View the Entire April 1998 Issue in PDF - Lawrence Livermore

April 1998

Lawrence

Livermore

National

Laboratory

Meeting the ASCI ChallengeMeeting the ASCI Challenge

Also in this issue: • Argus Advances in Security Technology• Supersensitive Superconductor Detectors

About the Review

The computer simulation of compressibleinstability and turbulent mixing on the coverwas produced at Lawrence Livermore on theAccelerated Strategic Computing Initiative(ASCI) Blue Pacific computer system usingadvanced codes developed at the Laboratory.The Department of Energy’s ASCI program is afundamental component of DOE’s StockpileStewardship Program. A cooperative effortinvolving the three national weapons laboratories,private industry, and academia, ASCI is rapidlyfulfilling its mission of substantially increasingthe nation’s computational power so that numericsimulations can contribute to the assurance ofthe safety and reliability of our nuclear deterrentwithout testing. Turn to p. 4 for a report onASCI’s progress in meeting its stockpilestewardship responsibilities.

• •

About the Cover

Lawrence Livermore National Laboratory is operated by the University of California for theDepartment of Energy. At Livermore, we focus science and technology on assuring our nation’s security.We also apply that expertise to solve other important national problems in energy, bioscience, and theenvironment. Science & Technology Review is published ten times a year to communicate, to a broadaudience, the Laboratory’s scientific and technological accomplishments in fulfilling its primary missions.The publication’s goal is to help readers understand these accomplishments and appreciate their value tothe individual citizen, the nation, and the world.

Please address any correspondence (including name and address changes) to S&TR, Mail Stop L-664,Lawrence Livermore National Laboratory, P.O. Box 808, Livermore, California 94551, or telephone (925) 422-8961. Our electronic mail address is [email protected].

Prepared by LLNL under contractNo. W-7405-Eng-48

S&TR is available on the Internet athttp://www.llnl.gov/str. As references becomeavailable on the Internet, they will be interactivelylinked to the footnote references at the end of eacharticle. If you desire more detailed informationabout an article, click on any reference that is incolor at the end of the article, and you will connectautomatically with the reference.

Electronic Access

We want to know what you think of ourpublication. Please use the enclosed survey formto give us your feedback.

What Do You Think?

SCIENTIFIC EDITOR

J. Smart

MANAGING EDITOR

Sam Hunter

PUBLICATION EDITOR

Dean Wheatcraft

WRITERS

Sam Hunter and Gloria Wilt

ART DIRECTOR AND DESIGNER

Kitty Tinsley

INTERNET DESIGNER

Kitty Tinsley

COMPOSITOR

Louisa Cardoza

PROOFREADER

Al Miguel

S&TR is a Director’s Office publication,produced by the Technical InformationDepartment, under the direction of the Office of Policy, Planning, and SpecialStudies.

2 The Laboratory in the News

3 Commentary by Michael Anastasio and David Cooper The Accelerated Strategic Computing Initiative: A Daunting Technical Challenge

Feature4 Computer Simulations in Support of National Security

The Accelerated Strategic Computing Initiative is multiplying computing power and vastly improving simulation technology as part of DOE’s Stockpile Stewardship Program, which seeks to support assessment and certification of the nation’s nuclear deterrent in the absence of nuclear testing.

Research Highlights13 Argus Oversees and Protects All16 Of Josephson Junctions, Quasi-particles, and Cooper Pairs

19 Patents and Awards

20 Abstract

S&TR Staff April 1998

LawrenceLivermoreNationalLaboratory

Printed in the United States of America

Available fromNational Technical Information ServiceU.S. Department of Commerce5285 Port Royal RoadSpringfield, Virginia 22161

UCRL-52000-98-4Distribution Category UC-700April 1998

Page 4

Page 16

Page 13

2 The Laboratory in the News

Science & Technology Review April 1998

Laboratory spotlights movie special effectsIn a presentation open to the public at Lawrence Livermore,

three computer graphic artists from Industrial Light and Magic, aspecial-effects company based in San Rafael, California, explainedhow they work movie magic behind the scenes on some of today’shottest films. Their presentation, “The Digital Creature Feature:Putting It All Together,” walked the audience through a typicalsequence of events, from the director’s first phone call to thecomputer modeling work to assembling all the elements thatmake up the final shots as we see them on the screen.

The free presentation was part of the Laboratory’s “Scienceon Saturday” lecture series, a nine-week series of talks gearedto middle and high school students, and their teachers, parents,and chaperones. Topics are selected from the forefront ofscientific research, covering a variety of disciplines.Contact: Dolores Doyle (925) 422-5056 ([email protected]).

Laboratory produces largest optical crystalScientists at Livermore have succeeded in using a “rapid-

growth” method to produce the world’s largest single-crystaloptical elements. The pyramid-shaped KDP (potassium dihydrogenphosphate) crystal, measuring about 3 feet tall and over 20 incheswide at the base and weighing nearly 500 pounds, was grown in a6-foot-high tank filled with nearly a ton of supersaturated solution.

The fast-growth method was pioneered in Russia by NataliaZaitseva at Moscow State University and perfected at Livermoreby Zaitseva and Laboratory scientists over the past few years. Itallowed scientists to grow the record crystal in six weeks. Previousmethods would have required a growing period of 12 to 24 monthsto achieve the same result. Slices of the KDP crystals will becritical components of the world’s largest laser, the NationalIgnition Facility, currently under construction at Livermore.Contact: Natalia Zaitseva (925) 423-1505 ([email protected]).

Scientists fine-tune mine detectionLawrence Livermore scientists are working with Ukrainian

scientists to evaluate a new device to detect land mines. “This may be a solution to a small part of a large problem,”

says Arthur Toor, physics project leader for Lawrence Livermore.An estimated 100 million land mines in 70 countries cause26,000 casualties each year.

Toor and his Livermore colleague David Eimerl are workingwith Ukrainian scientists Gregory Pekarsky and Vitaly Bystritskito evaluate a new detection device that is similar to one the Russianmilitary is using. The device is used like a metal detector, but itstechnology involves detecting large amounts of hydrogen. Whenit detects a large amount of hydrogen, its gauge rises, indicatingthat a land mine may be present. Then the spot is tested for certainhigh levels of nitrogen that indicate a land mine is present.

In January, the scientists demonstrated Pekarsky’stechnology at the Buried Objects Detection Facility at theNevada Test Site, a secure area with 296 defused mines.Contact: Arthur Toor (925) 422-0953 ([email protected]).

UC-managed labs rated excellentThe three national laboratories operated by the University

of California for the Department of Energy received overallratings of excellent from DOE for fiscal year 1997, based onannual assessments of the laboratories by the University aswell as the Department’s own reviews.

