APS, May 2009
Th R
The
Role
l off US G
Groups iin
LHC Physics
Dan Green
Fermilab
APS Meeting, May 2-5, 2009
1
Outline
Introduction and History
The Problem
The Realities
The Evolution to Worldwide Physics
Analysis
APS Meeting, May 2-5, 2009
2
Scientists and Physics
Faraday could work at home in his laboratory. The Curies
could too but they processed tons of pitchblende
pitchblende.
APS Meeting, May 2-5, 2009
3
Accelerator Based Research
Then the tools became larger. Still, they could be
operated on a university campus.
APS Meeting, May 2-5, 2009
4
National Laboratories
Post WW II
laboratories became
national facilities.
y
learned to
Physicists
“commute”
APS Meeting, May 2-5, 2009
5
The Large Hadron Collider
The LHC
collaborations are
truly world wide
and enormous by all
measures
ATLAS
APS Meeting, May 2-5, 2009
6
The Problem
The detectors have 100 million channels
Theyy need to be aligned
g
, calibrated and
monitored. There are large data bases which
must be accessed.
The data volume, even after zero suppression,
is enormous; 1 MB/event, 100 Hz of events or
109 MB/yr
MB/ ( 1 million
illi CDs/yr)
CD / )
The US groups are only 1/3 of the LHC
collaborations and they live an ocean away
away. In
addition only about 1/3 of US physicists can be
at CERN long term at any given time.
APS Meeting, May 2-5, 2009
7
Science Magazine
APS Meeting, May 2-5, 2009
8
The Realities
LHC experiments
need to engage
g g
the entire
collaboration in
order to
simultaneously
operate, analyze
and upgrade
these complex
devices.
The necessities of the LHC experiments
mean that US groups will be fully
engaged
d and
d enabled
bl d – e.g. Remote
R
t
Operations Center
APS Meeting, May 2-5, 2009
9
Remote Operations
Exploit the fact that the collaborative tools are falling
into place – for example Web Based Monitoring,
ELog, webcams, teleconferencing. The US is about 7
hours out of phase with CERN. Therefore, stand
g shifts. With China and India on
remote “evening”
board full 24/7 coverage is easily possible.
APS Meeting, May 2-5, 2009
10
Tiered Computing
LHC data analysis
required worldwide,
grid computing
grid,
computing. First
pass is done at CERN
and then the data is sent
to “T1” centers
centers. BNL
and FNAL are national
computing centers for
US ATLAS and
d US
CMS. “T2” are centers
located at specific
universities.
i
i i Finally,
Fi ll
“T3” are located at all
collaborating
institutions.
APS Meeting, May 2-5, 2009
11
Data Operations
The enormous data
sets require world
wide computing.
CERN sends data to
several national “T1”
Centers for analysis.
analysis
Data flow and
monitoring of
computing must be
24/7. There is a
permanent video link
between CERNDESY-and FNAL for
US CMS
APS Meeting, May 2-5, 2009
12
University Groups are “T2” or “T3”
US ATLAS
APS Meeting, May 2-5, 2009
13
The Evolution of Work
The reality of the LHC
experiments is that all
physicists
h i i t cannott live
li att
CERN while the work of
all physicists is needed
as the number of jobs
soon will increase
dramatically
(operations, analysis
and upgrade). The T1
and T2 centers are
becoming nucleation
points for ‘local” work.
APS Meeting, May 2-5, 2009
14
Collaborative Tools
INDICO is a tool for agenda
creation. It also serves as a
“memory” since all the
posted talks are p
p
persistent.
APS Meeting, May 2-5, 2009
15
Collaborative Tools - II
The LHC experiments use
a teleconferencing tool
called EVO. This allows all
LHC collaborators with
web access to participate
remotely in meetings. ILC
uses WEBEX
APS Meeting, May 2-5, 2009
16
Collaborative Tools - III
This research modality is not limited to particle physics.
APS Meeting, May 2-5, 2009
17
US Groups in LHC
US CMS iis similar.
i il F
Fraction
ti iis slightly
li htl llarger ~ 1/3
APS Meeting, May 2-5, 2009
18
Nucleation at Laboratories
US ATLAS has Argonne, Brookhaven
Lawrence Berkeley and SLAC as
collaborating laboratories. A dispersed
support model is favored.
US CMS has Fermilab as the major
laboratory (LLNL is a small effort ).
N t
Naturally,
ll with
ith a different
diff
t sociology,
i l
they favor a central nucleation. Thus in
US CMS there is a single large ROC and
LPC. These were part of the US CMS
plan ab initio.
baseline p
APS Meeting, May 2-5, 2009
19
US ATLAS Analysis Support Efforts
SLAC is a T2
APS Meeting, May 2-5, 2009
20
Analysis Support
“Local” US
ATLAS analysis
support.
t Favor
F
hires at
Universities which
split the costs.
