API documentation

Eval

cortecs.eval.eval_polynomial.eval_polynomial(temperature, pressure, wavelength, temperatures, pressures, wavelengths, fitter_results, **kwargs)

evaluates the polynomial at a given temperature and pressure.

Inputs

temperature: float

temperature in K.

pressure: float

pressure in bar.

wavelength: float

wavelength in microns.

temperatures: array-like

temperature grid.

pressures: array-like

pressure grid.

wavelengths: array-like

wavelength grid.

fitter_results: array-like

the PCA coefficients.

returns:

xsec_int – The cross section at the given temperature and pressure.

rtype:

float

For evaluating the PCA fit at a given temperature, pressure, and wavelength.

cortecs.eval.eval_pca.eval_pca(temperature, pressure, wavelength, T, P, wl, fitter_results, fit_axis='pressure', **kwargs)

Evaluates the PCA fit at a given temperature, pressure, and wavelength.

Inputs

temperature: float

The temperature to evaluate at.

pressure: float

The pressure to evaluate at.

wavelength: float

The wavelength to evaluate at.

cortecs.eval.eval_pca.eval_pca_ind_wav(first_ind, second_ind, vectors, pca_coeffs)

Evaluates the PCA fit at a given temperature and pressure.

Unfortunately, not all GPUs will support a simpler dot product, I believe, Also, we cannot loop over n_components explicitly because JAX functions require static loop lengths.

Inputs

first_ind: int

The index of the first axis quantity (default temperature) to evaluate at.

second_ind: int

The index of the second axis quantity (default pressure) to evaluate at.

vectors: array

The PCA vectors.

pca_coeffs: array

The PCA coefficients.

n_components: int

The number of PCA components used in the fitting

This file holds the classes for evaluating opacity data as trained by a neural network.

maybe call all of these eval_poly.py, eval_neural_net.py, etc.?

author: @arjunsavel

cortecs.eval.eval_neural_net.eval_neural_net(T, P, temperatures=None, pressures=None, wavelengths=None, n_layers=None, weights=None, biases=None, **kwargs)

evaluates the neural network at a given temperature and pressure.

Inputs

T: float

The temperature to evaluate at.

P: float

The pressure to evaluate at.

n_layers: int

The number of layers in the neural network.

weights: list

The weights of the neural network.

biases: list

The biases of the neural network.

cortecs.eval.eval_neural_net.feed_forward(x, n_layers, weights, biases)

feed forward neural network. this is a function that takes in the input, weights, and biases and returns the output.

Inputs

x: array-like

the input to the neural network.

n_layers: int

the number of layers in the neural network.

weights: list

the weights of the neural network.

biases: list

cortecs.eval.eval_neural_net.feed_forward_equal_layer_sizes(x, n_layers, weights, biases)

feed forward neural network. this is a function that takes in the input, weights, and biases and returns the output.

This function only works if all layers have the same size.

Inputs

x: array-like

the input to the neural network.

n_layers: int

the number of layers in the neural network.

weights: list

the weights of the neural network.

biases: list

Fit

The high-level API for fitting. Requires the Opac object.

class cortecs.fit.fit.Fitter(opac, method='pca', **fitter_kwargs)

Bases: object

fits the opacity data to a neural network.

todo: fit CIA. only on one dimension, because there’s no pressure dependence.

fit(parallel=False, verbose=1)

fits the opacity data to a neural network. loops over all wavelengths. Can loop over a list of species? …should be parallelized!

Inputs

parallel: bool

whether to parallelize the fitting.

fit_parallel()

fits the opacity data with a given method. Parallel. :return:

fit_serial(verbose=0)

fits the opacity data with a given method. Serial. :return: todo: keep in mind that the PCA method is not actually independent of wavelength.

method_dict = {'neural_net': <function fit_neural_net>, 'pca': <function fit_pca>, 'polynomial': <function fit_polynomial>}

prep_method_dict = {'neural_net': <function prep_neural_net>, 'pca': <function prep_pca>, 'polynomial': <function prep_polynomial>}

save(savename)

saves the fitter results.

save_method_dict = {'neural_net': <function save_neural_net>, 'pca': <function save_pca>, 'polynomial': <function save_polynomial>}

update_pbar(arg)

Updates a tqdm progress bar.

