import warnings
import numpy as np
import pandas as pd
__all__ = ['Orbits']
[docs]class Orbits(object):
"""Orbits reads, checks for required values, and stores orbit parameters for moving objects.
Instantiate the class and then use readOrbits or setOrbits to set the orbit values.
self.orbits stores the orbital parameters, as a pandas dataframe.
self.dataCols defines the columns required, although objId, H, g, and sed_filename are optional.
"""
def __init__(self):
self.orbits = None
self.orb_format = None
# Specify the required columns/values in the self.orbits dataframe.
# Which columns are required depends on self.orb_format.
self.dataCols = {}
self.dataCols['COM'] = ['objId', 'q', 'e', 'inc', 'Omega', 'argPeri',
'tPeri', 'epoch', 'H', 'g', 'sed_filename']
self.dataCols['KEP'] = ['objId', 'a', 'e', 'inc', 'Omega', 'argPeri',
'meanAnomaly', 'epoch', 'H', 'g', 'sed_filename']
self.dataCols['CAR'] = ['objId', 'x', 'y', 'z', 'xdot', 'ydot', 'zdot',
'epoch', 'H', 'g', 'sed_filename']
def __len__(self):
return len(self.orbits)
def __getitem__(self, i):
orb = Orbits()
orb.setOrbits(self.orbits.iloc[i])
return orb
def __iter__(self):
for i, orbit in self.orbits.iterrows():
orb = Orbits()
orb.setOrbits(orbit)
yield orb
def __eq__(self, otherOrbits):
if isinstance(otherOrbits, Orbits):
if self.orb_format != otherOrbits.orb_format:
return False
for col in self.dataCols[self.orb_format]:
if col == 'objId':
continue
if not self.orbits[col].equals(otherOrbits.orbits[col]):
return False
else:
return True
else:
return False
def __neq__(self, otherOrbits):
if self == otherOrbits:
return False
else:
return True
[docs] def setOrbits(self, orbits):
"""Set and validate orbital parameters contain all required values.
Sets self.orbits and self.orb_format.
If objid is not present in orbits, a sequential series of integers will be used.
If H is not present in orbits, a default value of 20 will be used.
If g is not present in orbits, a default value of 0.15 will be used.
If sed_filename is not present in orbits, either C or S type will be assigned,
according to the semi-major axis value.
Parameters
----------
orbits : pandas.DataFrame, pandas.Series or numpy.ndarray
Array-like object containing orbital parameter information.
"""
# Do we have a single item or multiples?
if isinstance(orbits, pd.Series):
# Passed a single SSO in Series, convert to a DataFrame.
orbits = pd.DataFrame([orbits])
elif isinstance(orbits, np.ndarray):
# Passed a numpy array, convert to DataFrame.
orbits = pd.DataFrame.from_records(orbits)
elif isinstance(orbits, np.record):
# This was a single object in a numpy array and we should be a bit fancy.
orbits = pd.DataFrame.from_records([orbits], columns=orbits.dtype.names)
if 'index' in orbits:
del orbits['index']
nSso = len(orbits)
# Error if orbits is empty (this avoids hard-to-interpret error messages from pyoorb).
if nSso == 0:
raise ValueError('Length of the orbits dataframe was 0.')
# Discover which type of orbital parameters we have on disk.
self.orb_format = None
if 'FORMAT' in orbits:
if ~(orbits['FORMAT'] == orbits['FORMAT'].iloc[0]).all():
raise ValueError('All orbital elements in the set should have the same FORMAT.')
self.orb_format = orbits['FORMAT'].iloc[0]
# Backwards compatibility .. a bit. CART is deprecated, so swap it to CAR.
if self.orb_format == 'CART':
self.orb_format = 'CAR'
del orbits['FORMAT']
# Check that the orbit format is approximately right.
if self.orb_format == 'COM':
if 'q' not in orbits:
raise ValueError('The stated format was COM, but "q" not present in orbital elements?')
if self.orb_format == 'KEP':
if 'a' not in orbits:
raise ValueError('The stated format was KEP, but "a" not present in orbital elements?')
if self.orb_format == 'CAR':
if 'x' not in orbits:
raise ValueError('The stated format was CAR but "x" not present in orbital elements?')
if self.orb_format is None:
# Try to figure out the format, if it wasn't provided.
if 'q' in orbits:
self.orb_format = 'COM'
elif 'a' in orbits:
self.orb_format = 'KEP'
elif 'x' in orbits:
self.orb_format = 'CAR'
else:
raise ValueError("Can't determine orbital type, as neither q, a or x in input orbital elements.\n"
"Was attempting to base orbital element quantities on header row, "
"with columns: \n%s" % orbits.columns)
# Check that the orbit epoch is within a 'reasonable' range, to detect possible column mismatches.
general_epoch = orbits['epoch'].head(1).values[0]
# Look for epochs between 1800 and 2200 - this is primarily to check if people used MJD (and not JD).
expect_min_epoch = -21503.
