BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:la nuit des temps
DTSTART;VALUE=DATE-TIME:20200624T160000Z
DTEND;VALUE=DATE-TIME:20200624T180000Z
DTSTAMP;VALUE=DATE-TIME:20200713T084000Z
UID:indico-event-6259@indico.ijclab.in2p3.fr
DESCRIPTION:https://indico.ijclab.in2p3.fr/event/6259/
LOCATION:
URL:https://indico.ijclab.in2p3.fr/event/6259/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Learning to discover
DTSTART;VALUE=DATE-TIME:20200720T070000Z
DTEND;VALUE=DATE-TIME:20200731T150000Z
DTSTAMP;VALUE=DATE-TIME:20200713T084000Z
UID:indico-event-5999@indico.ijclab.in2p3.fr
DESCRIPTION:POSTPONED to first half of February 2021\n\nOverall program:
\n\n\n Mon 20 - Wed 22 July : Workshop on Graph Neural Network\n Thu 23 -
Sat 25 July morning : Workshop on Dealing with Uncertainties \n Mon 27
- Wed 29 July : Learning to Discover AI and Physics conference\n Thu
30 - Fri 31 July : Workshop on Generative Models\n\n\nLearning To Disco
ver is a program on Artificial Intelligence and High Energy Physics (HEP)
to take place at Institut Pascal Paris-Saclay 20-31 July 2020\, in its be
autiful new building. Over the two weeks\, three themes will be successive
ly tackled during innovation-oriented sessions of 2-3 days each\, with a
three three days general conference on AI and physics in the middle.\n\
nSince 2014\, the field of AI and HEP has grown exponentially. Physicists
have realised quickly the potential of AI to deal with the large amount of
complex data they are collecting and analysing. Many AI techniques have b
een put forward\, with scientific collaboration based on open data sets\,
challenges\, workshops and papers.\n\nLearning To Discover is a first-of-i
ts-kind program where participants will have access to deep technical ins
ights in advanced machine learning techniques\, and their application to p
article physics. During the event\, blending the concept of hackathon\, ha
nds-on and tutorials\, physicists who have attempted to apply machine lear
ning to specific challenges in HEP will expose their problem case and the
solutions they have arrived at so far. ML experts will expose the latest a
dvances in relevant techniques. Machine learning experts and ML-aware phys
icists will work hand in hand on existing datasets\, building upon and imp
roving existing solutions.Bibliography and open datasets will have been ma
de available months earlier in order to make the event most productive. Un
der this particular environment\, the participants will be able to discuss
and understand shortcomings of existing solutions and develop novel archi
tectures and methods to outperform on the specific problem. \nThree theme
s have tentatively been selected based on the one hand\, their interest fo
r HEP\, and the fact there is already a number of HEP teams working on it\
, on the other hand\, their importance in the Machine Learning field.\n\n\
nGraph Neural Network (GNN) : despite promising applications to several HE
P problems (tracking\, identification\, calorimeter reconstruction\, parti
cle flow\, …) there remains several bottlenecks in the models being used
. In particular the following topics are of interest: ways to extract grap
h-level information\, mechanisms for information propagation over the grap
h\, accurate graph generation\, evolution of graph topology within the m
odel\, ...\n\nDealing with Uncertainties : physicists ultimately write p
apers with measurements which always include an assessment of uncertaintie
s. How to evaluate the uncertainty on a trained model ? How to build confi
dence into a ML model and how to convince our peers about it ? How to deal
with the uncertainties on the input of the models to maximise the overall
accuracy ? Bayesian Networks\, Adversarial architectures and other techni
ques are being explored. \n\nGenerative Models : High Energy Physics is w
ell known for having built accurate simulators which are able to deliver s
imulation of real experiments\, from the Quantum Field Theory equations de
aling with the interactions of fundamental particles up to specific answer
s of electronic circuits. These simulations are accurate\, but not perfect
\, and very slow. Proof of Concept of using Generative Models (GAN\, VAE\,
...) are being developed but many issues are to be solved before they make
it to production. How to reach sufficient accuracy over a large parameter
space ? How to make sure multi-dimension correlations are well described
? How to deal with the irregularity of the detector (think of an image w
