ESCAPE SUMMER SCHOOL

Europe/Paris
Auditorium (LAPP)

Auditorium

LAPP

9 chemin de Bellevue, Annecy
Description

Group picture

Happy participants gathering for a virtual group picture



Due to the current situation in France and the rest of the world, the ESCAPE Summer School has been moved to an online event happening from 7 to 18 June 2021.

Registrations opening date: from 26 April to 31 May.
Acceptation to the school is automatic within the limit of available seats. All information regarding the school organisation will be sent to participants after the registration deadline.

The registration is now closed. However, all the school content will be freely available to everybody.

Please, go our lectures portal to find the necessary links:
https://escape2020.github.io/school2021/

 

School Program: the program of the school is devoted to project development for astrophysics, astroparticle physics & particle physics. The aim of the school is to provide theoretical and hands-on training on Data Science and Python development:

  • Coding environment and good code practices
  • Version control and collaborative development
  • Debugging and profiling
  • Python packaging
  • Scientific libraries for data science and analysis
  • Machine learning

The school is held as a continuation of the Asterics/Obelics summer school.

Course duration: From June 07 to June 18 2021

Organisation: All necessary information will be sent to participants after the registration deadline

Communication: a Slack community for the school is available for communication between participants, to ask questions to tutors, to get out, etc.

School venue: Online event.

Accommodation Arrangements:  None (online event)

School Fees: None, it's free!

Requirements: The school is open to all. However, basic knowledge of Python is required to benefit from most of the courses.

Previous editions of the school: 1st, 2nd, 3rd.

Testimonies from students and tutors from second edition:

Who is organizing this school: LAPP is the organizing this school in association with ESCAPE partners, PRACE. This international school is being organized in the framework of ESCAPE.

What is ESCAPE: European Science Cluster of Astronomy & Particle physics ESFRI research infrastructures (ESCAPE) aims to address the Open Science challenges shared by ESFRI facilities (CTA, ELT, EST, FAIR, HL-LHC, KM3Net, SKA) as well as other pan-European research infrastructures (CERN, ESO, JIVE, EGO) in astronomy and particle physics. (https://projectescape.eu)           

For more information visit https://www.projectescape.eu/about-us

ESCAPE partners: 31 European organisations with wealth expertise and experience on astronomy, astroparticle and particle physics, three fields contributing heavily to the final designs of the ESFRI projects.

These partners are committed to contribute to the construction of the data research open environment, which is represented by EOSC, and promote the application of FAIR principles for data stewardship. All partners help bringing together and engage a wide range of ESCAPE stakeholders.

More info: https://www.projectescape.eu/partners

About LAPP : Created in 1976, the LAPP is one of the 19 laboratories of the National Institute of Nuclear and Particle Physics (IN2P3). LAPP has about 150 researchers, professor, researchers, engineers, technicians, administrators, students and foreign visitors. The research carried out at the LAPP aims to study the physics of elementary particles and their fundamental interactions, as well as the exploration of the connections between the infinitely small and the infinitely large. LAPP is involved in LHC experiments with CERN (ATLAS, LHCb), neutrinos (STEREO,SuperNemo, Dune), astrophysics (HESS, CTA, AMS,LSST), Future colliders (ILC, CLIC) and gravitational waves (Virgo).

LAPP Website http://lapp.in2p3.fr/

LAPP Presentation video ( in English): https://youtu.be/Sk-xzDEiy7Q

LAPP Presentation video ( in French):  https://youtu.be/uPSXV4rAYPw

 
 
 
 
 
 
 
    • 09:00 09:10
      Welcome address 10m

      https://www.youtube.com/channel/UC05braEQdP2rCSUamHm9I_Q/live

      Speaker: Dr Thomas Vuillaume (LAPP, CNRS)
    • 09:10 09:30
      ESCAPE 20m

      https://www.youtube.com/channel/UC05braEQdP2rCSUamHm9I_Q/live

      Speaker: Dr Giovanni LAMANNA (LAPP - IN2P3/CNRS)
    • 09:30 10:30
      School organisation 1h

      https://www.youtube.com/channel/UC05braEQdP2rCSUamHm9I_Q/live

      Speaker: Dr Thomas Vuillaume (LAPP, CNRS)
    • 10:30 11:00
      Coffee break 30m
    • 11:00 12:30
      Coding environment, tools, good code practice for collaborative and continuous developments: Reproducible science

