def setup_class(self):
self.DataSpectrum = DataSpectrum.open(
"../data/WASP14/WASP-14_2009-06-15_04h13m57s_cb.spec.flux",
orders=np.array([22]))
self.Instrument = TRES()
self.HDF5Interface = HDF5Interface(
"../libraries/PHOENIX_submaster.hdf5")
stellar_Starting = {
"temp": 6000,
"logg": 4.05,
"Z": -0.4,
"vsini": 10.5,
"vz": 15.5,
"logOmega": -19.665
}
stellar_tuple = C.dictkeys_to_tuple(stellar_Starting)
cheb_tuple = ("c1", "c2", "c3")
cov_tuple = ("sigAmp", "logAmp", "l")
region_tuple = ("h", "loga", "mu", "sigma")
self.Model = Model(self.DataSpectrum,
self.Instrument,
self.HDF5Interface,
stellar_tuple=stellar_tuple,
cheb_tuple=cheb_tuple,
cov_tuple=cov_tuple,
region_tuple=region_tuple,
outdir="")
def __init__(self, DataSpectrum, Instrument, LibraryHA, LibraryLA,
parameters, deltaParameters):
'''Initialize the comparison object.
:param DataSpectrum: the spectrum that provides a wl grid + natural resolution
:type DataSpectrum: :obj:`grid_tools.DataSpectrum`
:param Instrument: the instrument object on which the DataSpectrum was acquired (ie, TRES, SPEX...)
:type Instrument: :obj:`grid_tools.Instrument`
:param LibraryHA: the path to the native resolution spectral library
:type LibraryHA: string
:param LibraryLA: the path to the approximate spectral library
:type LibraryLA: string
'''
self.DataSpectrum = DataSpectrum
self.Instrument = Instrument
self.HDF5InterfaceHA = HDF5Interface(LibraryHA)
self.HDF5InterfaceLA = HDF5Interface(LibraryLA)
print("Bounds of the grids are")
print("HA", self.HDF5InterfaceHA.bounds)
print("LA", self.HDF5InterfaceLA.bounds)
#If the DataSpectrum contains more than one order, we only take the first one. To get behavior with a
# different order, you should only load that via the DataSpectrum(orders=[22]) flag.
self.wl = self.DataSpectrum.wls[0]
self.fullModelLA = Model(self.DataSpectrum,
self.Instrument,
self.HDF5InterfaceLA,
stellar_tuple=("temp", "logg", "Z", "vsini",
"vz", "logOmega"),
cheb_tuple=("c1", "c2", "c3"),
cov_tuple=("sigAmp", "logAmp", "l"),
region_tuple=("loga", "mu", "sigma"))
self.modelLA = self.fullModelLA.OrderModels[0]
self.fullModelHA = ModelHA(self.DataSpectrum,
self.Instrument,
self.HDF5InterfaceHA,
stellar_tuple=("temp", "logg", "Z", "vsini",
"vz", "logOmega"),
cheb_tuple=("c1", "c2", "c3"),
cov_tuple=("sigAmp", "logAmp", "l"),
region_tuple=("loga", "mu", "sigma"))
self.modelHA = self.fullModelHA.OrderModels[0]
self.parameters = parameters
self.deltaParameters = deltaParameters
self.base = self.get_specHA(self.parameters)
self.baseLA = self.get_specLA(self.parameters)
self.approxResid = get_resid_spec(
self.base, self.baseLA) #modelHA - modelLA @ parameters
def main():
#Use argparse to determine if we've specified a config file
import argparse
parser = argparse.ArgumentParser(prog="plotly_model.py", description="Plot the model and residuals using plot.ly")
parser.add_argument("json", help="*.json file describing the model.")
parser.add_argument("params", help="*.yaml file specifying run parameters.")
# parser.add_argument("-o", "--output", help="*.html file for output")
args = parser.parse_args()
import json
import yaml
if args.json: #
#assert that we actually specified a *.json file
if ".json" not in args.json:
import sys
sys.exit("Must specify a *.json file.")
if args.params: #
#assert that we actually specified a *.yaml file
if ".yaml" in args.params:
yaml_file = args.params
f = open(args.params)
config = yaml.load(f)
f.close()
else:
import sys
sys.exit("Must specify a *.yaml file.")
yaml_file = args.params
from StellarSpectra.model import Model
from StellarSpectra.spectrum import DataSpectrum
from StellarSpectra.grid_tools import TRES, HDF5Interface
#Figure out what the relative path is to base
import StellarSpectra
base = StellarSpectra.__file__[:-26]
myDataSpectrum = DataSpectrum.open(base + config['data'], orders=config['orders'])
myInstrument = TRES()
myHDF5Interface = HDF5Interface(base + config['HDF5_path'])
myModel = Model.from_json(args.json, myDataSpectrum, myInstrument, myHDF5Interface)
for model in myModel.OrderModels:
#Get the data
wl, fl = model.get_data()
#Get the model flux
flm = model.get_spectrum()
#Get residuals
residuals = model.get_residuals()
name = "Order {}".format(model.order)
url = plotly_order(name, wl, fl, flm, residuals)
print(url)
def __init__(self, DataSpectrum, Instrument, LibraryHA, LibraryLA, parameters, deltaParameters):
"""Initialize the comparison object.
