gateau: an observation simulator for ground-based submillimeter astronomy with integral field units and kinetic inductance detectors

arXiv:2604.23305 · astro-ph.IM · Submitted 2026-04-25 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Astrophysics. The day in astrophysics: observations, models and what the data forces.

Vera: I'm Vera.

Jocelyn: And I'm Jocelyn, here with Subrahmanyan.

Vera: We're looking at the paper: "gateau: an observation simulator for ground-based submillimeter astronomy with integral field units and kinetic inductance detectors".

Jocelyn: Here's the gist of the paper. Submillimeter integral field units utilizing kinetic inductance detectors are a promising instrument architecture for studying galaxies and large-scale structure of the Universe This paper developed gateau, a modular, flexible, and efficient simulator for submm IFU observations of astronomical sources The simulator consists of a Python interface powered by a C/C++ backend that uses CUDA for GPU-acceleration Simulator Architecture gateau simulates observations by taking user input such as an astronomical source, a set of atmospheric screens, a scan pattern, and telescope and instrument parameters The source signal is propagated through a dynamical model of the atmosphere and optical path at the telescope using a customised radiative transfer cascade A dispersive element model, which can be a filterbank, grating, or user-supplied model, is used to calculate the total power per spectral channel for each spatial pixel on the IFU A physically motivated photon-noise model is used to add a white noise component to the received power Detector noise is added as temporally correlated pink noise The output is stored in the form of time-ordered datasets.

Vera: Atmospheric and Optical Modeling The atmosphere model currently used in gateau consists of user-supplied screens created by the Astronomical Radio Interferometer Simulator (ARIS) software The atmospheric screen is translated with a constant windspeed along the azimuthal direction, assuming a "frozen-flow" atmospheric fluctuation model To convert the pointing of some spaxel i at some time t to an (xi(t), yi(t)) coordinate on the moving PWV screen, a constant column height hcolumn above the telescope is assumed The atmospheric transmission at zenith ηatm,z(t, ν) is calculated using the Atmospheric Transmission at Microwaves (ATM) software and corrected for elevation-dependent airmass.

Jocelyn: Instrument and Noise Components The instrument is defined by setting the spectral and spatial characteristics, along with the material of the KID Currently, for materials gateau only supports hybrid NbTiN-Al KIDs.

Vera: The spectral channels of the instrument are modeled by Lorentzians ηfilt.

Paper summary: Vera: So, looking at the paper "gateau: an observation simulator for ground-based submillimeter astronomy with integral field units and kinetic inductance detectors", the main thing is that they developed this modular, flexible simulator.

Jocelyn: It’s a way to create time-ordered datasets that resemble real instrument output so researchers can test their analysis pipelines on synthetic data instead of just relying on observed data.

Subrahmanyan: The implication for us is that we can now test how our software handles the complex noise and atmospheric effects inherent in submillimeter IFU observations without needing to dedicate huge telescope time upfront.

Vera: They are using a GPU-accelerated C plus plus backend with CUDA to make this process efficient, which addresses the need for speed when simulating these kinds of complex observations.

Jocelyn: The work validates the entire pipeline by showing it can reproduce real observations of the atmosphere and Uranus with a median ratio agreement within a factor of two across all channels.

Subrahmanyan: It’s important that they also include physically motivated noise models, like white and pink noise for the detector, which shows they aren't just throwing random numbers at the problem; they are modeling what’s actually happening.

Vera: The limitations mentioned are that currently, for materials gateau only supports hybrid NbTiN al KIDs, so it’s not yet universal for all detector types.

Jocelyn: And another point is that they're still working on adding support for other atmosphere generators besides ARIS, which limits the immediate breadth of atmospheric modeling.

Subrahmanyan: Overall, this paper provides a suitable tool for simulating observations across different science cases for large KIDbased submm IFUs, making it a useful resource for cosmology and structure formation studies.

Conclusion: Vera: So, gateau is this simulator they built for submillimeter observations using kinetic inductance detectors and integral field units, right?

Jocelyn: Yeah, it’s a way to test how real data looks before you even get it from the telescope, so we can see what our analysis pipeline can actually do.