      https://www.youtube.com/watch?v=ZEcklfIp6Og

      Convener: Rachael Ainsworth
      • 11:00
        Reproducible science 1h 30m

        Making research results more accessible and reproducible can contribute to better and more efficient science, however widespread adoption of open research practices has not yet been achieved. Funding agencies (such as the European Commission Horizon 2020) are increasingly requiring research products (such as data, code and publications) to be made openly available in order to make results more accessible, transparent and reproducible. Recent studies have also shown that open research practices are associated with benefits to the researcher such as increases in citations, media attention, potential collaborators, job and funding opportunities. In this talk I will discuss the different aspects of Open Science, the barriers we face to practicing openly, how to "open" up your research workflow using open and transparent data and software services in order to reap the benefits associated with open research practices.

        Speaker: Rachael Ainsworth
    • 12:30 14:00
      Lunch break 1h 30m
    • 14:00 14:30
      Coding environment, tools, good code practice for collaborative and continuous developments: Environment setup

      https://www.youtube.com/watch?v=ZEcklfIp6Og

      Convener: Enrique Garcia Garcia
      • 14:00
        Environment setup 30m

        The installation of the ESCAPE school environment will be described and implemented from scratch in this hands-on session. We will go through the installation of conda - an open source package management system - and the dependencies needed to follow the lectures of the summer school.

        Speaker: Enrique Garcia Garcia
    • 14:30 16:00
      Coding environment, tools, good code practice for collaborative and continuous developments: Python and Notebooks

      https://www.youtube.com/watch?v=ZEcklfIp6Og

      Convener: Enrique Garcia Garcia
      • 14:30
        Python and Notebooks 1h

        In the last years, the evolution of the python editors has gone from a simple terminal to elaborated interactive environments. In this lecture we will see the basic usage of these tools and a basic tutorial to get used to them. Also, some useful features of Jupyter Notebooks will be presented.

        Speaker: Enrique Garcia Garcia
    • 09:00 10:30
      Coding environment, tools, good code practice for collaborative and continuous developments: Git

      https://www.youtube.com/watch?v=ZEcklfIp6Og

      Convener: Maximilian Nöthe
    • 10:30 11:00
      Coffee break 30m
    • 11:00 12:30
      Coding environment, tools, good code practice for collaborative and continuous developments: Git

      https://www.youtube.com/watch?v=ZEcklfIp6Og

      Convener: Maximilian Nöthe
    • 12:30 14:00
      Lunch break 1h 30m
    • 09:00 10:30
      Community specific analysis: Python introduction, useful packages and libraries for data scientists
      Convener: Tamas Gal
      • 09:00
        Library overview 1h 30m

        Python: this talk gives an overview about the Python language in the scientific context. It shows its strengths and weaknesses and also introduces a few important libraries which should be part of a scientist's toolbox.

        Numpy: Numpy has built the foundation of numerical computing in Python. Without this package, Python could not have reached such a level of popularity in data science. This course will teach the basics of Numpy and shows how to utilise it to solve numerical problems.

        Pandas: Pandas is a great library to work with tabular data and perform high-level statistical analyses with them. In this short lecture, we will explore how to load, transform, combine and analyse datasets using the powerful Dataframe structure.

        Matplotlib: Visualisation is a key component of scientific work. Matplotlib is one of the most popular libraries to create static and interactive graphs with Python and offers endless possibilities to tweak those in detail. This lecture is a short introduction to the basics of working with Matplotlib.

        Julia: Although Python has gained to lot of momentum as a scientific language in the past years, we cannot ignore the fact that it is by design not a language for scientific computing, it's the tools around Python which makes it so successful. Julia in contrast is a language which was created in 2012 and designed from the beginning for high performance while being as interactive and easy to use as Python. This talk gives an introduction to the language and shows why it might be the future of scientific computing.

        Speaker: Tamas Gal
    • 10:30 11:00
      Coffee break 30m
    • 11:00 12:30
      Community specific analysis: Numpy
      Convener: Tamas Gal
      • 11:00
        Numpy 1h 30m

        Numpy: Numpy has built the foundation of numerical computing in Python. Without this package, Python could not have reached such a level of popularity in data science. This course will teach the basics of Numpy and shows how to utilise it to solve numerical problems.ic computing.