:param DataSpectrum: the spectrum that provides a wl grid + natural resolution
:type DataSpectrum: :obj:`grid_tools.DataSpectrum`
:param Instrument: the instrument object on which the DataSpectrum was acquired (ie, TRES, SPEX...)
:type Instrument: :obj:`grid_tools.Instrument`
:param LibraryHA: the path to the native resolution spectral library
:type LibraryHA: string
:param LibraryLA: the path to the approximate spectral library
:type LibraryLA: string
"""
self.DataSpectrum = DataSpectrum
self.Instrument = Instrument
self.HDF5InterfaceHA = HDF5Interface(LibraryHA)
self.HDF5InterfaceLA = HDF5Interface(LibraryLA)
print("Bounds of the grids are")
print("HA", self.HDF5InterfaceHA.bounds)
print("LA", self.HDF5InterfaceLA.bounds)
# If the DataSpectrum contains more than one order, we only take the first one. To get behavior with a
# different order, you should only load that via the DataSpectrum(orders=[22]) flag.
self.wl = self.DataSpectrum.wls[0]
self.fullModelLA = Model(
self.DataSpectrum,
self.Instrument,
self.HDF5InterfaceLA,
stellar_tuple=("temp", "logg", "Z", "vsini", "vz", "logOmega"),
cheb_tuple=("c1", "c2", "c3"),
cov_tuple=("sigAmp", "logAmp", "l"),
region_tuple=("loga", "mu", "sigma"),
)
self.modelLA = self.fullModelLA.OrderModels[0]
self.fullModelHA = ModelHA(
self.DataSpectrum,
self.Instrument,
self.HDF5InterfaceHA,
stellar_tuple=("temp", "logg", "Z", "vsini", "vz", "logOmega"),
cheb_tuple=("c1", "c2", "c3"),
cov_tuple=("sigAmp", "logAmp", "l"),
region_tuple=("loga", "mu", "sigma"),
)
self.modelHA = self.fullModelHA.OrderModels[0]
self.parameters = parameters
self.deltaParameters = deltaParameters
self.base = self.get_specHA(self.parameters)
self.baseLA = self.get_specLA(self.parameters)
self.approxResid = get_resid_spec(self.base, self.baseLA) # modelHA - modelLA @ parameters
class TestModel:
def setup_class(self):
self.DataSpectrum = DataSpectrum.open("../data/WASP14/WASP-14_2009-06-15_04h13m57s_cb.spec.flux", orders=np.array([22]))
self.Instrument = TRES()
self.HDF5Interface = HDF5Interface("../libraries/PHOENIX_submaster.hdf5")
stellar_Starting = {"temp":6000, "logg":4.05, "Z":-0.4, "vsini":10.5, "vz":15.5, "logOmega":-19.665}
stellar_tuple = C.dictkeys_to_tuple(stellar_Starting)
cheb_tuple = ("c1", "c2", "c3")
cov_tuple = ("sigAmp", "logAmp", "l")
region_tuple = ("h", "loga", "mu", "sigma")
self.Model = Model(self.DataSpectrum, self.Instrument, self.HDF5Interface, stellar_tuple=stellar_tuple, cheb_tuple=cheb_tuple,
cov_tuple=cov_tuple, region_tuple=region_tuple, outdir="")
def test_update(self):
self.Model.OrderModels[0].update_Cheb({"c1": -0.017, "c2": -0.017, "c3": -0.003})
cov_Starting = {"sigAmp":1, "logAmp":-14.0, "l":0.15}
self.Model.OrderModels[0].update_Cov(cov_Starting)
params = {"temp":6005, "logg":4.05, "Z":-0.4, "vsini":10.5, "vz":15.5, "logOmega":-19.665}
self.Model.update_Model(params) #This also updates downsampled_fls
#For order in myModel, do evaluate, and sum the results.
def test_evaluate(self):
self.Model.evaluate()
def test_to_json(self):
self.Model.to_json()
def test_from_json(self):
newModel = Model.from_json("final_model.json", self.DataSpectrum, self.Instrument, self.HDF5Interface)
def setup_class(self):
self.DataSpectrum = DataSpectrum.open("../data/WASP14/WASP-14_2009-06-15_04h13m57s_cb.spec.flux", orders=np.array([22]))
self.Instrument = TRES()
self.HDF5Interface = HDF5Interface("../libraries/PHOENIX_submaster.hdf5")
stellar_Starting = {"temp":6000, "logg":4.05, "Z":-0.4, "vsini":10.5, "vz":15.5, "logOmega":-19.665}
stellar_tuple = C.dictkeys_to_tuple(stellar_Starting)
cheb_tuple = ("c1", "c2", "c3")
cov_tuple = ("sigAmp", "logAmp", "l")
region_tuple = ("h", "loga", "mu", "sigma")
self.Model = Model(self.DataSpectrum, self.Instrument, self.HDF5Interface, stellar_tuple=stellar_tuple, cheb_tuple=cheb_tuple,
cov_tuple=cov_tuple, region_tuple=region_tuple, outdir="")
class AccuracyComparison:
'''
Gather the data products necessary to make a test about accuracy of the reduced grid sizes.