Module for fitting polynomial functions in temperature and pressure to opacity functions.

cortecs.fit.fit_polynomial.fit_polynomial(Z, P, T, prep_res, plot=False)

fits a polynomial to the opacity data.

Inputs

Z:

(n_temp x n_pres) it’s an array.

P:

pressure grid

T:

temperature grid

prep_res:

(n_temp x n_pres) PCA components

plot:

(bool) whether to plot the fit.

savename:

(str) if not None, the PCA components will be saved to this filename.

rtype:

coeff: (nc x pixels) PCA coefficients

cortecs.fit.fit_polynomial.prep_polynomial(cross_section, **kargs)

cortecs.fit.fit_polynomial.save_polynomial(savename, coeff)

Saves the polynomial coefficients to a file.

Inputs

savename:

(str) if not None, the PCA components will be saved to this filename.

coeff:

(nc x ntemp) PCA coefficients

Module for performing PCA regression on opacity functions.

cortecs.fit.fit_pca.do_pca(cube, nc=3)

Does all the PCA steps. right now for a single spectrum.

Inputs

cube:

(ntemperature x npressure) array being fit with PCA. Generally, this is the opacity data at a single

wavelength as a function of temperature and pressure. :nc: (int) number of PCA components to keep. Increasing the number of components can make the reconstruction of opacity data more accurate, but it can also lead to overfitting and make the model size larger (i.e., decrease the compression factor).

cortecs.fit.fit_pca.do_svd(standardized_cube, nc, nx)

Does the SVD (singular value decomposition) on the standardized cube.

Inputs

standardized_cube:

(ntemperature x npressure) standardized (mean-substracted, standard deviation-divided) array for PCA.

nc:

(int) number of PCA components to keep in the reconstruction. Increasing the number of components can make the reconstruction

of opacity data more accurate, but it can also lead to overfitting and make the model size larger (i.e., decrease the compression factor).

cortecs.fit.fit_pca.fit_mlr(cube, X)

todo: check array shapes. Fits the MLR (multiple linear regression) to the input cube with the PCA outputs and normalizes accordingly.

Inputs

cube:

(ntemp x npressure) array being fit with PCA. Generally, this is the opacity data at a single

wavelength as a function of temperature and pressure. :X: (npressure x ncomponents) PCA components. these are set by functions such as do_pca.

returns:

• beta: (ncomponents x ntemp) PCA coefficients. These are the coefficients that are used to reconstruct the

• opacity data at a later stage (e.g., within an atmospheric forward model). They represent the “compressed” opacity.

cortecs.fit.fit_pca.fit_pca(cross_section, P, T, prep_res, fit_axis='pressure', **kwargs)

Fits the PCA to the opacity data.

Inputs

cross_section:

(ntemp x npressure) the array of cross-sections being fit.

P:

pressure grid

T:

temperature grid

xMat:

(npres x nc) PCA components

rtype:

beta: (nc x pixels) PCA coefficients

cortecs.fit.fit_pca.move_cross_section_axis(cross_section, fit_axis, dim=2)

todo: add docstring :param cross_section: :param fit_axis: :return:

cortecs.fit.fit_pca.prep_pca(cross_section, wav_ind=-1, nc=2, force_fit_constant=False, fit_axis='pressure')

Prepares the opacity data for PCA. That is, it calculates the PCA components to be fit along the entire dataset by fitting the PCA to a single wavelength.

todo: perform type checking for inputs.

