Calendar

Sep
6
mar
Jean-Marc Huré – Figures d’équilibres sphéroidales des fluides multicouches. @ B18N, Salle Univers
Sep 6 @ 11 h 00 min – 12 h 00 min
Séminaire

Speaker : Jean-Marc Huré (LAB)

Title: Figures d’équilibres sphéroidales des fluides multicouches.

J’exposerai les principaux éléments d’un formalisme vectoriel résolvant approximativement O(c) les configurations d’équilibre d’un fluide autogravitant hétérogène composé de L couches homogènes en rotation rigide et bordées par des sphéroides parfaits. Cette approche, basée sur un développement du potentiel gravitationnel sur le paramètre confocal c (plutôt que sur l’ellipticité e), permet des configurations très oblates, prolongeant ainsi les travaux classiques fin XIXe de l’école française. Ces solutions analytiques sont validées par les résultats numériques obtenus par la méthode SCF du champ auto-cohérent, et par l’établissement de l’équation du Viriel associée. Quelques perspectives ainsi qu’une application à la détermination de structures internes pour la planète Jupiter réalisant les principales observables seront présentées.

The slides will be in English, the seminar will be given in French.

Sep
13
mar
David Cornu – Winning the SKA Science Data Challenge 2 with a fast Deep Learning object detector @ B18N, Salle Univers
Sep 13 @ 11 h 00 min – 12 h 00 min
Séminaire

Speaker : David Cornu (Obs de Paris)

Title: Winning the SKA Science Data Challenge 2 with a fast Deep Learning object detector

Abstract:
With its 1 TB simulated data cube of HI line emission, the SKA Science Data Challenge 2 (SDC2) is getting closer to the difficulty of real upcoming SKA observation analysis. Even if the type of task to perform in the SKA SDCs are rather classical (detection, classification, parameter extraction, etc.) modern datasets have become heavily demanding for classical approaches due to size and dimensionality. It is not a surprise then, that many astronomers started to focus their work on Machine Learning approaches that demonstrated their efficiency in similar applications. However, hyperspectral images from astronomical interferometers are in fact very different from images used to train state-of-the-art pattern recognition algorithms, especially in terms of noise level, contrast, object size, class imbalance, spectral dimensionality, etc. As a direct consequence, these methods do not perform as well as expected when directly applied to astronomical datasets. In this context, the MINERVA (MachINe lEarning for Radioastronomy at obserVatoire de PAris) project has assembled a team to participate in the SDC2 with the objective of developing innovative Machine Learning methods that better suit the needs of astronomical images.

In this presentation, I will describe the work we have made on implementing a modern YOLO (You Only Look Once) CNN object detector inside our custom framework CIANNA (Convolutional Interactive Artificial Neural Networks by/for Astrophysicists) and describe the modifications and tuning that allowed us to reach the first place of the SKA SDC2. I will start by discussing the strengths and weaknesses of this type of method in comparison to more widely adopted Region-Based CNN detectors (Faster R-CNN, Mask R-CNN, …). I will also review the motivation and the effect of the numerous changes we made on the method (data quantization, 3D convolution, layer architecture, detection layout to manage blending, objectness decomposition, IoU selection, additional parameter inference, …) in order to apply it to both SDC1 and SDC2, and identify what are the present limits as well as some tracks for further improvements. I will detail the computational efficiency of the method (with GPU acceleration) and discuss its scaling capabilities for upcoming challenges or datasets. Finally, we will comment on how this methodology could be used to analyze the actual data from SKA pathfinders or any other similar astronomical dataset and how it could be used to merge knowledge and information from multiple datasets at the same time.

 

Sep
15
jeu
Codir 07 @ Salle Chandon
Sep 15 @ 10 h 00 min – 12 h 00 min
Institutionnel
Sep
20
mar
Olivier Lai – Optimising image quality at the telescope: Altering reality to fit our simulations @ B18N, Salle Univers
Sep 20 @ 11 h 00 min – 12 h 00 min
Séminaire

Speaker : Olivier Lai

Title: Optimising image quality at the telescope: Altering reality to fit our simulations

Self generated turbulence and dome seeing are starting to be recognised as strong contributors to the degradation of astronomical image quality, especially when using adaptive optics; besides the now infamous low wind effect, strange temporal power spectra, strong ground layer, low (temporal) frequency turbulence, high (spatial) frequency turbulence all point to local effects. Unfortunately accounting for and including all these sources of image degradation in our modelling seems intractable, because most of these effects are transient, stochastic and non-stationary, they all depend on too many environmental parameters to be causally interpreted (e.g. wind speed, direction, dome azimuth, heat sources, radiative cooling in the telescope structure and its (differential) thermal inertia, natural air temperature variations and partial venting).
It therefore seems useless, and on top of that very difficult, to estimate realistic AO performance from simulations that would try to incorporate all these effects. So instead of trying to align our simulations to reality, I propose that we try to align reality to our simulations. Indeed, we do not have to incorporate all these effects in our simulations: we can attempt to prevent them from occurring in the first place, and luckily all these local effects can be attenuated or averted at the source, using yet to be developed techniques such as intelligent dome design, cleverly positioned fans or precise real time thermal control including telescope structure heating. None of these are technologically challenging, but the control needs to be sensitive, active and real time, which requires an accurate sensing and mapping of the turbulence inside the telescope and dome.
In this talk, I will show some examples of intermittent turbulence related to environmental parameters on telescopes in Hawaii (CFTH, UH88), Arizona (LBT) and Antarctica (ASTEP), as a first step to implementing heuristic dome turbulence control.

Sep
27
mar
Comitech 07 @ Salle Chandon
Sep 27 @ 14 h 00 min – 16 h 00 min
Institutionnel
Oct
6
jeu
CLAB 04 @ Salle Univers
Oct 6 @ 9 h 30 min – 12 h 30 min
Institutionnel
Oct
20
jeu
Codir 08 @ Salle Chandon
Oct 20 @ 10 h 00 min – 12 h 00 min
Institutionnel
Nov
8
mar
Comitech 08 @ Salle Chandon
Nov 8 @ 14 h 00 min – 16 h 00 min
Institutionnel
Nov
10
jeu
AG des projets @ Salle Univers
Nov 10 @ 9 h 30 min – 17 h 00 min
Institutionnel
Nov
24
jeu
Codir 09 @ Salle Chandon
Nov 24 @ 10 h 00 min – 12 h 00 min
Institutionnel