'''
def __init__(self, DataSpectrum, Instrument, LibraryHA, LibraryLA,
parameters, deltaParameters):
'''Initialize the comparison object.
:param DataSpectrum: the spectrum that provides a wl grid + natural resolution
:type DataSpectrum: :obj:`grid_tools.DataSpectrum`
:param Instrument: the instrument object on which the DataSpectrum was acquired (ie, TRES, SPEX...)
:type Instrument: :obj:`grid_tools.Instrument`
:param LibraryHA: the path to the native resolution spectral library
:type LibraryHA: string
:param LibraryLA: the path to the approximate spectral library
:type LibraryLA: string
'''
self.DataSpectrum = DataSpectrum
self.Instrument = Instrument
self.HDF5InterfaceHA = HDF5Interface(LibraryHA)
self.HDF5InterfaceLA = HDF5Interface(LibraryLA)
print("Bounds of the grids are")
print("HA", self.HDF5InterfaceHA.bounds)
print("LA", self.HDF5InterfaceLA.bounds)
#If the DataSpectrum contains more than one order, we only take the first one. To get behavior with a
# different order, you should only load that via the DataSpectrum(orders=[22]) flag.
self.wl = self.DataSpectrum.wls[0]
self.fullModelLA = Model(self.DataSpectrum,
self.Instrument,
self.HDF5InterfaceLA,
stellar_tuple=("temp", "logg", "Z", "vsini",
"vz", "logOmega"),
cheb_tuple=("c1", "c2", "c3"),
cov_tuple=("sigAmp", "logAmp", "l"),
region_tuple=("loga", "mu", "sigma"))
self.modelLA = self.fullModelLA.OrderModels[0]
self.fullModelHA = ModelHA(self.DataSpectrum,
self.Instrument,
self.HDF5InterfaceHA,
stellar_tuple=("temp", "logg", "Z", "vsini",
"vz", "logOmega"),
cheb_tuple=("c1", "c2", "c3"),
cov_tuple=("sigAmp", "logAmp", "l"),
region_tuple=("loga", "mu", "sigma"))
self.modelHA = self.fullModelHA.OrderModels[0]
self.parameters = parameters
self.deltaParameters = deltaParameters
self.base = self.get_specHA(self.parameters)
self.baseLA = self.get_specLA(self.parameters)
self.approxResid = get_resid_spec(
self.base, self.baseLA) #modelHA - modelLA @ parameters
def get_specHA(self, parameters):
'''
Update the model and then query the spectrum
:param parameters: Dictionary of fundamental stellar parameters
:type parameters: dict
:returns: flux spectrum
'''
params = parameters.copy()
params.update({"vsini": 0., "vz": 0, "logOmega": 0.})
self.fullModelHA.update_Model(params)
return self.modelHA.get_spectrum()
def get_specLA(self, parameters):
'''
Update the model and then query the spectrum
:param parameters: Dictionary of fundamental stellar parameters
:type parameters: dict
:returns: flux spectrum
'''
params = parameters.copy()
params.update({"vsini": 0., "vz": 0, "logOmega": 0.})
self.fullModelLA.update_Model(params)
return self.modelLA.get_spectrum()
def createEnvelopeSpectrum(self, direction='both'):
'''
The parameters should always be specified at a grid point of the HDF5 file.
For this, do the deltaParameters interpolation.
Direction specifies whether to do interpolation up (+ 10 K, etc.), down (- 10 K), or
do both and then find the minimum envelope between the two.
For now, only up is implemented.
'''
#For each key, add the delta parameters
temp_params = self.parameters.copy()
temp_params["temp"] += self.deltaParameters["temp"]
temp_spec = get_resid_spec(self.base, self.get_specHA(temp_params))
logg_params = self.parameters.copy()
logg_params["logg"] += self.deltaParameters["logg"]
logg_spec = get_resid_spec(self.base, self.get_specHA(logg_params))
Z_params = self.parameters.copy()
Z_params["Z"] += self.deltaParameters["Z"]
Z_spec = get_resid_spec(self.base, self.get_specHA(Z_params))
self.envelope = get_min_spec([temp_spec, logg_spec, Z_spec])
def plot_quality(self):
'''
Visualize the quality of the interpolation.
Two-panel plot.