Inputs

cross_section:

(ntemp x npressure x nwavelength) the array of cross-sections being fit.

wav_ind:

(int) index of wavelength to fit. Ideally, this should be fit to a wavelength that has somewhat

representative temperature–pressure structure to the opacity function. If there is no temperature or wavelength dependence at this wavelength index but strong dependence elsewhere, for instance, the opacity function will be poorly reconstructed. :nc: (int) number of PCA components to keep. Increasing the number of components can make the reconstruction of opacity data more accurate, but it can also lead to overfitting and make the model size larger (i.e., decrease the compression factor). :force_fit_constant: (bool) if True, will allow the PCA to fit an opacity function without temperature and pressure dependence. This usually isn’t recommended, if these PCA vectors are to e used to fit other wavelengths that do have temperature and pressure dependence. :fit_axis: (str) the axis to fit against. determines the shape of the final vectors and components. if “best”, chooses the largest axis. otherwise, can select “temperature” or “pressure”.

rtype:

xMat: (n_exp x nc) PCA components

cortecs.fit.fit_pca.save_pca(savename, fit_results)

Saves the PCA components and coefficients to files

Inputs

savename:

(str) if not None, the PCA components will be saved to this filename.

fit_results:

contains the PCA coeefficients and vectors.

cortecs.fit.fit_pca.standardize_cube(input_array)

Prepares an array for PCA.

Inputs

input_array:

(ntemperature x npressure) array for PCA.

rtype:

standardized_cube: (ntemperature x npressure) standardized (mean-substracted, standard deviation-divided) array for PCA.

Trains a neural network to fit the opacity data.

cortecs.fit.fit_neural_net.fit_neural_net(cross_section, P, T, wl, n_layers=3, n_neurons=8, activation='sigmoid', learn_rate=0.04, loss='mean_squared_error', epochs=2000, verbose=0, sequential_model=None, plot=False)

trains a neural network to fit the opacity data.

Inputs

cross_section:

(ntemp x npressure) the array of cross-sections being fit.

P:

(array) pressure grid corresponding to the cross-sections.

T:

(array) temperature grid corresponding to the cross-sections.

wl:

(array) wavelength grid corresponding to the cross-sections.

n_layers:

(int) number of layers in the neural network. Increasing this number increases the flexibility of the model, but it also increases the number of parameters that need to be fit — and hence the model size and training time.

n_neurons:

(int) number of neurons in each layer. Increasing this number increases the flexibility of the model, but it also increases the number of parameters that need to be fit — and hence the model size and training time.

activation:

(str) activation function to use. See the keras documentation for more information: https://keras.io/api/layers/activations/

learn_rate:

(float) learning rate for the optimizer. Increasing this number can help the model converge faster, but it can also cause the model to diverge.

loss:

(str) loss function to use. See the keras documentation for more information: https://keras.io/api/losses/

epochs:

(int) number of epochs to train for. Increasing this number can help the model converge better, but it primarily makes the training time longer.

verbose:

(int) verbosity of the training.

sequential_model:

(keras.Sequential) if not None, this is the neural network to be trained.

plot:

(bool) whether to plot the loss.

returns:

• history: the history object from the keras fit method

• neural_network: the trained neural network

todo: implement all_wl

cortecs.fit.fit_neural_net.prep_neural_net(cross_section)

cortecs.fit.fit_neural_net.save_neural_net(filename, fit_results)

saves the neural network to a file.

Inputs

filename:

(str) filename to save the neural network to.

fit_results:

(keras.Sequential) the neural network to save.

rtype:

None:

cortecs.fit.fit_neural_net.unravel_data(x, y, z=None, tileboth=False)

unravels the data into a single column. takes log the log of quantities as well.

todo: move to a utils file?

Inputs

x: array-like

first dimension of data.

y: array-like

second dimension of data.

z: array-like

third dimension of data.

Opac

A few functions that can be used to chunk opacity files. Assumes that opacity files are in the RT code format, that each opacity file to be chunked is in its on folder, and that we are currently in that folder.

Example instructions / workflow:

>>> file = 'opacTiO.dat'
>>> chunk_wavelengths(file, wav_per_chunk=2598)
>>> filename = 'opacTiO'
>>> add_overlap(filename)

And that should work!

There’s some replicated (and likely unnecessary) code, but it hopefully shouldn’t be too confusing. Furthermore, these functions have not been subjected to robust unit testing, so they might not work right out of the box. Please let me know if that’s the case, and I’d be happy to debug :)

author: @arjunsavel

so far, only works for exo-transmit chunks. TODO: generalize for others.

cortecs.opac.chunking.add_lams(max_lam_to_add_ind, file, next_file)

Takes the first max_lam_to_add_ind opacities from next_file and appends them to file.

