`torchsky.extinction`¶

Stability: core. Names on this page follow the deprecation policy in contributing.

Milky Way and starburst dust laws on torch tensors.

This reference is generated from the package source. Names are grouped by the module that defines them; import every name from the public namespace torchsky.extinction unless the group text says otherwise.

Top-level namespace¶

Milky Way and starburst dust laws on torch tensors.

Wavelengths are Angstrom-native internally. Pass wavelength_unit='nm' at the boundary. A(λ) is in magnitudes; transmission is 10**(-0.4 A).

The infrared power law is extrapolated below the CCM89 x=0.3 cutoff so broadband work stays defined into the far-IR. Optical and UV pieces follow Cardelli, Clayton & Mathis (1989); odonnell94 replaces the optical polynomials only.

g23 (Gordon et al. 2023), f19 (Fitzpatrick et al. 2019), and gcc09 (Gordon, Cartledge & Clayton 2009) use the published coefficients as implemented in dust_extinction (K. Gordon, BSD 3-Clause), and fail closed outside their calibrated wavelength and R_V ranges instead of extrapolating.

a_lambda ¶

a_lambda(
    wave: Tensor,
    *,
    ebv: Tensor | float | None = None,
    a_v: Tensor | float | None = None,
    r_v: Tensor | float = 3.1,
    law: ExtinctionLaw = "ccm89",
    wavelength_unit: WavelengthUnit = "angstrom",
) -> Tensor

Extinction in magnitudes at wave.

Supply exactly one of ebv or a_v. A_V = R_V E(B-V).

apply_extinction ¶

apply_extinction(
    flux: Tensor,
    wavelength: Tensor | None = None,
    ext_params: Tensor | None = None,
    extinction_law_name: str = "ccm89",
    r_v: float = 3.1,
    *,
    wavelength_unit: WavelengthUnit = "angstrom",
) -> Tensor

Apply a dust screen: ext_params is E(B-V) broadcast against flux.

Parameters:

Name	Type	Description	Default
`flux`	`Tensor`	Flux-density tensor with wavelength on the trailing axis.	required
`wavelength`	`Tensor \| None`	Wavelength samples in `wavelength_unit`.	`None`
`ext_params`	`Tensor \| None`	Color excess `E(B-V)`, broadcastable against `flux`.	`None`
`extinction_law_name`	`str`	Supported extinction-law name.	`'ccm89'`
`r_v`	`float`	Total-to-selective extinction ratio `A(V) / E(B-V)`.	`3.1`
`wavelength_unit`	`WavelengthUnit`	`"angstrom"` or `"nm"`.	`'angstrom'`

Returns:

Type	Description
`Tensor`	Attenuated flux with the same shape, dtype, and device as `flux`.

Examples:

>>> import torch
>>> from torchsky.extinction import apply_extinction
>>> apply_extinction(torch.ones(1), torch.tensor([5500.0]), torch.tensor(0.0)).tolist()
[1.0]

calzetti ¶

calzetti(
    wave_angstrom: Tensor,
    a_v: Tensor,
    r_v: Tensor | float | None = None,
    *,
    wavelength_unit: WavelengthUnit = "angstrom",
) -> Tensor

Calzetti (2000) starburst A(λ). Default R_V = 4.05.

calzetti_a_lambda ¶

calzetti_a_lambda(
    wave_angstrom: Tensor, a_v: Tensor, r_v: Tensor
) -> Tensor

Calzetti (2000) A(λ). Default R_V for the law is 4.05.

calzetti_k_ebv ¶

calzetti_k_ebv(
    wave_angstrom: Tensor, r_v: Tensor
) -> Tensor

Calzetti (2000) k(λ) = A(λ)/E(B-V). λ clamped to 0.12–2.20 μm.

ccm89 ¶

ccm89(
    wave_angstrom: Tensor,
    a_v: Tensor | float,
    r_v: Tensor | float | None = None,
    *,
    wavelength_unit: WavelengthUnit = "angstrom",
) -> Tensor

CCM89 A(λ) for A_V; wavelengths default to Angstrom.

ccm89_ab ¶

ccm89_ab(x_um_inv: Tensor) -> tuple[Tensor, Tensor]

CCM89 a(x), b(x) for inverse-micron x = 1 / λ_μm.

