Partner laboratories and research teams

Researchers contributing to ECLAT

• Julien Girard

The institute

Some of the members of the LESIA HPA radio team are involved in the NenuFAR project, SKA’s pathfinder, producing data for a wide range of scientific cases in imaging and beamforming mode. The team also participates in LOFAR scientific projects and is involved in the European EXTRACT project.

Past, present and future activities serving SKAO

• Since January 2023, the EXTRACT project, “A distributed data-mining software platform for EXTReme dAta Across the Compute continuum”, aims to develop technical solutions for orchestrating the processing of data from radio telescopes on architectures. The solutions being explored include the management and dynamic allocation of computing resources and data partitioning. One of the objectives of the project is to test the scalability of future solutions for processing data stored on the SKA Regional Centres. The project covers the massive processing of imaging data as well as the detection and characterisation of transient sources detected in data streams (beamforming and imaging) using deep learning.
• September/October 2024: Finalisation of an initial Minimum Viable Product (MVP) covering the ingestion, orchestration and processing of radio imaging data.
• Using the NenuFAR scout to prepare the radio astronomy community to welcome SKA.
• Development of NenuFAR-DC (Data Centre) for object-based storage of NenuFAR radiotelescope data

Researchers contributing to ECLAT

• Carlo-Maria Zwölf, AI RI, scientific coordination and AI dissemination
• Florent Mertens, postdoctoral fellow and scientific partner
• David Cornu, postdoctoral fellow and scientific partner
• Gregory Sainton, AI RI, technical partner
• Aristide Doussot,HPC RI, technical partner
• Additional contributors: B. Semelin, P. Salomé, N. Moreau

The institute

The Galaxies and Cosmology team at LERMA (future LUX), Observatoire de Paris, PSL, primarily consists of researchers working on these themes. Part of the team uses radio interferometric data from major international instruments, and several researchers and engineers include SKA preparation in their research themes. 

Past, present and future activities serving SKAO

• Study of the cosmic dawn (CD) and the epoch of reionisation (EoR). Use of a high-resolution numerical simulation code (LICORICE) to produce databases (LORELI) of 21cm signal power spectra corresponding to different possible universes. ‘Simulation-based inference’ to reconstruct the properties of the EoR using several methods (BNN, NDE, etc.). Combination with radio astronomical observations to place upper limits on cosmological models. This theme will benefit greatly from the sensitivity of the future SKA. Several researchers in the team are members of the SKA CD/EoR working group, the LOFAR-EoR project and the NenuFAR CD project. 

• Processing and visualisation of massive data, participation in the development of processing chains for large radio instruments. Simulation of data and instrument processing chains. Development of innovative methods for the calibration and imaging of radio observations. Development of visualisation tools (YaFITS) and database exploration tools for ALMA (ARTEMIX). Participation in the SKAO orange team for the development of visualisation tools. Participation in the SKAO SCOOP team for the co-design of SRC/SDP and SKA processing chains. 

• Source detection and statistical analysis of galaxy properties using deep learning. Development of custom object detectors for the detection and characterisation of sources (YOLO-CIANNA) in massive radio surveys. Development of tools-services for real-time extraction of sources and to facilitate exploration of high-dimensional data. Development of statistical learning tools for data mining and statistical analysis of large galaxy surveys.  

• Participation in SKA SDCs. The group has participated (or rallied a posteriori) and consistently obtained 1st or 2nd place in SDC1 (MINERVA), SDC2 (MINERVA) and SDC3a (DOTSS-21). The group has currently registered two teams for SDC3b and intends to continue participating in future editions in order to stimulate the development of new analysis methods. 

Researchers contributing to ECLAT

• François Orieux, , “Application Algorithms” Coordinator
• Mohammed Nabil El Korso

The institute

The Signals and Systems Laboratory has renowned expertise in statistical signal processing, antenna processing, inverse problems and computational imaging. The laboratory has worked on developing new statistical signal processing methods, Bayesian and variational approaches, robust statistics and statistical learning for image estimation and reconstruction, particularly in inverse problems. The laboratory is involved in the SKA project through the ANR Dark-Era project (2021-2025), which focuses on HPC simulation for the design of a heterogeneous computing cluster for SKA, as well as several theses on data calibration and the acceleration of image reconstruction algorithms.