Subrahmanyan: It basically creates these time-ordered datasets that mimic what an actual observation would look like across all those different spatial pixels.

Vera: The authors, they’re using this modular design with a Python interface powered by a C plus plus backend that uses CUDA for GPU acceleration to make it run fast.

Jocelyn: Fast is key because these kinds of simulations have to be pretty complex, dealing with the atmosphere and the instrument noise all at once.

Subrahmanyan: It’s about simulating the whole path from the source signal through a dynamical model of the atmosphere, then hitting a dispersive element model for each spatial pixel.

Vera: And they add these physical noise models, like photon noise for white noise and temporally correlated pink noise for detector noise.

Jocelyn: That means they’re not just throwing random numbers in; they're modeling the actual physics of how the signal gets noisy.

Subrahmanyan: The validation against DESHIMA two point zero is interesting because they showed a median ratio agreement to within a factor of two for atmospheric and Uranus observations.

Vera: That suggests the simulation is actually capturing the key physical effects well enough for real-world science testing.

Jocelyn: It shows this tool can be used across different science cases, even simulating something like the Sunyaev-Zel’dovich effect in a galaxy cluster.

Subrahmanyan: The main point is that gateau is a suitable tool for testing various scenarios for large kinetic inductance detector based submillimeter integral field units.

Vera: So, this simulator gives us a solid workbench to develop and test our observational strategies before we commit telescope time.

A. Moerman, N. Soshnin, S. A. Brackenhoff, S. O. Dabironezare, K. Karatsu, L. H. Marting, S. A. H., e Roos M., B R., A Endo

Faculty of Electrical Engineering, Mathematics and Computer Science, Delft University of Technology · SRON—Netherlands Institute for Space Research · Huygens-Kamerlingh Onnes Laboratory, Leiden Institute of Physics, Leiden University · Faculty of Applied Sciences, Delft University of Technology · Leiden Observatory, Leiden University · Faculty of Aerospace Engineering, Delft University of Technology

astro-ph.IM

Submitted: 2026-04-25

Updated: 2026-04-25

Comments: 13 pages, 8 figures

Journal ref: A&A 713, A330 (2026)

DOI: 10.1051/0004-6361/202660606

Project page: https://tifuun.github.io/gateau

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 82/100

The gist: Submillimeter integral field units utilizing kinetic inductance detectors are a promising instrument architecture for studying galaxies and large-scale structure of the Universe This paper developed…

Key concepts

Submillimeter IFU
This refers to an instrument used in astronomy that observes a large area of the sky at submillimeter wavelengths using an integral field unit. These units allow astronomers to take detailed, spatially resolved images of astronomical objects, which is crucial for studying galaxies and the large-scale structure of the Universe.
Kinetic Inductance Detectors (KIDs)
KIDs are a type of detector used in this instrument architecture that measure radiation. They work by detecting changes in kinetic inductance caused by incident photons. This technology is promising for submillimeter astronomy because it offers high sensitivity, which is necessary for studying faint astronomical sources.
Atmospheric Modeling
Gateau incorporates models to simulate how the Earth's atmosphere affects the observation of a source. It uses user-supplied screens and models like ARIS to account for atmospheric transmission and movement, such as wind speed and elevation, ensuring accurate simulation of the signal received by the telescope.
Photon-Noise Model
This model is used to add white noise to the simulated received power, which represents fundamental quantum limits on measurement. It is physically motivated to ensure that the simulated noise accurately reflects real photon statistics during observation.

Terminology

Summary

Submillimeter integral field units utilizing kinetic inductance detectors are a promising instrument architecture for studying galaxies and large-scale structure of the Universe This paper developed gateau, a modular, flexible, and efficient simulator for submm IFU observations of astronomical sources The simulator consists of a Python interface powered by a C/C++ backend that uses CUDA for GPU-acceleration

Simulator Architecture

gateau simulates observations by taking user input such as an astronomical source, a set of atmospheric screens, a scan pattern, and telescope and instrument parameters The source signal is propagated through a dynamical model of the atmosphere and optical path at the telescope using a customised radiative transfer cascade A dispersive element model, which can be a filterbank, grating, or user-supplied model, is used to calculate the total power per spectral channel for each spatial pixel on the IFU A physically motivated photon-noise model is used to add a white noise component to the received power Detector noise is added as temporally correlated pink noise The output is stored in the form of time-ordered datasets.

