TaskA2008a.DVI

Study of Elementary Particle Physics at the LHC

Task A – ATLAS

The University of Michigan, Ann Arbor, MI 48105

UM ATLAS Personnel name list

1 Introduction

1.1 ATLAS at the University of Michigan

The Michigan ATLAS group was formed in 1998 and rapidly developed into a world-class pro-

gram. Our objectives for the program have been to accept major responsibilities and to play

leading roles in many important areas within the ATLAS experiment. Among these assumed

responsibilites we have designed and built large precision muon chambers, designed and built

muon detector frond-end electronics, have led the US detector integration, pre-commissioning,

and system tests at CERN both for H8 and X5 beam facilities and for detector preparations

in building 184. We have contributed broadly to the muon software development including cal-

ibration algorithms and strategy studies for muon final state physics, GRID computing R&D,

and collaborative tools development. Recently, we have assumed a new leadership role within

the ATLAS physics community, convenor of the muon combined performance group. Michigan

is one of the strongest institutes in the ATLAS experiment.

This document presents a short perespective on the challenges and opportunities for ATLAS

in the three years covered by this proposal. It then describes the university community we

bring to this endeavor, both the individuals working under the grant and the university facilities

that are available. We then turn to the work accomplished in the past three year grant period

by the University of Michigan ATLAS group. In the final sections we describe our plans and

commitments to ATLAS and the support we require to accomplish these tasks.

1.2 The ATLAS Experiment

The ATLAS experiment is well into the pre-commissioning phase, Phase I, and is entering the

detector integration and installation phase, Phase II. Fig. 1 shows the online picture of the

ATLAS cavern as of Feb. 11, 2005. In this image one can see that two barrel magnets, and

12 muon chambers (in the area of the detector supports) have been installed. In addition, the

barrel calorimeter has been assembled and is ready to be placed into the ATLAS detector. With

the first collisions due at the LHC on July 1, 2007, our highest priority has to be the delivery

of a top quality detector to ATLAS in the limited time remaining. The first of July 2007 has

become the target date, our focus, and our challenge. In a two year period everything must

be in place and working. The Michigan ATLAS group has fully committed and will continue

to shoulder the primary load for ATLAS endcap muon detector integration, commissioning and

calibration.

LHC is the first new energy frontier in a generation. The precise discoveries that will emerge

from the LHC are unknown. In spite of these uncertainites, we must be ready to fully exploit

the opportunity with top quality hardware and software. To do so requires that our attention

remain directed toward detector commissioning, trigger, calibration/data preparation, software

preparation for data acquision monitoring, event reconstruction, and for physics analysis. The

software must be fully developed and mature if we are to understand what is present in the

early data. We must also understand the SM physics landscape, as seen by ATLAS, prior to

LHC collisions. Only when those foundations are built and well understood, can we address,

with confidence, the most exciting and compelling issues in high energy physics:

• The mystery of the origin of mass and the nature of electroweak symmetry breaking;

• Whether supersymmetry existing, and at which energy scale such symmetry breaking?

• Whether quarks and leptons have further structure at much smaller distance scale?

Figure 1: The online picture of the ATLAS cavern as of Feb. 11,2005. Two barrel magnets have

been installed, and 12 muon chambers have be installed in the area of detector supports.

• Whether nature indeed has extra spatial dimensions?

• Whether new interaction force, such as Technicolor, and new gauge bosons existing?

1.3 Michigan’s Commitments

To extract answers to the above profound questions with LHC data, our group is committed

to continued strong leadership in muon detection focusing on the physics potential for new

discoveries with muon final states. Since we have established a large and highly experience team

working on muon detector construction, commissioning, calibration, and database development

which also possesses considerable expertise in computing and muon software, we propose to

merge these activities into a team sufficient to tackle the task of muon “object” qualification.

To optimize the quality of the “objects” we call muons, we must understand the detector, the

signal muons, the signals we conclude are backgrounds, and calibrate the measurements so that

the precision meets the specification required by the physics goals. These “objects” will be the

input to our physics analysis and to the analysis of others. We will take this service role seriously

including the need to document thoroughly algorithms and procedures developed.

1.4 Overview of the group

The Michigan ATLAS group comprises 25-30 people, including 7 faculty, 5 Research Scientists, 3

postdoctoral fellows, 4 engineers, 2 mechanical technicians, 3-6 students and 3-6 visiting scholars.

Four faculty, Chapman, Neal, Thun and Zhou are primarily working on the ATLAS Project.

