def itau(temp,
dens,
line,
n_min=5,
n_max=1000,
other='',
verbose=False,
value='itau',
location=LOCALDIR):
"""
Gives the integrated optical depth for a given temperature and density.
It assumes that the background radiation field dominates the continuum emission.
The emission measure is unity. The output units are Hz.
:param temp: Electron temperature. Must be a string of the form '8d1'.
:type temp: string
:param dens: Electron density.
:type dens: float
:param line: Line to load models for.
:type line: string
:param n_min: Minimum n value to include in the output. Default 1
:type n_min: int
:param n_max: Maximum n value to include in the output. Default 1500, Maximum allowed value 9900
:type n_max: int
:param other: String to search for different radiation fields and others.
:type other: string
:param verbose: Verbose output?
:type verbose: bool
:param value: ['itau'|'bbnMdn'|'None'] Value to output. itau will output the integrated optical depth. \
bbnMdn will output the :math:`\\beta_{n,n^{\\prime}}b_{n}` times the oscillator strenght :math:`M(\\Delta n)`. \
None will output the :math:`\\beta_{n,n^{\\prime}}b_{n}` values.
:type value: string
:returns: The principal quantum number and its asociated value.
"""
t = str2val(temp)
d = float(dens)
dn = fc.set_dn(line)
mdn_ = mdn(dn)
bbn = load_betabn(temp, dens, other, line, verbose, location=location)
nimin = best_match_indx(n_min, bbn[:, 0])
nimax = best_match_indx(n_max, bbn[:, 0])
n = bbn[nimin:nimax, 0]
b = bbn[nimin:nimax, 1]
if value == 'itau':
i = itau_norad(n, t, b, dn, mdn_)
elif value == 'bbnMdn':
i = b * dn * mdn_
else:
i = b
return n, i
def kappa_line(Te, ne, nion, Z, Tr, trans, n_max=1500):
"""
Computes the line absorption coefficient for CRRLs between levels :math:`n_{i}` and :math:`n_{f}`, :math:`n_{i}>n_{f}`.
This can only go up to :math:`n_{\\rm{max}}` 1500 because of the tables used for the Einstein Anm coefficients.
:param Te: Electron temperature of the gas. (K)
:type Te: float
:param ne: Electron density. (:math:`\\mbox{cm}^{-3}`)
:type ne: float
:param nion: Ion density. (:math:`\\mbox{cm}^{-3}`)
:type nion: float
:param Z: Electric charge of the atoms being considered.
:type Z: int
:param Tr: Temperature of the radiation field felt by the gas. This specifies the temperature of the field at 100 MHz. (K)
:type Tr: float
:param trans: Transition for which to compute the absorption coefficient.
:type trans: string
:param n_max: Maximum principal quantum number to include in the output.
:type n_max: int<1500
:returns:
:rtype: array
"""
cte = np.power(c, 2.) / (16. * np.pi) * np.power(
np.power(h, 2) / (2. * np.pi * m_e * k_B), 3. / 2.)
bn = load_bn(val2str(Te), ne, other='case_diffuse_{0}'.format(val2str(Tr)))
bn = bn[:np.where(bn[:, 0] == n_max)[0]]
Anfni = np.loadtxt('{0}/rates/einstein_Anm_{1}.txt'.format(
LOCALDIR, trans))
# Cut the Einstein Amn coefficients table to match the bn values
i_bn_i = best_match_indx(bn[0, 0], Anfni[:, 1])
i_bn_f = best_match_indx(bn[-1, 0], Anfni[:, 0])
Anfni = Anfni[i_bn_i:i_bn_f + 1]
ni = Anfni[:, 0]
nf = Anfni[:, 1]
omega_ni = 2 * np.power(ni, 2)
omega_i = 1.
xi_ni = xi(ni, Te, Z)
xi_nf = xi(nf, Te, Z)
exp_ni = np.exp(xi_ni.value)
exp_nf = np.exp(xi_nf.value)
#print len(Anfni), len(bn[1:,-1]), len(bn[:-1,-1]), len(omega_ni[:]), len(ni), len(exp_ni), len(exp_nf)
kl = cte.value / np.power(
Te, 3. / 2.) * ne * nion * Anfni[:, 2] * omega_ni[:] / omega_i * (
bn[1:, -1] * exp_ni - bn[:-1, -1] * exp_nf)
return kl
def get_line_mask(freq, reffreq, v0, dv):
"""
Return a mask with ranges where a line is expected in the given frequency range for \
a line with a given reference frequency at expected velocity v0 and line width dv0.
