[0, 1, 1, 0, 1],
[1, 0, 0, 1, 1],
[1, 1, 1, 1, 1]])
numpy.testing.assert_array_equal(ans1, tb.unique_columns(a, axis=1))
numpy.testing.assert_array_equal(ans0, tb.unique_columns(a, axis=0))
def test_unique_columns(self):
"""Make sure unique_columns gives the expected answers"""
a = numpy.array([[1, 1, 1, 0, 0, 0],
[0, 1, 1, 1, 0, 0],
[0, 1, 1, 1, 0, 0],
[1, 1, 1, 0, 0, 0],
[1, 1, 1, 1, 1, 0]])
ans1 = numpy.array([[0, 1, 1, 1, 0, 0],
[1, 1, 1, 0, 0, 0],
[1, 1, 1, 1, 1, 0]])
ans0 = numpy.array([[0, 0, 0, 0, 0],
[0, 0, 0, 0, 1],
[0, 1, 1, 0, 1],
[1, 0, 0, 1, 1],
[1, 1, 1, 1, 1]])
numpy.testing.assert_array_equal(ans1, tb.unique_columns(a, axis=1))
numpy.testing.assert_array_equal(ans0, tb.unique_columns(a, axis=0))
def simpleSpectrogram(*args, **kwargs):
"""
Plot a spectrogram given Z or X,Y,Z. This is a wrapper around pcolormesh()
that can handle Y being a 2d array of time dependent bins. Like in the
Van Allen Probes HOPE and MagEIS data files.
Parameters
==========
*args : 1 or 3 arraylike
Call Signatures::
simpleSpectrogram(Z, **kwargs)
simpleSpectrogram(X, Y, Z, **kwargs)
Other Parameters
================
zlog : bool
Plot the color with a log colorbar (default: True)
ylog : bool
Plot the Y axis with a log scale (default: True)
alpha : scalar (0-1)
The alpha blending value (default: None)
cmap : string
The name of the colormap to use (default: system default)
vmin : float
Minimum color value (default: Z.min(), if log non-zero min)
vmax : float
Maximum color value (default: Z.max())
ax : matplotlib.axes
Axes to plot the spectrogram on (default: None - new axes)
cb : bool
Plot a colorbar (default: True)
cbtitle : string
Label to go on the colorbar (default: None)
Returns
=======
ax : matplotlib.axes._subplots.AxesSubplot
Matplotlib axes object that the plot is on
"""
if len(args) not in [1,3]:
raise(TypeError("simpleSpectrogram, takes Z or X, Y, Z"))
if len(args) == 1: # just Z in, makeup an X and Y
Z = np.ma.masked_invalid(np.asarray(args[0])) # make sure not a dmarray or VarCopy
X = np.arange(Z.shape[0]+1)
Y = np.arange(Z.shape[1]+1)
else:
# we really want X and Y to be one larger than Z
X = np.asarray(args[0]) # make sure not a dmarray or VarCopy
Y = np.asarray(args[1]) # make sure not a dmarray or VarCopy
Z = np.ma.masked_invalid(np.asarray(args[2]))
if X.shape[0] == Z.shape[0]: # same length, expand X
X = tb.bin_center_to_edges(X) # hopefully evenly spaced
if len(Y.shape) == 1: # 1d, just use as axis
Y = tb.bin_center_to_edges(Y) # hopefully evenly spaced
elif len(Y.shape) == 2:
Y_orig = Y
# 2d this is time dependent and thus need to overplot several
Y = tb.unique_columns(Y, axis=1)
if Y.shape[1] == Z.shape[1]: # hopefully evenly spaced
Y_uniq = Y
Y_tmp = np.empty((Y.shape[0], Y.shape[1]+1), dtype=Y.dtype)
for ii in range(Y.shape[0]):
Y_tmp[ii] = tb.bin_center_to_edges(Y[ii])
Y = Y_tmp
# deal with all the default keywords
zlog = kwargs.pop('zlog', True)
ylog = kwargs.pop('ylog', True)
alpha = kwargs.pop('alpha', None)
cmap = kwargs.pop('cmap', None)
vmin = kwargs.pop('vmin', np.min(Z) if not zlog else np.min(Z[np.nonzero(Z)]))
vmax = kwargs.pop('vmax', np.max(Z))
ax = kwargs.pop('ax', None)
cb = kwargs.pop('cb', True)
cbtitle = kwargs.pop('cbtitle', None)
# the first case is that X, Y are 1d and Z is 2d, just make the plot
if ax is None:
fig = plt.figure()
ax = fig.add_subplot(111)
else:
fig = ax.get_figure()
Z = Z.filled(0)
if len(Y.shape) > 1:
for yy in Y_uniq:
# which indices in X have these values
ind = (yy == Y_orig).all(axis=1)
print(Y_orig[ind][0].min(), Y_orig[ind][0].max())
Y_tmp = np.zeros_like(Y_orig)
Y_tmp[ind] = Y_orig[ind][0]
Z_tmp = np.zeros_like(Z)
Z_tmp[ind] = Z[ind]
pc = ax.pcolormesh(X, Y_tmp[ind][0], Z_tmp.T,
norm=LogNorm(vmin=vmin, vmax=vmax), cmap=cmap,
alpha=alpha)
else:
pc = ax.pcolormesh(X, Y, Z.T, norm=LogNorm(vmin=vmin, vmax=vmax),
cmap=cmap,
alpha=alpha)
if ylog: ax.set_yscale('log')
if cb: # add a colorbar
cb_ = fig.colorbar(pc)
cb_.set_label(cbtitle)
return ax
def simpleSpectrogram(*args, **kwargs):
"""
Plot a spectrogram given Z or X,Y,Z. This is a wrapper around pcolormesh()
that can handle Y being a 2d array of time dependent bins. Like in the
Van Allen Probes HOPE and MagEIS data files.