Chapman has been the project leader for the ATLAS muon front-end readout certification

and for the design of the readout multiplexer (CSM), he is also leading the endcap Big-Wheel

commissioning at CERN (Phase II effort); Neal has been the project leader for US ATLAS

collaborative tools development, he also leads the ATLAS GRID computing effort at Michigan.

Thun has been the muon detector construction project leader and the chamber constrution and

quality assurance officer. He is leading the muon detector calibration effort for endcap muon

system. Zhou is the US ATLAS MDT project leader, she is leading the US ATLAS endcap

phase I work at CERN for single chamber integration and certifications. She will continue play

a leading role for ATLAS endcap phase II and phase III commissioning. Three additional UM

ATLAS faculty, Amidei, Campbell and Qian are currently focusing on the Tevatron experiments

(CDF and D0), and have planned transitions from Tevatron to LHC activity during the next

three years.

Five Research Scientists, Diehl, Goldfarb, McKee, Levin and Zhao are supported by DoE base

grant and are working full time on the ATLAS project. Zhao manages the MDT integration and

commissioning effort at CERN; Levin has led the US H8 test beam effort at CERN for the past

three years. He has just assumed leadership of the muon combined performance working group

within the ATLAS physics analysis community and will lead the activity as the convener for this

area; McKee has been the endcap MDT production and certification database coordinator and

is also the US ATLAS network R&D project leader. He has played leading roles at Michigan for

ATLAS GRID computing effort and for DC2 project. Goldfarb has served as the overall ATLAS

muon software database coordinator (2000 - 2002), and muon software coordinator (2002- now)

at CERN. Diehl oversees the muon detector construction and is the QC/QA control systems

officer. He has also taken the responsibility for the development of programs for endcap muon

calibration, particularly for handling the long tube wire sag RT function corrections.

Three Michigan postdoctoral fellows are resident at CERN, Avramidou, Biglietti and Fer-

retti. They are supported by non-base funds. Avramidou is working on test beam data analysis;

Biglietti is working on Monte Carlo event-generator program development for Λ

; and Ferretti is

working on MDT Phase I data processing and on the implementation of the MDT commissioning

database. He is also working on Λ

physics studies. Both Avramidou and Biglietti are leaving

in summer of 2005. We are in the process of replacing them with two new postdoctoral fellows

at CERN. One new postdoctoral fellow, Manuela Cirilli, has been identified. She is currently a

CERN Follow coordinating the ATLAS Muon Database programs. She will join the UM ATALS

team in July 2005.

Our group has a very outstanding engineering staff, mechanical engineer Weaverdyck and

electrical engineers, Ball, Dai, and Gregory. They are partially supported by our base program

and partially supported by Project funds. These engineers represents one of the strongest

assets we have provided the ATLAS experiment over the past years. They have handled many

complex technical issues. Gregory will leave the group in summer of 2005. Weaverdyck, Ball

and Dai will continue to provide crucial technical support to the ATLAS detector installation,

commissioning, and calibration activities. They all have truly remarkable records over past

15-25 years in experimental high energy physics.

Schick and Yanchula are very good mechanical technicians who were supported by our base

program. They were key personnel for the detector construction work at the University of

Michigan site. With the conclusion of the construction project at Ann Arbor, they are both

leaving the UM ATLAS group.

With Project support, we hired many students and invited several visiting scholars from

Institute of High Energy Physics in Beijing to help build and test the muon detectors. Our group

has been very successful in training and using undergraduate students in hardware projects at

Ann Arbor and at CERN. Many undergraduate students entered first class graduate schools

after working with us and gaining enormous laboratory experience.

2 Λ

study in ATLAS

One focus of the UM ATLAS physics effort is the study of Λ

hyperons in the ATLAS experiment,

with special interest on measuring its polarization and its parity violating α

parameter. The

is the lightest baryon containing a b quark, and since its discovery in 1991 by the UA1

Collaboration (Phys. Lett. B 243, (1991) 540-548) it has created a great deal of interest.

Besides the so−called Λ

lifetime puzzle (Phys. Rev. D 68, (2003) 114006), the Λ

has been

subject of various theoretical studies ranging from proposed tests of CP violation (Z. Phys. C

56, (1992) 129), T violation tests and new physics studies (Phys. Rev. D 65, (2002) 091502),

measurement of top quark spin correlation functions (Eur. Phys. J. C 19, (2001) 323) and

the extraction of the weak phase γ of the CKM matrix (Phys. Rev. D 65, (2002) 073029).