Inputs:

max_lam_to_add_ind:

(int) number of wavelength points to add from one file to the other.

file:

(str) (str) path to file to which wavelength points are being added.

next_file:

(str) path to file from which wavelength points are being drawn.

Outputs:

None

Side effects:

Modifies file.

cortecs.opac.chunking.add_overlap(filename, v_max=11463.5)

Adds overlap from file n+1 to file n. The last file has nothing added to it. This step is necessary for the Doppler-on version of the RT code.

*Assumes that the current directory only has files labeled ‘filename*.dat’*

Inputs:

filename:

(str) “base name” of the opacity chunks. e.g., ‘opacFe’, corresponding to ‘opacFe*.dat’.

v_max:

(float, default 11463.5) maximum velocity that will be Doppler-shifted to in the RT code. Twice this value is used to calculate how much overlap to include (+ an extra 20 wavelength points, to be safe!). The default comes from Tad’s WASP-76b GCM outputs. [m/s]

Output:

None

Side effects:

Modifies every ‘filename*.dat’ file.

cortecs.opac.chunking.chunk_wavelengths(file, nchunks=None, wav_per_chunk=None, adjust_wavelengths=False, loader='exotransmit')

Performs wavelength-chunking. todo: yell if the chunked results are already in the directory?

Inputs

file:

path to file to be chunked. e.g., ‘opacFe.dat’.

nchunks:

(int or None) Number of chunks to use, splitting the opacity file into roughly even chunks. If None, wav_per_chunk must be specified.

wav_per_chunk:

(int or None) Number of wavelengths per chunk. If None, nchunks must be specified.

adjust_wavelengths:

(bool) whether or not to scale from CGS to MKS. For most situations, can be kept false.

Outputs

None

Side effects

Creates a number of files in same directory as file, titled file*.dat.

cortecs.opac.chunking.get_header(file)

Gets the header of a file.

Inputs:

file:

(str) path to file whose header we want.

Outputs:

header of file (string)

cortecs.opac.chunking.write_to_file(line, file, file_suffix, numfiles=None)

Writes (appends, really) a line to a file.

Inputs:

line:

(str) line to write to file.

file:

(str) base path to file to be written to. e.g., ‘opacFe’

file_suffix:

(int) suffix to add, usually chunk number. e.g. 5

Outputs:

None

Side effects:

Writes a line to a file!

Functions that can be used to interpolate CIA to a higher resolution and subsequently chunk it up. Example instructions / workflow: >>> reference_file = ‘../opacFe/opacFe.dat’ >>> CIA_file = ‘opacCIA.dat’ >>> interpolate_cia(CIA_file, reference_file) >>> CIA_file = ‘opacCIA_highres.dat’ >>> ref_file_base = ‘../opacFe/opacFe’ >>> chunk_wavelengths_cia(CIA_file, ref_file_base) And that should work!

todo: generalize to PLATON CIA.

author: @arjunsavel

cortecs.opac.interpolate_cia.add_line_string_species(line_string, species_dict, i, buffer)

Adds the species to the line string. Inputs ——

line_string:

(str) the line string to be modified.

species_dict:

(dict) dictionary of species to be interpolated.

i:

(int) index of the line string.

buffer:

(str) buffer between each species.

Outputs

None

Side effects

Modifies line_string in place.

cortecs.opac.interpolate_cia.build_new_string(interped_wavelengths, interped_temps, reference_species, species_dict_interped)

Builds the new string that will be written to the new CIA file. Inputs ——

interped_wavelengths:

(list) list of wavelengths that have been interpolated.

interped_temps:

(list) list of temperatures that have been interpolated.

reference_species:

(list) list of species that have been interpolated.

buffer:

(str) buffer between each species.

Outputs

new_string:

(list) list of strings that will be written to the new CIA file.