Atmospheric and Optical Modeling

The atmosphere model currently used in gateau consists of user-supplied screens created by the Astronomical Radio Interferometer Simulator (ARIS) software The atmospheric screen is translated with a constant windspeed along the azimuthal direction, assuming a frozen-flow atmospheric fluctuation model To convert the pointing of some spaxel i at some time t to an (xi(t), yi(t)) coordinate on the moving PWV screen, a constant column height hcolumn above the telescope is assumed The atmospheric transmission at zenith ηatm,z(t, ν) is calculated using the Atmospheric Transmission at Microwaves (ATM) software and corrected for elevation-dependent airmass.

Instrument and Noise Components

The instrument is defined by setting the spectral and spatial characteristics, along with the material of the KID Currently, for materials gateau only supports hybrid NbTiN-Al KIDs. The spectral channels of the instrument are modeled by Lorentzians ηfilt. p. 3

Where the paper says it

For the hybrid KIDs, this corresponds to a cutoff frequency of about 90 GHz. The spectral channels of the instrument are modeled by Lorentzians ηfilt, defined by a central frequency f and resolving power R: ηfilt(f, ν) = ηpeak(f) 4R2 (ν/ f − 1)2 , (6) where ηpeak(f) is the peak height of each Lorentzian in isolation, i.

White noise is added to the output using a physically motivated photon noise model. p. 0
Where the paper says it

A physically motivated photon-noise model is used to add a white noise component to the received power. Detector noise is added as temporally correlated pink noise.

User-defined temporally correlated pink noise can also be included. p. 1
Where the paper says it

White noise is added to the output using a physicallymotivated photon noise model. User-defined temporally correlated pink noise can also be included.

Validation and Performance

Gateau was validated against observations with DESHIMA 2.0, a superconducting, ultra-wideband spectrometer utilizing KIDs We show that we can reproduce real observations of the atmosphere and Uranus with gateau simulations. p. 0

Where the paper says it

0, a superconducting, ultra-wideband spectrometer utilising KIDs and on-chip filterbank technology. We show that we can reproduce real observations of the atmosphere and Uranus with gateau simulations.

The median ratio across all channels shows a much closer agreement between DESHIMA 2. p. 6
Where the paper says it

4, we see that, on a channel-to-channel comparison, the ratios vary substantially up to an order of magnitude. However, the median ratio across all channels shows a much closer agreement between DESHIMA 2.

0 and gateau, to within a factor of 2. p. 6
Where the paper says it

However, the median ratio across all channels shows a much closer agreement between DESHIMA 2. 0 and gateau, to within a factor of 2.

Example Use Case

We present an example use case to simulate an observation of a galaxy cluster through the Sunyaev-Zel’dovich effect The per-spaxel simulation time is orders of magnitude less than the observation time itself We conclude that gateau is a suitable tool for simulating observations of different science cases for large KIDbased submm IFUs.

Future Directions

Future work will involve adding the option of defining and scanning a source in equatorial coordinates allowing for the testing of scanning strategies Another improvement will be to extend the atmosphere module to also work for atmosphere generators other than ARIS.

Software Availability

The gateau software, as well as the official documentation, is publicly available at 10.5281/zenodo.17183878. The MockSZ software is publicly available at 10. p. 0

Where the paper says it

To this end, we have developed gateau, a modular, flexible, and efficient simulator for submm IFU observations of astronomical sources. The simulator consists of a Python interface, powered by a C/C++ backend that uses CUDA for GPU-acceleration, and is publicly available and fully open-source.

5281/zenodo.8365066.

Acknowledgements

This work is supported by the European Union (ERC Consolidator Grant No. 101043486 TIFUUN). p. 1

Where the paper says it

The ϕ and θ coordinates are defined in a horizontal coordinate system centered on the telescope to be simulated and throughout the simulation the source remains stationary with respect to the telescope. 1 https://tifuun.

