1) Grid on Go and Motorola work Y 2) EpiSims and Travis; 3) Bloch Baker/Travis nice people in jobs which there is NO chance of success No use of my expertise either locally or globally (I am "bored" and being driven into a corner) Not obvious that anybody wants CSIT CSIT is 50% SCRI -- 50% Business staff ineffective 50% researchers are "lepers" Best work on IBM machine is by Lepers .... SCRI hated by faculty and so revenge of FSU implies CSIT not wanted There are implementation issues Academic program Is CSIT core in CSIT or in CS/Math Foss has won battle but of course CS harmed Not clear if any solution 4) Ballard 5) Epoch -------------------------------------- Bernard SCEC $50K Rundle Maui meeting DRC Could develop ensemble models and use statistical analysis as in climate data assimilation in weather adjoint code (Note weather is real time; climate is more like current TRANSSIMS) TRANSIMS does have feedback!!!!!! They don't use data assimiliation as don't have data --------------------------------------- What about real time versus long term (policy/planning) Wants real time crisis management Systems Integration? Technology versus Applications ITEA FEMA Nuclear work seems out of place -- not really decision applications Simulation science Highly Controversial but probably right Needs to make case -- could change DMSO C4I Infrastructure rather pedestrian SOFIA simply wrong (CB Java Zealot) DoD work dubious technology -- typical of Lockheed Martin National Health Science is unclearly thought out All could benefit from CCS Simulation Science needs to filled out Relate SDS to other approaches Establish what can be done Software and Alogorithmic architecture ADD needed additional capabilities Why does nobody else do SDS? Problems much larger and so "microscopic methods" appropriate. Event Driven is Macrosopic Implications for applications done with other techniques Optimization,AI -- D1 does not cover everything! Information Technology System Integration Data C4I Object standards in XML can make many things much easier (Did not hear this mentioned once) IIT software infrastructure Architecture to enable Re-use Use Internet technologies Need to establish expertise Pat Saunders of T&E recommended ------------------------------------------------------------------- We saw Exciting and very important Approach to Complex Systems Simulation You Will need Complete Simulation Architecture SDS/IIT Revolutionary components Quantify applicability of SDS and IIT Information Technology (C4I Systsms of systems/Data Intensive/Computational Grids SDS/IIT/Other best of practice Integration) XML Internet retarget SOFIA? Consensus as to approaches on Infrastructure / Complex Systems simulations Create community 100's times bigger CFD computational science community than SDS ... Market areas you can simulate ------------------------------------------------------------------- Two major simulation approachs CFD / Bunch of particle: step in time or iteration --> MPP works well to arbitary size machines as naturally synchronized by time or iteration number (ASCI) Bunch of interacting objects with dynamic events with no time synchronization Cant run in parallel as nodes dont know when events occur Event Driven Simulation: (DMSO) Conservative -- look at all events and see what is allowed Optimistic -- go ahead and use rollback Nothing scales well. Use by others for transportation and epidemological simulations SDS realizes can break systems like traffic simulation into a) layers -- Gas / Electric // People/tanks/aircraft... b) objects which plan locally and can plan in parallel c) classic CFD like massively parallel microsimulation Maybe difference is size of systems large systems must have this type of structure. So what is SDS abstraction What problems canNOT be done with it What is simulation software architecture Scientific American article Introduction ------------------------------------------------------------------- Who are their peers Outstanding is top 10% and pushing state of art 3(4) areas Quality of science and engineering Relevance to National Needs and agency missions Programmatic Performance and Planning Performance in technical development and operation of major research facilities Relevance and Appropriateness in light of Laboratory's mission Data re peers; customers UC Management contract renewed for 3 years UC will never compete! Fee at risk ($5M a year of this goes to LANL) UC Evaluation DoE Evaluation UC doesn't address DoE programs New program review committee uses DRC Need customer feedback Shift from "Big Science" to "Customer Satisfaction" So ask "Why Science" Threat Reduction a lot of Sponsors D division agile; not unique; recruiting problems IT and globalization --> new threats China increasing strength -- political uncertaintities Russia -- weak but has lots of weopen expertise and old weopens New nuclear licenses! Biothreat (B and D division and Chemistry division) Defense Science and Technology Missile defense needs new technology military Space Critical Infrastructure Protection D division lead Recruiting and Retention 48 years old mean Outstanding/Excellent As good as peers Outstanding better than expected Did they answer questions above what customer wanted Political Changes NOW Darrell: D division is outstanding Best job Most exciting ever More on Cyber Security C4I for JBREWS Aging force structure upgrade electronics Motorola Understand components Don't understand systems No failures!! Plutonium facility construction cancelled LANL too expensive Crisis Management education NDU Curriculum Elective class -- very successful (June 30 USA Today) Cancelled this year BUT issue is technical not programmatic failures Human Factors Thrust GONE people still here Intellectual Property Lab understaffed in legal area Better retention than rest of lab LDRD 3% which is average DoD funding increasing D-1 Systems -- Statistics Certifying processes IIT Combines Experts Simulations Test data 3 new DoD projects Motorola P and G This is ITEA? Motorola IP Reference Transsims gives people position on second by second basis Cell phone packets BASIS Decision support Spanish Flu in 1918 killed 20 to 40 million people Electrical Grid -- have national database Could simulate national grid Taxonomy of decision making No one approach Analysis of Los Alamos fire decision process (controlled Burn) Dallas traffic simulation done down to person