The findings, which a University spokesman said areconsistent with those of previous years, stem from a five-yearcontract signed by DOE and the University in 1992, whichpioneered the concept of performance-based management fornonprofit operators of DOE laboratories. The UC President’sCouncil on National Laboratories characterized the scienceand technology activities at Los Alamos and Livermore asoutstanding, the highest ratings available. Lawrence BerkeleyNational Laboratory received an excellent rating. The UCLaboratory Administration Office concluded that theadministration and operations performance of each of the threelaboratories exceeds expectations, which the University saidis equivalent to an excellent rating. Based on these Universityassessments, DOE gave overall ratings of excellent to allthree laboratories.Contact: Public Affairs Office (925) 422-4599 ([email protected]).

NIF mammoth didn’t live aloneThe ancient mammoth excavated from Lawrence Livermore’s

National Ignition Facility construction site had loads of company,including at least one other mammal that apparently made ameal out of the mammoth, a scientist says.

Paleontologist C. Bruce Hanson found tooth marks on anupper leg bone of the elephant-like animal, which died roughly10,000 years ago. “There are a few carnivores living then thatcould have been responsible,” Hanson said. “I have to take alook and see if there’s any chance of matching up the toothmarks with the known suspects.”

The most likely culprit would have been a dire wolf similar insize to modern wolves, he said. But saber-toothed tigers and atype of lion that lived in the ancient, stream-crossed valley mighthave killed the mammoth or simply gnawed on its carcass.

Two other clusters of bones lie on the superlaser site, Hansonsaid. One includes the skull of an ancient horse, probably fromthe Pleistocene era, and the other cluster includes several ribsand a shoulder blade that appear to have come from a giantground sloth, which probably stood 10 feet tall on its hind legs.Contact: Public Affairs Office (925) 422-4599 ([email protected]).

HE Department of Energy’s Stockpile Stewardship Programis designed to maintain high confidence in the safety,

reliability, and performance of the U.S. nuclear weaponsstockpile into the indefinite future under a ComprehensiveTest Ban Treaty. This objective requires a fundamentalunderstanding of nuclear weapons science and technologythrough an integrated program incorporating archival nucleartest data, high-fidelity nonnuclear experiments, and advancedcomputer simulations. The Accelerated Strategic ComputingInitiative (ASCI) was developed to provide the simulationcapabilities required to meet the stockpile assessment andcertification requirements in this uncharted environment. Akey element of this initiative is an aggressive plan to acquirecomputing systems that are substantially beyond those thatcan be expected simply from market forces. It also representssystems that are (1) at least an order of magnitude greater inspeed and memory than any currently in use and (2) three ordersof magnitude greater in capability than servers used by mostindustry and commercial firms and some governmentlaboratories. Unprecedented advances will also be necessaryin several areas of computer science and technology to integratethe data transmission, networking, storage, and visualizationneeds of the ASCI simulation codes.

To meet this tremendous challenge requires much more thanjust developing a hardware environment. New simulation codesmust be developed that incorporate more fundamental science,

T

3


Commentary by Michael Anastasio and David Cooper

a more accurate database describing the properties of materialsunder nuclear weapon conditions, and new algorithms that takeadvantage of the massively parallel architectures of the ASCIcomputers. And the ASCI program’s extremely short deadlineamplifies the challenge—the target for assembling this high-fidelity three-dimensional simulation capability is the year 2004.

The ASCI approach to accelerating the nation’s computersimulation capability reflects the fact that expertise in themany disciplines required to do this job does not necessarilyreside within any single laboratory. Many of the importantcontributions to our nation’s technological future must befound within the wider circle of scientific expertise representedin a variety of ASCI partnerships—involving the Sandia, LosAlamos, and Lawrence Livermore national laboratories andfeaturing collaborations with the commercial computerindustry and some of the nation’s leading universities.

Here at Livermore, participation in the ASCI programextends across two directorates: the Defense and NuclearTechnologies Directorate and the Computation Directorate.This issue’s feature article, which begins on p. 4, describesASCI, a critical element of DOE’s Stockpile StewardshipProgram.

■ Michael Anastasio is Associate Director, Defense and Nuclear Technologies.■ David Cooper is Associate Director, Computation, and the Laboratory’s

Chief Information Officer.

The Accelerated Strategic Computing Initiative:A Daunting Technical Challenge

IME is running out on the U.S. nuclear weaponsstockpile. As the weapons age beyond their design

lifetimes, important questions arise: Are the weapons stillsafe? Will they still perform reliably? How long will theycontinue to be reliable? What maintenance and retrofittingshould be prescribed to extend their working life? Thesequestions must be answered with confidence as long asnuclear deterrence remains an essential part of U.S.national security policy.

With the U.S. commitment to the Comprehensive TestBan Treaty, the viability of the U.S. nuclear arsenal can nolonger be determined through underground nuclear testing.Thus, new approaches are being taken to maintain andpreserve the U.S. nuclear deterrent through DOE’sStockpile Stewardship Program.

One key component of the multifaceted StockpileStewardship Program is the Accelerated StrategicComputing Initiative (ASCI), an effort to pushcomputational power far beyond present capabilities soscientists can simulate the aging of U.S. nuclear weaponsand predict their performance. To calculate in precise detailall the complex events of a thermonuclear explosion requirescomputational power that does not yet exist, nor would itexist any time soon without the ASCI push, even at computerdevelopment speeds predicted by Moore’s Law (thatcomputer power doubles about every two years). ASCI’sgoal is to put such a high-fidelity simulation capabilityin place in the near future. To do that, the Americancomputer industry must dramatically speed up the pace ofcomputational development. Currently, computing’s top

T


The Accelerated Strategic

Computing Initiative is making

significant progress toward

meeting its major challenge:

dramatically increasing the

nation’s computing power as

a necessary contribution to

the assurance by scientists

of the safety and reliability

of our nuclear deterrent in

the absence of testing.

Computer Simulations in Support of National

Security

5


speed is 1.8 teraflops, that is, 1.8 trillionfloating-point (arithmetic) operationsper second. This speed must increase toat least 100 teraflops by 2004, growththat must be coordinated with a host ofaccomplishments in code developmentand networking.

Why is this accelerated schedulenecessary? Not only are weapons aging,so are the nuclear weapons experts withexperience in designing and testing them.The Stockpile Stewardship Programmust have this high-fidelity, three-dimensional simulation capability inplace before that expertise is gone.“It’s a tremendously ambitious goal,especially under such a short schedule,”says Randy Christensen, ASCI’s deputyprogram leader at Lawrence LivermoreNational Laboratory. Christensendescribes the work as something akin to“trying to get a computer code to run ina few days a simulation that would havetaken so long with current capability thatit would not have been attempted.”