ith pixels of different sizes and shapes) ? \n \n\nLearning To Discover
AI and Physics conference will deal with the most recent topics at the i
ntersection of the two fields. It will propose a few keynote talks from le
ading scientists in the field as well as wide sample of on-going devel
opments. Submission of abstracts is open.\n\nLearning To Discover will be
hosted at the new Université Paris-Saclay Institut Pascal\, an ideal venu
e for such interactions to take place. Located in Orsay (close to Paris) i
n the heart of Université Paris-Saclay\, it has offices for up to 60 peop
le (40 the first week of the workshop)\, meeting rooms\, amphitheaters (u
p to 120 seats) and all amenities for a successful workshop. In the usual
Institut Pascal format\, lunches and social programme is free. Accommodati
on can be booked in the area on Institut Pascal expense for a large fracti
on of the participants\; travel is covered only if needed. See "Institut
Pascal" item in the menu bar for more practical details.\n\n\nA special i
ssue of Computing and Software for the Big Science (a refereed journal) wi
ll be prepared with the most relevant contributions with a submission dead
line 4 months after the workshop\, leaving ample time for finalisation of
the studies showcased at the workshop. A number of workshops concerning AI
and Physics have been organised in recent years\, notably “Deep Learnin
g for the Physical Science” at NeurIPS 2017 (https://dl4physicalsciences
.github.io/)\, “Machine Learning and the Physical Science” at Neurips
2019 (https://ml4physicalsciences.github.io) and “AI and physics” at A
MLD 2020 in Lausanne (https://appliedmldays.org/tracks/ai-physics). Studie
s developed at “Learning To Discover” will have a high chance at being
accepted at future such events.\n\n \nhttps://indico.ijclab.in2p3.fr/eve
nt/5999/
LOCATION:Institut Pascal
URL:https://indico.ijclab.in2p3.fr/event/5999/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Conformal bootstrap and statistical models
DTSTART;VALUE=DATE-TIME:20210503T070000Z
DTEND;VALUE=DATE-TIME:20210528T160000Z
DTSTAMP;VALUE=DATE-TIME:20200713T084000Z
UID:indico-event-5833@indico.ijclab.in2p3.fr
DESCRIPTION:A platform where experts in conformal bootstrap techniques and
in statistical physics can discuss and solve specific mathematical and ph
ysical problems. \n\nWeek 1: Status of the conformal bootstrap\n\nLeading
experts on the conformal bootstrap will give introductory courses on:\n\n\
n Conformal symmetry and conformally invariant QFTs.\n Input and output da
ta of the conformal bootstrap.\n Analytical and numerical tools.\n Pros an
d cons of various techniques.\n\n\nWeek 2: Statistical physics targets\n\n
Discussing various statistical models\, with an emphasis on the aspects th
at are relevant to a bootstrap approach.\n\n\n Symmetries\, unitarity.\n S
pace of states\, critical exponents\, correlation functions.\n Which obser
vables can be computed with a good precision?\n Families of models\, margi
nal deformations\, toy models.\n Outstanding physical questions\, interest
of the critical limit for these questions.\n\n\nExamples may include: per
colation\, depinning\, loop-erased random walks\, sandpiles\, Chalker-Codd
ington network for the integer quantum Hall effect.\n\nWeek 3: Bootstrap f
or structural phase transitions\n\nStructural phase transitions are descri
bed by unitary theories\, which makes them accessible to existing numerica
l bootstrap methods. However\, they are challenging because of their intri
cate group theory and phenomenology. First we will review:\n\n\n Necessary
group theory tools.\n Main results from the renormalization group approac
h.\n Experimental situation.\n\n\nThen we will apply the bootstrap approac
h.\n\nWeek 4: Non-unitary bootstrap methods\n\nThe best-developed numerica
l bootstrap techniques rely on unitarity\, which is however not available
in logarithmic CFTs\, the critical Potts model\, percolation\, or disorder
ed systems. We will explore two approaches to the non-unitary bootstrap:\n
\n\n The perturbative epsilon expansion.\n Gliozzi's method and extremal f
lows.\n\n\nWe may start with a family of unitary models such as the O(N) m
odel\, and vary the parameter continuously to a non-unitary or logarithmic
fixed point.\n\n\n\nhttps://indico.ijclab.in2p3.fr/event/5833/
LOCATION:Institut Pascal\, Orsay\, France
URL:https://indico.ijclab.in2p3.fr/event/5833/
END:VEVENT
END:VCALENDAR