Side effects

None

cortecs.opac.interpolate_cia.check_temp_grid(df, real_temperature_grid, cia_file)

Checks that the temperature grid of the CIA file is the same as the reference file. Inputs ——

df:

(pd.DataFrame) dataframe with the CIA data.

real_temperature_grid:

(np.array) temperature values for which the opacity file had been computed.

Outputs

None

Side effects

Raises a ValueError if the temperature grid is not the same.

cortecs.opac.interpolate_cia.check_wav_grid(reference_opac, df)

Checks that the wavelength grid of the CIA file is the same as the reference file. Inputs ——

reference_opac:

(Opac) Opac object with the wavelength grid of interest.

df:

(pd.DataFrame) dataframe with the CIA data.

Outputs

None

Side effects

Raises a ValueError if the wavelength grid is not the same.

cortecs.opac.interpolate_cia.chunk_wavelengths_cia(file, ref_file_base, numfiles)

Performs chunking based on the reference file’s wavelength chunking. Inputs ——-

file:

(str) path to CIA file that should be chunked. e.g., opacCIA_highres.dat.

ref_file_base:

(str) path base to a set of reference files that are already chunked on the desired wavelength grid. e.g., ../opacFe/opacFe

cortecs.opac.interpolate_cia.get_wav_per_chunk(file_suffix, ref_file_base)

Grabs the number of wavelengths of a given chunk. Inputs ——

file_suffix:

(int) number corresponding to the given chunk. e.g., 1.

ref_file_base:

(str) path base to a set of reference files that are already chunked on the desired wavelength grid. e.g., ../opacFe/opacFe

Outputs

len_grid:

(int) number of wavelengths in chunk.

cortecs.opac.interpolate_cia.initialize_species_dict(species_dict, wavelength_grid)

Initializes a dictionary of species to be interpolated. Inputs ——

species_dict:

(dict) dictionary of species to be interpolated.

Outputs

None

Side effects

Modifies species_dict in place.

cortecs.opac.interpolate_cia.interpolate_cia(cia_file, reference_file, outfile=None, loader='exotransmit', load_kwargs=None)

Interpolates a CIA file to a higher resolution, using the wavelength grid of a reference file. Note: This function assumes that the CIA file has Hels, HeHs, CH4CH4s, H2Hes, H2CH4s, H2Hs, H2H2s, and CO2CO2s.

TODO: add 2d interpolation. Inputs ——

CIA_file:

(str) path to CIA file to be interpolated. e.g., ‘opacCIA/opacCIA.dat’

reference_file:

(str) path to opacity file with the wavelength grid of interest. e.g., ‘opacFe/opacFe.dat’

Outputs

None

Side effects

Creates a file with “hires” attached to the end of CIA_file that has been interpolated to higher resolution.

cortecs.opac.interpolate_cia.prepend_line(file_name, line)

Insert given string as a new line at the beginning of a file. Used to fix the header within the chunked CIA files.

Inputs

file_name:

(str) path to file to be modified.

line:

(str) line to be inserted.

Outputs

None

Side effects

Modifies the file in place.

Reads opacity data from various sources.

author: @arjunsavel

cortecs.opac.io.get_empty_species_dict(CIA_file, verbose=False)

returns a species dictioanry given a CIA file

Inputs

CIA_file:

(str) path to CIA file. e.g., ‘opacCIA/opacCIA.dat’

verbose:

(bool) whether to print out the species that are likely in the file.

Outputs

species_dict:

(dict) dictionary of species.

cortecs.opac.io.get_n_species(CIA_file, verbose=False)

Returns the number of species in a CIA file. Inputs ——-

CIA_file:

(str) path to CIA file. e.g., ‘opacCIA/opacCIA.dat’

Outputs

n_species:

(int) number of species in the CIA file.

class cortecs.opac.io.loader_base(wl_key='wno', T_key='T', P_key='P', cross_section_key='xsec', wl_style='wno', temperature_axis=0, pressure_axis=1, wavelength_axis=2)

Bases: object

loads in opacity data from various sources. To be passed on to Opac object.

todo: tutorial on how to use this?

load(filename)

loads in opacity data from various sources. To be passed on to Opac object.