Past, present and future activities serving SKAO

• ANR Dark-Era on HPC prototyping for SKA.
• Nicolas Monnier’s thesis in collaboration with ATOS (Eviden)
• G2G algorithm for accelerating gridding and degridding for image reconstruction.
  [1] N. Monnier, F. Orieux, N. Gac, C. Tasse, E. Raffin, and D. Guibert, “Fast Grid to Grid Interpolation for Radio Interferometric Imaging,” Astronomy and Computing, vol. 45, p. 100767, Nov. 2023, doi: 10.1016/j.ascom.2023.100767.
• Jianhua Wang’s thesis : Antenna desing for interferometer arrays, soutenance prévue 2025)
• Nawel Arab’s thesis : Maximum likelihood and informed learning for dynamic imaging in radio astronomy (defence due in 2025)
• Yassine Mhiri’s thesis : Contributions to calibration and imaging methods for radio interferometers in the presence of interference (soutenue 2023)

The laboratory is interested in continuing its work in:

• Computational imaging
• Bayesian approach
• Machine learning for image reconstruction
• Unsupervised approach and self-calibration
• Quantifying uncertainties

The laboratory is interested in collaborations to:

• Test algorithms
• Develop new applications for its algorithms
• Receive new issues related to data analysis and processing algorithms, whether using statistical or machine learning approaches.

Researchers contributing to ECLAT

• N. Gac
• L. Bacharah
• P. Larzabal
• I Vin
• T. Rodet

The institute

Computational imaging for radio astronomy is a cross-disciplinary theme of the SICOIA and SIEU teams in the Information and Energy Technology Systems and Applications (SATIE) laboratory. SATIE is drawing on its expertise in both signal processing and algorithm-architecture matching to meet the computational challenges of new-generation radio telescopes. In signal processing, work focuses on imaging techniques, dynamic imaging, calibration, finding optimal geometries, switching sub-networks and informed learning.
The aim of the work on algorithm-architecture matching is to better understand and scale up algorithms. Working closely with L2S, SATIE brings this dual expertise to ECLAT joint lab and the ANR Dark-Era consortium. With a focus on the application side of its work, SATIE is developing a collaboration with the Nançay Observatory (ORN), where NenuFAR, an SKA precursor, is located.

Researchers contributing to ECLAT

• Jean-François Nezan, IETR Scientific Lead, Calculation models, video compression, and signal processing for astronomy
• Karol Desnos, Calculation models, AI, and memory optimisation
• Mickael Dardaillon, Calculation models, Compilation and reconfigurable architecture
• Maxime Pelcat, Cybersecurity and sustainable electronics calculation models
• Daniel Ménard, Approximate computation and energy consumption
• Christophe Moy, Software Radio, Cognitive Radio
• Jordane Lorandel, Embedded system architecture (MCU, SoC, FPGA)

The institute

• IETR: Institute of Electronics and Digital Technologies
• UMR CNRS 6164
• Disciplines: Antennas and microwave devices, EMC and electromagnetism, multifunctional materials, microtechnologies and microsensors, digital communications, remote sensing and multimodal imaging, image analysis and processing, automation, energy systems, lifecycle and eco-design of electrical systems

Activities carried out serving SKAO

• With AUT (NZ) and Kalray: Evaluation of MPPA on SKA CSP algorithms (FFT – correlation), Dataflow modelling of the SEP processing chain for SKA SDP
• With the Paris Observatory: Dataflow modelling of calibration algorithms (Cyril Tasse – KillMS)​

Ongoing activities serving SKAO

• ANR Dark-Era project: Extension of dataflow simulation tools (PREESM) for heterogeneous HPC targets (L2S, INRIA, OP, OCA)​
• MSCA Rising Stars project: ​
• Application modelling with partners MicroGATE and CSIRO; ​
• Extension of dataflow programming models/OpenMP with BSC; ​
• Design and programming of dedicated computing systems with the Paris Observatory

Future activities serving SKAO

• Use case: Embedded AI and SPML (Signal Processing + Machine Learning) systems​
• Integration of dataflow methods in MLIR-based compilation chains
• Hardware: GPU, heterogeneous HPC systems (CPU/GPU/FPGA), computing continuum​

Researchers contributing to ECLAT

• François Bodoin, Links with work carried out under the Numpex/Exa-AToW project
• Mathis Certenais, Work on the DDF pipeline as a use case for data logistics

The institute

Irisa, which is structured into seven scientific departments, has positioned itself as a centre of research excellence. Its scientific priorities include bioinformatics, systems security, new software architectures, virtual reality, mass data analysis and artificial intelligence.