Top: HA and LA spectrum
Bottom: Residual between HA + LA spectrum and the HA spectrum error bounds for deltaParameters
'''
self.createEnvelopeSpectrum()
fig, ax = plt.subplots(nrows=2, figsize=(8, 6), sharex=True)
ax[0].plot(self.wl, self.base, "b", label="HA")
ax[0].plot(self.wl, self.baseLA, "r", label="LA")
ax[0].legend()
ax[0].set_ylabel(r"$\propto f_\lambda$")
ax[0].set_title(
"Temp={temp:} logg={logg:} Z={Z:}".format(**self.parameters))
ax[1].semilogy(self.wl, self.approxResid, "k", label="(HA - LA)/HA")
ax[1].semilogy(self.wl, self.envelope, "b", label="Interp Envelope")
ax[1].legend()
ax[1].set_xlabel(r"$\lambda$\AA")
ax[1].set_ylabel("fractional error")
return fig
myHDF5Interface = HDF5Interface(config['HDF5_path'])
stellar_Starting = config['stellar_params']
stellar_tuple = C.dictkeys_to_tuple(stellar_Starting)
cheb_Starting = config['cheb_params']
cheb_tuple = ("logc0", "c1", "c2", "c3")
cov_Starting = config['cov_params']
cov_tuple = C.dictkeys_to_cov_global_tuple(cov_Starting)
region_tuple = ("h", "loga", "mu", "sigma")
region_MH_cov = np.array([0.05, 0.04, 0.02, 0.02])**2 * np.identity(
len(region_tuple))
myModel = Model.from_json(args.json, myDataSpectrum, myInstrument,
myHDF5Interface)
model = myModel.OrderModels[0]
#Get the data
wl, fl = model.get_data()
#Get the model flux
flm = model.get_spectrum()
#Get residuals
residuals = model.get_residuals()
#Get the Chebyshev spectrum
cheb = model.get_Cheb()
def main():
#Use argparse to determine if we've specified a config file
import argparse
parser = argparse.ArgumentParser(
prog="flot_model.py",
description="Plot the model and residuals using flot.")
parser.add_argument("json", help="*.json file describing the model.")
parser.add_argument("params",
help="*.yaml file specifying run parameters.")
# parser.add_argument("-o", "--output", help="*.html file for output")
args = parser.parse_args()
import json
import yaml
if args.json: #
#assert that we actually specified a *.json file
if ".json" not in args.json:
import sys
sys.exit("Must specify a *.json file.")
if args.params: #
#assert that we actually specified a *.yaml file
if ".yaml" in args.params:
yaml_file = args.params
f = open(args.params)
config = yaml.load(f)
f.close()
else:
import sys
sys.exit("Must specify a *.yaml file.")
yaml_file = args.params
from StellarSpectra.model import Model
from StellarSpectra.spectrum import DataSpectrum
from StellarSpectra.grid_tools import TRES, HDF5Interface
#Figure out what the relative path is to base
import StellarSpectra
base = StellarSpectra.__file__[:-26]
myDataSpectrum = DataSpectrum.open(base + config['data'],
orders=config['orders'])
myInstrument = TRES()
myHDF5Interface = HDF5Interface(base + config['HDF5_path'])
myModel = Model.from_json(args.json, myDataSpectrum, myInstrument,
myHDF5Interface)
for model in myModel.OrderModels:
#If an order has regions, read these out from model_final.json
region_dict = model.get_regions_dict()
print("Region dict", region_dict)
#loop through these to determine the wavelength of each
wl_regions = [value["mu"] for value in region_dict.values()]
#Make vertical markings at the location of the wl_regions.
#Get the data, sigmas, and mask
wl, fl, sigma, mask = model.get_data()
#Get the model flux
flm = model.get_spectrum()
#Get chebyshev
cheb = model.get_Cheb()
name = "Order {}".format(model.order)
plot_data = order_json(wl, fl, sigma, mask, flm, cheb)
plot_data.update({"wl_regions": wl_regions})
print(plot_data['wl_regions'])
render_template(base, plot_data)
myHDF5Interface = HDF5Interface(config['HDF5_path'])
stellar_Starting = config['stellar_params']
stellar_tuple = C.dictkeys_to_tuple(stellar_Starting)
cheb_Starting = config['cheb_params']
cheb_tuple = ("logc0", "c1", "c2", "c3")
cov_Starting = config['cov_params']
cov_tuple = C.dictkeys_to_cov_global_tuple(cov_Starting)
region_tuple = ("h", "loga", "mu", "sigma")
region_MH_cov = np.array([0.05, 0.04, 0.02, 0.02])**2 * np.identity(len(region_tuple))
myModel = Model.from_json(args.json, myDataSpectrum, myInstrument, myHDF5Interface)
model = myModel.OrderModels[0]
#Get the data
wl, fl = model.get_data()
#Get the model flux
flm = model.get_spectrum()
#Get residuals
residuals = model.get_residuals()
#Get the Chebyshev spectrum
cheb = model.get_Cheb()
#Use the model_final.json to figure out how many orders there are
from StellarSpectra.model import Model
from StellarSpectra.spectrum import DataSpectrum
from StellarSpectra.grid_tools import TRES, HDF5Interface
#Figure out what the relative path is to base
import StellarSpectra
base = StellarSpectra.__file__[:-26]
myDataSpectrum = DataSpectrum.open(base + config['data'],
orders=config['orders'])
myInstrument = TRES()
myHDF5Interface = HDF5Interface(base + config['HDF5_path'])
myModel = Model.from_json(args.run + "/model_final.json", myDataSpectrum,
myInstrument, myHDF5Interface)
orders = [orderModel.order for orderModel in myModel.OrderModels]
flot_plots = {22: "Hi"}
#If the Jinja templater is going to work, it needs a list of orders. It also needs a list of how many regions
# are in each order
# each order, there is dictionary
#of global
#Set the categories as the decomposition of the run directory, excluding
#output and the "run00" directory.
