аЯрЁБс;ўџ ўџџџ}џџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџ§џџџzўџџџ{  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxy|ўџџџўџџџ~§џџџ€Rџџџџџџџџ РF€ѓ№_єtМРCompObjџџџџџџџџџџџџ\WordDocumentџџџџџџџџQObjectPoolџџџџ]X_єtМ]X_єtМўџџџ ўџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџўџџџџџ РFMicrosoft Word 6.0 DocumentўџџџNB6WWord.Document.6;џў ўџ ђŸ…рOљhЋ‘+'Гй0х˜р0Lh n z †’š ЂЎЫ гнџџџџџџџџџџџџџџџџџџџџџџџџ@Macintosh HD:Microsoft Office:Microsoft Word 6:Templates:NormalHInfrastructure for distributed Heterogeneous Collaborative Computation.Theory Center StaffмЅhOР eъQчlllllllцццццц №vц)Offffffff&xXа)lfvwffff)fllfffffflflf€†рllllff›f Infrastructure for distributed Heterogeneous Collaborative Computation. Dave Lifka Cornell Theory Center Cornell University Ithaca, NY 14853 (607) 254-8621 lifka@tc.cornell.edu Geoffrey Fox NorthEast Parallel Architecture Center, Physics Department, Computer Science Department Syracuse University Phone: (315) 443-2163 Fax: (315) 443-4741 gcf@nova.npac.syr.edu http://www.npac.syr.edu, S.Lennart Johnsson Department of Computer Science University of Houston, Houston, TX 77204-3475 Phone: 713-743-3371 e-mail: Johnsson@cs.uh.edu http://www.cs.uh.edu/~johnsson B. Montgomery Pettitt Department of Computer Science Department of Theoretical and Physical Chemistry University of Houston, Houston, TX 77204-3475 Phone: 713-743-3263 email:Pettitt@uh.edu Abstract There is general expectation that computational science will make increasing and critical use of distributed heterogeneous environments. However this implies the development and integration of new technologies from several different disciplines and their testing on complex applications. In this proposal we bring together experts from distributed computing, virtual environments, high speed networking and web based collaboration. These are integrated in terms of carefully chosen applications with which we will be able to motivate and assess the technology base. We believe that any successful system must be able to harness multiple distributed computers, data analysis and visualization subsystems as well as a distributed team of users. This requires a multidisciplinary approach in any successful approach to this problem. This scenario can be expected both for academic applications in terms of both nationally and internationally distributed research teams, and in distance and computer-aided learning environments, and for Industrial problems as envisaged in multidisciplinary analysis and design for next generation manufacturing. We build on the Advanced Resource Management System (ARMS, from the Cornell Theory Center), an advanced resource management tool that has already been designed to handle compute, datastorage and visualization systems. The latter will span virtual environments from advanced CAVETM capabilities at the Cornell Theory Center and the University of Houston, as well as conventional workstation-based analysis. A centerpiece of this proposal will be the integration of advanced networking with ARMS which will be lead valuable insight to Internet 2 and NGI as this develops. TANGOsim (Syracuse University) is a recently developed Java Server-based collaborative environment that includes an event-driven simulator with the conventional services offered in comparable systems offered by the National Center for Supercomputing Applications(Habanero) and IBM. TANGOsim links applications written in any language. It can be integrated naturally with the compute services offered by ARMS, which is also based on a similar Web technology base. These base technologies will be tied together with industrial (medical, seismic, and manufacturing) and academic applications involving all members of the proposal team. We have included a strong networking research group from Cornell, which will allow us to assess the results of effort from application, computing, and networking perspectives. We believe that we have put together a team that combines leading-edge expertise and technologies in the diverse components needed in future distributed metacomputing environments. Further, in each case, base technologies (ARMS, TANGOsim, virtual environments) are well developed and this proposal will fund their integration and extension of capabilites. The partners in this proposal are the Cornell Theory Center (CTC), the University of Houston (UH) and Syracuse University (NPAC/Syracuse). In the following we describe the component technologies ARMS, TANGOsim, Virtual Environments and Networking. These are integrated by the applications described at the end with a significant assessment effort. Keywords: Distributed Resource Management, collaborative environment 1. Vision and Motivation Here we have two scenarions, a) classroom individual and collaborative analysis of data or interactive simulation: need to schedule compute power, classroom, video server, network resources, distributed (wide area) file systems/data bases writers: gcf, anne. Here we present two motivating scenarios that both drive the development of the proposed collaborative infrastructure and provide a forum for evaluation and assessment of it. The latter aspect will be discussed later in the proposal. The Virtual Classroom and Computer-Aided Learning There are a multitude of ways in which the growing power of high-speed networks and Web Software may play a role in the revolutionizing of education. Two of these are to provide the ability of students to interact with computational simulations and analysis in the classroom, and to provide a framework for distance education. The ability to schedule resources that are required in both cases, data resources, visualisation resources, together with the collaboration components of education and training will be provided by the resulting infrastructure gained by integrating ARMS and TANGOsim. This impact will be important at all levels of education K-12, undergraduate, graduate, continuing education and institutional training. Currently there is very little experience as to what features of both hardware and software are important and it is probable that there is no one answer but different applications will have different requirements and varying success. Syracuse has already integrated its HPCC and Web course system WebWisdom with Tango to deliver training in classic synchronous style. This application is particularly demanding of network