expect_max_epoch = 124594.
if general_epoch < expect_min_epoch or general_epoch > expect_max_epoch:
raise ValueError("The epoch detected for this orbit is odd - %f. "
"Expecting a value between %.1f and %.1f (MJD!)" % (general_epoch,
expect_min_epoch,
expect_max_epoch))
# If these columns are not available in the input data, auto-generate them.
if 'objId' not in orbits:
orbits['objId'] = np.arange(0, nSso, 1)
if 'H' not in orbits:
orbits['H'] = np.zeros(nSso) + 20.0
if 'g' not in orbits:
orbits['g'] = np.zeros(nSso) + 0.15
if 'sed_filename' not in orbits:
orbits['sed_filename'] = self.assignSed(orbits)
# Make sure we gave all the columns we need.
for col in self.dataCols[self.orb_format]:
if col not in orbits:
raise ValueError('Missing required orbital element %s for orbital format type %s'
% (col, self.orb_format))
# Check to see if we have duplicates.
if len(orbits['objId'].unique()) != nSso:
warnings.warn('There are duplicates in the orbit objId values' +
' - was this intended? (continuing).')
# All is good.
self.orbits = orbits
[docs] def assignSed(self, orbits, randomSeed=None):
"""Assign either a C or S type SED, depending on the semi-major axis of the object.
P(C type) = 0 (a<2); 0.5*a - 1 (2<a<4); 1 (a > 4),
based on figure 23 from Ivezic et al 2001 (AJ, 122, 2749).
Parameters
----------
orbits : pandas.DataFrame, pandas.Series or numpy.ndarray
Array-like object containing orbital parameter information.
Returns
-------
numpy.ndarray
Array containing the SED type for each object in 'orbits'.
"""
# using fig. 23 from Ivezic et al. 2001 (AJ, 122, 2749),
# we can approximate the sed types with a simple linear form:
# p(C) = 0 for a<2
# p(C) = 0.5*a-1 for 2<a<4
# p(C) = 1 for a>4
# where a is semi-major axis, and p(C) is the probability that
# an asteroid is C type, with p(S)=1-p(C) for S types.
if 'a' in orbits:
a = orbits['a']
elif 'q' in orbits:
a = orbits['q'] / (1 - orbits['e'])
elif 'x' in orbits:
# This definitely isn't right, but it's a placeholder to make it work for now.
a = np.sqrt(orbits['x']**2 + orbits['y']**2 + orbits['z']**2)
else:
raise ValueError('Need either a or q (plus e) in orbit data frame.')
if not hasattr(self, "_rng"):
if randomSeed is not None:
self._rng = np.random.RandomState(randomSeed)
else:
self._rng = np.random.RandomState(42)
chance = self._rng.random_sample(len(orbits))
prob_c = 0.5 * a - 1.0
# if chance <= prob_c:
sedvals = np.where(chance <= prob_c, 'C.dat', 'S.dat')
return sedvals
[docs] def readOrbits(self, orbitfile, delim=None, skiprows=None):
"""Read orbits from a file, generating a pandas dataframe containing columns matching dataCols,
for the appropriate orbital parameter format (currently accepts COM, KEP or CAR formats).
After reading and standardizing the column names, calls self.setOrbits to validate the
orbital parameters. Expects angles in orbital element formats to be in degrees.
Note that readOrbits uses pandas.read_csv to read the data file with the orbital parameters.
Thus, it should have column headers specifying the column names ..
unless skiprows = -1 or there is just no header line at all.
in which case it is assumed to be a standard DES format file, with no header line.
Parameters
----------
orbitfile : str
The name of the input file containing orbital parameter information.
delim : str, optional
The delimiter for the input orbit file. Default is None, will use delim_whitespace=True.
skiprows : int, optional
The number of rows to skip before reading the header information for pandas.
Default is None, which will trigger a check of the file to look for the header columns.
"""
names = None
# If skiprows is set, then we will assume the user has handled this so that the
# first line read has the header information.
# But, if skiprows is not set, then we have to do some checking to see if there is
# header information and which row it might start in.
if skiprows is None:
skiprows = -1
# Figure out whether the header is in the first line, or if there are rows to skip.
# We need to do a bit of juggling to do this before pandas reads the whole orbit file though.
with open(orbitfile, 'r') as fp:
headervalues = None
for line in fp:
values = line.split()
try:
# If it is a valid orbit line, we expect column 3 to be a number.
float(values[3])
# And if it worked, we're done here (it's an orbit) - go on to parsing header values.
break
except (ValueError, IndexError):
# This wasn't a valid number or there wasn't anything in the third value.
# So this is either the header line or it's a comment line before the header columns.
skiprows += 1
headervalues = values
if headervalues is not None: # (and skiprows > -1)
