slicersim 1.4.0

Realistic simulations of integral field spectrograph observations — the Lazuli image slicer simulator.

slicersim turns a spectrum on the sky into the spectrum you would actually measure. It propagates a scene through the telescope, the spectrograph and the detector, adds every noise source, and tells you how long you need to expose to reach a given signal-to-noise.

It is the reference simulator of the Lazuli Space Observatory image slicer, but its core is generic and applies to any slicer or micro-lens array IFU.

📖 Documentation: https://slicersim.readthedocs.io

• Exposure time calculator: exposure time and detector read-out mode to reach a requested signal-to-noise, in one call.
• Realistic spectra and cubes: photon, read-out, dark, thermal and background noise all accounted for.
• Built-in sources: Type Ia supernovae (SALT, Twins Embedding), kilonovae (POSSIS), CalSpec standards, black bodies, power laws, or your own spectrum.
• Every knob is accessible: detector, spectrograph sampling, throughput, scene… and see at once how each noise source contributes.

Installation

pip install slicersim

Python ≥ 3.10 is required. See the installation page for development installs.

Exposure time in one line

>>> import numpy as np
>>> import slicersim

Any spectrum, here a flat spectrum normalised to B = 21 mag:

>>> lbda_ref = np.linspace(3000, 20_000, 500)  # [Å], larger than the Lazuli bandpass
>>> flux_ref = np.ones(lbda_ref.shape)         # flat spectrum
>>> exptime, target = slicersim.lazuli_etc(lbda_ref, flux_ref, snr=20, mag=21, band="bessellb")
>>> print(round(exptime, 2))  # [s]
328.28

A Type Ia supernova (SALT model):

>>> exptime, snia = slicersim.lazuli_sn_etc(snr=20, model="salt", redshift=1.2, x1=1.5, c=0.2, phase=-2.2)
>>> print(round(exptime, 2))  # [s]
10142.72
>>> snia.get_readout_config()
{'nmd': (64, 8, 0), 'nramps': 7}

Quick look

import slicersim

# a SN Ia (see also LazuliKilonova, LazuliCalSpec, LazuliBlackBody, LazuliTarget(lbda, flux)...)
snia = slicersim.LazuliSupernova(redshift=0.8, c=0.2, x1=-1.2, phase=1.5)

# set the detector read-out to reach a signal-to-noise of 20
_ = snia.setup_to_snr(20)

# the expected observed spectrum
lbda, flux_1, variance_1 = snia.get_spectrum(unit="adu")

# change the target properties
# (the read-out is not updated: this new target will not have snr=20)
snia.change_properties(redshift=1.2)
lbda, flux_2, variance_2 = snia.get_spectrum(unit="adu")

and show the simulated spectra:

import matplotlib.pyplot as plt
import numpy as np

fig, ax = plt.subplots(figsize=[7, 3])
for flux, variance, label in [(flux_1, variance_1, "z=0.8"), (flux_2, variance_2, "z=1.2")]:
    ax.plot(lbda, flux)
    ax.fill_between(lbda, flux - np.sqrt(variance), flux + np.sqrt(variance),
                    alpha=0.3, label=label)

ax.legend(frameon=False, fontsize="small")
ax.set(xlabel=r"wavelength [$\AA$]", ylabel="flux [ADU]")

Going deeper: the Simulation

Every target is backed by a Simulation, which chains a scene (what is observed), a telescope, a spectrograph and a detector. It gives access to every parameter and to the origin of the variance:

import slicersim

config = slicersim.get_config(instrument="lazuli_cbe.toml")
sim = slicersim.Simulation.from_config(config)

# set the target you want
sim.update(pointsource__redshift=1.0, x1=0, c=0)

# find the read-out mode reaching a mean SNR of 20 in [4000, 6800] Å rest-frame
new_config, snr, exptime = sim.fetch_snr(20, lbda_range=[4000, 6800], frame="rest")
sim.update(**new_config)

# expected spectrum [ADU] and the contribution of each noise source
lbda, flux, variance = sim.get_spectrum()
fig = sim.show_variance_sources(flux_calibrated=False)

sim.mutable_parameters lists everything you can change with sim.update(...).

Tutorials

All tutorials are rendered in the documentation.

Beginner

• Exposure time calculator: exposure time and spectrum in one call.
• Lazuli targets: supernovae, CalSpec stars or any spectrum; reach a SNR and get the read-out mode.
• Changing target properties: redshift, phase, magnitude.
• Origin of the variance: which noise source dominates.
• Kilonovae: POSSIS models, viewing angle and phase.
• Populations with Sample: exposure time and data volume across many targets.

Advanced

• Detector and spectrograph and MACC read-out modes
• Instrument properties: QE, throughput, spectral resolution
• Dispersion resolution and resolution vs. spot size
• CalSpec standard stars
• Building a scene and adding a host galaxy

Citing

If slicersim is useful for your work, please cite Rigault et al. (2026).

Credits

Developed by M. Rigault, adapted from the original MLAPerf (v0.18.0) developed by Y. Copin and M. Rigault. Released under the Apache 2.0 license.