Past, present and future activities serving SKAO

The Exa-AToW roadmap for the SKA SRC project presents a strategy for testing interoperable solutions for executing large-scale workflows with the SKA processing pipeline (or its precursors). The aim is to develop workflow simulation methods to improve the orchestration of multiple processing pipelines on distributed architectures. This approach includes the study of existing example pipelines, the extraction of runtime data, the rewriting of workflows in an appropriate language (such as CWL), and the implementation of agent-based simulations representing computing resources and workflow tasks. Various scenarios will be tested, including the competition of multiple workflows for system resources, the impact of failures in the processing chain, variations in network bandwidth, and the estimation of solution time in relation to the carbon footprint. Short-term tasks include studying current implementations, expressing pipelines in workflow languages, and implementing agent-based simulations, while medium-term tasks aim to integrate optimised versions of pipelines and run them on HPC machines.

Researchers contributing to ECLAT

• Gabriel Antoniu, Coordinator of the PEPR NumPEx Exa-DoST project, under which the KerData team is involved in ECLAT
• François Tessier, Responsible for batch 1 of the Exa-DoST project (PEPR NumPEx)

The institute

KerData s a joint research team of the Inria Centre at Rennes University, INSA Rennes and IRISA laboratory. Its research focuses on data storage and processing on the Edge-Cloud-HPC continuum.

Past, present and future activities serving SKAO

• Exa-DoST: As part of ECLAT, the KerData team is mainly involved in collaborative activities with the CNRS, the Observatoire de Paris and the Observatoire de la Côte d’Azur as part of the Exa-DoST project of the NumPEx PEPR (2023-2029). In this project, SKA is one of the two application demonstrators driving joint research into the challenges of scaling storage architectures, input/output management and in situ analysis. Learn more.
• Summer 2024: Ugo Thay’s engineering internship in the context described above on a study of the input-output schematics of Quartical (a component used in the SKA pipeline).
• InPEx: The InPEx initiative is part of the international collaborative actions of the NumPEx PEPR. KerData (Gabriel Antoniu) is co-leading the working group dedicated to the challenges of the digital continuum in the post-exaflop age. At the InPEx workshop held in Barcelona in July 2024, SKA was presented as a representative use case.

Researchers contributing to ECLAT

• Antsa Rasamoela, Link with Numpex/Exa-DoST project work
• Valentin hazard, Benchmarking works on CPU and I/O for DDFacet and killMS imaging software

The institute

The SE2I (Electronic Systems and Instrumental Computing) team at the Bordeaux Astrophysics Laboratory (LAB) has technical expertise in digital signal processing for radio astronomy, particularly through its contributions to the construction of the ALMA radio telescope. Over the last decade, LAB has led several development studies funded by ESO to explore various digital technologies in preparation for second-generation ALMA instrumentation. This expertise, primarily based on FPGA technology, now also includes CPUs and GPUs. One of the main challenges for the next generation is high-speed data transfer between different hardware resources within a heterogeneous system. The SE2I team has gained solid experience in data transport from digitizers to FPGAs and between FPGAs, and is now exploring data transfer between FPGAs and GPUs. Lastly, the team contributes to the Exa-DoST project under the PEPR NumPEx framework by studying and designing an exascale-level application demonstrator tailored to SKAO’s data processing requirements.

Past, present and future activities serving SKAO

• Study and development of a suite of tools dedicated to benchmarking radio imaging applications:
  • CPU performance benchmarking applied to the DDF pipeline (DDFacet + killMS) software, using processor load trace tools.
  • I/O benchmarking for DDFacet through an M2 internship by Iheb Becher (2024). Development of visualisation and I/O trace tools using Darshan API.
• Study of a reproducible deployment solution for DDFacet and killMS software (Spack, Guix, etc.).