:param freq: Frequency axis where the line is located.
:type freq: numpy array or list
:param reffreq: Reference frequency for the line.
:type reffreq: float
:param v0: Velocity of the line.
:type v0: float, km/s
:param dv: Velocity range to mask.
:type dv: float, km/s
:returns: Mask centered at the line center and width `dv0` referenced to the input `freq`.
"""
f0 = vel2freq(reffreq, v0 * 1e3)
df0 = dv2df(reffreq * 1e6, dv0 * 1e3)
df = abs(freq[0] - freq[1])
f0_indx = utils.best_match_indx(f0, freq, df / 2.0)
mindx0 = f0_indx - df0 / df / 1e6
mindxf = f0_indx + df0 / df / 1e6
return [mindx0, mindxf]
def load_bn(te,
ne,
tr='',
ncrit='1.5d3',
n_min=5,
n_max=1000,
verbose=False,
location=LOCALDIR):
"""
Loads the bn values from the CRRL models.
:param te: Electron temperature of the model.
:type te: string
:param ne: Electron density of the model.
:type ne: string
:param other: Radiation field of the model or any other string with model characteristics.
:type other: string
:param verbose: Verbose output?
:type verbose: bool
:returns: The :math:`b_{n}` value for the given model conditions.
:rtype: array
"""
#LOCALDIR = os.path.dirname(os.path.realpath(__file__))
if tr == '-' or tr == '' or tr == 0:
model_file = 'Carbon_opt_T_{0}_ne_{1}_ncrit_{2}_vriens_delta_500_vrinc_nmax_9900_dat'.format(
te, ne, ncrit)
if verbose:
print("Loading {0}".format(model_file))
else:
model_file = 'Carbon_opt_T_{0}_ne_{1}_ncrit_{2}_{3}_vriens_delta_500_vrinc_nmax_9900_dat'.format(
te, ne, ncrit, tr)
if verbose:
print("Loading {0}".format(model_file))
model_path = glob.glob('{0}/{1}'.format(location, model_file))[0]
if verbose:
print("Loaded {0}".format(model_path))
bn = np.loadtxt(model_path)
nimin = best_match_indx(n_min, bn[:, 0])
nimax = best_match_indx(n_max, bn[:, 0])
bn = bn[nimin:nimax + 1]
return bn
def load_bn_h(te, ne, other='', n_min=5, n_max=1000, verbose=False):
"""
Loads the bn values from the HRRL models.
:param te: Electron temperature of the model.
:type te: string
:param ne: Electron density of the model.
:type ne: string
:param other: Radiation field of the model or any other string with model characteristics.
:type other: string
:param verbose: Verbose output?
:type verbose: bool
:returns: The :math:`b_{n}` value for the given model conditions.
:rtype: array
"""
#LOCALDIR = os.path.dirname(os.path.realpath(__file__))
if other == '-' or other == '':
mod_file = 'H_bn2/Hydrogen_opt_T_{1}_ne_{2}_ncrit_8d2_vriens_delta_500_vrinc_nmax_9900_dat'.format(
LOCALDIR, te, ne)
if verbose:
print("Loading {0}".format(mod_file))
mod_file = glob.glob(
'{0}/H_bn2/Hydrogen_opt_T_{1}_ne_{2}*_ncrit_8d2_vriens_delta_500_vrinc_nmax_9900_dat'
.format(LOCALDIR, te, ne))[0]
else:
mod_file = 'H_bn2/Hydrogen_opt_T_{1}_ne_{2}_ncrit_8d2_{3}_vriens_delta_500_vrinc_nmax_9900_dat'.format(
LOCALDIR, te, ne, other)
if verbose:
print("Loading {0}".format(mod_file))
mod_file = glob.glob(
'{0}/H_bn2/Hydrogen_opt_T_{1}_ne_{2}*_ncrit_8d2_{3}_vriens_delta_500_vrinc_nmax_9900_dat'
.format(LOCALDIR, te, ne, other))[0]
if verbose:
print("Loaded {0}".format(mod_file))
bn = np.loadtxt(mod_file)
nimin = best_match_indx(n_min, bn[:, 0])