    • 12:30 14:00
      Lunch break 1h 30m
    • 14:00 15:30
      Community specific analysis: Pandas
      Convener: Tamas Gal
      • 14:00
        Pandas 1h 30m

        Pandas: Pandas is a great library to work with tabular data and perform high-level statistical analyses with them. In this short lecture, we will explore how to load, transform, combine and analyse datasets using the powerful Dataframe structure.

        Speaker: Tamas Gal
    • 15:30 16:30
      Community specific analysis: Matplotlib
      Convener: Tamas Gal
      • 15:30
        Matplotlib 1h

        Matplotlib: Visualisation is a key component of scientific work. Matplotlib is one of the most popular libraries to create static and interactive graphs with Python and offers endless possibilities to tweak those in detail. This lecture is a short introduction to the basics of working with Matplotlib.

        Speaker: Tamas Gal
    • 09:00 10:30
      Community specific analysis: Reproducible science in practice
      Convener: Arturo Sanchez Pineda (LAPP)
      • 09:00
        Reproducible science in practice 1h 30m

        It refers to a series of principles, techniques, tools and practical considerations that allow the documentation, recording and preservation of data analysis pipelines — enhancing the possibilities of collaborations across borders and increasing the probabilities of replicating results by others (and yourself) in the future. Reproducibility involves using standard and well-established protocols to ensure that your code will survive outside your computer, the passing of time and that others will be able to use it as a starting point for new analysis. We will explore several of those tools: from the use of version control and Jupyter notebook in the cloud to prepare and encapsulate software environments (VM, containers) and the usage of good practices regarding licences, citation and DOIs.

        Speaker: Arturo Sanchez Pineda (LAPP)
    • 10:30 11:00
      Coffee break 30m
    • 11:00 12:30
      Community specific analysis: Reproducible science in practice
      Convener: Arturo Sanchez Pineda (LAPP)
      • 11:00
        Reproducible science in practice 1h 30m

        It refers to a series of principles, techniques, tools and practical considerations that allow the documentation, recording and preservation of data analysis pipelines — enhancing the possibilities of collaborations across borders and increasing the probabilities of replicating results by others (and yourself) in the future. Reproducibility involves using standard and well-established protocols to ensure that your code will survive outside your computer, the passing of time and that others will be able to use it as a starting point for new analysis. We will explore several of those tools: from the use of version control and Jupyter notebook in the cloud to prepare and encapsulate software environments (VM, containers) and the usage of good practices regarding licences, citation and DOIs.

        Speaker: Arturo Sanchez Pineda (LAPP)
    • 12:30 14:00
      Lunch break 1h 30m
    • 14:00 15:30
      Coding environment, tools, good code practice for collaborative and continuous developments: Test driven devs - unit tests and continuous integration

      https://www.youtube.com/watch?v=ZEcklfIp6Og

      Convener: Maximilian Nöthe
      • 14:00
        Test driven developments and continuous integration 1h 30m
        Speaker: Maximilian Nöthe
    • 15:30 16:00
      Coffee break 30m
    • 16:00 17:00
      Coding environment, tools, good code practice for collaborative and continuous developments: Packaging

      https://www.youtube.com/watch?v=ZEcklfIp6Og

      Convener: Maximilian Nöthe
    • 09:00 10:30
      Community specific analysis: Scipy
      Convener: Axel Donath
    • 10:30 11:00
      Coffee break 30m
    • 11:00 12:30
      Community specific analysis: Astropy
      Convener: Axel Donath
    • 12:30 14:00
      Lunch break 1h 30m
    • 14:00 15:00
      Webinar: AI in Cosmological Experiments 1h

      Abstract:
      Detailed observations of the present contents of the Universe are consistent with the Lambda + Cold Dark Matter model, subject to some ‘tensions'. To probe these components further the Dark Energy Survey and  larger galaxy surveys require new statistical approaches. In particular the role of Artificial Intelligence, or preferably Augmented Intelligence, is crucial for analysing forthcoming surveys (e.g. Rubin-LSST and Euclid) of billions of galaxies. This also requires new ways to train the next generation of scientists for the challenges ahead.