Researchers contributing to ECLAT

• Olivier Aumage, Supervision of a Master 2 computer science course at the University of Bordeaux on task scheduling in the DDFACET software

The institute

• Inria Centre at Bordeaux University, STORM team.
• The STORM team is working on the design of methods and tools for optimising intensive computing applications, involving in particular paralleling and scheduling techniques via task-based execution media such as StreamPU and StarPU. The STORM team is involved in the NumPEx PEPR, in the Exa-SofT focus project.

Past, present and future activities serving SKAO

• 2024 activities — Supervision of a Master 2 Computer Science internship at Bordeaux University:
  • Packaging of DDFACET software dependencies with the GNU GUIX package manager to enable direct and reproducible installation of DDFACET on computing machines supporting GNU GUIX without using a Docker / Singularity type container.
  • Preliminary exploration of interfacing DDFACET software with StreamPU (ex-AFF3CT Core) task-based execution support for pipelined execution on multicore architecture.

Researchers contributing to ECLAT

• Francieli Zanon-Boito
• Luan Teylo

The institute

• Inria Centre at Bordeaux University
• Data management in HPC platforms and applications

Past, present and future activities serving SKAO

Working with the Laboratoire d’Astrophysique de Bordeaux to characterise the input/output (I/O) behaviour of the DDFacet application. We have carried out executions on the PlaFRIM platform using the Darshan profiling tool and have identified that the read time corresponds to around 15% of the execution time, and that the application writes a lot of data to a so-called ‘cache’ directory. New experiments will be necessary because the Darshan tool has not proved suitable for a Python application such as DDFacet. This work will continue in the context of NumPEx – PC3 Exa-DoST with DDFacet and other SKA pipeline applications.

Researchers contributing to ECLAT

• Christian Perez, R&D Manager
• Anass Serhani, HPC Research Engineer

The team

The Inria-CNRS-ENSL-Univ Lyon I Avalon team based at LIP, ENS Lyon, focuses on efficiently executing parallel and/or distributed applications on parallel and/or distributed resources, such as supercomputers or clouds. The challenge is to define application models, systems, and algorithms to execute applications under user constraints (price, performance, etc.) and system administrator constraints (maximising resource use, minimising energy consumption, etc.). The team particularly tackles profiling and modelling energy consumption and data access, data management, component-based application modelling, and application placement and scheduling. Theoretical results are validated through simulations or experiments on Grid’5000 or production platforms.

Past and present activities serving SKAO

• Observer at the Critical Design Review (CDR) of the SDP consortium in January 2019 for France
• Participation in SKAO activities since 2020
• Active involvement in software co-design in SKAO’s Agile Software Engineering organisation: Development of a multi-level benchmarking suite, covering basic testing functions to entire pipelines/workflows, and Benchmarking of radio astronomy pipelines for self-calibration and imaging (DP3, WSClean, RASCIL,…)
• Participation in various WGs for initiating the French node (SRC-FR) in the SKA Regional Centres (SRCNet) network

Future activities serving SKAO

• Participation in the European ODISSEE project, particularly for SKA-SLICES links
• Energy characterisation and optimisation of radio astronomy pipelines.
• Lifecycle analysis and sustainability of Science Processing Centres (SPC).

Researchers contributing to ECLAT

• Frédéric Vivien

The institute

The joint CNRS-ENS de Lyon-INRIA-UCBL ROMA project-team aims to define scheduling models, algorithms and strategies to optimise the execution of applications on high-performance computing platforms. More specifically, ROMA seeks to obtain the ‘best’ possible performance from the user’s point of view (e.g. the shortest execution time) while using resources as efficiently as possible (e.g. minimising energy consumption). The work carried out by the team ranges from theoretical studies to the development of software tools.

Past, present and future activities serving SKAO

• Study of the complexity of scheduling radio astronomical observations (Mathis Lamiroy’s Master 2 internship, 2023) and future study of parametric complexity (Maher Mallem’s post-doc, starting in November 2024).
• Predictive maintenance planning as part of the ODISSEE project
• Previous work that can serve as a basis:
  • Optimisation of the placement of so-called checkpoint backup points by taking into account fault predictions. Guillaume Aupy, Yves Robert, Frédéric Vivien, Dounia Zaidouni : “Checkpointing algorithms and fault prediction.”  J. Parallel Distributed Comput. 74(2): 2048-2064 (2014)
  • Steady-state scheduling (Loris Marchal’s thesis, etc.)