nimax = best_match_indx(n_max, bn[:, 0])
bn = bn[nimin:nimax + 1]
return bn
def mask_cube_(vel, data, vel_rngs, offset=10.):
"""
"""
logger = logging.getLogger(__name__)
vel_indx = np.empty(vel_rngs.shape, dtype=int)
mdata = deepcopy(data)
mdata = np.ma.masked_invalid(mdata)
for i, velrng in enumerate(vel_rngs):
vel_indx[i][0] = utils.best_match_indx(velrng[0], vel)
vel_indx[i][1] = utils.best_match_indx(velrng[1], vel)
logger.info('Channels selected to be masked: {0} -- {1}'.format(
vel_indx[i][0], vel_indx[i][1] + 1))
mdata[vel_indx[i][0]:vel_indx[i][1] + 1].mask = True
return mdata
def mask_cube(vel, data, vel_rngs, offset=10.):
"""
"""
logger = logging.getLogger(__name__)
vel_indx = np.empty(vel_rngs.shape, dtype=int)
nvel_indx = np.empty(vel_rngs.shape, dtype=int)
chns = np.zeros(len(vel_rngs))
nchns = np.zeros(len(vel_rngs))
nvel = []
extend = nvel.extend
logger.info("Velocity ranges for masking: ", vel_rngs)
for i, velrng in enumerate(vel_rngs):
vel_indx[i][0] = utils.best_match_indx(velrng[0], vel)
vel_indx[i][1] = utils.best_match_indx(velrng[1], vel)
nvel.extend(vel[vel_indx[i][0]:vel_indx[i][1] + 1])
nvel_indx[i][0] = utils.best_match_indx(velrng[0], nvel)
nvel_indx[i][1] = utils.best_match_indx(velrng[1], nvel)
chns[i] = vel_indx[i][1] - vel_indx[i][0] + 1
nchns[i] = nvel_indx[i][1] - nvel_indx[i][0] + 1
mdata = np.ones(((int(sum(chns)), ) + data.shape[1:])) * offset
logger.info('Masked data shape: {0}'.format(mdata.shape))
logger.info('Will select the unmasked data.')
for i in range(len(vel_indx)):
mdata[nvel_indx[i][0]:nvel_indx[i][1] +
1] = data[vel_indx[i][0]:vel_indx[i][1] + 1]
return nvel, mdata
def pressure_broad_coefs(Te):
"""
Defines the values of the constants :math:`a` and :math:`\\gamma` that go into the collisional broadening formula
of Salgado et al. (2017).
:param Te: Electron temperature.
:type Te: float
:returns: The values of :math:`a` and :math:`\\gamma`.
:rtype: list
"""
te = [
10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 200, 300, 400, 500, 600, 700,
800, 900, 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000,
20000, 30000
]
te_indx = utils.best_match_indx(Te, te)
a = [
-10.974098, -10.669695, -10.494541, -10.370271, -10.273172, -10.191374,
-10.124309, -10.064037, -10.010153, -9.9613006, -9.6200366, -9.4001678,
-9.2336349, -9.0848840, -8.9690170, -8.8686695, -8.7802238, -8.7012421,
-8.6299908, -8.2718376, -8.0093937, -7.8344941, -7.7083367, -7.6126791,
-7.5375720, -7.4770500, -7.4272885, -7.3857095, -7.1811733, -7.1132522
]
gammac = [
5.4821631, 5.4354009, 5.4071360, 5.3861013, 5.3689105, 5.3535398,
5.3409679, 5.3290318, 5.3180304, 5.3077770, 5.2283700, 5.1700702,
5.1224893, 5.0770049, 5.0408369, 5.0086342, 4.9796105, 4.9532071,
4.9290080, 4.8063682, 4.7057576, 4.6356118, 4.5831746, 4.5421547,
4.5090104, 4.4815675, 4.4584053, 4.4385507, 4.3290786, 4.2814240
]
a_func = interpolate.interp1d(te,
a,
kind='linear',
bounds_error=True,
fill_value=0.0)
g_func = interpolate.interp1d(te,
gammac,
kind='linear',
bounds_error=True,
fill_value=0.0)
return [a_func(Te), g_func(Te)]
def lookup_freq(n, line):
"""
Returns the frequency of a line given the transition number n.