#For example, output/WASP14/Kurucz/22/run01 becomes categories="WASP14 Kurucz 22"
categories = " ".join(args.run.split("/")[1:-1])
templateVars = {
import numpy as np
from StellarSpectra.model import Model
from StellarSpectra.spectrum import DataSpectrum
from StellarSpectra.grid_tools import TRES, HDF5Interface
from StellarSpectra import utils
myDataSpectrum = DataSpectrum.open("../../data/WASP14/WASP14-2009-06-14.hdf5", orders=np.array([21,22,23]))
myInstrument = TRES()
myHDF5Interface = HDF5Interface("../../libraries/PHOENIX_TRES_F.hdf5")
#Load a model using the JSON file
myModel = Model.from_json("WASP14_PHOENIX_model0_final.json", myDataSpectrum, myInstrument, myHDF5Interface)
myOrderModel = myModel.OrderModels[1]
model_flux = myOrderModel.get_spectrum()
spec = myModel.get_data()
wl = spec.wls[1]
fl = spec.fls[1]
model_fl = myOrderModel.get_spectrum()
residuals = fl - model_fl
mask = spec.masks[1]
cov = myModel.OrderModels[1].get_Cov().todense()
np.save("PHOENIX_covariance_matrix.npy", cov)
import sys; sys.exit()
def test_from_json(self):
newModel = Model.from_json("final_model.json", self.DataSpectrum,
self.Instrument, self.HDF5Interface)
import numpy as np
from StellarSpectra.model import Model
from StellarSpectra.spectrum import DataSpectrum
from StellarSpectra.grid_tools import TRES, HDF5Interface
from StellarSpectra import utils
myDataSpectrum = DataSpectrum.open("../../data/WASP14/WASP14-2009-06-14.hdf5",
orders=np.array([21, 22, 23]))
myInstrument = TRES()
myHDF5Interface = HDF5Interface("../../libraries/PHOENIX_TRES_F.hdf5")
#Load a model using the JSON file
myModel = Model.from_json("WASP14_PHOENIX_model0_final.json", myDataSpectrum,
myInstrument, myHDF5Interface)
myOrderModel = myModel.OrderModels[1]
model_flux = myOrderModel.get_spectrum()
spec = myModel.get_data()
wl = spec.wls[1]
fl = spec.fls[1]
model_fl = myOrderModel.get_spectrum()
residuals = fl - model_fl
mask = spec.masks[1]
cov = myModel.OrderModels[1].get_Cov().todense()
np.save("PHOENIX_covariance_matrix.npy", cov)
def main():
#Use argparse to determine if we've specified a config file
import argparse
parser = argparse.ArgumentParser(
prog="plotly_model.py",
description="Plot the model and residuals using plot.ly")
parser.add_argument("json", help="*.json file describing the model.")
parser.add_argument("params",
help="*.yaml file specifying run parameters.")
# parser.add_argument("-o", "--output", help="*.html file for output")
args = parser.parse_args()
import json
import yaml
if args.json: #
#assert that we actually specified a *.json file
if ".json" not in args.json:
import sys
sys.exit("Must specify a *.json file.")
if args.params: #
#assert that we actually specified a *.yaml file
if ".yaml" in args.params:
yaml_file = args.params
f = open(args.params)
config = yaml.load(f)
f.close()
else:
import sys
sys.exit("Must specify a *.yaml file.")
