Introduction to data analysis techniques and virtual laboratories
Monday 22 March – Thursday 25 March 2021
4-day online meeting
- Lectures and tutorials on data analysis techniques, modelling of atmospheres and disks
- Exchange between science and arts
Day 1, 22 March 2021: Quality Assessment of Complex Observational Data
Morning
9:00h – 9:10h GMT Welcome
9:10h – 10:10h GMT Lecture by Dr Jeroen Bouwman (MPIA Heidelberg) “Spectro-photometric calibration: general concepts and applications to observations of protoplenetary disks and planetary transits”
10:10h – 10:20h GMT – Break
10:20h – 10:40h GMT – Q&A with Dr Jeroen Bouwman
10:40h GMT – Short intro to CHAMELEON School 1 social events
Afternoon
13:30h – 16:30h GMT – Hands-on Session on archival data by Prof Inga Kamp (University of Groningen):
Finding and using archival data from different telescopes and observatories, combining and displaying data across wavelengths and comparing with existing literature for some specific object (exoplanet or disk). Students will work in groups of two. They will present their results at the end of the afternoon.
Schedule:
13:30h – 15:00h GMT group work
15:00h – 15:15h GMT coffee break
15:15h – 16:30h GMT presentations by each group (10 min each)
Students’ preparation task: preselect two objects (incl. one as back-up) that benefit your research.
Evening
19:00h – ca. 21:00h GMT – Prof Katrien Kolenberg (KU Leuven): Ex(p)oplanet artists’ presentation of work-in-progress
The platform Ex(p)oplanet has been set up by Marc Horemans, a Belgian artist. Marc leads the Multidisciplinary Studio Projectatelier of SLAC, Academy of Fine Arts Leuven, Belgium. Within the studio, art works are produced that are stimulated by research going on within CHAMELEON.
The presenting artists are:
Wilma Geyskens, Kelly Willems, Monik Myle, Susanne Stoop, Lieve Lissens, Lou Bielen, Lies Daenen, Mien Van Kerckhoven, Renild D’haese, Hilde Ghesquiere, Christiane Aerts, Kristin De Vis, Johan Thimpont, Patrick De Nys, Mia Van Nieuwenhove, Josée Van Goubergen, Ingrid Degroodt, Ross Olson, Elisabeth Vanhoutte.
From l-r: Hilde Ghesquiere: Atmosphere 1 (Photograph), Christiane Aerts: Study for Day & Night on a Exoplanet (Drawing), Elisabeth Vanhoutte: Surface (Plaster).
From l-r: Kelly Willems: Still Points (Drawing), Mien Van Kerckhoven: Object (Projection).
Day 2, 23 March 2021: Data Analysis Techniques – Focus: Neural Networks
Morning
8:00h – 8:45h GMT – Lecture by Prof Johan Suykens (KU Leuven, AI institute): “Deep Learning and Kernel Machines: a Unifying Perspective”
8:45h – 9:10h GMT – Q&A with Prof Johan Suykens
9:20h – ca. 9:40h GMT – Introduction to the hands-on machine learning afternoon session (Till Käufer, Francisco Ardevol Martinez)
Afternoon
13:30h – 15:30h GMT – Hands-on session on machine learning by Dr Michiel Min (SRON), Till Käufer, Francisco Ardevol Martinez
During this hands-on session, the participants will learn how to build a Neural Network (NN) and use it to solve one of two astronomical problems.
Extracting exoplanet parameters using spectra:
The participants will be provided with a dataset containing synthetic HST/WFC3 transmission spectra of exoplanets and their corresponding parameters. The parameters defining the forward models are isothermal temperature; H2O, HCN and NH3 mixing ratios; and a grey cloud opacity. The spectra consist of 13 wavelength bins between 0.8 and 1.6 microns, which match those of the WASP-12b observation in Kreidberg et al. (2015). This observation is also provided to retrieve on once your machine learning algorithms have been trained.
Emulating SED modelling of protoplanetary disks:
The goal is to predict Spectral Energy distributions (SEDs) for protoplanetary disks based on disk parameters. The conventional way of predicting SEDs is to run a modelling code (ProDiMo, MCFOST, ..) to simulate a disk and then calculating how the output would look like. If you want to calculate a lot of SEDs this can take a lot of time. A potential solution for that problem is to train a Machine Learning algorithm to predict SEDs based on input parameters of disks. For doing that the algorithm must be trained on a large dataset of models (DENT-grid Kamp et al. 2011), where the SED and the input parameters are known.