Specific physics interest in the α

parameter studies derives from its ability to serve as a test

for various heavy quark factorization models and perturbative QCD (PQCD). Λ

studies are

also of interest because of the continuing mystery of why hyperons have consistently displayed

large polarizations when produced at energies even up to several hundred GeV and at large p

where most models predict zero polarization. It is not known if these effects can be explained

by some not yet understood effect of existing physics or if they point to new physics altogether.

polarization holds the possibility of illuminating just how polarized b quarks are produced

and, indeed, it may have relevance to how fermions are produced in all pp induced processes.

The ATLAS-wide Λ

Working Group is led by Homer Neal and includes Eduard De La

Cruz Burelo (Michigan), Natasha Panikashvili (Michigan), Shlomit Tarem (Technion Israel)

and Maria Smizanski (Lancaster UK and ATLAS B-Physics Group Coordinator). We expect to

acquire approximately 35,000 Λ

’s which decay into four final-state charged particles through

the reactions Λ

→ J/ψ(μ

−

)Λ(pπ). Angular correlations amongst the four final state charged

particles (see Figure ??) is sufficient to determine both the parity violating parameter α

and

the polarization of the Λ

itself, at a statistical level of a few percent.

Figure 2: Caption for angulardistribution

During the last years the Λ

working group has faced many challenges to demonstrate that

the polarization of polarized Λ

’s can indeed be measured in the ATLAS detector, and through

studying the feasibility of this measurement the Λ

working group has contributed to the overall

efforts on understanding the expected response of the ATLAS detector. This work is summarized

in a recently submitted final version of the so-call ATLAS CSC-13 note (CSC is for Computing

System Commissioning). Homer Neal is the primary editor for this note that reports on the Λ

polarization measurement studies as part of an overall ATLAS effort to test the latest release of

the ATLAS software environment just before the start up of the LHC. As part of this work, a

new Monte Carlo generator, EvtGen, was adapted to generate polarized particles in the ATLAS

experiment. More than 400,000 polarized Λ

’s inside the ATLAS detector have been generated

with different polarizations, and reconstructed to demonstrate that we can recover the known

polarization. We study the expected background composition for our Λ

signal (see Figure ??).

In addition, we report on the fitting techniques developed to deal with multidimensional distri-

bution fits commonly found in polarization measurements. We also studied the effects of small

statistics on the fit results, a situation to be faced in the beginning of the ATLAS experiment.

This note will be published as part of an ATLAS report book on the CSC studies.

As we prepare for the measurement of Λ

polarization in the ATLAS experiment we are cog-

nizant of the fact that a modest event sample is already available for analysis in DØ experiment.

Indeed, we have been able to extract roughly 100 clean Λ

events from DØ data, have proceeded

to make a new determination of the Λ

lifetime and have submitted a PRL publication with the

new results. We have invested considerable effort in attempting to make a rough measurement

in DØ of the α

parameter of the Λ

, since this factor determines the sensitivity for our AT-

LAS studies and knowing its magnitude would be helpful in planning our strategy for ATLAS

running. We had found that we need approximately a factor of 5 more data than previously

existed in the DØ data sample to make a meaningful determination. Indeed, it was this need

for more events that caused us to push for the re-processing of DØ data, which subsequently led

to our discovery of the recently announced Ξ

−

. More details of the Ξ

−

analysis can be found

in the DØ section of this report. However, it is important to notice that those techniques we

have developed in the discovery of the Ξ

−

will clearly have relevance in the search for new heavy

baryons in ATLAS. These baryons, in addition to its own importance as new particles, could

be present as background of our Λ

reconstruction for polarization studies, and then its search

becomes also relevant for our analysis.

It is to be further noted that techniques we have developed for analyzing the four final state

Invariant Mass (MeV/c

)

5000

5200

5400

5600

5800

6000

200

400

600

800

1000

1200

All

→

J/ψ +

→

J/ψ +

(Λ

γ)

B→ J/ψ

⊕ Λ

Figure 3: Angular distribution of Λ

→ J/ψ(μ

−

)Λ(pπ

−

) decays.

charged particles from Λ

to extract its spin information will also have relevance to Higgs →

4l studies (with four charged particles in the final state in each case), both for the purpose of

confirming the JCP of the Higgs, as well as a possible discovery tool if it can be shown that the

angular correlations can help discriminate against background. We plan to explore this topic as

part of our overall Michigan Higgs effort.