Inputs

filenamestr

name of file to load

Outputs

wlnp.ndarray

wavelengths

Tnp.ndarray

temperatures

Pnp.ndarray

pressures

cross_sectionnp.ndarray

cross sections for the opacity

class cortecs.opac.io.loader_chimera(wl_key='wno', T_key='T', P_key='P', cross_section_key='xsec', wl_style='wno', temperature_axis=0, pressure_axis=1, wavelength_axis=2)

Bases: loader_base

loads in opacity data that are produced with the CHIMERA code.

class cortecs.opac.io.loader_exotransmit(wl_key='wno', T_key='T', P_key='P', cross_section_key='xsec', wl_style='wno', temperature_axis=0, pressure_axis=1, wavelength_axis=2)

Bases: loader_base

loads in opacity data in the exo-transmit format. This format is also used by PLATON, I believe.

get_lams_and_opacities(file)

Takes in an opacity file and returns an array of all wavelengths within the file.

Returns the opacities, as well — so that the file is only iterated through once.

Inputs:

file:

(str) path to opacity file.

Outputs:

wavelengths:

(numpy.array) individual wavelength points within the opacity file [m]

get_t_p(file)

Gets the temperatures and pressures of a file from its header.

Inputs:

file:

(str) path to file whose header we want.

Outputs:

header of file (string)

load(filename)

Loads file.

Inputs

filenamestr

name of file to load

Outputs

wlnp.ndarray

wavelengths

Tnp.ndarray

temperatures

Pnp.ndarray

pressures

class cortecs.opac.io.loader_exotransmit_cia(wl_key='wno', T_key='T', P_key='P', cross_section_key='xsec', wl_style='wno', temperature_axis=0, pressure_axis=1, wavelength_axis=2)

Bases: loader_base

loads in opacity data that are used with the PLATON code’s collision-induced absorption..

check_line_break(line, temperature)

Checks whether the given line should be skipped

load(cross_sec_filename, verbose=False)

loads in opacity data that’s built for exo-transmit. To be passed on to Opac object.

Inputs

CIA_file:

(str) path to CIA file. e.g., ‘opacCIA/opacCIA.dat’

Outputs

df:

(pd.DataFrame) dataframe with the CIA data.

class cortecs.opac.io.loader_helios(wl_key='wno', T_key='T', P_key='P', cross_section_key='xsec', wl_style='wno', temperature_axis=0, pressure_axis=1, wavelength_axis=2)

Bases: loader_base

loads in opacity data that are produced with the HELIOS ktable function.

P_key = 'pressures'

T_key = 'temperatures'

cross_section_key = 'opacities'

wl_key = 'wavelengths'

wl_style = 'wl'

class cortecs.opac.io.loader_platon(wl_key='wno', T_key='T', P_key='P', cross_section_key='xsec', wl_style='wno', temperature_axis=0, pressure_axis=1, wavelength_axis=2)

Bases: loader_base

loads in opacity data that are used with the PLATON code.

load(cross_sec_filename, T_filename='', P_filename='', wl_filename='')

loads in opacity data that’s built for PLATON. To be passed on to Opac object.

The temperature grid, pressure grid, and wavelength grid are saved as separate files for PLATON.

Inputs

filenamestr

name of file to load

cross_sec_filenamestr

name of cross section file

T_filenamestr

name of temperature file

P_filenamestr

name of pressure file

wl_filenamestr

name of wavelength file

species_weight_dict = {'C2H2': 26.038, 'C2H4': 28.054, 'CH4': 16.04, 'CH4+': 16.04, 'CH4-': 16.04, 'CO': 28.01, 'CO2': 44.009, 'FeH': 55.845, 'H': 1.008, 'H+': 1.008, 'H-': 1.008, 'H2': 2.016, 'H2+': 2.016, 'H2-': 2.016, 'H2O': 18.015, 'H2O+': 18.015, 'H2O-': 18.015, 'H2S': 34.08, 'H3+': 3.024, 'H3-': 3.024, 'HCN': 27.026, 'He': 4.003, 'He+': 4.003, 'K': 39.098, 'NH3': 17.031, 'Na': 22.99, 'PH3': 33.997, 'TiO': 63.866, 'VO': 66.94, 'e-': 0.00054858}

class cortecs.opac.io.loader_platon_cia(wl_key='wno', T_key='T', P_key='P', cross_section_key='xsec', wl_style='wno', temperature_axis=0, pressure_axis=1, wavelength_axis=2)

Bases: loader_base

loads in opacity data that are used with the PLATON code’s collision-induced absorption..

load(cross_sec_filename, T_filename, wl_filename, species_name)

loads in opacity data that’s built for PLATON. To be passed on to Opac object.