:param n: Principal quantum number to look up for.
:type n: int
:param line: Line for which the frequency is desired.
:type line: string
:returns: Frequency of line(n).
:rtype: float
"""
qns, freqs = load_ref(line)
indx = utils.best_match_indx(n, qns)
return freqs[indx]
def get_line_mask2(freq, reffreq, dv):
"""
Return a mask with ranges where a line is expected in the given frequency range for \
a line with a given reference frequency and line width dv.
:param freq: Frequency axis where the line is located.
:type freq: numpy array or list
:param reffreq: Reference frequency for the line.
:type reffreq: float
:param dv: Velocity range to mask.
:type dv: float, km/s
:returns: Mask centered at the line center and width `dv0` referenced to the input `freq`.
"""
df = dv2df(reffreq, dv * 1e3)
df_chan = utils.get_min_sep(freq)
f0_indx = utils.best_match_indx(reffreq, np.asarray(freq))
f_mini = int(f0_indx - df / df_chan)
if f_mini < 0:
f_mini = 0
f_maxi = int(f0_indx + df / df_chan)
return [f_mini, f_maxi]
def stack_cubes(cubes,
outfits,
vmax,
vmin,
dv,
weight_list=None,
v_axis=3,
overwrite=False,
algo='channel'):
"""
"""
logger = logging.getLogger(__name__)
cubel = cubes #glob.glob(cubes)
logger.debug(cubel)
if dv == 0:
logger.info('Velocity width not specified.'),
logger.info('Will try to determine from input data.')
for i, cube in enumerate(cubel):
hdu = fits.open(cube)
head = hdu[0].header
x = crrls.get_axis(head, v_axis)
if i == 0:
dv = utils.get_min_sep(x)
vmax_min = max(x)
vmin_max = min(x)
else:
dv = max(dv, utils.get_min_sep(x))
vmax_min = min(vmax_min, max(x))
vmin_max = max(vmin_max, min(x))
logger.debug('Cube: {0}'.format(cube))
logger.debug('Cube velocity limits: {0} {1} {2}'.format(
min(x), max(x), utils.get_min_sep(x)))
logger.info('Will use a velocity width of {0}'.format(dv))
# Check velocity ranges to avoid latter crashes.
if vmax_min < vmax:
logger.info('Requested maximum velocity is larger '\
'than one of the cubes velocity axis.')
#logger.info('Conflicting cube: {0}'.format(cube))
logger.info('v_max={0}, v_max_min={1}'.format(vmax, vmax_min))
logger.info('Will now exit')
# sys.exit(1)
if vmin_max > vmin:
logger.info('Requested minimum velocity is smaller '\
'than one of the cubes velocity axis.')
#logger.info('Conflicting cube: {0}'.format(cube))
logger.info('v_min={0}, v_min_max={1}'.format(vmin, vmin_max))
logger.info('Will now exit')
# sys.exit(1)
shape = hdu[0].shape
if len(shape) > 3:
logger.info('Will drop first axis.')
s = 1
else:
s = 0
nvaxis = np.arange(vmin, vmax + dv, dv)
stack = np.ma.zeros((len(nvaxis), ) + shape[s + 1:])
covrg = np.zeros((len(nvaxis), ) + shape[s + 1:])
# Check if there is a weight list
if weight_list:
try:
wl = np.loadtxt(weight_list,
dtype=[('fits', np.str_, 256), ('w', np.float64)])
except ValueError:
wl = np.loadtxt(weight_list,
dtype=[('fits', np.str_, 256),
('w', np.str_, 256)])
weight = np.ma.zeros((len(nvaxis), ) + shape[s + 1:])
logger.info('Output stack will have dimensions: {0}'.format(stack.shape))
for i, cube in enumerate(cubel):
logger.info('Adding cube {0}/{1}'.format(i, len(cubel) - 1))
# Load the data
hdu = fits.open(cube)
head = hdu[0].header
data = np.ma.masked_invalid(hdu[0].data)
# Check the weight
if weight_list:
w = wl['w'][np.where(wl['fits'] == cube)]
else:
w = 1
logger.info('Will use a weight of {0}.'.format(w))
try:
aux_weight = np.ones(stack.shape) * w
except TypeError:
w = np.ma.masked_invalid(fits.open(w[0])[0].data)
#aux_weight = np.ones(shape[s+1:])*w
if len(data.shape) > 3:
logger.info('Will drop first axis.')