3 Tier-2 and Tier-3 Computing

The basic ATLAS computing philosophy is based on a grid-linked system of tiered computing

centers. This philosophy has emphasized the leveraged acquisition of computing hardware and

operational personnel with little regard for mission-oriented computing activities. In the latter

half of 2006 our group, along with the ATLAS group at Michigan State, was chosen to host the

ATLAS Great Lakes Tier-2 center (AGLT2). This facility first came on-line in September 2006

utilizing donated CPU cycles at the Center for Advanced Computing (CAC) in the UM Engi-

neering College. In April 2007 our own Tier-2 facility housed in the College of LS&A Computing

Room came on-line and our use of the CAC terminated. As with the CAC, job submission to

this system is via USATLAS panda and is regulated by the USATLAS Resource Allocation

Committee (RAC). We operate the AGLT2 center primarily as a service to US ATLAS, with

strictly limited access for private physics analysis. As is true for other ATLAS groups, we are

obligated to pursue our physics goals through a local Tier-3 center.

3.1 Tier-3 at Michigan

We intend to participate fully in the ATLAS physics program with particular emphasis on

diboson physics, Higgs searches, lambda-b and SUSY studies. As responsible and dedicated

scientists, we must plan to implement sufficient computing resources for our own use within

a local ATLAS Tier-3 facility. We count on our funding agencies for the appropriate support

in acquiring and operating the necessary computational facilities. Based upon current ATLAS

estimates a typical ATLAS physicist should minimally require 15-25 SI2K of processing power

Invariant Mass (MeV/c

)

5100 5200 5300 5400 5500 5600 5700 5800 5900 6000 6100

Events / ( 50 )

100

150

200

250

Invariant Mass (MeV/c

)

5100 5200 5300 5400 5500 5600 5700 5800 5900 6000 6100

Events / ( 50 )

100

150

200

250

→

ψ Λ

→

ψ Σ

(

Λ γ

)

Figure 4: Angular distribution of Λ

→ J/ψ(μ

−

)Λ(pπ

−

) decays.

and 2.5TB of disk space to support local Tier-3 research needs at LHC startup. At current

pricing, upgrade, maintenance and support is about $7.5K / physicist / 3 year period. For a

group of our size, the total cost will be about $150K over three years or $50K/year. This is a

modest amount, especially considering that in past years our ATLAS group was supported at

the $90K/year level for such computing resources.

We have fully utilized the HYPNOS cluster acquired three years ago (now called UMROCKS)

to support our extensive physics simulation activities. UMROCKS is described in detail in the

section below. During the last three years we have had to deal with UMROCKs age, dropping

from a high point of 220 processors to 120 currently. A significant effort has gone into repair

and replacement of failing components, especially as this systems age past their useful life, and

we expect UMROCKS to end its useful life by the end of 2008. New purchases (outlined below

and funded by DOE and the University of Michigan) have augmented our Tier-3 and Muon

Analysis Center capacity during the last three years to provide our large physics group with

sufficient capacity to support our efforts. We will need to continue to upgrade and maintain

these resources during the coming years.

3.1.1 The UMROCKs Cluster

The UMROCKS computing cluster, formerly an NPACI resource, was transferred to the Physics

department into an appropriately modified laboratory space (2268 Randall). The UMROCKS

cluster today provides 60 nodes of dual AMD Athlon MP 2000 (and 2400) processors, 2 GB

of RAM per node, two 100 GB hard disks per node, a one-TB RAID array and a FastEthernet

switch for interconnections. While this cluster is over five and a half years old, it is still a useful

computing resource capable of running ATLAS simulation, reconstruction, and analysis jobs

though its lack of memory (below the ATLAS specification of 2GB/processor) is becoming more

problematic. UMROCKS has given us extensive experience in system and cluster management.

We have learned a number of lessons about reliability and robustness that we have integrated into

our operations and future equipment planning. Understanding how best to organize and manage

storage, networks and computing resources is critical to delivering a successful infrastructure to

support our ATLAS work.

3.1.2 Tier-3 Purchases

During the last three years we have augmented our UMROCKS cluster by purchasing additional

computational and storage nodes as well as providing workstations for our physicists, including

graduate students. New Tier-3 resources include 20 dual dual-core Opteron nodes, 8 dual quad-

core Intel nodes, 5 storage nodes of varying sizes providing ≈ 60TB of space and approximately

10 workstations located at Michigan and CERN. We have been able to cost effectively deploy

and maintain these Tier-3 resources by leveraging our existing Tier-2 provisioning, monitoring

and management systems. This was possible because the University allocated us Tier-3 space,

co-located with our Tier-2 center.

3.1.3 Tier-3 Plans

Because of the age of the UMROCKS nodes and the ever increasing demand for computational

power and storage required to meet our ATLAS computing needs we are continuing our plan

to acquire and deploy equipment in constant dollars per year. The University has been very

supportive of our needs and contributed funding during the last three years (now complete) to

help us acquire our initial required Tier-3 resources. We plan to provide newer, more capable

equipment each year, both to meet expanding demands (as LHC turns on and ramps up) as

well as to replace outdated or failing equipment. By spreading out the purchases we are able

to reap the benefits implicit in Moores law increases in capability for constant dollar amounts.