The temperature grid, pressure grid, and wavelength grid are saved as separate files for PLATON.

Inputs

filenamestr

name of file to load

cross_sec_filenamestr

name of cross section file

T_filenamestr

name of temperature file

wl_filenamestr

name of wavelength file

species_nametuples

name of two colliding species. E.g., (‘H2’, ‘CH4’). todo: all at once?

class cortecs.opac.io.writer_base

Bases: object

class cortecs.opac.io.writer_exotransmit_cia

Bases: writer_base

does writing for the CIA object, takng in an opac object.

append_line_string(new_string, i, temp)

write(opac, outfile, verbose=False)

loads in opacity data that’s built for exo-transmit. To be passed on to Opac object.

Inputs

CIA_file:

(str) path to CIA file. e.g., ‘opacCIA/opacCIA.dat’

Outputs

df:

(pd.DataFrame) dataframe with the CIA data.

this file holds the class describing opacity data. hmm…maybe I want the loader in a different file?

author: @arjunsavel

class cortecs.opac.opac.Opac(filename, loader='chimera', load_kwargs=None, loader_kwargs=None)

Bases: object

this class holds the opacity data and provides methods for evaluating the opacity at a given temperature and pressure.

Everything’s already been loaded into memory, so this is just a wrapper for the data.

todo: use the different loaders

P = None

T = None

copy()

method_dict = {'chimera': <class 'cortecs.opac.io.loader_chimera'>, 'exotransmit': <class 'cortecs.opac.io.loader_exotransmit'>, 'helios': <class 'cortecs.opac.io.loader_helios'>, 'platon': <class 'cortecs.opac.io.loader_platon'>}

wl = None

class cortecs.opac.opac.Opac_cia(filename, loader='platon_cia', view='default')

Bases: Opac

this class holds the opacity data and provides methods for evaluating the opacity at a given temperature and pressure.

Everything’s already been loaded into memory, so this is just a wrapper for the data.

join_cross_section(opac)

joins another opacity’s cross-section data to this one. :param opac: :return:

method_dict = {'exotransmit_cia': <class 'cortecs.opac.io.loader_exotransmit_cia'>, 'platon_cia': <class 'cortecs.opac.io.loader_platon_cia'>}

Optimize

Holds object for optimizing the fits.

class cortecs.opt.opt.Optimizer(fitter, **optim_kwargs)

Bases: object

for optimizing the fits.

method_dict = {'neural_net': <function optimize_neural_net>, 'pca': <function optimize_pca>, 'polynomial': <function optimize_polynomial>}

optimize(max_size, max_evaluations, **kwargs)

optimizes the fit. :return:

Performs simple optimization of neural network hyperparameters.

cortecs.opt.optimize_neural_net.optimize_neural_net(max_size, max_evaluations, opac, min_layers=2, min_neurons=2, max_layers=3, max_neurons=13, min_learn_rate=0.01, max_learn_rate=0.1)

performs simple optimization of neural network hyperparameters.

Inputs

max_size: float

maximum size of file in kB.

max_evaluations: int

maximum number of evaluations of the neural network.

Performs simple optimization of PCA hyperparameters — i.e., number of components and wavelength index for computing eigenvectors.

cortecs.opt.optimize_pca.optimize_pca(max_size, max_evaluations, opac, min_components=3, max_components=5, wav_ind_start=3573)

Inputs

max_size: float

maximum size of file in kB.

max_evaluations: int

maximum number of evaluations of the fitter

cortecs.opt.optimize_polynomial.optimize_polynomial(*args, **kwargs)