A Repository is prepared, where all the necessary information can be found. At the end, all participants will have a working NN on their computer and should be able to use such a network to solve other problems.
Evening
19:00h GMT – Prof Petra Rudolf (Zernike Institute for Advanced Materials, University of Groningen): “How to keep women (and men) in science”
Abstract: The EU report on Women in Science identifies different risks which can result in the drop out of women from academic life at different career stages: at the beginning (PhD phase) it is mainly the lack of support from the supervisor, in the postdoc phase the problem of reconciling career and motherhood is most acute, then the risk shifts to lack of career expectations and even at the associate/full professor level there remains a risk due to isolation and exclusion. I shall discuss these issues based on studies from Europe, the U.S.A. and Japan. Based on the trends which emerge from these studies, I shall also suggest some measures to be taken to ensure that women stay in science. From this it will be obvious that all measures which favour women in science are good for men as well.
Day 3, 24 March 2021: Cutting-edge of Complex Modelling – Exoplanet Atmosphere Virtual Laboratories
Morning
9:00h – 10:00h GMT + 15 mins Q&A – Lecture by Dr Ludmila Carone (MPIA Heidelberg) on “Challenges in 3D modelling of exoplanet atmospheres”
10:15h-10:30h GMT – Break
10:30h – 11:30h GMT + 15 mins Q&A – Lecture by Dr Christiane Helling (University of St Andrews) on “Cloud formation in diverse exoplanet atmosphere environments”
Afternoon
13:30h – 16:00h GMT – Hands-on session on simulating planet observables using the ARCiS retrieval framework by Dr Michiel Min (SRON)
We will go into detail on how to run the atmospheric modelling code ARCiS in forward modelling mode. We will discuss various levels of complexity in the modelling.
Detailed program:
13:30h-14:00h GMT – Introduction ARCiS
14:00h-14:30h GMT – Split up in groups to perform 1D computations of transit spectra
14:30h-15:00h GMT – Explaining ARCiS 3D mode
15:00h-16:00h GMT – Split up in groups to perform 3D computations of phase curves and realistic images of your favourite exoplanet.
Student preparation: Install ARCiS and think of 1 or 2 exoplanets you would like to model.
Evening
19:00h GMT – The CHAMELEONs’ Quiz
Day 4, 25 March 2021: Cutting-edge of Complex Modelling – Planet-forming Disk Virtual Laboratories
Morning
Lecture by Dr Peter Woitke (University of St Andrews) on “Physical structure and processes”
8:00h GMT – Part 1
8:25h GMT – splinter group session 1 (groups of 4-5)
8:30h GMT – back to main call, Q&A
8:40h GMT – short break
8:45h GMT – Part 2
9:10h GMT – discussion
9:20h-9:30h GMT – Break
Lecture by Prof Inga Kamp (University of Groningen) on “Chemical structure and processes”
9:30h GMT – Part 1
9:55h GMT – splinter group session 2 (groups of 4-5)
10:00h GMT – back to main call, Q&A
10:10h GMT – short break
10:15h GMT – Part 2
10:40h GMT – discussion
10:50h GMT – end of session
Afternoon
13:30h – 16:30h GMT – Hands-on using Virtual Laboratories by Prof Inga Kamp (University of Groningen) and Dr Peter Woitke (University of St Andrews)
Preparation:
In preparation for the hands-on activities on Thursday, we would like to ask you to install two software packages. These software packages require more basic ones
to be functional, in particular python3, git, gfortran (or better ifort) and make.
prodimopy: https://bitbucket.org/cheesyog/prodimopy/src/master/
GGChem: https://github.com/pw31/GGchem
You find an introduction into ProDiMo and prodimopy here.
If you have any problems installing our software, please let us know in advance.
Schedule:
13:30h GMT – introduction of the problems (Peter/Inga)
14:00h GMT – splinter group work (2-3 participants/group)
15:15h GMT – coffee break
15:30h GMT – short presentations from each group on their findings, and discussion
16:30h GMT – end of session
Each student will need to select one problem to work on during this session – please put your name on this sign-up sheet.
The problems will be based on pre-calculated results from our ProDiMo disc models, which you can access here, password: ProDiMo4fun
The four problems are: (extended description)
1) How do ice lines in the disk change with stellar luminosity and dust properties?
2) How does the ice/rock ratio change depending on dust settling?
3) How does the water snowline and CO ice line change as a function of time?
4) What is the mineral composition of the dust assuming phase equilibrium?
Round-up and Goodbye!