The $50K/year from DOE will allow us to maintain our Tier-3 capability as we move forward.

A typical storage node costs $25K. Workstations are about $1.5K. Compute nodes providing

8 job slots cost $5K. Planned Tier-3 expenses are 4 workstations/year, one storage node/year,

three compute nodes and the remainder for parts and maintenance.

3.2 Michigan ATLAS Muon Alignment and Calibration Center

As noted in the Calibration Center section, the University of Michigan was selected as one of

three ATLAS Muon Alignment and Calibration centers (along with Munich and Rome) because

of our expertise in the endcap Muon system, our significant computing capabilities and our

excellent network infrastructure. Thus Michigan has a unique responsibility within the US in

being both a Tier-2 center and an ATLAS Muon Alignment and Calibration Center. This

imposes some additional requirements on a computing site. The calibration and alignment

constants must be determined within 24 hours of the raw data being acquired at the collider

to enable the Tier-0 center to complete a first-pass data reconstruction within 48 hours. This

requires real-time deadline limited processing to ensure that we are able to provide the required

parameters to the Tier-0.

Our estimates of the computing power required to produce the needed alignment and calibra-

tion parameters are about 200 current processors each with 2 GB of RAM and around 100TB

of disk to store data and maintain a history of datasets. This should be provided by Tier-2

resources allocated to us by the USATLAS RAC (Resource Allocation Committee).

3.3 Michigan-MSU ATLAS Great Lakes Tier-2

As mentioned above, the University of Michigan and Michigan State ATLAS groups proposed

and were awarded one of the final two US ATLAS Tier-2 sites. The ATLAS Great Lakes Tier-2

Figure 5: Caption for feb-gratia-table2

(AGLT2) came on-line in September, 2006. Shawn McKee is the Director (at 50% time), Chip

Brock of Michigan State is the Co-Director, and Bob Ball is the overall Tier-2 Manager (at 100%

time).

Our Tier-2 hardware is partially based upon our experience with prototype hardware pur-

chased for our Tier-3 system. The prototyping tests on one storage node and four dual dual-core

processors nodes, and later on an additional 10 processor nodes, helped to refine our specifi-

cations on both mother-board hardware and storage implementation. Subsequently our first

round of Tier-2 purchases was bid out and resulted in the purchase of equipment from Dell

Computers, consisting of a single 20TB disk server and 20 dual quad-core processing nodes with

2GB of RAM per processor core. In fall of 2007 we purchased an additional 52 dual quad-core

systems and 5 new 40TB disk servers at UM and 55 dual quad-core systems and an additional

5 new 40TB storage servers for the MSU site. These installations at UM and MSU benefit from

excellent networking, strong institutional support and significant local computational, grid, sim-

ulation and analysis expertise. Because of MiLR (Michigan Lambda Rail) our Tier-2 site will be

unique within ATLAS in having redundant 10 Gigabit Ethernet links between our distributed

Michigan-MSU sites and Chicago.

Since the new MSU component of AGLT2 came online in December 2007 the AGLT2 has

performed near the top of all US Tier-2 sites (including CMS). Using the most recent statistics

available (February 2008) as an example, our site finished a close second in the most production

processing out of 12 US sites (ATLAS and CMS) and third out of 36 ATLAS Tier-2s worldwide.

The excellent network connectivity has allowed us to surpass all other USATLAS Tier-2 sites

in terms of fetching and providing access to ATLAS AOD and NTuple analysis objects. In

summer of 2008 we will further expand both our Tier-2 and Tier-3 capability with the purchase

of additional disk and computation servers for the beginning of ATLAS running.

3.4 Calibration Center

The ATLAS group at Michigan has joined with groups from Rome and Munich to provide Muon

Calibration Centers for the ATLAS experiment. These centers will provide the MDT calibration

constants to keep the muon spectrometer operating at its full potential: 1% momentum reso-

lution @100 GeV/c rising to 10% at 1 TeV/c. The calibrations required by the MDT tube are

the timing offset (”T0”), and time to space function (”RT function”). The T0 represents the

time measured for particles passing with next the the MDT wire, where the drift time is zero.

The RT function converts drift-times to drift position. The calibration centers will provide daily

updates to the MDT calibration constants for use with the prompt reconstruction of ATLAS

data at the CERN tier-0 computing center. Michigan will concentrate on the endcap detector

calibrations, while Rome and Munich will concentrate on the barrel region.