Orchestrating IntegrationASCI is reaching for computational

powers in the hundreds of teraflops, butthe ASCI challenge demands more thanhardware. Meeting it will require carefulintegration of the major elements of anational effort: platform development,applications development, problem-solving environment, and strategicalliances—coordinated work conductedat three national laboratories inpartnership with the commercialsupercomputer industry and the nation’sgreat universities (Figure 1).

To ensure this balanced development,ASCI planning began with a “oneprogram–three laboratories” approach.Project leaders at each laboratory,guided by the DOE’s Office of theAssistant Secretary for DefensePrograms, are implementing thiscollaboration and extending it to ASCI’sindustrial and academic partners. Theoverriding challenge for the ASCI scopeof work is to synchronize the varioustechnological developments with eachother. For example, sufficient platformpower must be delivered in time to runnew advanced codes, and networkingcapabilities must enable the variousparts of the system to behave as if theywere one. The success of ASCI dependson this integration as much as itdepends on the success of ASCI’sindividual elements.

Developing the PlatformASCI’s computer hardware is being

developed by a consortium of threenational laboratories and a select groupof industrial partners in a prime exampleof government–industry cooperation.The national laboratories—LawrenceLivermore, Los Alamos, and Sandia—

are each teamed with a majorcommercial computer manufacturer—IBM, Silicon Graphics–Cray, and Intel,respectively—to design and buildparallel, supercomputing platformscapable of teraflops speeds.

The development of infrastructuretechnologies seeks to tap all availableresources to make these computerplatforms perform the kind of high-fidelity simulation that stockpilestewardship requires. ASCI has aPathForward component, a programthat invites computer companies tocollaborate in developing requiredtechnologies. For instance, theprogram’s first PathForward contracts,announced on February 3, 1998,awarded more than $50 million overfour years to four major U.S. computercompanies to develop and engineerhigh-bandwidth and low-latencytechnologies for the interconnection of 10,000 commodity processors thatare needed to build the 30-teraflopscomputer. (See box, p. 6.) As a resultof this effort, subsequent collaborationsinvolving other agencies, academia,and industry are expected.

Applications

Platforms

Problem-solving environment

Alliances

Tri-lab integration

Figure 1. Meeting the challenge of ASCI requirescareful integration of the major elements of theprogram across three national laboratories andASCI’s industrial and academic partners.

ASCI and National Security

final nuclear yield and the effects ofchanges introduced by remanufacturing(perhaps using different materials andfabrication methods) or defects broughton by aging. In addition, they mustsimulate weapon behavior in a widevariety of abnormal conditions toexamine weapon safety issues in anyconceivable accident scenario. If thisweren’t difficult enough, the new codesmust provide a level of fidelity to theactual behavior of weapons that is muchhigher than their predecessors provided.

The major challenges facing thedevelopers of these advanced simulationcodes are to base them on rigorous, first-principles physics and eliminate many

6



of the numerical approximations andsimplified physics that limit the fidelityof current codes; make them runefficiently on emerging high-performance computer architectures;validate their usefulness by means ofnonnuclear experiments and archivalnuclear test data; and do all of these intime to meet stockpile needs.

Meeting these challenges requiresthe coordinated efforts of over ahundred physicists, engineers, andcomputer scientists organized into manyteams. Some teams create the advancedweapon simulation codes, writing andintegrating hundreds of smaller programsthat treat individual aspects of weaponbehavior into a single, powerfulsimulation engine that can model anentire weapon. Other teams are devotedto developing the advanced numericalalgorithms that will allow these codesto run quickly on machines consistingof thousands of individual processors—a feat never before achieved withprograms this complex. Still others aredeveloping much improved models forthe physics of nuclear weapon operationor for the behavior of weapon materialsunder the extreme conditions of anuclear explosion. Both the scale (thelargest teams have about 20 people) andthe degree of integration demanded bythis complex effort have required a muchgreater level of planning and coordinationthan was needed in the past.

One example of the advancedsimulation capabilities being developedin the ASCI program is its materialmodeling program. Enormously powerfulASCI computers are being used tocarry out very accurate, first-principlescalculations of material behavior at theatomic and molecular level. Thisinformation is then used to createaccurate and detailed models of materialbehavior at larger and larger lengthscales until we have a model that can

Developing ApplicationsASCI is an applications-driven

program. Unprecedented computerpower with a first-rate computingenvironment is required to do ASCI’sstockpile stewardship job, which is torun new computer codes programmedwith all of the accumulated scientificknowledge necessary to simulate thelong-term viability of our weaponssystems. The new generation of advancedsimulation codes being developed in theASCI program must cover a wide rangeof events and describe many complexphysical phenomena. They must addressthe weapon systems’ normal performancefrom high-explosive initiation through

PathForward Contracts Awarded February 1998

Industrial Partner PathForward Project

Digital Equipment Corporation Develop and demonstrate a processor (DEC) interconnect capable of tying Maynard, Massachusetts together 256 Digital UNIX-based

AlphaServer symmetric multiprocessing (SMP) nodes.

International Business Machines Develop future high-speed, low-(IBM) latency, scalable switching Poughkeepsie, New York technology to support systems that

scale to 100 teraflops.

Silicon Graphics–Cray Research Develop and evaluate advanced (SGI/Cray) signaling and interconnect Chippewa Falls, Wisconsin techniques. The technology will be

used in future routers, switches, communication lines, channels, and interconnects.

Sun Microsystems Perform hardware and software (SUN) viability assessments by constructing Chelmsford, Massachusetts interconnect fabric and verifying

scalability and correctness of the interconnect monitoring facilities, resource management, and message- passing interface (MPI) capabilities.

7



be used directly in the weaponsimulation codes (Figure 2). Thiscomputational approach to materialmodeling has already produced a muchbetter understanding of the phasechanges in actinides (the chemicalfamily of plutonium and uranium). Thenew approach is expected to be appliedto many weapons materials, rangingfrom plutonium to high explosives.When fully developed, it will become apowerful tool for understanding and

predicting the behavior of any material(for example, alloys used in airplaneconstruction, steel in bridges), not justthose used in nuclear weapons.

Developing the InfrastructureIn addition to platform and

applications development, ASCI is alsodeveloping a powerful computerinfrastructure. A high-performanceproblem-solving environment must beavailable to support and manage the

Years

Minutes

Milliseconds

Microseconds

Nanoseconds

Femtoseconds

Distance, meters 10–10 10–9 10–8 10–6 10–2 1

Level 6: System validation Examples: fires, aging, explosions

Level 5: Continuum models

Level 4: Turbulence and mix simulations

Level 3: Molecular/atomic-level simulations

Level 2: Atomic-physics opacity simulations

Level 1: Quantum mechanical simulations of electronic bonding in materials

Aging

Fires

Explosions

Figure 2. ASCI is providing DOE’s StockpileStewardship Program with a hierarchy ofmodels and modeling methods to enablepredictive capability for all processes relevantto weapon performance.