Regions follow Cardelli, Clayton & Mathis (1989): IR x<1.1 (power law extrapolated below x=0.3), optical 1.1<=x<3.3, UV 3.3<=x<8, far-UV 8<=x<=10. Beyond x=10 the far-UV polynomials are evaluated at the x=10 boundary; the law is not defined there.

f19_a_lambda ¶

f19_a_lambda(
    wave_angstrom: Tensor, a_v: Tensor, r_v: Tensor
) -> Tensor

Fitzpatrick et al. (2019) A(λ); 1150 Å to 3.3 μm, 2 <= R_V <= 6.

F19 is normalized at 5500 Å: a_v and r_v are A(55) and R(55), with k(λ-55) = k_3.02 + 0.990 Δk (R(55) - 3.10).

f99_a_lambda ¶

f99_a_lambda(
    wave_angstrom: Tensor, a_v: Tensor, r_v: Tensor
) -> Tensor

Fitzpatrick (1999) A(λ); wavelengths in Angstrom.

fitzpatrick99 ¶

fitzpatrick99(
    wave_angstrom: Tensor,
    a_v: Tensor,
    r_v: Tensor | float | None = None,
    *,
    wavelength_unit: WavelengthUnit = "angstrom",
) -> Tensor

Fitzpatrick (1999) A(λ) for A_V; wavelengths default to Angstrom.

g23_a_lambda ¶

g23_a_lambda(
    wave_angstrom: Tensor, a_v: Tensor, r_v: Tensor
) -> Tensor

Gordon et al. (2023) A(λ); 912 Å to 32 μm, 2.3 <= R_V <= 5.6.

gcc09_a_lambda ¶

gcc09_a_lambda(
    wave_angstrom: Tensor, a_v: Tensor, r_v: Tensor
) -> Tensor

Gordon, Cartledge & Clayton (2009, with 2014 erratum) ultraviolet A(λ).

Defined only for 909-3030 Å (3.3 <= x <= 11) and 2 <= R_V <= 6; the law was not derived for the optical or infrared.

k_lambda ¶

k_lambda(
    wave: Tensor,
    *,
    law: ExtinctionLaw = "ccm89",
    r_v: Tensor | float = 3.1,
    wavelength_unit: WavelengthUnit = "angstrom",
) -> Tensor

Named k(λ) = A(λ)/A_V so k(λ_V) ≈ 1. Host screens use this at λ_rest.

odonnell94 ¶

odonnell94(
    wave_angstrom: Tensor,
    a_v: Tensor,
    r_v: Tensor | float | None = None,
    *,
    wavelength_unit: WavelengthUnit = "angstrom",
) -> Tensor

O'Donnell (1994) A(λ) for A_V; wavelengths default to Angstrom.

odonnell94_ab ¶

odonnell94_ab(x_um_inv: Tensor) -> tuple[Tensor, Tensor]

O'Donnell (1994) optical update; CCM89 in UV and IR.

transmission ¶

transmission(
    wave: Tensor,
    ebv: Tensor | float,
    *,
    r_v: Tensor | float = 3.1,
    law: ExtinctionLaw = "ccm89",
    wavelength_unit: WavelengthUnit = "angstrom",
) -> Tensor

Multiplicative flux attenuation 10**(-0.4 A).

A is clamped at zero: a dust screen cannot emit.

Typing¶

ExtinctionLaw

typing.Literal['ccm89', 'odonnell94', 'odonnell', 'f99', 'fitzpatrick99', 'calzetti', 'g23', 'f19', 'gcc09']

WavelengthUnit

typing.Literal['angstrom', 'nm']