FY09 at Duke,
Iowa State and
NYU. Try to
ramp up support
att the
th 3 ASC and
d
at CERN (US
physicists
resident)
APS Meeting, May 2-5, 2009
21
The LHC Physics Center
The LPC has been part of the US CMS baseline plan from
the beginning. Note the Physics Support Group – ramp up
APS Meeting, May 2-5, 2009
22
Guest and Visitor Programs
There are several
G&V programs to
support US CMS
visitors to the LPC
APS Meeting, May 2-5, 2009
23
Experts on Call
It is important to
have experts
available
il bl for
f help
h l
in real time. In
addition, the
experts should be
in your time zone.
As of now
now, CMS
has 2 teams – one
in CERN and one
i FNAL ( 4 US
in
CMS – LPC
Physics Support
Team).
APS Meeting, May 2-5, 2009
24
Web Based Mentoring
In order to be up to
date the LPC has
date,
started a “US CMS
Mentors Program”
which takes
Facebook as a
model. In this wayy
senior US CMS
physicists can help
mentor graduate
students and
postdocs remotely.
APS Meeting, May 2-5, 2009
25
LPC Hosts US CMS Workshops
“Local”
Local Workshops in the US are well attended
attended.
APS Meeting, May 2-5, 2009
26
Metrics - I
JTERM-III, 39 Institutions
APS Meeting, May 2-5, 2009
PKU
Virginia
Purdue
Calumet
Princeton
Texas A&M
Texas Tech
CMU
Wisconsin
Rockefeller
Kansas State
Tata
Iowa
JHU
UIC
CBPF
FIT
FIU
FSU
Florida
UCR
UCSB
Rutgers
UCLA
Colorado
Nebraska
Brown
Wayne State
Vanderbilt
UC Davis
Boston
Caltech
Puerto Rico
Martyland
Cukuriva
OSU
How well is US CMS
engaged in physics
analysis? Metrics hard
to specify. Well
attended
W k h
Workshops/Tutorials
/T t i l
with a good sampling of
US CMS groups. The
LPC is reaching the US
CMS community of 46
university groups, quite
widely it seems.
27
Metrics - II
Last 2 years – 6 month bins
APS Meeting, May 2-5, 2009
Mine data in
INDICO to see
what fraction
of CMS talks
at CMS
meetings are
given by US
physicists. The
fraction is
roughly
hl pro
rata, or 1/3.
28
Metrics - III
Mine CMS Conference Committee data to look at
conference talks sorted by nationality
nationality. The US is roughly
pro rata and gave the most talks of any national group.
APS Meeting, May 2-5, 2009
29
Metrics - IV
Physics Coordinator
P.Sphicas
Deputies:
J.Incandela*, R.Tenchini
e/γ
Forward
Physics
Muons
* = US CMS
QCD
C.Seez
P. Meridiani
R. Bellan
M.Mulders
K. Piotrzkowski
N.Varelas*
V. O’Dell*
Jets/MissET
b-tagging
Heavy Ions
EWK
D.Elvira*
P. Schleper
ParticleFlow/τ
C. Bernet
S. Gennai
W. Adam
T.Boccali
O.Kodolova
G. Roland*
Tracking
K. Burkett *
B. Mangano*
R.Chierici
F. Stoeckli
A. Korytov*
C. Mariotti
SUSY
J.Alcaraz
S.Dasu*
O.Buchmuller
J. Richman*
Top
p
Exotica
C.Campagnari*
T. Christiansen
MC generators
Higgs
A. De Roeck
G. Landsberg*
B physics
P. Eerola
J Ol
J.Olsen*
*
Due to the Tevatron experience of US physicists the
representation of US physicists in CMS physics groups is more
than pro rata In the core analyses (yellow) US physicists
comprise more than half the conveners
APS Meeting, May 2-5, 2009
30
Summary
The LHC experiments are of unprecedented size and
complexity, as are the collaborations
In order to operate
operate, analyze and upgrade the
experiments , all of the collaborators must be both
engaged and enabled.
T enable
To
bl fully
f ll - tiered
ti d computing,
ti
grid
id tools,
t l remote
t
operations centers, distributed data operations and
collaborative tools must be fully deployed.
US groups are fully active in the LHC experiments and
participate proportionally.
Metrics for US p
participation
p
indicate that US ATLAS
and US CMS strategies to engage fully have, to date,
been successful. However, we have yet to take first
data.
APS Meeting, May 2-5, 2009
31
APS Meeting, May 2-5, 2009
32
APS Meeting, May 2-5, 2009
33
APS Meeting, May 2-5, 2009
34