Scalable network

High-speed machine

Reality engine

Tape controller and transport

Disk controller and transport

Wide-area network

Local-area network

Network of stations

High-speed switch

High-speed switch

Massively parallel processor

Figure 3. (a) A high-performance problem-solving environment manages the workflowand communications among all ASCIcomputers. (b) The result is ASCI’s ability tobring three-dimensional images resultingfrom calculations to scientists on theirdesktops regardless of the physical locationof the processors doing the work. Thisdistance-computing option features remotecaching of simulation data.

Wide-area network

Mass storage

Local diskGeometry engine

Simulation application

Graphics rendering engine

DesktopCompute engine

(a)

(b)

8 ASCI and National Security

9



of billions of bytes per second. AnotherASCI team is writing scalable inputand output software to move data fromcomputer to computer and reducecongestion between computers andstorage. The changes resulting from itsimprovements will be tantamount tomoving busloads of data, as compared tocarloads—a sort of mass transit for data.

Weapons scientists will be confrontedwith analyzing and understandingoverwhelmingly large amounts of dataderived from three-dimensionalnumerical models. To help them, ASCIis developing advanced tools andtechniques for computer visualization,wherein stored data sets are read into acomputer, processed into smaller datasets, and then rendered into images.The development of visualization toolsfor use across three national laboratorieswill require close collaboration withregard to programming language,organization, and data-formattingstandards. The Livermore team isfocusing on how to reduce data sets forvisualization—because they surely willbecome larger and larger—through theuse of such techniques as resampling,

workflow and the communicationsbetween all the ASCI machines. At anytime, over 700 classified and unclassifiedcode developers and testers may beaccessing ASCI computers, either fromwithin the national laboratories or viathe Internet. A scalable networkarchitecture, in which individualcomputers are connected by very high-speed switches into one system, makesthis high-demand access possible.With such a configuration, the networkis, in effect, the computer (Figure 3).

Allowing large numbers of computersto communicate over a network as ifthey were a single system requiressophisticated new tools to performscientific data management, resourceallocation, parallel input and output,ultrahigh-speed and high-capacityintelligent storage, and visualization.These capabilities must be layered intothe computer architecture, betweenuser and hardware, so that the two caninteract effectively and transparently.The applications integrate the computingenvironment and allow users, forexample, to access a file at any of thethree national laboratory sites as if itwere a local file or to share a local filewith collaborators at any ASCI site.

At Livermore, ASCI staff areperforming numerous projects to developthis integrated computing environment.One team is working on a science-datamanagement tool that organizes,retrieves, and shares data. An importantobjective of this tool is to reduce theamount of data needed for browsingterascale data sets. Another team isdeveloping data storage that will offera vast storage repository for keepingdata available and safe 24 hours a day.The repository will store petabytes(quadrillions of bytes) of information,equivalent to one hundred times thecontents of the Library of Congress.The storage device will also rapidlydeliver information to users, at a rate

multiresolution representation, featureextraction, pattern recognition,subsetting, and probing.

While the fast, powerful machinesand complex computer codes garnermost of the headlines, this problem-solving-environment effort isfundamental to fulfilling the ASCIchallenge. As we come to understandthat “the network is the computer,”the significance of this element of theASCI program comes sharply into focus.

In Pursuit of 100 TeraflopsThe 100-teraflops milestone, the

entry-level computing capabilityneeded to fulfill stockpile stewardshiprequirements, is ASCI’s goal for 2004(Figure 4). Fulfilling it will requireenough computational power to runcalculations distributed over10,000 processors, which is justenough to conduct three-dimensionalweapons simulations at a level ofcomplexity that matches the currentunderstanding of weapons physics.While this computing capacity is notthe final goal, it is already 100,000 timesmore than the computing power used

1996 1997 1998 1999 2000 2001 2002 2003 2004

1.8 teraflops

Com

putin

g ca

pabi

lity

3.2 teraflops

10 teraflops

100 teraflops

30 teraflops 100 teraflops is the entry-level computing capability needed to fulfill

stockpile stewardship requirements.

Figure 4. The ASCI goal is to achieve the 100-teraflops (trillion-floating-point-operations-per-second)threshold by 2004.


by weapons scientists today,represented by Livermore’s J-90 Craycomputer. At 100 teraflops, all of thecalculations used to develop the U.S.nuclear stockpile from the beginningcould be completed in less than twominutes.

ASCI’s approach to the 100-teraflopsgoal has been to use off-the-shelf,mass-market components in innovativeways. It aggregates the processorsdeveloped for use in desktop computersand workstations to scaleup computingpower. It is this approach that makesASCI development cost-effective; andleveraging of commercially availablecomponents will encourage technologydevelopment in the commercial sector.The mass-market approach will takeadvanced modeling and simulationsinto the computational mainstream foruniversal PC use.

Improvements to ASCI power willoccur over five generations of high-performance computers. To ensuresuccess, multiple-platform developmentapproaches are being attempted. Thisstrategy will reduce risk, allow fasterprogress, and result in greater breadthof computing capability. For example,the Sandia/Intel Red machine, whichwas put on line in August 1995, hasachieved 1.8-teraflops speed (currentlythe world’s fastest) and is now beingused for both code development andsimulation. The Lawrence Livermore/IBM Blue Pacific and the Los Alamos/Silicon Graphics–Cray Blue Mountainsystems, which resulted from technicalbids awarded in late 1996, are alreadyrunning calculations.

Blue Pacific was delivered toLawrence Livermore on September 20,1996, with a thousand times morepower than Livermore’s existing CrayYMP supercomputer (Figure 5a). TheLawrence Livermore/IBM teaminstalled and powered up the systemand had it running calculations withintwo weeks. Already, it is conducting

Figure 5. (a) The IBM Blue Pacific computer arrived at Livermore on September 20, 1996, justtwo months after the IBM/Livermore partnership was announced by the White House and aboutsix weeks after the contract was signed. (b) The initial-delivery system has already begunsignificant calculations in important areas of stockpile stewardship such as three-dimensionalmodeling of material properties, turbulence, and weapon effects. Upgrades will bring systempower to 3.28 teraflops (trillion floating-point operations per second) by 1999, with an option toupgrade to 10 teraflops in fiscal year 2000.

