def __init__(self, data, err=None, parameters=dict()):
""" TODO """
self.data = data
self.err = err
self.parameters = parameters
nparticles = self.data.shape[0]
default_priors = np.zeros_like(self.data)
default_priors[:,0] = np.zeros(nparticles) - np.log(2*np.pi)
default_priors[:,1] = np.log(np.cos(data[:,1])/2)
# distance range: 1-200 kpc
default_priors[:,2] = -np.log(np.log(200.)) - np.log(data[:,2])
p = NormalPrior(0., 0.306814 / data[:,2]) # 300 km/s at d
default_priors[:,3] = p.logpdf(data[:,3])
default_priors[:,4] = p.logpdf(data[:,4])
p = NormalPrior(0., 0.306814) # 300 km/s
default_priors[:,5] = p.logpdf(data[:,5])
self.default_priors = default_priors
def __init__(self, data, err=None, parameters=dict()):
""" TODO """
self.data = data
self.err = err
self.parameters = parameters
nparticles = self.data.shape[0]
default_priors = np.zeros_like(self.data)
default_priors[:, 0] = np.zeros(nparticles) - np.log(2 * np.pi)
default_priors[:, 1] = np.log(np.cos(data[:, 1]) / 2)
# distance range: 1-200 kpc
default_priors[:, 2] = -np.log(np.log(200.)) - np.log(data[:, 2])
p = NormalPrior(0., 0.306814 / data[:, 2]) # 300 km/s at d
default_priors[:, 3] = p.logpdf(data[:, 3])
default_priors[:, 4] = p.logpdf(data[:, 4])
p = NormalPrior(0., 0.306814) # 300 km/s
default_priors[:, 5] = p.logpdf(data[:, 5])
self.default_priors = default_priors
def __init__(self, lbd, pmv, units,
lbd_err=None, pmv_err=None, frozen=list()):
l,b,d = lbd
mul,mub,vr = pmv
# validate input quantities
shape = None
coords = [l,b,d,mul,mub,vr]
for c in coords:
if not hasattr(c, 'unit'):
raise TypeError("Coordinate and velocity information must be Astropy"
"Quantity objects.")
if shape is None:
shape = c.shape
else:
if c.shape != shape:
raise ValueError("All input coordinates must have same shape.")
# validate frozen parameter names
for name in frozen:
if name not in _names:
raise ValueError("Parameter name '{}' is not a valid coordinate. Valid"
" names are: {}".format(name,','.join(_names)))
# validate input uncertainties
if lbd_err is None:
lbd_err = [None, None, None]
l_err,b_err,d_err = lbd_err
if pmv_err is None:
pmv_err = [None, None, None]
mul_err,mub_err,vr_err = lbd_err
errs = [l_err,b_err,d_err,mul_err,mub_err,vr_err]
for name,err in zip(_names, errs):
if name in frozen:
continue
if err is None:
raise ValueError("Uncertainty is None for '{}' but that parameter"
" is not frozen.".format(name))
if not hasattr(err, 'unit'):
raise TypeError("Coordinate and velocity uncertainties must be Astropy"
"Quantity objects.")
# containers
n = l.shape[0]
self.data = np.ones((n,6))*np.nan
self.err = np.ones((n,6))*np.nan
# TODO: right now this makes the assumption that each coordinate has Gaussian,
# uncorrelated uncertainties?
for i,name,c,err in zip(range(6),_names,coords,errs):
self.data[:,i] = c.decompose(units).value
if err is None:
self.err[:,i] = np.zeros(n)
else:
self.err[:,i] = err.decompose(units).value
# priors on coordinates
default_priors = np.zeros_like(self.data)
default_priors[:,0] = np.zeros(n) - np.log(2*np.pi)
default_priors[:,1] = np.log(np.cos(self.data[:,1])/2)
# distance range: 1-100 kpc
default_priors[:,2] = -np.log(np.log(100.)) - np.log(self.data[:,2])
p = NormalPrior(0., 0.306814 / self.data[:,2]) # 300 km/s at d
default_priors[:,3] = p.logpdf(self.data[:,3])
default_priors[:,4] = p.logpdf(self.data[:,4])
p = NormalPrior(0., 0.306814) # 300 km/s
default_priors[:,5] = p.logpdf(self.data[:,5])
self.default_priors = default_priors