A. M. thanks Michiel Darcis for discussions on the validity of using the frozen-flow atmospheric model. p. 3
Where the paper says it

At the same time, the atmosphere screen is translated with a constant windspeed uw along the azimuthal direction. This assumes a "frozen-flow" atmospheric fluctuation model, which has been found to be a generally valid description using microwave and optical observations (Lay 1997; Poyneer et al.

We thank Y. Asaki for kindly providing us with the ARIS software. p. 2
Where the paper says it

These assumptions reduce the input parameter set, whilst still providing a realistic model that captures the telescope main beam. The atmosphere model currently used in gateau consists of user-supplied screens created by the Astronomical Radio Interferometer Simulator (ARIS) (Asaki et al.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. GPU acceleration for simulation efficiency allows for orders of magnitude faster simulations than observation time itself, enabling rapid end-to-end testing of instrument design and observation strategies.

  2. The simulator can reproduce real observations of the atmosphere and Uranus, allowing quantitative assessment of reduction pipelines, filterbank parameters, and observation strategies to be tested against ground truth data produced by the simulator.

  3. Gateau can simulate long observations in timespans orders of magnitude smaller than the actual observation time, highlighting its applicability and efficiency for complex scientific cases.

  4. The system provides a flexible Python interface powered by a C/C++ backend using CUDA for GPU-acceleration, combining ease of Python scripting with raw computing power to handle large IFUs.

  5. The simulator models the radiative transfer of an arbitrary astronomical source signal through a dynamical model of the atmosphere and optical path at the telescope, down to the detector.

  6. It can model various optical components including reflectors, dielectric lenses and cryostat windows using a customized radiative transfer cascade defined by the user.

  7. The system includes physical photon noise models and temporally correlated pink noise modeling to add white noise and readout or TLS noise components to the received power TODs.

  8. The simulator handles dispersive elements such as filterbanks, gratings, or user-supplied models, allowing for accurate calculation of total power per spectral channel on each spatial pixel.

  9. It implements a physically motivated atmospheric model using user-supplied screens created by ARIS software to simulate attenuation and contamination of the incoming astronomical source signal.

  10. The system can calculate time dependent atmospheric transmission at zenith ηatm,z(t, ν) using a 2 dimensional lookup table for computational efficiency, assuming only PWV affects this quantity.

  11. The simulator allows defining custom scan patterns and telescope pointing offsets, enabling the simulation of complex sky surveys and testing scanning strategies for extended sources.

  12. It enables the simulation of faint sources, such as galaxy clusters through the Sunyaev-Zel’dovich effect, by generating mock SZ maps based on isothermal-beta models.

  13. The system includes data reduction steps like common mode subtraction and 2 dimensional binning to process simulated TODs into final flux density maps, allowing testing of these pipelines against real observations.

Abstract

Submillimeter (submm) integral field units (IFUs) utilising kinetic inductance detectors (KIDs) are a promising instrument architecture for the study of galaxies, galaxy clusters, and the large-scale structure of the Universe. In order to design successful experiments targeting these science cases, several aspects such as instrument design, observation and calibration strategies, and data reduction pipelines must be collectively developed, tested, and optimised. This can be achieved through end-to-end simulations of the experiment, allowing for quantitative assessment of the aforementioned aspects. To this end, we have developed gateau, a modular, flexible, and efficient simulator for submm IFU observations of astronomical sources. The simulator consists of a Python interface, powered by a C/C++ backend that uses CUDA for GPU-acceleration, and is publicly available and fully open-source. gateau simulates observations by taking user input such as an astronomical source, a set of atmospheric screens, a scan pattern, and telescope and instrument parameters. It then propagates the source signal to the detectors. A physically motivated photon-noise model is used to add a white noise component to the received power. Detector noise is added as temporally correlated pink noise. The output is stored in the form of time-ordered datasets. We validated gateau against observations with DESHIMA 2.0, a superconducting, ultra-wideband spectrometer utilising KIDs and on-chip filterbank technology. We show that we can reproduce real observations of the atmosphere and Uranus with gateau simulations. Lastly, we present a use case to show how gateau can simulate long observations in timespans orders of magnitude smaller than the observation time itself, highlighting its applicability and efficiency.

Related papers