Michigan, Rome, and Munich have developed the calibration plan over the past few years

and full implementation is nearing completion. The plan, described in detail below, is to use a

special high-statistics data stream from ATLAS, which is transferred to the calibration centers

for processing at the Tier-2 computing centers at each site. Full calibrations are done daily and

results are stored in a local calibration database. A validation program is run, and changed

constants are uploaded to the central ATLAS ”conditions” database at CERN, all with a 24

hour turn-around time.

The ordinary ATLAS data stream collects about 3 million muon tracks per day which is

not adequate to do daily muon system calibrations. For this reason, the calibrations will be

done using a special data stream of muon-system only data which is extracted from the ATLAS

level-2 trigger processors, collecting 30 million muon tracks per day. The calibration stream

has been tested in cosmic-ray running of the past year and is working for the barrel-region of

the muon spectrometer. The endcap data collection is currently under test, and we expect it

to be ready by the time of first beam collisions in July, 2008. The calibration data stream will

be transferred to the calibration centers via the standard ATLAS distributed data management

(DDM) system.

At the calibration centers the calibration data stream will be processed by the ATLAS data

analysis program, Athena, which produces ”calibration ntuples”, which contain reconstructed

muon tracks which are the input the calibration process. When adequate statistics have been

collected, the calibration algorithms are run: first the T0 calibrations in which the rising and

falling edges of the drift-time spectrum are fit; secondly the RT function calibration which is

done by finding the RT function which minimizes the tracking residuals. These calibrations have

been developed and optimized over the past few years.. The computing requirements for are

estimated to be about 2000 CPU-hours per day per site, which is to be supplied by the ATLAS

tier-2 computing centers at each site.

The daily calibration results will be stored in a local calibration database, and a sepa-

rate validation program will be run to monitor calibration results and to check for update

constants. Updated constants will be transferred to the CERN conditions database. The cali-

bration database will be replicated to a central calibration database at CERN using the Oracle

”Streams” technology.

The Michigan group has also developed several backup calibration procedures in case any

of the pieces of regular calibration plan is not functioning. As an alternative to the calibration

stream, Daniel Levin, Andrew Eppig, and Tiesheng Daits have developed a method of stripping

ATLAS data files to remove all but muon-system data which results in a greatly reduced file size.

We also have a alternative data transfer path using the ”Ultralight” network which Shawn McKee

has set up. This system uses a protected, dedicated channel on the US-LHCnet trans-Atlantic

optical fiber to transmit data at 288 Megabits/second. We also have a backup calibration

program, Mutrak, developed by Daniel Levin capable of doing the T0 and RT calibrations.

These redundant methods provide a useful cross-check to the Athena-based calibrations, and

give us confidence that we will be able to meet our goal of providing daily muon calibrations.

To complement the calibration center, Michigan, led by Daniel Levin, has built a “gas-

monitor” chamber, a miniature MDT chamber, which has been installed in the ATLAS gas

supply facility at CERN. This chamber monitors both input and output MDT gas supply by

measuring the RT function and maximum drift time every few hours using cosmic-rays. These

measurements allow us to quickly detect changes in the gas mixture as well as providing an

excellent baseline RT function. We have been pleased to see that throughout 2007/2008 the

max drift times from input and output gas lines have converged, indicating that the MDT gas

system contains few contaminants or large leaks.

Edward Diehl is the overall calibration center coordinator at Michigan. He has been re-

sponsible for installing and testing the required software, and has provided calibrations for the

cosmic-ray commissioning runs which have taking place through out the year. Shawn Mckee has

provided additional support setting up both standard ATLAS DDM and backup Ultralight data

transfer processes. Shawn has also set up a local Oracle calibration database and configured it

to be replicated at CERN using Oracle streams. Daniel Levin has developed much of the backup

calibration procedures, and in addition has provided default RT functions derived from the gas

monitor chamber.

The major pieces of the calibration plan have been developed and tested, but there is still

some further development and integration required to bring the full plan to fruition. We fully

expect to have the calibration center function by the time of ATLAS beam data collisions.

3.5 Collaborative Tools

Several members of the University of Michigan ATLAS high-energy physics group and mem-

bers of the School of Information have been quite active in providing collaborative tool support

services to the ATLAS collaboration, and in conducting cutting-edge research designed to pro-

vide better tools. The effort is led by Homer Neal, under the umbrella of the UM ATLAS

Collaboratory Project (ACP), and includes Steve Goldfarb and Jeremy Herr who coordinate

the web recording of ATLAS tutorials and serve various LHC-wide roles, and Mitch McLachlan

who manages the tutorial recordings. Technical accomplishments to date include transatlantic

videoconferencing tests with QoS, advances in authentication techniques, development of web

lecture recording protocols, the first call for a lecture object standard, and the development of

speaker tracking systems.