# There is a header, but we also need to check if there is a comment key at the start
# of the proper header line.
# ... Because this varies as well, and is sometimes separated from header columns.
linestart = headervalues[0]
if linestart == '#' or linestart == '!!' or linestart == '##':
names = headervalues[1:]
else:
names = headervalues
# Add 1 to skiprows, so that we skip the header column line.
skiprows += 1
# So now skiprows is a value. If it is -1, then there is no header information.
if skiprows == -1:
# No header; assume it's a typical DES file -
# we'll assign the column names based on the FORMAT.
names_COM = ('objId', 'FORMAT', 'q', 'e', 'i', 'node', 'argperi', 't_p',
'H', 'epoch', 'INDEX', 'N_PAR', 'MOID', 'COMPCODE')
names_KEP = ('objId', 'FORMAT', 'a', 'e', 'i', 'node', 'argperi', 'meanAnomaly',
'H', 'epoch', 'INDEX', 'N_PAR', 'MOID', 'COMPCODE')
names_CAR = ('objId', 'FORMAT', 'x', 'y', 'z', 'xdot', 'ydot', 'zdot',
'H', 'epoch', 'INDEX', 'N_PAR', 'MOID', 'COMPCODE')
# First use names_COM, and then change if required.
orbits = pd.read_csv(orbitfile, delim_whitespace=True, header=None, names=names_COM)
if orbits['FORMAT'][0] == 'KEP':
orbits.columns = names_KEP
elif orbits['FORMAT'][0] == 'CAR':
orbits.columns = names_CAR
else:
if delim is None:
orbits = pd.read_csv(orbitfile, delim_whitespace=True, skiprows=skiprows,
names=names)
else:
orbits = pd.read_csv(orbitfile, sep=delim, skiprows=skiprows, names=names)
# Drop some columns that are typically present in DES files but that we don't need.
if 'INDEX' in orbits:
del orbits['INDEX']
if 'N_PAR' in orbits:
del orbits['N_PAR']
if 'MOID' in orbits:
del orbits['MOID']
if 'COMPCODE' in orbits:
del orbits['COMPCODE']
if 'tmp' in orbits:
del orbits['tmp']
# Normalize the column names to standard values and identify the orbital element types.
ssoCols = orbits.columns.values.tolist()
# These are the alternative possibilities for various column headers
# (depending on file version, origin, etc.)
# that might need remapping from the on-file values to our standardized values.
altNames = {}
altNames['objId'] = ['objId', 'objid', '!!ObjID', '!!OID', '!!S3MID', 'OID', 'S3MID'
'objid(int)', 'full_name', '#name']
altNames['q'] = ['q']
altNames['a'] = ['a']
altNames['e'] = ['e', 'ecc']
altNames['inc'] = ['inc', 'i', 'i(deg)', 'incl']
altNames['Omega'] = ['Omega', 'omega', 'node', 'om', 'node(deg)',
'BigOmega', 'Omega/node', 'longNode']
altNames['argPeri'] = ['argPeri', 'argperi', 'omega/argperi', 'w', 'argperi(deg)', 'peri']
altNames['tPeri'] = ['tPeri', 't_p', 'timeperi', 't_peri', 'T_peri']
altNames['epoch'] = ['epoch', 't_0', 'Epoch', 'epoch_mjd']
altNames['H'] = ['H', 'magH', 'magHv', 'Hv', 'H_v']
altNames['g'] = ['g', 'phaseV', 'phase', 'gV', 'phase_g', 'G']
altNames['meanAnomaly'] = ['meanAnomaly', 'meanAnom', 'M', 'ma']
altNames['sed_filename'] = ['sed_filename', 'sed']
altNames['xdot'] = ['xdot', 'xDot']
altNames['ydot'] = ['ydot', 'yDot']
altNames['zdot'] = ['zdot', 'zDot']
# Update column names that match any of the alternatives above.
for name, alternatives in altNames.items():
intersection = list(set(alternatives) & set(ssoCols))
if len(intersection) > 1:
raise ValueError('Received too many possible matches to %s in orbit file %s'
% (name, orbitfile))
if len(intersection) == 1:
idx = ssoCols.index(intersection[0])
ssoCols[idx] = name
# Assign the new column names back to the orbits dataframe.
orbits.columns = ssoCols
# Validate and assign orbits to self.
self.setOrbits(orbits)
[docs] def updateOrbits(self, neworb):
"""Update existing orbits with new values, leaving OrbitIds, H, g, and sed_filenames in place.
Example use: transform orbital parameters (using PyOrbEphemerides) and then replace original values.
Example use: propagate orbital parameters (using PyOrbEphemerides) and then replace original values.
Parameters
----------
neworb: pandas.DataFrame
"""
col_orig = ['objId', 'otype', 'model', 'H', 'g', 'sed_filename']
new_order = ['objId', 'otype'] + [n for n in neworb.columns] + ['H', 'g', 'model', 'sed_filename']
updated_orbits = neworb.join(self.orbits[col_orig])[new_order]
self.setOrbits(updated_orbits)