data = data[0]
# Get the cube axes
ra = crrls.get_axis(head, 1)
de = crrls.get_axis(head, 2)
ve = crrls.get_axis(head, v_axis)
logger.info('RA axis limits: {0} {1}'.format(min(ra), max(ra)))
logger.info('DEC axis limits: {0} {1}'.format(min(de), max(de)))
logger.info('VEL axis limits: {0} {1}'.format(min(ve), max(ve)))
vmin_idx = utils.best_match_indx(vmin, ve)
vmax_idx = utils.best_match_indx(vmax, ve)
[vmin_idx, vmax_idx] = sorted([vmin_idx, vmax_idx])
if vmin_idx > 0: vmin_idx -= 1
if vmax_idx < len(ve): vmax_idx += 1
data_ = data[vmin_idx:vmax_idx]
ve_ = ve[vmin_idx:vmax_idx]
# Check that the axes are in ascending order
if ve_[0] > ve_[1]:
vs = -1
else:
vs = 1
if ra[0] > ra[1]:
vr = -1
else:
vr = 1
# Interpolate the data
interp = RegularGridInterpolator((ve_[::vs], de, ra[::vr]),
data_[::vs, :, ::vr])
# Add the data to the stack
if 'vector' in algo.lower():
logger.info('Will use one vector to reconstruct the cube')
pts = np.array([[nvaxis[k], de[j], ra[i]] \
for k in range(len(nvaxis)) \
for j in range(len(de)) \
for i in range(len(ra))])
aux_weight = np.ones(stack.shape) * w
stack += interp(pts).reshape(stack.shape) * aux_weight
elif 'channel' in algo.lower():
logger.info('Will reconstruct the cube one channel at a time')
for k in range(len(nvaxis)):
pts = np.array([[nvaxis[k], de[j], ra[i]] \
for j in range(len(de)) \
for i in range(len(ra[::vr]))])
newshape = shape[s + 1:] # Only spatial dimensions
aux_weight = np.ones(newshape) * w
aux_cov = np.ones(newshape)
try:
stack_ = np.ma.masked_invalid(
interp(pts).reshape(newshape) * aux_weight)
stack_.fill_value = 0.
aux_weight[stack_.mask] = 0
aux_cov[stack_.mask] = 0
weight[k] += aux_weight
covrg[k] += aux_cov
stack[k] += stack_.filled()
except ValueError:
logger.info('Point outside range: ' \
'vel={0}, ra={1}..{2}, dec={3}..{4}'.format(pts[0,0],
min(pts[:,2]),
max(pts[:,2]),
min(pts[:,1]),
max(pts[:,1])))
else:
logger.info('Cube reconstruction algorithm unrecognized.')
logger.info('Will exit now.')
sys.exit(1)
# Divide by the number of input cubes to get the mean
stack = stack / weight
stack.fill_value = np.nan
weight.fill_value = np.nan
# Write to a fits file
hdulist = fits.PrimaryHDU(stack.filled())
# Copy the header from the first channel image
hdulist.header = fits.open(cubel[0])[0].header.copy()
hdulist.header['CTYPE3'] = 'VELO'
hdulist.header['CRVAL3'] = nvaxis[0]
hdulist.header['CDELT3'] = dv
hdulist.header['CRPIX3'] = 1
hdulist.header['CUNIT3'] = 'm/s'
hdulist.writeto(outfits, overwrite=overwrite)
stack_head = hdulist.header.copy()
hdulist = fits.PrimaryHDU(weight.filled())
hdulist.header = stack_head
hdulist.header['CTYPE3'] = 'VELO'
hdulist.header['CRVAL3'] = nvaxis[0]
hdulist.header['CDELT3'] = dv
hdulist.header['CRPIX3'] = 1
hdulist.header['CUNIT3'] = 'm/s'
hdulist.writeto(outfits.split('.fits')[0] + '_weight.fits',
overwrite=overwrite)
hdulist = fits.PrimaryHDU(covrg)
hdulist.header = stack_head
hdulist.header['CTYPE3'] = 'VELO'
hdulist.header['CRVAL3'] = nvaxis[0]
hdulist.header['CDELT3'] = dv
hdulist.header['CRPIX3'] = 1
hdulist.header['CUNIT3'] = 'm/s'
hdulist.writeto(outfits.split('.fits')[0] + '_coverage.fits',
overwrite=overwrite)
def stack_interpol(specs, output, vmax, vmin, dv, x_col, y_col, weight, weight_list=None, weight_list_cols='0,1'):
"""
"""
logger = logging.getLogger(__name__)
#specs = glob.glob(spec)
# If only one file is passed, it probably contains the list
#if len(specs) == 1:
#specs = np.genfromtxt(specs[0], dtype=str)