Based on earlier studies of ATLAS collaborative tool needs organized by our group, CERN

recently established a LCG RTAG (Requirements and Technical Assessment Group) on collabo-

rative tools for the LHC. Steven Goldfarb of our Michigan group was chosen to chair this entity.

The final RTAG report was endorsed by the spokespersons of the four LHC experiments. Since

that time, Steve has been invited to present the findings and recommendations of the group to

various international conferences, including HEPiX 2005 (Stanford), ESnet 2005 (LBNL), CHEP

2006 (Mumbai), and ViDeNet 2006 (Atlanta), and LHC forums, including ACCU (CERN Users

Group) and the ATLAS Overview Plenary in Stockholm. The RTAG report has clearly served

to energize discussions within CERN about the need for collaborative tool support.

A major conference on collaborative tools was organized by a group of LHC Users in 2006,

in which Michigan played a key role, to focus on the need for special tools to support LHC

experiments, and on the state of the art in collaborative tool R/D. The conference was held in

Geneva and the output has been helpful in generating a more focussed effort in addressing the

needs of the LHC collaborations.

ATLAS, CERN and the LHC have taken several steps toward adopting several of the recom-

mendations emerging from these events. This has included consolidation of responsibilities for

infrastructure under CERN IT, the installation of prototype video conferencing facilities, invest-

ment in an MCU complementary to ESnet through the HERMES collaboration, automation of

phone conferencing with the possibility of integration with video conferencing, and the creation

of the RCTF (Remote Collaboration Task Force) at CERN with representation from each of the

LHC experiments and IT. Steve Goldfarb currently serves as the ATLAS representative on that

body.

The Michigan group has received a special allocation from ATLAS management to record

its plenary sessions and software tutorials for posting on the Web to provide access to members

of the collaboration not able to be present in person. Reports that we have from colleagues

in the collaboration have reassured us that our archives are being heavily used and that they

are serving a clear need. We know of no other technology that is as effective in providing

high-quality, inexpensive web lecture recordings as ours and we are constantly seeking ways to

improve it. Our recording activity has been extended by the ATLAS administration through

2008.

Finally, we note the very successful operation of the dual videoconferencing system the

Michigan ATLAS group established at CERN and in Ann Arbor. This robust link has turned

out to be a key component in the massive task of maintaining strong communications between

staff located in Ann Arbor and CERN, for purposes of coordinating the commissioning of the

endcap muon system, dealing with the development of computing clusters, and for meetings

with other colleagues. Indeed, we have been under tremendous pressures to share our facilities

with colleagues from other institutions – which we do whenever possible – even though Michigan

colleagues alone frequently encounter scheduling conflicts with each other. We believe the high

usage, and the numerous productive sessions made possible by these facilities daily, attest to the

wisdom of investments in tools to assist US physicists to operate in a new environment where

their principal activities are based in distant sites.

3.5.1 ATLAS Web Lecture Capture Activities in 2007

One important service provided by the University of Michigan ATLAS Group to the Collabora-

tion is the recording and web posting of talks and tutorials of vital interest to the collaboration

members. This helps keep the users abreast of key events even if they are not able to be physi-

cally present at each event. In 2007 and the first two months of 2008, 191 lectures were recorded

at 11 events. This required a total of eight trips from Michigan: one to Glasgow, one to Stanford,

and six to Geneva. Primry funding for this activity is provided by the ATLAS Management via

a grant to the University of Michigan through USATLAS.

Details of the recording sessions are provided below.

The graph below illustrates the continued growth in the holdings of the ACP archives.

Statistics concerning the usage of the archives by ATLAS Collaboration members are given

below:

Note the drop in usage for the most recent ATLAS Week, 2008 February. All this usage data

is for the RealPlayer lectures. February 2008 is the first ATLAS Week for which we have posted

all the talks as Flash lectures in a timely fashion. A quick survey of the web logs indicates that

usage of the new Flash lectures is radically higher than the RealPlayer versions (73 times as

high).

3.5.2 LHC Collaborative Tool Committee Efforts

Steven Goldfarb currently serves as the ATLAS representative to the LHC Remote Collaboration

Task Force (RCTF) [1]. This is the primary group at CERN charged with overseeing the

development, implementation and maintenance of collaborative tool facilities and services for

the LHC. The existence of this body, which comprises a mix of CERN IT developers and

representatives of the LHC experiments, was one of the primary recommendations of LCG

RTAG 12 (LHC Computing Grid Requirements and Technical Assessment Group), chaired by

Steve in 2005, with the charge of assessing LHC readiness, in terms of collaborative tools [2].