Parameters
==========
*args : 1 or 3 arraylike
Call Signatures::
simpleSpectrogram(Z, **kwargs)
simpleSpectrogram(X, Y, Z, **kwargs)
Other Parameters
================
zlog : bool
Plot the color with a log colorbar (default: True)
ylog : bool
Plot the Y axis with a log scale (default: True)
alpha : scalar (0-1)
The alpha blending value (default: None)
cmap : string
The name of the colormap to use (default: system default)
vmin : float
Minimum color value (default: Z.min(), if log non-zero min)
vmax : float
Maximum color value (default: Z.max())
ax : matplotlib.axes
Axes to plot the spectrogram on (default: None - new axes)
cb : bool
Plot a colorbar (default: True)
cbtitle : string
Label to go on the colorbar (default: None)
Returns
=======
ax : matplotlib.axes._subplots.AxesSubplot
Matplotlib axes object that the plot is on
"""
if len(args) not in [1, 3]:
raise (TypeError("simpleSpectrogram, takes Z or X, Y, Z"))
if len(args) == 1: # just Z in, makeup an X and Y
Z = np.ma.masked_invalid(np.asarray(
args[0])) # make sure not a dmarray or VarCopy
X = np.arange(Z.shape[0] + 1)
Y = np.arange(Z.shape[1] + 1)
else:
# we really want X and Y to be one larger than Z
X = np.asarray(args[0]) # make sure not a dmarray or VarCopy
Y = np.asarray(args[1]) # make sure not a dmarray or VarCopy
Z = np.ma.masked_invalid(np.asarray(args[2]))
if X.shape[0] == Z.shape[0]: # same length, expand X
X = tb.bin_center_to_edges(X) # hopefully evenly spaced
if len(Y.shape) == 1: # 1d, just use as axis
if Y.shape[0] == Z.shape[1]: # same length, expand Y
Y = tb.bin_center_to_edges(Y) # hopefully evenly spaced
elif len(Y.shape) == 2:
Y_orig = Y
# 2d this is time dependent and thus need to overplot several
Y = tb.unique_columns(Y, axis=1)
if Y.shape[1] == Z.shape[1]: # hopefully evenly spaced
Y_uniq = Y
Y_tmp = np.empty((Y.shape[0], Y.shape[1] + 1), dtype=Y.dtype)
for ii in range(Y.shape[0]):
Y_tmp[ii] = tb.bin_center_to_edges(Y[ii])
Y = Y_tmp
# deal with all the default keywords
zlog = kwargs.pop('zlog', True)
ylog = kwargs.pop('ylog', True)
alpha = kwargs.pop('alpha', None)
cmap = kwargs.pop('cmap', None)
vmin = kwargs.pop('vmin',
np.min(Z) if not zlog else np.min(Z[np.nonzero(Z)]))
vmax = kwargs.pop('vmax', np.max(Z))
ax = kwargs.pop('ax', None)
cb = kwargs.pop('cb', True)
cbtitle = kwargs.pop('cbtitle', None)
# the first case is that X, Y are 1d and Z is 2d, just make the plot
if ax is None:
fig = plt.figure()
ax = fig.add_subplot(111)
else:
fig = ax.get_figure()
Z = Z.filled(0)
if len(Y.shape) > 1:
for yy in Y_uniq:
# which indices in X have these values
ind = (yy == Y_orig).all(axis=1)
print(Y_orig[ind][0].min(), Y_orig[ind][0].max())
Y_tmp = np.zeros_like(Y_orig)
Y_tmp[ind] = Y_orig[ind][0]
Z_tmp = np.zeros_like(Z)
Z_tmp[ind] = Z[ind]
pc = ax.pcolormesh(X,
Y_tmp[ind][0],
Z_tmp.T,
norm=LogNorm(vmin=vmin, vmax=vmax),
cmap=cmap,
alpha=alpha)
else:
pc = ax.pcolormesh(X,
Y,
Z.T,
norm=LogNorm(vmin=vmin, vmax=vmax),
cmap=cmap,
alpha=alpha)
if ylog: ax.set_yscale('log')
if cb: # add a colorbar
cb_ = fig.colorbar(pc)
if cbtitle is not None:
cb_.set_label(cbtitle)
return ax