# Setup weight list columns, if need be.
if weight_list:
wlc0,wlc1 = list(map(int, weight_list_cols.split(',')))
print(wlc0,wlc1)
logger.info('Will use the following files as input: ')
logger.info(specs)
logger.info('The stack consists of {0} lines'.format(len(specs)))
# Determine velocity resolution of stack.
if dv == 0:
logger.info('Will determine the velocity resolution of the stack.')
for i,s in enumerate(specs):
data = np.loadtxt(s)
x = data[:,x_col]
if i == 0:
dv = utils.get_min_sep(x)
else:
dv = max(dv, utils.get_min_sep(x))
logger.info('Spectrum {0} has velocity resolution of: {1}'.format(s, utils.get_min_sep(x)))
logger.info('Stack velocity resolution: {0}'.format(dv))
# Setup the grid for the stack.
xgrid = np.arange(vmin, vmax, dv)
ygrid = np.zeros(len(xgrid)) # the temperatures
zgrid = np.zeros(len(xgrid)) # the number of tb points in every stacked channel
for i,s in enumerate(specs):
logger.info('Working on file: {0}'.format(s))
data = np.loadtxt(s)
x = data[:,x_col]
y = data[:,y_col]
# Sort by velocity
o = x.argsort()
x = x[o]
y = y[o]
# Catch NaNs and invalid values:
mask_x = np.ma.masked_equal(x, -9999).mask
mask_y = np.isnan(y)
mask = mask_x | mask_y
# Interpolate non masked ranges indepently.
my = np.ma.masked_where(mask, y)
mx = np.ma.masked_where(mask, x)
valid = np.ma.flatnotmasked_contiguous(mx)
y_aux = np.zeros(len(xgrid))
if not isinstance(valid, slice):
for j,rng in enumerate(valid):
#print "slice {0}: {1}".format(i, rng)
if len(x[rng]) > 1:
interp_y = interpolate.interp1d(x[rng], y[rng],
kind='linear',
bounds_error=False,
fill_value=0.0)
y_aux += interp_y(xgrid)
elif not np.isnan(x[rng]):
#print x[rng]
y_aux[utils.best_match_indx(x[rng], xgrid)] += y[rng]
else:
#print "slice: {0}".format(valid)
interp_y = interpolate.interp1d(x[valid], y[valid],
kind='linear',
bounds_error=False,
fill_value=0.0)
y_aux += interp_y(xgrid)
# Check which channels have data.
ychan = [1 if ch != 0 else 0 for ch in y_aux]
# Determine the weight.
w = None
if not weight:
w = np.ones(len(xgrid))
elif weight == 'list':
wl = np.loadtxt(weight_list, dtype=str)
try:
w = float(wl[:,wlc1][np.where(wl[:,wlc0] == s)[0][0]])
except IndexError:
logger.error("Weight for spectrum {} not found in {} .".format(s, weight_list))
logger.error("Will not include it in the stack.")
continue
elif weight == 'sigma':
w = 1./crrls.get_rms(my)
elif weight == 'sigma2':
w = 1./np.power(crrls.get_rms(my), 2)
logger.info('Will use a weight of: {0}'.format(w))
# Stack!
zgrid = zgrid + np.multiply(ychan, w)
ygrid = ygrid + y_aux*np.multiply(ychan, w)
# Divide by the total weight to preserve optical depth
ygrid = np.divide(ygrid, zgrid)
logger.info('Saving stack to: {0}'.format(output))
np.savetxt(output, np.c_[xgrid, ygrid, zgrid], header="x axis, " \
"stacked y axis, " \
"y axis weight")