(a)

(b)

10 ASCI and National Security


some of the most detailed codesimulations to date.

The Blue Pacific initial-deliverysystem, which arrived in 340 refrigerator-sized crates, takes up a significant portionof Livermore’s computing machineroom space, operates at 136 gigaflops,and has 67 gigabytes of memory and2.5 terabytes of storage (Figure 5b).Initially, each of its 512 nodes containedone processor. During March 1998,these nodes were replaced with four-way symmetrical multiprocessors,quadrupling the number of processors. A further improvement will endow itwith thousands of significantly improvedprocessor nodes for the ASCI productionmodel. These reduced-instruction-setcomputing microprocessors operate at a peak of 800 megaflops and, in thisconfiguration, will bring the system to a total of 3.28 teraflops.

In that three-teraflops configuration,the Blue Pacific’s “SustainedStewardship Teraflops” system alonewould more than fill up all the space inLivermore’s current machine room. Forthat reason, construction crews are nowbuilding and wiring new space toaccommodate it. In new, larger quarters,workers have been installing electricpower, replacing air handlers andcoolers, and hooking up new fans as partof necessary building upgrades. Thenumbers are impressive: 12,000 squarefeet of building extension,5.65 megawatts of power, 11 tons of airconditioning, 16 air handlers that replacethe air four times per minute, andcontrollers that keep the temperaturebetween 52° and 72°F at all times. Thismachine is scheduled to be installed inMarch or April of 1999 (Figure 6).

Involving AcademiaAlthough work on weapons physics

is classified, work on the methods andtechniques for predictive materialsmodels encompasses unclassifiedresearch activities. ASCI thus can pursue

a strategy of scientific exchange withacademic institutions that will morerapidly establish the viability of large-scale computational simulation andadvance simulation technology. Thisstrategy is embodied in the AcademicStrategic Alliances Program. Theprogram invites the nation’s bestscientists and engineers to help developthe computational tools needed toapply numerical simulation to real-world problems. In this way, a broaderscientific expertise is at work makingthe case for simulation; simulationalgorithms are tested over a broad rangeof problems; and the independentlyproduced simulations provide a peerreview that helps validate stockpilestewardship simulations (see box below).

Computers Changed It AllIn the short span of time since

computers came into general use, the

nature of problem-solving has changed,by first becoming reliant on computers,and then becoming constrained by thelimits of computer power. ASCI willdevelop technologies that will makecomputational capability no longer thelimiting factor in solving huge problems.Just as important, ASCI will change thefundamental way scientists and engineerssolve problems, moving toward fullintegration of numerical simulationwith scientific understanding garneredover decades of experimentation.

In the stockpile stewardship arena, theASCI effort will support high-confidenceassessments and stockpile certificationthrough higher fidelity simulations.Throughout American science andindustry, new products and technologiescan be developed at reduced risk andcost. Advanced simulation technologieswill allow scientists and engineers to dosuch things as study the workings of

The Academic Strategic Alliances Program

In July 1997, the Academic Strategic Alliances Program awarded Level I fundsto five universities to perform scientific modeling to establish and validate modelingand simulation as viable scientific methodologies.• Stanford University will develop simulation technology for power generation and fordesigning gas turbine engines that are used in aircraft, locomotives, and boats. Thistechnology is applicable to simulating high-explosive detonation and ignition.• At their computational facility for simulating the dynamic response of materials, theCalifornia Institute of Technology will investigate the effect of shock waves inducedby high explosives on various materials in different phases.• The University of Chicago will simulate and analyze astrophysical thermonuclearflashes. • The University of Utah at Salt Lake will provide a set of tools to simulate accidentalfires and explosions. • The University of Illinois at Urbana/Champaign will focus on detailed, whole-systemsimulation of solid-propellant rockets. This effort will increase the understanding ofshock physics and the quantum chemistry of energetic materials, as well as the effectsof aging and other deterioration.

These Level I projects are part of a 10-year program, in which projects can berenewed after five years. Also under the Alliances program, smaller research projectsare being funded at universities across the country as Level II and III collaborations.

ASCI and National Security 11

disease molecules, so they can designdrugs that combat the disease; observethe effects of car crashes without anactual crash; and model global weather todetermine how human activities might beaffecting it. The uses are limitless, andtheir benefits would more than justifythis investment in high-end computing,even beyond the benefits of ASCI’sprincipal national-security objective.

—Gloria Wilt

Key Words: Academic Strategic AlliancesProgram, Accelerated Strategic ComputingInitiative (ASCI), computer infrastructure,computer platform, parallel computing,PathForward, problem-solving environment,Stockpile Stewardship Program, simulation,teraflops, weapons codes.

12



For further information contact Randy B. Christensen (925) 423-3054([email protected]).

Figure 6. Rendering of the3.28-teraflops IBM BluePacific “SustainedStewardship Teraflops”system in its new homebeing constructed atLawrence Livermore. Themachine is scheduled forinstallation in early 1999.

RANDY CHRISTENSEN is Deputy Program Leader of theDepartment of Energy’s Accelerated Strategic ComputingInitiative (ASCI). He has broad management responsibilitieswithin the ASCI program as well as specific responsibility forapplications development. He holds a B.S. in physics from UtahState University and an M.S. and Ph.D. in physics from theUniversity of Illinois. Following a postdoctoral fellowship at the

Joint Institute for Laboratory Astrophysics (1978–1981), he joined LawrenceLivermore National Laboratory as a code physicist in the Defense and NuclearTechnologies Directorate. He held a number of leadership positions in thatdirectorate before becoming Deputy Associate Director of the ComputationDirectorate in 1992, where his responsibilities included management of theLivermore Computing Center.

About the Scientist

13


ROM outsideLawrence Livermore

National Laboratory, thepublic sees a site protectedby chain link fences andguards at entry gates. But thisDepartment of Energy nationallaboratory, home to a variety of classified research,requires much higher level security measures. Therefore, it isguarded as well by a sophisticated, computerized securitysystem called Argus. Argus was designed, engineered, andinstalled at Livermore and is continually being upgraded andenhanced. It is also available to other Department of Energy andDepartment of Defense facilities.

Although named for the hundred-eyed monster of Greekmyth, Argus security comprises much more than visualcapabilities. A highly interconnected network engineeredwith comprehensive security features, Argus lives up to suchstringent security requirements that DOE’s Office of Safeguardsand Security has cited it as the standard for physical securitysystems protecting facilities where the consequences of intrusionare significant. In addition to Lawrence Livermore, the Argussystem has been installed at three other DOE sites and at oneDOD site to protect top-priority assets or nuclear material.

