bandpasses

The bandpasses of the instrument involved in the Lemaitre dataset: MeagCam, HSC and ZTF.

• The MegaCam bandpasses were re-determined from the decommissioned MegaCam filters by M. Betoule and a team from the Laboratoire des matériaux avancés (LMA). These determinations differ slightly from the passband models that were published along with the SNLS and JLA papers. They are all non-radial. They do not (yet?) include per-CCD quantum efficiency determinations.

• The ZTF bandpasses were re-assembled by P. Rosnet from a series of bench measurements sent by the ZTF team at Caltech. These passbands differ from the models sncosmo (ztfg, ztfr and ztfi), which contains only averaged passbands, which account for (1) the QE difference betweem the single- and double-coated CCDs (2) the passband radial variations due to the coating non-uniformities and to the variations of the incident beam angle w.r.t. the filter surface.

• The HSC passbands are based on the measurements performed by Kawanomoto et al. These passband models are radial. They do not (yet?) include any per-CCD quantum efficiency determinations.

The passband models are distributed with sncosmo>=2.12, which is able to handle non-radial variations as well as per-CCD quantum efficiencies.

This module contains the raw measurements and the code that was used to prepare the sncosmo release. It also contain code that automatically registers the Lemaitre passbands in sncosmo when the module is imported.

We also use the bbf module for fast computation of broadband fluxes. Within bbf, the passbands are projected on a wavelength and position-dependent spline basis, which is cached as a binary hdf5 file. The bbf module is

Getting started

Installation

Using pixi (recommended)

pixi handles all compiled dependencies including scikit-sparse. First add it from conda-forge, then the package itself:

pixi add "scikit-sparse<0.5"
# Install the lastest release on Pypi
pixi add --pypi lemaitre-bandpasses 
# Or from the latest development version on Gitlab
# (UV_GIT_LFS=1 is required due to a known pixi/uv issue: https://github.com/prefix-dev/pixi/issues/2000)
UV_GIT_LFS=1 pixi add --git https://gitlab.in2p3.fr/lemaitre/bandpasses.git --branch main --pypi lemaitre-bandpasses

To work from a local clone instead:

git clone git@gitlab.in2p3.fr:lemaitre/bandpasses.git
cd bandpasses && pixi install

Using pip

scikit-sparse requires the SuiteSparse C library; install it before running pip:

• Linux / macOS — conda: conda install "scikit-sparse<0.5"
• Linux (Debian/Ubuntu) — apt: sudo apt-get install libsuitesparse-dev
• macOS — Homebrew: brew install suite-sparse

Then:

pip install lemaitre-bandpasses
# or from a local clone:
git clone git@gitlab.in2p3.fr:lemaitre/bandpasses.git && cd bandpasses && pip install -e .

Instantiating bandpasses from sncosmo

import sncosmo

# getting the average ZTF/MegaCam6/HSC bandpasses
for name in ['ztf::g', 'ztf::r', 'ztf::I']:
    band = sncosmo.get_bandpass(name)
for name in ['megacam6::' + b for b in ['g', 'r', 'i2', 'z']]:
    band = sncosmo.get_bandpass(name)
for name in ['hsc::' + b for b in ['g', 'r', 'r2', 'i', 'i2', 'z', 'Y']]:
    band = sncosmo.get_bandpass(name)

# getting, e.g. the ZTF r-passband at a given position
r = sncosmo.get_bandpass('ztf::r', x=724, y=2829, sensor_id=55)

# this function can be vectorized
r = sncosmo.get_bandpass('ztf::r', x=[724., 1802, 222.], y=[42., 58., 2512], sensor_id=[5, 42, 22])

Working with bbf.FilterLibs

from lemaitre import bandpasses

flib = bandpasses.get_filterlib(rebuild=False)

The filterlib contains all the lemaitre passbands, projected on specific (adapted) spline bases. Then it is possible to use it to compute efficiently broadband fluxes.

First, load a stellar library. A stellar library is like a FilterLib: it consists in a collection of spectra, projected on a spline basis:

import bbf.stellarlib.pickles
pickles = bbf.stellarlib.pickles.fetch()

With that in hand, one can compute broadband fluxes on all (average passbands) in the library:

fluxes = bbf.flux(flib, pickles)

from bbf.magsys import SpecMagSys

ms = SpecMagSys('AB')
mags = bbf.mag(flib, pickles, magsys=ms)

# or just
mags = bbf.mag(flib, pickles, 'AB')

We may be in a situation where we have actually one passband instance per measurement. bbf handles that:

nmeas = 10_000

star = np.random.choice(np.arange(len(pickles)), size=nmeas
x = np.random.rand(3000., 3000., nmeas)
y = np.random.rand(3000., 3000., nmeas)
sensor_id = np.random.choice(np.arange(64), size=nmeas)

mags = bbf.mags(flib, pickles, star, x, y, sensor_id, magsys='AB')