Reference Stars¶
The ReferenceSpectra class represents the spectra of the star
being observed. Here is an overview of the class, as well as some of its important attributes and methods:
Overview¶
-
class
chemicalc.reference_spectra.ReferenceSpectra(reference: str, init_res: float = 300000, scale_by_iron: bool = False, alpha_included: bool = True, **kwargs)¶ Object for spectra of a specific reference star
- Parameters
reference (str) – Name of reference star to load (e.g., ‘RGB_m1.5’)
init_res (str) – Initial resolution of high-res reference spectra. Only 300000 is presently included for default spectra. Can be approximate if using custom reference spectra.
scale_by_iron (bool) – If true, scales all elemental abundances by [Fe/H]
alpha_included (bool) – If true, will include an alpha label after the atmospheric parameters and before the other elements (i.e., between v_micro and Li)
**kwargs – see below
- Keyword Arguments
ref_spec_file (str) – Full path to file of reference spectra
ref_label_file (str) – Full path to file of reference spectra labels
- Variables
resolution (Dict[Union[str,float],int]) – Dictionary of resolving powers for each instrument
ref_spec_file – Full path to file of reference spectra
ref_label_file – Full path to file of reference spectra labels
wavelength (Dict[Union[str,float],np.ndarray]) – Dictionary of wavelength arrays for each instrument
spectra (Dict[Union[str,float],np.ndarray]) – Dictionary of spectral grids for each instrument
labels (pd.DataFrame) – Labels corresponding to each spectrum in the grid
gradients (Dict[Union[str,float],pd.DataFrame]) – Dictionary of partial derivatives for each instrument
nspectra (int) – Number of spectra included in the grid
nlabels (int) – Number of labels included
Attributes¶
Spectra¶
ReferenceSpectra.spectra
is a Python dictionary, each entry of which consists of a grid of spectra generated around the stellar labels of the reference star
Each spectral grid is a \(n_{\text{spectra}} \times n_{\text{pixel}}\) numpy.ndarray.
The order of spectra in the pre-computed spectral grids is described below.
Upon initialization, the dictionary consists of only one entry with the key “init” that corresponds to the initial grid
of high-resolution (\(R \sim 300000\)) normalized spectra.
It is read in from reference_spectra_300000.h5 by default or from ref_spec_file if provided.
With each call to ReferenceSpectra.convolve(),
the dictionary is populated with a new grid of spectra that have been convolved and subsampled to match
the resolution and wavelength array of the passed InstConfig object.
The key for the new spectral grid is the name of the instrument configuration, InstConfig.name.
Wavelength¶
ReferenceSpectra.wavelength is a
Python dictionary containing the wavelength array (as a size times n_{text{pixel}} numpy.ndarray)
corresponding to spectral grids.
Upon initialization, the dictionary consists of only one entry with the key “init” that corresponds to the wavelength
array for the initial high-resolution spectral grid.
It is read in from reference_spectra_300000.h5 by default or from ref_spec_file if provided.
With each call to ReferenceSpectra.convolve(),
the dictionary is populated with a new entry corresponding to the wavelength array of the passed
InstConfig object.
The key for the new wavelength array is the name of the instrument configuration, InstConfig.name.
Labels¶
ReferenceSpectra.labels is a
pandas.DataFrame of shape \(n_{\text{labels}} \times n_{\text{spectra}}\),
containing the stellar labels 1 corresponding to each spectrum in the grid.
The DataFrame’s row names (indices) consist of the stellar labels included.
The DataFrame’s column names consist of (arbitrary) ID’s for the individual spectra in the grid.
It is read in from reference_labels.h5 by default or from ref_label_file if provided.
Gradients¶
ReferenceSpectra.gradients is initially an empty Python dictionary.
With each call to ReferenceSpectra.calc_gradient(),
partial derivatives of the spectrum with respect to each stellar label is calculated for one of the sets of spectral grids.
They are added as a new entry in the form of an \(n_{\text{label}} \times n_{\text{pixel}}\) pandas.DataFrame.
The key for the new spectral gradient DataFrame is the name of the instrument configuration, InstConfig.name.
The DataFrame’s row names (indices) consist of the stellar labels included.
The DataFrame’s column names consist of the wavelength of each pixel.
Methods¶
Convolve Spectral Grids¶
-
ReferenceSpectra.convolve(instrument, name: Optional[str] = None) → None¶ Convolves spectra to instrument resolution and samples onto instrument’s wavelength grid
- Parameters
instrument (InstConfig) – Instrument object to convolve and sample spectra onto
name (str) – Name to give spectra. If None, defaults to name of instrument
- Returns
Calculate Spectral Gradients¶
-
ReferenceSpectra.calc_gradient(name: Union[str, chemicalc.instruments.InstConfig], symmetric: bool = True, ref_included: bool = True) → None¶ Calculates gradients of the reference spectra with respect to each label.
- Parameters
name (Union[str,InstConfig]) – Name of convolved spectra to calculate gradient for. Will also accept an InstConfig object and use InstConfig.name.
symmetric (bool) – If True, calculates symmetric gradient around reference labels
ref_included (bool) – If True, expects first spectra to be reference spectra w/ no offsets to any labels. Required for symmetric=False.