As it monitors and controls entry into the Laboratory’s high-security buildings, Argus is simultaneously monitoring theentire site for security threats and can alert and direct securityforces to those threats. Argus security is all-encompassing andomnipresent, but it is surprisingly noninvasive. Employees ofLawrence Livermore enter and move about the Laboratorycampus with relative ease. Yet, the Laboratory’s Top Secretdocuments, materials, and facilities are thoroughly protected,intruders can be detected in real time, and intrusions andemergencies get instantaneous response from police andinvestigative personnel. The Laboratory is provided withmaximum security 24 hours a day, 7 days a week.

F

This security results from a software system thatcomprises some 1.5 million lines of code, offering a widerange of security features. Extensive features are necessary,because Argus must accommodate many differentconfigurations of security rules within one security complex,and sometimes one complex may have multiple geographicallocations (for example, Livermore’s Argus system controlsthe main site and the nearby Site 300 high-explosives testingfacility). Moreover, Argus must be reconfigurable at anytime. Extensive features also translate into flexibility andsimplicity for end users. That’s important because everyauthorized person in a high-security site accesses andinterfaces with the Argus system. To ensure that designers,operators, and users understand Argus, DOE’s CentralTraining Academy in Albuquerque, New Mexico, has 14 classes available, ranging from one hour to one week,that cover the complete set of Argus features.

While protecting a security complex, Argus also protectsitself. A high degree of redundancy has been incorporated toprevent system failure, and tamper-indicating devices anddata encryption have been used throughout to protect

Research Highlights

Figure 1. Remote-access panels are the primary interface between users and the Argus system and are the means by which badged employees enter controlled buildings and areas.

Argus Oversees and Protects All

14


Security Technology

employees are “badged,” they are enrolled into the Argussystem and can then use RAPs to gain entry into controlledbuildings and areas on site. They swipe their badges, whichhave been coded with a unique identification number and adecryption key. The RAP communicates the badge informationto the AFP, another microprocessor-based device, which verifiesit against locally stored encrypted access authorization databases.

Argus software allows access based on credentials(determined by a badge), a user’s identity (determined bypersonal identification number and biometrics), clearancelevel, and privilege. Although access rules can be veryrestrictive, the access system provides flexibility by being ableto make fine distinctions within those rules. Thus, it mightallow a person into a high-security building within a classifiedarea but prevent that person from entering an even higher levelsecurity vault within that building. The access system alsoallows changes in user privileges, within rule confines; forexample, regular users can be enrolled to escort visitorsthrough high-security areas. The system eliminates the needfor labor-intensive badge checking, and it monitors, tracks,and logs all badge usage.

In addition to controlling and monitoring the RAP accesscontrols, AFPs also control and monitor the networks ofthousands of electronic sensors and other surveillance equipmentthat comprise the alarm stations of a security complex.

The AFP determines the status of security in the alarmstation by polling its sensors, controls station operating mode(that is, whether the station is open or secured, in maintenance,etc.), and provides entry authorization via the RAP interface.Alarm station caretakers can also use the RAP to modifyaccess lists, change the rules of the alarm station, andauthorize maintenance on the station.

Alarm stations are of many types—outdoor perimeterexclusion zones, normal interior rooms, vaults of concrete orsteel, or even entire buildings. They can have sensors andsurveillance equipment installed on walls, floors, and ceilings.Because as many AFP modules can be installed as necessaryto monitor alarm stations, site security is scalable. At the sametime, its modularity restricts problems and makes maintenanceand diagnostic work easier.

Real-Time Command and ControlOccasionally at the Laboratory, police cars with flashing

lights and howling sirens speed through the streets in responseto an alarm or other security incident. They have beendispatched by security personnel who monitor site security24 hours a day from Argus consoles (Figure 3). The consoles

surveillance equipment and data from intruders and thieves.Insider threats to weaken the system have been addressedwith a comprehensive set of system-enforced and proceduralmeasures, including consistency checking, captive accounts,and a rule prohibiting people from working alone.

How Users Work with ArgusArgus is implemented through four integrated computer

subsystems. One subsystem controls access into buildings andareas. Another monitors alarms and sensors installed throughoutthe site. A third integrates and displays security data so securitypersonnel can assess and control incidents. The fourth providescentral computing and data storage to support the overall systemconfiguration and databases. These four elements providewhat Greg Davis, the manager of Livermore’s Argus program,calls a “God’s eye view” of the site. They connect into a real-time, interactive security assessment and response system.

User interactions with the Argus network are made possiblethrough two hardware components: a remote access panel (RAP)and the Argus field processor (AFP).

The RAP (Figure 1) is a microprocessor-based, programmableinput–output device connected to an AFP (Figure 2). It is theprimary user interface to the Argus system. When Livermore

Figure 2. Argus field processors are microprocessor-based devicesthat verify badge information transmitted from remote access panelsagainst locally stored encrypted access authorization databases.

15


Security Technology

integrate and display graphical data from controlled entrywaysand alarm stations, and they are linked to telephone, radio,and intercom systems. They provide Livermore security staffwith a real-time command-and-control capability.

At each console workstation, an operator controls twohigh-resolution, color display screens that show maps ofsecurity areas and the security equipment contained in them(sensors, entry control devices, cameras). The system displaylists any security anomalies that are occurring and indicatesthe security status of surveillance equipment by color code.Green, for example, indicates normal or secure, while redindicates a potential security threat, an alarm, or a failure.When security anomalies occur, an operator is alerted by thelists and can view them on the screens; the views can beenlarged or adjusted for seeing additional details.

Operators may also be able to zoom in on the anomaly.Consoles can be linked to closed circuit televisions. Consolevideo subsystems have computer-controlled switches capableof delivering signals from any linked television camerasimultaneously to any display monitor and to all recordingdevices. Video options also include pan–tilt–zoom camerasand video motion detectors.

The consoles are ergonomically designed, providingcomfort and ease of use to operators. The number of consolesin operation depends on site requirements and operatorworkloads; Argus can support any number of workstationswithout degradation.

Continuing Improvements, Ever More UsesThe installation of Argus at a major DOE nuclear weapons

storage and dismantlement site is nearing completion. There,

Argus was modified to accommodate access authorizationprocedures that require observation of the two-person rulefor entry and exit. In addition to RAPs, the entry portals havedevices that read stored hand-geometry data, and booths mayhave special detectors to monitor the transport of sensitivematerials. To serve this site and other users, Argus programstaff are developing a 24-hour help line.

