Presentatie Freya Blekman
-
Upload
vlaamse-vereniging-voor-bibliotheek-archief-amp-documentatie-vzw -
Category
Data & Analytics
-
view
450 -
download
0
description
Transcript of Presentatie Freya Blekman
Managing the data of the Large Hadron Collider
(and other particle physics experiments)
Prof. Dr. Freya Blekman
Interuniversity Institute for High Energies Vrije Universiteit Brussel
O
H C
νe
u d
e ≈
The “Standard Model”
§ Over the last ~100 years: The combination of Quantum Field Theory and discovery of many particles has led to
§ The Standard Model of Particle Physics § With a new “Periodic Table” of fundamental elements
Matter p
articles
Force particles
One of the greatest achievements of 20th Century Science
The Standard Model!
The Large Hadron Collider
General Purpose, pp, heavy ions
CMS
ATLAS
General Purpose: pp, heavy ions
Compact Muon Solenoid (CMS)
Silicon
Pixels
c c c
µ+
e+
γ, πo
K-, π-,p,…
ν
Muon detectors
Hadron calorimeter
Crystal Electromagnetic
calorimeter
4 Tesla
Solenoid
All Silicon Strip
Tracker
Ko→ π+π-, …etc
Quite a camera § CMS is like a camera with 90 Million pixels § But no ordinary camera § It can take up to 40 million pictures per second § The pictures are 3 dimensional § And at 15 million kilograms, it’s not very portable
§ LHC data challenge: The problem is that we cannot store all the pictures we can take so we have to choose the good ones fast!
Experimental Challenges – Big Data in Particle Physics
§ Collisions are frequent § Beams cross ~ 16.5 million times per second at
present § About 20-‐30 pairs of protons collide each
crossing § Interesting collisions are rare -‐
§ less than 1 per 10 billion for some of the most interesting ones
§ We record only about 400 events per second.
§ We must pick the good ones and decide fast!
§ Decision (‘trigger’) levels § A first analysis is done in a few millionths of a
second and temporarily holds 100,000 pictures of the 16,500,000
§ A final analysis takes ~ 0.1 second and we use ~10000 computers
§ We still end up with lots of data – 1 GB per second!
Symmetry magazine’s summary infographic of LHC data volumes
CERN
Data distribution § Grid connects >100,000 processors in 34 countries
22 Petabytes in 2011
CMS data in Belgium § In Flanders: CMS T2 hosted at VUB § Alternative T2 at UCL
§ Access to all CMS members all over the world § And main working node for all Flemish (+ ULB/UMons) particle physicists
§ Brussels Computing cluster (Tier 2 computer center): Consist of modular PCs 440 TB storage space (and growing) for Belgian users
2.2 PB storage space for CMS 19 TeraFLOPS (FLoating-‐point Operations Per Second) Funding agencies: FRS-‐FNRS (ULB, UMons) FWO-‐BigScience – Vlaams Supercomputing Centrum (VUB)
Other CMS data
DBTA Workshop on Big Data, Cloud Data Management and NoSQL Big Data Management at CERN: The CMS Example
Other CMS Documents"
x 4000 people … for many decades
J.A. Coarasa (CERN) 25!
Other CMS data
DBTA Workshop on Big Data, Cloud Data Management and NoSQL Big Data Management at CERN: The CMS Example
Other CMS Documents: Size"
A printed pile of all CMS documents that are already in a managed system
= 1.0 x (Empire State building)
Plus we have almost the same amount spread all over the place (PCs, afs, dfs,
various websites …)
J.A. Coarasa (CERN) 26!
LHC open data? § LHC and CERN have very strict policies regarding publication of their results § ALL journal publications (including those in Nature/Science) are made public
§ Publishing in open access journals the norm
§ However, most of our data is only accessible to those in the collaboration
§ Exception: there are datasets available for education use § http://physicsmasterclasses.org/index.php
Secondary school student accessing public CMS data at Vrije Universiteit Brussel
Open data in (astro) particle physics § The IceCube experiment is another particle physics experiment studying elementary particles of astrophysical origin
§ Based at the South Pole, IceCube includes Belgian scientists from VUB/ULB/UGent/Umons
§ IceCube data is analysed with the same cluster in Brussels as mentioned before
Extreme High energy neutrinos § One of the most exciting IceCube results involves the observation of outrageously high-‐energy neutrinos from cosmic origin
§ Evidence for High-‐Energy Extraterrestrial Neutrinos at the IceCube Detector, IceCube Collaboration, Science 342, 1242856 (2013). DOI: 10.1126/science.1242856
§ After publication, the IceCube collaboration has made this data available to the scientific community
§ http://icecube.wisc.edu/science/data
§ Working through 40 million collisions per second provides a daunting challenge processing huge amounts of data
§ Journal publications of LHC experiments all public
§ Other experiments such as IceCube also make some of their datasets public after publication
Outlook and Conclusion
pp physics at the LHC corresponds to conditions around here
HI physics at the LHC corresponds to conditions around here
Where the largest and smallest things meet
The Dark Side § We now know that only ~5% of the energy in the universe is ordinary matter (remember E=mc2).
§ 25% is dark matter § SUSY theories can happily predict this amount
§ There are other possibilities but SUSY is a favorite § Provides great dark matter candidates (e.g. Neutralino or Gravitino)
§ Leads to remarkable unification of field strengths § And it fixes the Higgs mass problem
How would we see the Higgs Boson ? Simulation – to predict and design detector – and to compare to what we actually see
NB: These old plots correspond to ~50 times more sensitivity than we have now (20x more data, 2x the energy)!
§ all channels together: comb. significance: 4.9 σ
§ expected significance for SM Higgs: 5.9 σ
Characterization of excess near 125 GeV
26
[GeV]4lm
Eve
nts
/ 3 G
eV
0
2
4
6
8
10
12
[GeV]4lm
Eve
nts
/ 3 G
eV
0
2
4
6
8
10
12 Data
Z+X
*,ZZaZ
=126 GeVHm
µ, 2e2µ7 TeV 4e, 4µ, 2e2µ8 TeV 4e, 4
CMS Preliminary -1 = 8 TeV, L = 5.26 fbs ; -1 = 7 TeV, L = 5.05 fbs
[GeV]4lm80 100 120 140 160 180
Standard Model Higgs Decays
§ The SM Higgs is unstable § Decays “instantly” in a number of ways with very well known probabilities
(called Branching Fractions or Ratios that sum up to 1). § Branching ratios change with mass as seen here § Some decay modes are more easily seen than others Firstly if they end with electrons, muons, or photons
Supersymmetry
What made us so sure about the Higgs?
§ The Brout-‐Englert-‐Higgs theory has predictable consequences § It predicts very heavy force particles that carry the weak nuclear force known as the W+, W-‐ and Zo
§ The W+, W-‐ should both have a mass of 80.4 GeV Note that the proton has a mass of 1 GeV so these are very heavy fundamental particles!
§ The Zo should have a mass of 91.1 GeV § We find these predicted particles & measure their masses § For instance, the Zo should decay to two muons. We can measure their momenta and reconstruct the Zo mass.
§ If we do this for many Zo particles, a distribution of the mass values we get should have a very predictable shape.