yaml_file = args.params
from StellarSpectra.model import Model
from StellarSpectra.spectrum import DataSpectrum
from StellarSpectra.grid_tools import TRES, HDF5Interface
#Figure out what the relative path is to base
import StellarSpectra
base = StellarSpectra.__file__[:-26]
myDataSpectrum = DataSpectrum.open(base + config['data'],
orders=config['orders'])
myInstrument = TRES()
myHDF5Interface = HDF5Interface(base + config['HDF5_path'])
myModel = Model.from_json(args.json, myDataSpectrum, myInstrument,
myHDF5Interface)
for model in myModel.OrderModels:
#Get the data
wl, fl = model.get_data()
#Get the model flux
flm = model.get_spectrum()
#Get residuals
residuals = model.get_residuals()
name = "Order {}".format(model.order)
url = plotly_order(name, wl, fl, flm, residuals)
print(url)
import numpy as np
from StellarSpectra.model import Model
from StellarSpectra.spectrum import DataSpectrum
from StellarSpectra.grid_tools import TRES, SPEX, HDF5Interface
from StellarSpectra import utils
myDataSpectrum = DataSpectrum.open("../../data/WASP14/WASP14-2009-06-14.hdf5",
orders=np.array([21, 22, 23]))
myInstrument = TRES()
myHDF5Interface = HDF5Interface("../../libraries/Kurucz_TRES.hdf5")
#Load a model using the JSON file
#Taken from:
#/n/home07/iczekala/StellarSpectra/output/WASP14/Kurucz/21_22_23/logg/cov/2014-08-06/run18
myModel = Model.from_json("WASP14_Kurucz_logg_model_final.json",
myDataSpectrum, myInstrument, myHDF5Interface)
myOrderModel = myModel.OrderModels[1]
model_flux = myOrderModel.get_spectrum()
spec = myModel.get_data()
wl = spec.wls[1]
fl = spec.fls[1]
model_fl = myOrderModel.get_spectrum()
residuals = fl - model_fl
mask = spec.masks[1]
cov = myModel.OrderModels[1].get_Cov().todense()
draws = utils.random_draws(cov, num=50)
class AccuracyComparison:
"""
Gather the data products necessary to make a test about accuracy of the reduced grid sizes.
"""
def __init__(self, DataSpectrum, Instrument, LibraryHA, LibraryLA, parameters, deltaParameters):
"""Initialize the comparison object.
:param DataSpectrum: the spectrum that provides a wl grid + natural resolution
:type DataSpectrum: :obj:`grid_tools.DataSpectrum`
:param Instrument: the instrument object on which the DataSpectrum was acquired (ie, TRES, SPEX...)
:type Instrument: :obj:`grid_tools.Instrument`
:param LibraryHA: the path to the native resolution spectral library
:type LibraryHA: string
:param LibraryLA: the path to the approximate spectral library
:type LibraryLA: string
"""
self.DataSpectrum = DataSpectrum
self.Instrument = Instrument
self.HDF5InterfaceHA = HDF5Interface(LibraryHA)
self.HDF5InterfaceLA = HDF5Interface(LibraryLA)
print("Bounds of the grids are")
print("HA", self.HDF5InterfaceHA.bounds)
print("LA", self.HDF5InterfaceLA.bounds)
# If the DataSpectrum contains more than one order, we only take the first one. To get behavior with a
# different order, you should only load that via the DataSpectrum(orders=[22]) flag.
self.wl = self.DataSpectrum.wls[0]
self.fullModelLA = Model(
self.DataSpectrum,
self.Instrument,
self.HDF5InterfaceLA,
stellar_tuple=("temp", "logg", "Z", "vsini", "vz", "logOmega"),
cheb_tuple=("c1", "c2", "c3"),
cov_tuple=("sigAmp", "logAmp", "l"),
region_tuple=("loga", "mu", "sigma"),
)
self.modelLA = self.fullModelLA.OrderModels[0]
self.fullModelHA = ModelHA(
self.DataSpectrum,
self.Instrument,
self.HDF5InterfaceHA,
stellar_tuple=("temp", "logg", "Z", "vsini", "vz", "logOmega"),
cheb_tuple=("c1", "c2", "c3"),
cov_tuple=("sigAmp", "logAmp", "l"),
region_tuple=("loga", "mu", "sigma"),
)
self.modelHA = self.fullModelHA.OrderModels[0]
self.parameters = parameters
self.deltaParameters = deltaParameters
self.base = self.get_specHA(self.parameters)
self.baseLA = self.get_specLA(self.parameters)
self.approxResid = get_resid_spec(self.base, self.baseLA) # modelHA - modelLA @ parameters
def get_specHA(self, parameters):
"""
Update the model and then query the spectrum
:param parameters: Dictionary of fundamental stellar parameters
:type parameters: dict
:returns: flux spectrum
"""
params = parameters.copy()
params.update({"vsini": 0.0, "vz": 0, "logOmega": 0.0})
self.fullModelHA.update_Model(params)
return self.modelHA.get_spectrum()
def get_specLA(self, parameters):
"""
Update the model and then query the spectrum
:param parameters: Dictionary of fundamental stellar parameters
:type parameters: dict
:returns: flux spectrum
"""
params = parameters.copy()
params.update({"vsini": 0.0, "vz": 0, "logOmega": 0.0})
self.fullModelLA.update_Model(params)
return self.modelLA.get_spectrum()
def createEnvelopeSpectrum(self, direction="both"):
"""
The parameters should always be specified at a grid point of the HDF5 file.