They are also moving ahead to evolve Argus to the nexttechnological level, with such features as topology-independentnetwork-based sensors and capability to simulate intrusionsand attacks. In the first, Argus staff are in the midst ofdeveloping a neuron chip that can be embedded into sensors,adding the capability to communicate with sensors instead ofmerely receiving signals from them. This feature will enhanceAFP line supervision of alarm stations, enhance sensor security,and dramatically reduce installation costs. In the second,Argus staff are beginning research and development toendow Argus with simulation capabilities that can be used inconjunction with conflict simulation exercises. Argus consoleoperators will soon be able to detect simulated attacks andsend virtual security dispatches to contain and control them.Such simulation would hone a site’s emergency responsetactics and provide realistic training to console operators.

—Gloria Wilt

Key Words: Argus, Argus field processor (AFP), remote accesspanel (RAP), security technology.

For further information contact Gregory Davis (925) 422-4028 ([email protected])

Figure 3. Securitypersonnel monitorsite security 24 hoursa day from Argusconsoles like thisone at LawrenceLivermore.

16

F Josephson junction brings tomind an intersection of two small

back roads, it’s time to change gearsand think science. This term, alongwith quasi-particle and Cooper pair,is part of the large area ofsuperconductors.

Simon Labov and his colleaguesin Lawrence Livermore’s Physicsand Space Technology Directoratesay these concepts and discoveriesshow great promise for applicationsin areas such as wirelesscommunication, energy storage,and medical diagnostics. Labovand his fellow researchers are usingsuperconductors to create a newgeneration of supersensitivedetectors for nondestructiveevaluation and astrophysics.

When ordinary metal conductselectricity, the electrons carryingthe current collide with imperfectionsin the metal, thereby creatingresistance. But when asuperconducting material is cooledto its critical temperature, electrons pair off into Cooper pairs,named for Leon Cooper, one of the scientists who won a 1972Nobel Prize in physics for explaining the now widely acceptedtheory. Any movement of one electron is matched by equaland opposite movement of the other. As a result, they don’thit the imperfections, no electrical resistance is generated,and electrons flow freely, without the addition of more energy.

But to put these theories to practical use in detectorsrequires a Josephson junction. Named for Brian Josephson,who described the theory when he was a graduate student atCambridge University in 1962, a Josephson junction is twopieces of superconducting material linked by a weak insulatingbarrier. When an x ray hits a Josephson junction, the Cooperpairs break up, and quasi-particles are created. These quasi-particles, which are electronlike or holelike excitations in the

I

superconductor, can tunnel through the weak insulating barrierof the Josephson junction, producing a pulse of electrical current.By measuring the number of Cooper pairs that are broken,scientists can determine the energy of the x ray up to ten timesbetter than with conventional technology and can identify thematerial that emitted the x ray. These superconducting-tunnel-junction (STJ) detectors also work with optical, ultravioletgamma-ray photons and large biomolecules. Labov and histeam are working to use this new technology in applicationsfor analyzing all of these particles.

Measuring Large, Slow MoleculesThe Livermore group has, for example, teamed with

scientists at Lawrence Berkeley National Laboratory and acommercial firm, Conductus Inc., of Sunnyvale, California,

Research Highlights


Stephan Friedrich inserts a sample in the Lawrence Livermore superconducting-tunnel-junction (STJ)detector cryostat at the Stanford Synchrotron Radiation Laboratory. X rays from the synchrotron enter thecryostat through the beamline on the left and strike the sample, producing x-ray fluorescence, which ismeasured by the STJ detector.

Of Josephson Junctions, Quasi-particles, and Cooper Pairs

17


Josephson Junctions

to measure massive, slow-moving macromolecules in DNAresearch. In a typical time-of-flight mass spectrometer using amicrochannel-plate (MCP) ion detector, large ions move tooslowly to be efficiently detected. Using an STJ detector, theteam found that they could achieve close to 100% detectionefficiency for all ions, including the slow, massivemacromolecules. “A comparison of count rates obtainedwith both detectors indicated a hundred to a thousand timeshigher detection efficiency per unit area for the STJ detectorat 66,000 atomic-mass units,” Labov says. “For highermolecular masses, we expect an even higher relativeefficiency for cryogenic detectors because MCPs show arapid decline in detection efficiency as ion mass increases.”

Even more exciting, STJ detectors can measureindependently the mass and charge of the molecule. CurrentMCP detector technology cannot measure the charge of themolecule, and this inability often causes confusion ininterpreting mass spectrometer data. According to Labov,if nonfragmenting ionization techniques can be perfected,cryogenic detectors could make possible the rapid analysisof large DNA molecules for the Human Genome Projectand might be used to analyze intact microorganisms toidentify viruses or biological weapons materials.

High Resolution for Soft X RaysIn another experiment using an STJ, Labov again teamed

with Conductus and seven other Lawrence Livermorescientists to study energy resolution for soft x rays withenergies between 70 and 700 electron volts. The resultsshowed that STJ detectors can operate at count ratesapproaching those of semiconductor detectors while still

providing significant improvement in energy resolution for softx rays. “In this region, the STJ detector provides about ten timesbetter resolution,” Labov adds.

Astronomers also are looking to STJs as single-photondetectors of both x rays and visible wavelengths. In the visibleband, silicon-based, charge-coupled devices cannot measure aphoton’s energy, but STJs can. One photon, depending on itsenergy, can generate thousands of quasi-particles. By measuringthe photon’s energy, STJ detectors will allow astronomers tostudy galaxies and stars that are barely bright enough to be seenwith the largest telescopes.

Detecting Impurities as Semiconductors ShrinkAs semiconductor devices continue to shrink, the industry

needs to detect and identify small amounts of contamination onthe devices. Microanalysis systems with conventional energy-dispersive spectrometers “excite” contamination on chips withfairly high (10-kiloelectron-volt) energy, which results in thesurrounding material also being excited. When the surroundingmaterial is excited, a flood of unwanted signals or noise iscreated, making it impossible to detect the contamination. ButSTJ detectors can operate with excitation energies of lessthan 2 kiloelectron volts, which produce signals from thecontamination only, allowing the imperfections to be detected.

Helping to Enforce NonproliferationOne of the Laboratory’s important missions is to help guard

against the proliferation of nuclear weapons. Labov and histeam are conducting a research and development project thatinvolves using a superconducting tantalum detector to improvegamma-ray resolution. This technology provides better

0

5,000

10,000

15,000

20,000

25,000

0 100 200 300 400 500 600 700

Co

un

ts

Energy, eV

Carbon K

Nitrogen K

Titanium

Lα, Lβ

Oxygen K

Electronic pulser

Scattered incident beam

10-eV STJ- detector resolution 100-eV silicon- detector resolution

X-ray fluorescence measured with asuperconducting-tunnel-junction (STJ)detector shows ten times better energyresolution than can be achieved with aconventional silicon detector. A titaniumnitride sample was irradiated by 600-electron-volt (eV) x-ray photons at theStanford Synchrotron RadiationLaboratory. The STJ detector cleanlyseparates the titanium L lines from thenitrogen K line and the backgroundoxygen K line.

diagnostic capability, particularly when there are largeamounts of one isotope and small amounts of another. Forexample, when small quantities of nuclear materials arepresent, most of the gamma rays detected will be frombackground sources. Conventional detectors aren’t sensitiveenough to distinguish clearly between gamma radiation fromthe background source and from the nuclear material.