      Prof. Ofer Lahav is Perren Chair of Astronomy in the Astrophysics Group at University College London (UCL) and Vice-Dean (International) of the UCL Faculty of Mathematical and Physical Sciences (MAPS). He is also Co-Director of the STFC-funded Centre for Doctoral Training in Data Intensive Science at UCL. Ofer's research area is Observational Cosmology, in particular probing Dark Matter and Dark Energy. His work involves Machine Learning for Big Data.
      https://www.ucl.ac.uk/astrophysics/professor-ofer-lahav

      Speaker: Prof. Ofer Lahav (University College London)
    • 15:00 16:00
      Other: Discussion with teachers
    • 09:00 10:30
      Profile, debug and optimise: Debugging and profiling
      Convener: Karl KOSACK (CEA Saclay)
    • 10:30 11:00
      Coffee break 30m
    • 11:00 12:30
      Profile, debug and optimise: Optimisation and parallelism in Python
      Convener: Karl KOSACK (CEA Saclay)
    • 12:30 14:00
      Lunch break 1h 30m
    • 14:00 15:00
      Community specific analysis: Gammapy
      Convener: Axel Donath
    • 15:00 15:30
      Coffee break 30m
    • 15:30 17:30
      Community specific analysis: Scikit-HEP
      Convener: Dr Eduardo Rodrigues (University of Liverpool)
      • 15:30
        Community specific analysis: Scikit-HEP 1h 30m

        Data analysis in High Energy Physics (HEP) has evolved considerably in recent years. In particular, the role of Python has gained much momentum, sharing at present the show with C++ as a language of choice. Several (community) domain-specific projects have seen the day, providing (HEP) data analysis packages that profit from, and talk to well with, the huge Python scientific ecosystem, which navigates around NumPy and friends.
        This lecture introduces the Scikit-HEP project, which I started in late 2016 with a few colleagues from various backgrounds and domains of expertise.
        Scikit-HEP is a community-driven and community-oriented project with the aim of providing Particle Physics at large with a Big Data ecosystem for analysis in Python. It has developed considerably in the past couple of years, and is now part of the official software stack of the experiments ATLAS, Belle II, CMS and KM3NeT.
        In this lecture ample time will be provided to "play around" with the material, in Jupyter notebooks.

    • 10:00 16:00
      IN2P3's 50th anniversary - free day
    • 09:00 10:30
      Big data for big science: Spark
      Conveners: Dr Julien Peloton (CNRS-IJCLab) , frederic gillardo
    • 10:30 11:00
      Coffee break 30m
    • 11:00 12:30
      Big data for big science: Spark
      Conveners: Julien Peloton (CNRS-IJCLab) , frederic gillardo
    • 12:30 14:00
      Lunch break 1h 30m
    • 14:00 15:00
      Other: Introduction to Julia
      Convener: Tamas Gal
      • 14:00
        Introduction to Julia 1h

        Julia: Although Python has gained to lot of momentum as a scientific language in the past years, we cannot ignore the fact that it is by design not a language for scientific computing, it's the tools around Python which makes it so successful. Julia in contrast is a language which was created in 2012 and designed from the beginning for high performance while being as interactive and easy to use as Python. This talk gives an introduction to the language and shows why it might be the future of scientific computing.

        Speaker: Tamas Gal
    • 15:00 16:00
      Community specific analysis: Analysis Workflow from the KM3NeT Open Data Center Auditorium

      Auditorium

      LAPP

      9 chemin de Bellevue, Annecy

      Data analysis workflows from our communities

      Convener: Jutta Schnabel (FAU Erlangen (ECAP))
    • 16:00 17:00
      Community specific analysis: An introduction to gravitational wave data analysis
      Convener: Alberto Iess (INFN Roma Tor Vergata)
    • 09:00 10:30
      Machine Learning: Introduction to machine learning
      Conveners: Benson Muite, Claudia Beleites, Martino Sorbaro
      • 09:00
        Machine Learning 1h

        In this class, we will give an introduction to machine learning algorithms, in a data-oriented and coding-oriented way. We will explain what it means for an algorithm to learn, and the main categories of ML problems. We will then cover, with code examples, how to approach a dataset for analysis. In particular we will take a closer look at: automated clustering of data points; supervised classification; supervised regression; how to test a model and choose hyperparameters; dimensionality reduction. The workshop will be as hands-on as possible. A basic knowledge of Python is required.