For this, do the deltaParameters interpolation.
Direction specifies whether to do interpolation up (+ 10 K, etc.), down (- 10 K), or
do both and then find the minimum envelope between the two.
For now, only up is implemented.
"""
# For each key, add the delta parameters
temp_params = self.parameters.copy()
temp_params["temp"] += self.deltaParameters["temp"]
temp_spec = get_resid_spec(self.base, self.get_specHA(temp_params))
logg_params = self.parameters.copy()
logg_params["logg"] += self.deltaParameters["logg"]
logg_spec = get_resid_spec(self.base, self.get_specHA(logg_params))
Z_params = self.parameters.copy()
Z_params["Z"] += self.deltaParameters["Z"]
Z_spec = get_resid_spec(self.base, self.get_specHA(Z_params))
self.envelope = get_min_spec([temp_spec, logg_spec, Z_spec])
def plot_quality(self):
"""
Visualize the quality of the interpolation.
Two-panel plot.
Top: HA and LA spectrum
Bottom: Residual between HA + LA spectrum and the HA spectrum error bounds for deltaParameters
"""
self.createEnvelopeSpectrum()
fig, ax = plt.subplots(nrows=2, figsize=(8, 6), sharex=True)
ax[0].plot(self.wl, self.base, "b", label="HA")
ax[0].plot(self.wl, self.baseLA, "r", label="LA")
ax[0].legend()
ax[0].set_ylabel(r"$\propto f_\lambda$")
ax[0].set_title("Temp={temp:} logg={logg:} Z={Z:}".format(**self.parameters))
ax[1].semilogy(self.wl, self.approxResid, "k", label="(HA - LA)/HA")
ax[1].semilogy(self.wl, self.envelope, "b", label="Interp Envelope")
ax[1].legend()
ax[1].set_xlabel(r"$\lambda$\AA")
ax[1].set_ylabel("fractional error")
return fig
import numpy as np
from StellarSpectra.model import Model
from StellarSpectra.spectrum import DataSpectrum
from StellarSpectra.grid_tools import TRES, SPEX, HDF5Interface
from StellarSpectra import utils
myDataSpectrum = DataSpectrum.open("../../data/WASP14/WASP14-2009-06-14.hdf5", orders=np.array([21, 22, 23]))
myInstrument = TRES()
myHDF5Interface = HDF5Interface("../../libraries/Kurucz_TRES.hdf5")
# Load a model using the JSON file
# Taken from:
# /n/home07/iczekala/StellarSpectra/output/WASP14/Kurucz/21_22_23/logg/cov/2014-08-06/run18
myModel = Model.from_json("WASP14_Kurucz_logg_model_final.json", myDataSpectrum, myInstrument, myHDF5Interface)
myOrderModel = myModel.OrderModels[1]
model_flux = myOrderModel.get_spectrum()
spec = myModel.get_data()
wl = spec.wls[1]
fl = spec.fls[1]
model_fl = myOrderModel.get_spectrum()
residuals = fl - model_fl
mask = spec.masks[1]
cov = myModel.OrderModels[1].get_Cov().todense()
draws = utils.random_draws(cov, num=50)
def test_from_json(self):
newModel = Model.from_json("final_model.json", self.DataSpectrum, self.Instrument, self.HDF5Interface)
class TestModel:
def setup_class(self):
self.DataSpectrum = DataSpectrum.open(
"../data/WASP14/WASP-14_2009-06-15_04h13m57s_cb.spec.flux",
orders=np.array([22]))
self.Instrument = TRES()
self.HDF5Interface = HDF5Interface(
"../libraries/PHOENIX_submaster.hdf5")
stellar_Starting = {
"temp": 6000,
"logg": 4.05,
"Z": -0.4,
"vsini": 10.5,
"vz": 15.5,
"logOmega": -19.665
}
stellar_tuple = C.dictkeys_to_tuple(stellar_Starting)
cheb_tuple = ("c1", "c2", "c3")
cov_tuple = ("sigAmp", "logAmp", "l")
region_tuple = ("h", "loga", "mu", "sigma")
self.Model = Model(self.DataSpectrum,
self.Instrument,
self.HDF5Interface,
stellar_tuple=stellar_tuple,
cheb_tuple=cheb_tuple,
cov_tuple=cov_tuple,
region_tuple=region_tuple,
outdir="")
def test_update(self):
self.Model.OrderModels[0].update_Cheb({
"c1": -0.017,
"c2": -0.017,
"c3": -0.003
})
cov_Starting = {"sigAmp": 1, "logAmp": -14.0, "l": 0.15}
self.Model.OrderModels[0].update_Cov(cov_Starting)
params = {
"temp": 6005,
"logg": 4.05,
"Z": -0.4,
"vsini": 10.5,
"vz": 15.5,
"logOmega": -19.665
}
self.Model.update_Model(params) #This also updates downsampled_fls
#For order in myModel, do evaluate, and sum the results.