The team’s high-resolution, superconducting spectrometercan detect special nuclear materials by isolating emissionsfrom different radioisotopes. For example, if an inspectorsuspected that a heavily shielded barrel of spent plutoniumfrom a reactor plant also contained weapons-gradeplutonium, the superconducting spectrometer can measurethe composition of the materials in the barrel much moreaccurately than a conventional detector. The technology alsoholds promise in environmental monitoring for the analysisof trace contaminants because it can detect levels thatconventional detectors would miss.

Looking toward the FutureTraditional energy-dispersive and wavelength-dispersive

spectrometers are fully developed technologies, leaving little

room for significant performance improvements. Cryogenicdetectors are still in a developmental stage, with significantprogress having been made over the past few years. STJdetectors, although an “old” concept, are now better able toresolve low-energy x rays without sacrificing count-ratecapability, and the x-ray collection efficiency of thesedetectors can be increased by orders of magnitude withfocusing x-ray optics, which concentrate the x rays on thedetector. These developments could greatly increase the useof these detectors in a wide range of applications.

—Sam Hunter

Key Words: atomic spectroscopy, Cooper pairs, detectors, Josephsonjunction, mass spectrometry, quasi-particles.

18


For further information contact Simon Labov (925) 423-3818 ([email protected]).

19


Each month in this space we report on the patents issued to and/orthe awards received by Laboratory employees. Our goal is toshowcase the distinguished scientific and technical achievements ofour employees as well as to indicate the scale and scope of thework done at the Laboratory.

Patents and Awards

Patent issued to

Bernard T. MerrittGary R. DreifuerstRichard R. Post

George E. VogtlinBernard T. MerrittMark C. HsiaoP. Henrik WallmanBernardino M.Penetrante

Patent title, number, and date of issue

Halbach Array DCMotor/Generator

U.S. Patent 5,705,902January 6, 1998

Plasma-Assisted CatalyticReduction System

U.S. Patent 5,711,147January 27, 1998

Patents

David Leary, head of the Laboratory’s Business ServicesDepartment, received a Hammer Award for his efforts as partof a Department of Energy team to streamline propertymanagement regulations and performance evaluation, so thatthe latter is not regulation-based but outcome-based as it is inprivate industry. The award was given to a jointDOE/contractor value-based self-assessment team of whichLeary was a member. The Hammer awards were created byVice President Al Gore to recognize special achievements inthe efforts to reinvent government by cutting red tape andmaking government more efficient. They are given toindividuals and teams that have made significantcontributions in support of the President’s NationalPerformance Review principles.

Stan Trost has been named an Executive Fellow of the Institute

of Electrical and Electronics Engineers and will serve one yearwith the Federal Communications Commission (FCC) inWashington, D.C. Trost will work in the FCC’s Office of Plansand Policy on a broad range of deregulation issues resultingfrom the Telecommunications Act of 1996. Among thederegulation issues the FCC faces are the Year 2000 problemand critical communication infrastructure, reliability, openness,and service issues related to the Internet. Currently, Trost is aprogram manager in the Nonproliferation, Arms Control, andInternational Security Directorate. Previously, he was deputyassociate director for electronics engineering and has held avariety of development and leadership positions focusing oninformation technology.

Awards

Summary of disclosure

This direct-current (DC) motor/generator based on a Halbacharray of permanent magnets does not use ferrous materials; thus,the only losses are winding losses and losses due to bearings andwindage. The rotor is on the outside of the machine, and the statorforms the inside of the machine. The rotor contains an array ofpermanent magnets that provide a uniform field. The windings ofthe motor are placed in or on the stator. The stator windings areswitched or commuted to provide a DC motor/generator.

This inexpensive and reliable system for reducing NOx emissions,particularly in gasoline-powered vehicles and engines withoxygen-rich exhaust, combines nonthermal plasma gas treatmentwith selective catalytic reduction. A first reactor converts NO toNO2, and a second reactor reduces the output of the first toproduce an exhaust with reduced NOx emissions.

20


Computer Simulations in Support ofNational Security

The Accelerated Strategic Computing Initiative (ASCI) isan ambitious component of the DOE Stockpile StewardshipProgram to maintain the safety and reliability of the nation’snuclear weapons. ASCI’s goal is to provide the computingpower and simulation technology that will allow weaponsscientists to support assessment and certification of thenation’s nuclear deterrent from their computer desktops.To provide this capability, the ASCI program is developingcomputer platforms powerful enough for teraflops calculations,writing three-dimensional weapons codes that reflect ourunderstanding of the physics of the entire weapons system,and developing computer infrastructure that can run thecodes on the platforms. In the long term, ASCI’s achievements,developed to support our national security interests, willchange the fundamental way scientists and engineers—andeventually, all of us—solve problems. Because it is a scientificgrand challenge with a short schedule, ASCI work must bewell planned and well coordinated throughout the threeparticipant national laboratories—Livermore, Los Alamos, andSandia. It must also take advantage of the world-class expertiseof America’s universities and ASCI’s industrial partners.Contact:Randy B. Christensen (925) 423-3054 ([email protected]).

Abstract


© 1998. The Regents of the University of California. All rights reserved. This document has been authored by the The Regents of the University of California under Contract No. W-7405-Eng-48 with the U.S. Government. To request permission to use any material contained in this document, please submit your request in writing to the Technical Information Department,Publication Services Group, Lawrence Livermore National Laboratory, P.O. Box 808, Livermore, California 94551, or to our electronic mail address [email protected].

This document was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor the University of California norany of their employees makes any warranty, expressed or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus,product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name,trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or the University ofCalifornia. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or the University of California and shall not be usedfor advertising or product endorsement purposes.

U.S. Government Printing Office: 1998/683-051-60042

Science &

Technology Review

Lawrence Liverm

ore National Laboratory

P.O. B

ox 808, L-664Liverm

ore, California 94551

Printed on recycled paper.

Nonprofit O

rg.U

. S. P

ostage

PAID

Livermore, C

AP

ermit N

o. 154

View the Entire April 1998 Issue in PDF - Lawrence Livermore

Documents

Transcript of View the Entire April 1998 Issue in PDF - Lawrence Livermore