        Speakers: Claudia Beleites, Martino Sorbaro
    • 10:30 11:00
      Coffee break 30m
    • 11:00 12:30
      Machine Learning: Introduction to machine learning
      Conveners: Benson Muite, Claudia Beleites, Martino Sorbaro
      • 11:00
        Machine Learning 1h

        In this class, we will give an introduction to machine learning algorithms, in a data-oriented and coding-oriented way. We will explain what it means for an algorithm to learn, and the main categories of ML problems. We will then cover, with code examples, how to approach a dataset for analysis. In particular we will take a closer look at: automated clustering of data points; supervised classification; supervised regression; how to test a model and choose hyperparameters; dimensionality reduction. The workshop will be as hands-on as possible. A basic knowledge of Python is required.

        Speaker: Martino Sorbaro
    • 12:30 14:00
      Lunch break 1h 30m
    • 14:00 16:00
      Machine Learning: Introduction to machine learning
      Conveners: Benson Muite, Claudia Beleites, Martino Sorbaro
      • 14:00
        Machine Learning 1h

        In this class, we will give an introduction to machine learning algorithms, in a data-oriented and coding-oriented way. We will explain what it means for an algorithm to learn, and the main categories of ML problems. We will then cover, with code examples, how to approach a dataset for analysis. In particular we will take a closer look at: automated clustering of data points; supervised classification; supervised regression; how to test a model and choose hyperparameters; dimensionality reduction. The workshop will be as hands-on as possible. A basic knowledge of Python is required.

        Speaker: Martino Sorbaro
    • 09:00 10:30
      Machine Learning: Introduction to deep learning
      Convener: Mikaël Jacquemont
      • 09:00
        Introduction to Deep learning 1h

        Deep learning is leading the artificial intelligence revolution allowed by the world of data we are living in. It is a powerful method that automatically learns to address tasks from the data, with minimal preprocessing.
        This “Introduction to deep learning” lecture aims to give an insight on the fundamentals of deep learning. From the artificial neuron to famous deep architectures via the learning process, we will give an overview of the essential components of deep learning. We will also open the black box of neural networks to better understand their behavior. A substantial part of the lecture will be dedicated to practical hands on exercises.
        Key concepts studied:
        • Deep learning fundamentals
        ◦ architecture building blocks, from the neuron to convolution
        ◦ learning to address tasks, the gradient descent and the backpropagation algorithms
        • Going (a bit) deeper
        ◦ famous architectures
        ◦ transfer learning
        • Introduction to explainability of the neural networks (GradCam)
        • Some useful tools for deep learning (PyTorch, Lightning and Tensorboard)

    • 10:30 11:00
      Coffee break 30m
    • 11:00 12:30
      Machine Learning: Introduction to deep learning
      Convener: Mikaël Jacquemont
      • 11:00
        Introduction to Deep learning 1h

        Deep learning is leading the artificial intelligence revolution allowed by the world of data we are living in. It is a powerful method that automatically learns to address tasks from the data, with minimal preprocessing.
        This “Introduction to deep learning” lecture aims to give an insight on the fundamentals of deep learning. From the artificial neuron to famous deep architectures via the learning process, we will give an overview of the essential components of deep learning. We will also open the black box of neural networks to better understand their behavior. A substantial part of the lecture will be dedicated to practical hands on exercises.
        Key concepts studied:
        • Deep learning fundamentals
        ◦ architecture building blocks, from the neuron to convolution
        ◦ learning to address tasks, the gradient descent and the backpropagation algorithms
        • Going (a bit) deeper
        ◦ famous architectures
        ◦ transfer learning
        • Introduction to explainability of the neural networks (GradCam)
        • Some useful tools for deep learning (PyTorch, Lightning and Tensorboard)

    • 12:30 13:00
      Farewell 30m Auditorium

      Auditorium

      LAPP

      9 chemin de Bellevue, Annecy
      Speaker: Dr Thomas Vuillaume (LAPP, CNRS)