Researchers contributing to ECLAT

• André Ferrari, Professor, Lagrange Laboratory, participation in Dark Era and SEAMS projects
• Shan Mignot, Project Manager IR, Lagrange Laboratory, coordination with SKAO developments for SDP software and hardware acquisition, participation in Dark Era, SEAMS, and Exa-DoST projects
• Alain O’Miniussi, HPC IR, UAR Galilée, participation in Exa-DoST project
• Simon Prunet, CR, Lagrange Laboratory, participation in Dark Era project
• Sunrise Wang, postdoctoral fellow, Lagrange Laboratory, participation in Dark Era and Exa-DoST projects

The institute

The Observatoire de la Côte d’Azur is an EPSCP Grand Establishment, « component institution » of Université Côte d’Azur. The Observatoire de la Côte d’Azur groups and oversees Earth and Universe sciences research activities of Université Côte d’Azur. Its missions include research, observation, education, and the dissemination of knowledge in these fields.

The Observatoire de la Côte d’Azur is co-supervisor of three joint research units(Artemis, Géoazur, Lagrange) and one service unit (Galilée). In this way, the OCA is building an institutional strategy that is in line with the main thrusts of the Earth and Universe sciences at national and European level. The OCA’s mission is to contribute to the advancement of knowledge of the universe through the systematic acquisition of observational data and the development and use of appropriate theoretical, experimental and technical resources in the fields of astronomy, geosciences, related sciences and their applications; to provide services related to the institution’s research activity; to contribute to the initial and continuing training of students and all research staff; to contribute to the dissemination of knowledge, particularly among teaching staff and users of the public education service; and to implement international cooperation activities, particularly at European level.

Past, present and future activities serving SKAO

Since 2008, the Observatoire de la Côte d’Azur has been deeply involved in cutting-edge research on cosmology and extragalactic astrophysics through SKA and its precursors, as well as the development of new interferometric imaging techniques in radio astronomy.

The OCA contributed to the design of SKA’s two telescopes during the pre-construction phase, participating in key committees organised by the SKAO office, such as:

• SKA-Athena Synergy Team
• SKA Science Assessment Team « Impact of SKA-Low antenna frequency coverage »
• Critical Design Review du consortium Low-Frequency Aperture Array de SKA en charge de concevoir SKA-LOW
• Council Preparatory Task Force
• Board de SKA Organisation

Today, the OCA plays a major role in the construction of the SKA by steering the hardware-software co-design of the future SKAO supercomputers that will produce the scientific data distributed to the community. In addition to technical and financial feasibility, this effort is aimed at sustainable management of this sub-system in terms of both its energy consumption and the environmental impact of its life cycle.

The OCA is one of the five founding academic partners of SKA-France coordination in 2016 (OCA, OP-PSL, Université de Bordeaux and Université d’Orléans, with CNRS/INSU as lead partner) and one of the signatories of the ‘Maison SKA-France’ academic partnership between 2018 and 2020. Chiara Ferrari of the OCA has been in charge of SKA-France since its initial phases and is one of the two French representatives on the SKAO Board, giving the OCA an important role in the governance of the project at national and international level.

List of the main SKAO projects in which the OCA is involved:

• ANR J.CJ.C OPALES project « nOn-thermal Processes in gALaxy clustErS » – 2009-2014 (P.I. C. Ferrari, OCA)
• ANR Generic MAGELLAN project « Machine learning methods for the very large arrays in radio astronomy » – 2014-2018 (P.I. A. Ferrari, OCA)
• H2020 AENEAS project « Advanced European Network of E-infrastructures for Astronomy with the SKA »- 2017-2019 (P.I. for France and Chair of the project’s G.A.: C. Ferrari, OCA)
• ANR Dark-Era project to develop and evaluate algorithmic and hardware solutions for Science Data Processors – 2021-2025
• ANR-SNFS SEAMS « Sustainable & Energy Aware Methods for SKA » – 2024-2027
• Horizon Europe SPECTRUM project «Computing Strategy for Data-intensive Science Infrastructures in Europe» – 2024-2026 (P.I. pour la France: C. Ferrari, OCA)
• PEPR NumPEx, PC3 Exa-DoST « Data-oriented Software and Tools for the Exascale» 2023-2028