Figure 6: Caption for atlasrecordings

Figure 7: Caption for jeremy1

Figure 8: Caption for Jeremy2

As ATLAS representative, Steve communicates requirements and feedback from the col-

laboration to the developers, analyzes potential solutions, and helps to set priorities for the

implementation of facilities at CERN. He reports activities and plans to the ATLAS top-level

management and directly to the collaboration during Overview Week plenary sessions, via a

dedicated mailing list [3] and a weekly ATLAS eNews column [4].

Over the past few years, the RCTF has implemented several key RTAG recommendations:

centralization of collaborative tool development under IT, implementation of web-based 24/7

phone conferencing, central documentation for commonly used facilities, and integration of ser-

vices, such as event management software, room booking, and audio/video conferencing tools.

One of the most important developments over the past year has been the drafting and signing

of Memoranda of Understanding between the experiments and IT to cover the installation and

maintenance of video conferencing facilities in all of the meeting rooms at CERN commonly

used by the LHC. The installation, based on important resource committments from the col-

laborations and CERN, will be completed in 2008, making it possible for remote physicists

(such as ourselves) to stay informed and to actively participate in the important and exciting

decision-making processes that will accompany LHC start-up.

[1] RCTF Home: https://cern.ch/twiki/bin/view/RCTF/WebHome [2] LCG RTAG 12 Final

Report (2005): http://cern.ch/lcg/PEB/Documents/RTAG12-Report.doc [3] ATLAS Collabo-

rative Tool Hypernews: https://hypernews.cern.ch/HyperNews/Atlas/get/collaborativeTools.html

[4] ATLAS eNews: http://cern.ch/atlas-service-enews/index.html

3.5.3 CHEP Publications and ATLAS News

Over the past two years, our group summarized and reported its research, development, policy

findings and recommendations relevant to the LHC and HEP, to the International Conference

on Computing in High Energy and Nuclear Physics (CHEP). At CHEP 2006, in Mumbai, India,

Jeremy Herr presented a detailed and ground-breaking technical report on our efforts to develop

a robotic camera for web-based lecture archiving [1]. Steven Goldfarb presented a description

of the Web Lecture Archive Project [2], a joint initiative between CERN and the University

of Michigan, to record a comprehensive archive of synchronized web-based lectures in support

of HEP. He also presented the findings of RTAG 12 [3] (described above) and elaborated his

recommendations for action, in terms of policy and development, for the LHC. The conference

presented an excellent opportunity for us to report our work and the presence of this somewhat

novel material to CHEP sparked a great deal of interest.

At CHEP 2007, in Victoria BC, Canada, a complete session was dedicated to Collaborative

Tools for the very first time, and our group took advantage of the opportunity. Jeremy presented

a summary of the University of Michigan lecture archiving program [4], including an extensive

overview of the ATLAS and the LHC archives, and describing key, recent technical develop-

ments. Steve gave an update to his 2006 report on RTAG 12 [5], describing action taken since

publication of the group’s findings, and charting a course to complete implementation of the

recommendations, in time for LHC start up. He also prepared a presentation summarizing the

outcome of the ”Shaping Collaboration 2006” workshop[6], held in Geneva in December 2006,

addressing issues facing the LHC, in terms of collaborative tools, at the approach of start up.

This workshop [7] was organized by a committee chaired by Homer Neal.

In addition to conference presentations and published proceedings, we present the work of

our group to the ATLAS Collaboration during Overview Week plenary sessions, and in an-

nouncements to the general collaboration mailing list and ATLAS eNews [8].

3.5.4 Remote Teaching Facility

(to be filled in)

4 Education and Outreach

The University of Michigan ATLAS Group has taken quite seriously its obligation to participate

in education and outreach. We are aware that the principal center for research in high energy

physics is migrating to a abortaory at a foreign site that is not easily accessable to most U.S.

students. We operate, under a grant from the National Science Foundation, the only REU site

program for the U.S. at CERN. Over the past 8 years over xxxx students have participated

in a summer-long research and training program at CERN, in association with the prestigious

CERN Summer Student Program. The International Programs Division of the NSF has recog-

nized our program as an exemplar program and has featured it on their homepage for a year.

Students spend their mornings in lectures in the CERN Summer Student Program, and spend

the remiander of their time working on real physics and computer science problems in total

immersion with their assigned research groups. Mentors from these groups have uniformly re-

ported that the students make genuine positive contributions to the work of these groups and

that their impact extends far beyond the students period of residence at CERN.

Our students are selected on a nationally compet