Electricity National transportation Epidemiology Bio-attack Los Alamos Bio defense for the Nation Barrett: Started in C4I National Sim Center domestic version of National Military Command Center Policy to Real Time Darrell: Speedes is doomed JWARS must fail before anything will happen Bernard What about data assimilation P and G Reliable Engineering for Diapers model before building NTNAC one hour time slots Transsims one second Can link them Agent -- as trucks do their own planning e.g when they arrive they interchange info with terminal to Nuclear renaissance in country How else can you get enough power! Fusion and Solar are always power of the future! WFO Work for Others Press goal: 25% research Sequential Dynamical Systems local event based globally time stepped ---------------------------------------- Need document successes -- decisions made rather than how wonderful technology is Divison poised for "greater things" How do we present this Domestic Surety Nice "National Structure" diagram Richard Clark in NSC in charge Only group abled to simulate coupled infrastructue $20 million a year Olympic simulations for communication, power etc. DoE funded Cheryl Burger Civilian National Security What are questions NISAC will give more direction to D division NISRD -- Chem-Bio Non profileration BASIS Nuclear Explosion Detection Profileration Detection $220M a year total budget Asked for major increase; currently 4.6% decrease Chem-Bio anthrax hardest as a spore improvement in first response in this area in metropolitan areas LLNL ahead in Biology LANL ahead in modelling Transsims not fielded yet Portland (at LANL Thursday Friday) Dallas not continued with unbias activity survey by weighting with population survey IAAP products Joint Program office DoD classified Used by commanders in operational planning Electric, Natural Gas, Infrastructure Electra -- Use Object oriented Database (Access IM6) Static focus as not enough data Use Arcview GIS Utilities use switched fault tolerant networks SOFIA Coarse Grain (Multi-Grid) Can run in SDS mode claims to be actor based discrete event / time stepped mode Presenter has never heard of SPEEDES Model interactions between infrastructure Alternatives Modelica modeling language Complex Interactive Networks/Systems Initiative Used in Florida for electricity/natural gas Agrees unproven scalable has "proved" physics can be done by message passing ElectriSims 10,000 generators 100,000 transmission lines 70,000 substations Synthetic population in TRANSSIMS relevant Could use in New Mexico Parallel power simulator based on texas Code SDS view of demand classic HPC sparse matrix view of power simulations JBREWS DTRA project manaaged by joint program office User interface with standard GIS based tools Unclear technology competence "Java open architecture" D-3 (not working with CCS) C4I system for BASIS Not deployed as sensors failed in ACTD from DTRA SNCP LANL Communication LLNL GOOD Reviews BASIS Reuse SNCP as command center Olympics as example of "special event" Very practical deployment plan EpiSims Decide where to test passengers on aircraft Recent Ebola/TB cases Use TRANSSIMS strategy of human simulation Note olympics will need special data cities will be updated every year not clear if can do data assimilation Ireland AAA Advanced Accelerator Applications $68M 80% nuclear expertise of laboratory in D division 2050 80% electric power from nuclear Can burn plutonium in accelerators --> fixed Pu store South Africa First reactor order since 1970's NRC has no expertise Paul White Russian Nonprofileration Program Manager Edward Arthur AAA Transmutation as a nuclear waste management tool Rao Cost of Nuclear Power: 3 cents a KWH doubled by regulatory costs LANL and SANDIA have expertise and no conflict of interest Simulate insulation fiber flow in water Chodak Non proliferation fuels Russia China .. nuclear power and export Burn Plutonium in advanced reactors Burn rest in accelerators Define proliferation risk BMD camps think proliferation has failed and uses this to justify BMD They want proliferation to continue! Martin safe secure ecologically-sound storage do not design or make anything 100 year building life; 50 year container Plutonium; HeU support decision makers MILITARY simulation work gone down due to JSIMS JWARS Darrell hates HLA Stockpile Stewardship Very programmatic -- LLNL > D division > Sandia W76 is a Phase 6 weopen 15 year refurbishment time 60 year life thereafter Every year certification of weopen make weopens closer to underground test GTS Gas Transfer System Projects tend to be limited in size <= $2M Programs $20 to $50M Missions $150M or more $800M Mission in nuclear 30% of Transsims funding in CCS D Division functions Customer Interface; Proposal making; Program development Science and Technical Quality recruiting Running business Los Alamos inventing new thrust areas New division vision Real time simulations track people (data assimilation) '00 publication numbers are 18 months 179 professionals leads to 125 to 150 papers per year IIT Directed graph Nodes are people or simulations with probability inputs and outputs Noser Claims relevant in sensor fusion National Health Security PSM (Personal Monitor) difficult issues military applications easier In a time of crisis, everybody wants to be cured Not a very good presentation One good science is sequencing flu virus 4G Networks -- super bandwidth wireless Stochastic SDS Interesting mathematical systems Need to add telecommunication and distributed computing to portfolio optical networks infinite bandwidth R/F limiting routing tables too big for conventional methods can use Transsims to generate cell phone use FCC has base station data million transmitters on 128 node Linux cluster in real time I don't understand what is simulation what is new network technology Space Fission Power Melt ice on Europa Defend against asteroids Faster better Cheaper NASA plan failed (Mars crashes) Cassini launched RTG -- Space Fission can be safer than this It costs $400M for a nuclear test so DONT do 5 year life Needed by astronauts to get power on Mars and fast voyage Nuclear to Electric 20 % efficiency 50% extra inefficiency Use on Pluto mission around 75 Kilowatt electric power SP10 failed 10 year old effort cost $400M ----------------------------------------------------------------- Wrap up presentations not polished No good traffic planning education