def test_evaluate(self):
self.Model.evaluate()
def test_to_json(self):
self.Model.to_json()
def test_from_json(self):
newModel = Model.from_json("final_model.json", self.DataSpectrum,
self.Instrument, self.HDF5Interface)
import numpy as np
from StellarSpectra.model import Model
from StellarSpectra.spectrum import DataSpectrum
from StellarSpectra.grid_tools import SPEX, HDF5Interface
from StellarSpectra import utils
myDataSpectrum = DataSpectrum.open("../../data/Gl51/Gl51RA.hdf5", orders=np.array([0]))
myInstrument = SPEX()
myHDF5Interface = HDF5Interface("../../libraries/PHOENIX_SPEX_M.hdf5")
#Load a model using the JSON file
#Taken from:
# /home/ian/Grad/Research/Disks/StellarSpectra/output/Gl51/PHOENIX/RA/region/logg/4_8sig/
myModel = Model.from_json("Gl51_model0_final.json", myDataSpectrum, myInstrument, myHDF5Interface)
myOrderModel = myModel.OrderModels[0]
model_flux = myOrderModel.get_spectrum()
spec = myModel.get_data()
wl = spec.wls[0]
fl = spec.fls[0]
model_fl = myOrderModel.get_spectrum()
residuals = fl - model_fl
mask = spec.masks[0]
cov = myModel.OrderModels[0].get_Cov().todense()
np.save("Gl51_covariance_matrix.npy", cov)
import sys
sys.exit()
def main():
#Use argparse to determine if we've specified a config file
import argparse
parser = argparse.ArgumentParser(prog="flot_model.py", description="Plot the model and residuals using flot.")
parser.add_argument("json", help="*.json file describing the model.")
parser.add_argument("params", help="*.yaml file specifying run parameters.")
# parser.add_argument("-o", "--output", help="*.html file for output")
args = parser.parse_args()
import json
import yaml
if args.json: #
#assert that we actually specified a *.json file
if ".json" not in args.json:
import sys
sys.exit("Must specify a *.json file.")
if args.params: #
#assert that we actually specified a *.yaml file
if ".yaml" in args.params:
yaml_file = args.params
f = open(args.params)
config = yaml.load(f)
f.close()
else:
import sys
sys.exit("Must specify a *.yaml file.")
yaml_file = args.params
from StellarSpectra.model import Model
from StellarSpectra.spectrum import DataSpectrum
from StellarSpectra.grid_tools import TRES, HDF5Interface
#Figure out what the relative path is to base
import StellarSpectra
base = StellarSpectra.__file__[:-26]
myDataSpectrum = DataSpectrum.open(base + config['data'], orders=config['orders'])
myInstrument = TRES()
myHDF5Interface = HDF5Interface(base + config['HDF5_path'])
myModel = Model.from_json(args.json, myDataSpectrum, myInstrument, myHDF5Interface)
for model in myModel.OrderModels:
#If an order has regions, read these out from model_final.json
region_dict = model.get_regions_dict()
print("Region dict", region_dict)
#loop through these to determine the wavelength of each
wl_regions = [value["mu"] for value in region_dict.values()]
#Make vertical markings at the location of the wl_regions.
#Get the data, sigmas, and mask
wl, fl, sigma, mask = model.get_data()
#Get the model flux
flm = model.get_spectrum()
#Get chebyshev
cheb = model.get_Cheb()
name = "Order {}".format(model.order)
plot_data = order_json(wl, fl, sigma, mask, flm, cheb)
plot_data.update({"wl_regions":wl_regions})
print(plot_data['wl_regions'])
render_template(base, plot_data)
# Use the model_final.json to figure out how many orders there are
from StellarSpectra.model import Model
from StellarSpectra.spectrum import DataSpectrum
from StellarSpectra.grid_tools import TRES, HDF5Interface
# Figure out what the relative path is to base
import StellarSpectra
base = StellarSpectra.__file__[:-26]
myDataSpectrum = DataSpectrum.open(base + config["data"], orders=config["orders"])
myInstrument = TRES()
myHDF5Interface = HDF5Interface(base + config["HDF5_path"])
myModel = Model.from_json(args.run + "/model_final.json", myDataSpectrum, myInstrument, myHDF5Interface)
orders = [orderModel.order for orderModel in myModel.OrderModels]
flot_plots = {22: "Hi"}
# If the Jinja templater is going to work, it needs a list of orders. It also needs a list of how many regions
# are in each order
# each order, there is dictionary
# of global
# Set the categories as the decomposition of the run directory, excluding
# output and the "run00" directory.
# For example, output/WASP14/Kurucz/22/run01 becomes categories="WASP14 Kurucz 22"
categories = " ".join(args.run.split("/")[1:-1])