Deep Learning and its conceptual differences to classical machine learning have been largely overlooked in the community. The broad hypothesis motivating our current projects at the ESO AI Lab is that letting the abundant real astrophysical data speak for itself, with minimal supervision and no labels, can reveal interesting patterns which may facilitate discovery of novel physical relationships. This talk is about our first step, where we seek to interpret the representations a deep convolutional neural network chooses to learn, and find correlations in them with current physical understanding. We train an encoder-decoder architecture on the self-supervised auxiliary task of reconstruction to allow it to learn general representations without bias towards any specific task. As a case study, we apply this framework to HARPS, a dataset of ~270000 stellar spectra, each of which comprising ~300000 dimensions. We find that the network clearly assigns specific nodes to estimate (notions of) parameters such as radial velocity and effective temperature without being asked to do so, all in a completely physics-agnostic process. This supports the first part of our hypothesis. Moreover, we find with high confidence that there are ~4 more independently informative dimensions that do not show a direct correlation with our validation parameters, presenting potential room for future studies.

If the title looks scary to you then this talk is made for you. The aim is to define all the words the title contains and give you an overview of each concepts and how they interact with one another. The starting point will be the definition of a segmented mirror, why are they needed and what new challenges they introduce in term of optics. The optical alignment of a segmented mirror requires dedicated tools, methods and time. Among the optical alignment operations to be done is the co-phasing (or phasing for short). A definition is given and ways of measuring it is exposed. Finally, the approach of the ELT to co-phasing its own segmented mirror is explained based on the available published information.

Comets are among the most intriguing objects in the Solar System. Traditionally, they were considered to be well-preserved planetesimals that have remained unchanged for billions of years in the outer Solar System and have therefore retained key evidence from the epoch of planet formation. Nevertheless, we now know that Jupiter-family comets (JFCs) have had a rich dynamical history which has taken them from planetesimals in the protoplanetary disk, through long residences as Trans-Neptunian objects (TNOs), followed by a migration through the Centaur region between Jupiter and Neptune to become active or dormant comets making regular passes through the inner solar system. Since comet evolution is very closely related to the environment surrounding them, tracing the changes comets have experienced since formation is key for understanding the conditions in the outer Solar system throughout its history.

In the beginning of this talk, I will introduce the current understanding of the outer Solar System formation and evolution, emphasizing the latest progress in this field driven largely by the Rosetta and New Horizons space missions. I will then proceed to highlight the key role of telescope observations in addressing some of the remaining questions. I will discuss our most recent results on comet and TNOs photometry, as well as our efforts to design space missions that will address the missing knowledge in the coming decades.

For a radio interferometer like ALMA, the path from the antennas to science data weaves through an interconnected system of many synchronised instruments, operating in optical, analogue and digital electronic domains, distributed between the antennas and a central location. In this lecture I will cover some of the important principles of a heterodyne radio interferometer, and take a tour through the ALMA instrument system from antennas to archived data. As you will have seen from the science results, ALMA has been a game changer, transforming mm/sub-mm observations by an order of magnitude or more in resolution, sensitivity and image fidelity. Yet this is achieved using technologies which were state of the art a decade or more ago, and built to a tight construction deadline. ALMA now has a vision for a major instrument upgrade over the next decade: the ALMA2030 roadmap. I will conclude the lecture by discussing how the roadmap is being translated into ambitious but hopefully achievable specifications, and how this may be implemented based on current technology development in key areas.

System engineering (SE), a relatively recently developed domain of knowledge, has even more recently become an integral part of ground-based astronomy projects. More so now that we are building ginormous telescopes and their instruments. ESO itself includes a SE Department in its own organigram since (only) 2014.

Like the name suggests, SE is a field of engineering devoted to management of multidisciplinary projects, or systems, where all components must be made to work together – possibly flawlessly and, in astronomy, for typical lifetimes of 20+ years. If you ask 10 people - any 10 people, at ESO or outside - “what is System Engineering?” you will probably receive 10++ different answers. In this talk, I will propose my own view on it. It evolved in the past 25 years during which I slowly transformed from astronomer to system engineer (did I?). Spoiler alert: my definition is that “SE is just about problem solving”.  A deceptively simple and enticing definition, it’s clearly a trap! Ultimately, in my 35 minutes, I hope to convince the youngest among you that it may be a viable career choice for an astronomer, and I will shamelessly give you a rosy glimpse into the fun parts, like requirement analysis, inventive problem solving or trade-off studies.  For the rest of the story, you will have to invite me again!

For an overview and schedule of past and future lectures: http://www.eso.org/~rvanderb/KES/kes.html

This KES lecture will cover the current status of ESO’s Extremely Large Telescope and of the powerful instrument suite that is being designed and built to meet the science case for the ELT. In addition, I aim to give some insight into how such ambitious programmes, like the ELT, come to life. The challenge of building the instruments is greater than anything the ground-based astronomical community has previously attempted, thanks to the size of the telescope and the drive towards reaching diffraction limited observations. The basic concept and scientific capabilities of the first six instruments that are being developed will be summarised.

As astronomy enters the regime of precision science, astronomical instruments (and their detectors) are pushed ever closer to their technological limits. The quality and noise performance of detectors has improved over the years to the point that small systematic effects can now dominate the error budgets of astronomical instruments. I will go over some common systematic effects found in astronomical detectors, demonstrate what impacts they could have on various science cases, and give some practical strategies for how you, the astronomer, can get involved in ensuring that detector systematic effects don’t limit the science you want to do.

Exploring the low surface brightness (LSB) universe is one of the most challenging tasks in the era of the deep imaging and spectroscopic surveys. It is however a crucial ingredient to map the mass assembly of galaxies at all scales and all environments and thus constrain their formation within the Lambda-Cold Dark Matter paradigm. In this framework, clusters of galaxies are expected to grow over time by accreting smaller groups. During the infall process, the material stripped from the galaxy outskirts builds up the stellar halos and the intra-cluster light (ICL). These are extended (> 10 Re), diffuse and very faint (μ_g > 26 mag/arcsec^2) components made of stars stripped from satellite galaxies, also in the form of streams and tidal tails, with multiple stellar population and complex kinematics, which are still growing at the present epoch.

In the past, the main limitation of the above studies is the small field of view and angular resolution of the CCD images.

The advent of wide-field cameras allows to overcome these limits and thus to study the very out and faint regions of galaxies. Therefore, on the observational side, a big effort was made in the recent years to develop deep photometric surveys aimed at studying galaxy structures out to the regions of the stellar halos (e.g. Ferrarese et al. 2012; van Dokkum et al. 2014; Duc et al. 2015; Munoz et al. 2015; Merritt et al. 2016; Trujillo & Fliri 2016; Mihos et al. 2017; Iodice et al. 2019). With the current observing facilities the galaxy outskirts and intra-cluster regions can be effectively probed down to a surface brightness limit of ~ 31 mag/arcsec^2 in the g band for nearby galaxies (≤ 50 Mpc).

In this mini-series of KES lectures I would like to

- review the main progresses made on both observational and theoretical sides on this scientific topic;

- show the main steps to derive from the deep images a set of observables to be directly compared with the theoretical predictions on the mass assembly. This part will include details on observation strategies and tools adopted in deep surveys;

- compare results from the analysis of the deep images with simulations.