- Returns
Zero Out Gradients¶
-
ReferenceSpectra.zero_gradients(name: Union[str, chemicalc.instruments.InstConfig], labels: Union[str, List[str]])¶ Sets gradients of a spectrum to zero for the specified labels. This is equivalent to setting a delta-function prior on those labels (i.e., holding them fixed).
- Parameters
name (Union[str,InstConfig]) – Name of spectra to apply gradient zeroing to. Will also accept an InstConfig object and use InstConfig.name.
labels (Union[str,List[str]]) – List of labels for which to zero gradients
- Returns
Mask Out Gradients¶
-
ReferenceSpectra.mask_wavelength(name: Union[str, chemicalc.instruments.InstConfig], regions: List[Tuple[float, float]]) → None¶ Masks the information content of a spectrum by setting the gradient to zero within the bounds of the mask. Can be used to mimic the masking of skylines, non-LTE lines, or detector gaps.
- Parameters
name (Union[str,InstConfig]) – Name of the spectra to apply mask to
regions (List[Tuple[float,float]]) – List of wavelength bounds on the regions to mask.
- Returns
Get Names of Spectral Grids¶
-
ReferenceSpectra.get_names() → List[str]¶ Get names of all spectra contained in this object
- Return List[str]
List of spectra names that this object contains.
Pre-Computed Reference Stars¶
The spectral grids for the following Reference Stars have been pre-computed using atlas12 1D-LTE model atmospheres
and the synthe radiative transfer code.
Note
See Section 3.2 of Sandford et al. (In Press) for a detailed description of the spectral generation
Name |
\(M_{V}\) (Vega) |
\(T_{eff}\) |
\(\\log(g)\) |
\(v_{turb}\) (km/s) |
\(\log(Z)\) |
[X/H] |
\(\alpha\) included? |
|---|---|---|---|---|---|---|---|
RGB_m0.5 |
-0.5 |
4200 |
1.5 |
2.0 |
-0.5 |
Solar |
No |
RGB_m1.0 |
-0.5 |
4530 |
1.7 |
1.9 |
-1.0 |
Solar |
Yes |
RGB_m1.5 |
-0.5 |
4750 |
1.8 |
1.9 |
-1.5 |
Solar |
Yes |
RGB_m2.0 |
-0.5 |
4920 |
1.9 |
1.9 |
-2.0 |
Solar |
Yes |
RGB_m2.5 |
-0.5 |
5050 |
1.9 |
1.9 |
-2.5 |
Solar |
Yes |
MSTO_m1.5 |
+3.5 |
6650 |
4.1 |
1.2 |
-1.5 |
Solar |
Yes |
TRGB_m1.5 |
-2.5 |
4070 |
0.5 |
2.3 |
-1.5 |
Solar |
Yes |
KGiant |
— |
4800 |
2.5 |
1.7 |
0.0 |
Solar |
No |
Ph_k0i_m0.0 |
— |
4500 |
1.0 |
2.2 |
0.0 |
Solar |
Yes |
Ph_k0i_m1.0 |
— |
4500 |
1.0 |
2.2 |
-1.0 |
Solar |
Yes |
Ph_k5iii_m0.0 |
— |
4800 |
1.5 |
2.0 |
0.0 |
Solar |
Yes |
Ph_k5iii_m1.0 |
— |
4800 |
1.5 |
2.0 |
-1.0 |
Solar |
Yes |
Pre-Computed Spectral Grid Organization¶
The first spectrum in each grid is generated with the stellar labels in the above table.
The next 2 spectra are generated with \(T_{eff}\) offset by \(\pm 50\) K.
The next 2 spectra are generated with \(\\log(g)\) offset by \(\pm 0.1\) dex.
The next 2 spectra are generated with \(v_{turb}\) offset by \(\pm 0.1\) km/s.
The next 2 spectra are generated with [\(\alpha\)/H] offset by \(\pm 0.05\) dex (if “Yes” in the “\(\alpha\) included?” column above).
The next 97x2 spectra are generated with [X/H] offset by \(\pm 0.05\) dex, where X refers to elements with atomic numbers between 3 (Li) and 99 (Es).
In total, these spectral grids will consist of 203 (201) spectra if the offsets to [\(\alpha\)/H] are (not) included.
Custom Reference Stars¶
Some Chem-I-Calc users may wish to use their own spectral grids to calculate the CRLBs for additional reference stars or with
alternative spectral models (e.g., with 3D, non-LTE atmospheres).
This can be done by calling ReferenceSpectra with the keyword
arguments: ref_spec_file and ref_label_file. At present, these files must be in the same format as the default
Chem-I-Calc files (reference_spectra_300000.h5 and reference_labels.h5).
Warning
Use of custom spectral grids has not been thoroughly tested. We welcome feedback and code contributions to improve this functionality (see Contributing to Chem-I-Calc.)
Footnotes
- 1
“Stellar Labels” broadly encompasses both atmospheric parameters (e.g., effective temperature, surface gravity, and microturbulent velocity) and elemental abundances.