responsiveness for users. They have also successfully worked with the Syracuse Physics department (http://www.phy.syr.edu/courses/modsim.html) on first integrating basic Web resources, then Java applets andnow Tango collaboration technology into their introductory course PHY105/106 "Science for the 21st.Century". This course for non-science majors sustains great student interest (over 250 enrolled each semester)as it is organized into interesting modules (such as Mind and Machines or the Search for ExtraterrestrialIntelligence) and the Web allows material to be very topical. Further using the Web in the class teaches general skills. In teaching computer science, one needs to incorporate programming laboratories both into the actual delivery of the course and homework. This was emphasized to us when teaching Java, JavaScript and VRML where this is essentially automatic due to the natural Web integration of the languages. We found the ability to go through exemplar codes and illustrate their execution in real-time, an invaluable aid to teaching. Again students obviously prefer to learn such languages where success is not measured by dull text output -- printf("HelloWorld") -- but rather by the production of an interactive Web pages that can be shared with their peers. This motivated us to develop virtual programming laboratories (VPL) for both HPF and MPI (http://www.npac.syr.edu/projects/VPL/) and PERL (http://www.npac.syr.edu/users/gcf/perlvpl/). These are successful but still need more experimentation and our latest VPL is now available with integrated Java visualization of data and performance visualization. Syracuse has also linked the current WebWisdom to the collaborative system Tango and shown how a complex JavaScript teacher interface can be linked to a simpler student interface which has just the selected material and not the full capability of easily selecting material to teach. This ability to support multiple views of the same basic application is an important feature of collaborative systems. This set-up (teacher and student both with their own independent computer) is particularly helpful in teaching programming as it avoids visibility problems that occur if you try to go through programs which cannot easily fit on an overhead with font sizes that display in a classroom with a single display. Further we note that our educational applications have needed a link of Tango to C++ (3D GIS), Java (Physics Simulations) and HTML and JavaScript (web delivery). Such a multilingual interface appears to be a critical feature of collaboration systems. Since the summer of 1995, the Cornell Theory Center has been providing distance education in high performance computing to a broad audience of scientists, researchers, students and educators through the Virtual Workshop Project (http://www.tc.cornell.edu /Edu/VW). A fundamental goal has been to provide an interactive learning experience. Within the VW, CTC has integrated the Virtual Programming Laboratory (VPL) which was originally developed at NPAC as above. This allows the participants to try the techniques and methods they have been reading about without leaving their Web browser. They will receive results and feedback from the programs they execute in their browser window. The computer-based activity is entirely integrated with the course materials. Any activity is possible. It is possible to work on a computer program, run a simulation model, start an application package -- even one with a graphical interface, or access a database. A key technological breakthrough for enabling the integration of the VPL has been development of a secure web server daemon, Sweb. Sweb allows participants to log on to a remote computer from a Web browser. With Sweb, a user can access the remote computer as an authenticated user. Each user can create files in their own directories. This enables individual exploration of a problem, and also allows the learner to interrupt a session and return at a later time. Sweb has been integrated into the FIG VPL/SWEB Until this point in time the CTC Virtual Workshops have been asynchronous but with the integration and development of TANGOsim in the Virtual Workshop environment, we will create a truly Virtual Classroom environment. ARMS will enable advanced reservation of networking resources so that during a Virtual Workshop or class session users will be able to interact as they would in a physical meeting or class room. We will use the digital server technology developed by NPAC which is already being used in their ATM K-12 experiment -- the Living SchoolBook (ref). The key to this education application is the collaborative ability combined with the resource allocation. The resources range depending on the class being taught, but may include network bandwidth, Virtual Reality equipment, compute resources, data stores and video servers. (b) research collaboration: need, VR, network, distributed data resource s, compute etc Writers: Monte [THIS IS FROM THE PREPROPOSAL - NEEDS TO BE REWRITTEN] Structural Biology and Drug Design Structure-based drug development is an area being greatly impacted by recent developments in computational and datamining techniques together with visual and synthetic environments. Biomolecules are complex three-dimensional geometric structures built from thousands of atoms. By simulating the forces between atoms, researchers are beginning to be able to predict how well small candidate drug molecules can bind to these giant structures. Successful prediction of a new lead for drug development not only involves reasonably accurate modeling of molecular forces but also an understanding of the shape of the active site pocket and a good deal of chemical intuition not easily codified within a computer algorithm. Live interaction with a simulation while it is happening brings human pattern recognition skills, expert knowledge and adaptability into the picture. This application requires computational resources for real-time computational, access to data sources and CAVE resources and the network resource connecting them. The enhanced interactive and collaborative environments will provide a development environment for this and other pratical problems involving genetic and protein engineering. 2. Applications requirements Here we address some of the requirements that are something by the above application areas. We identify explicitly the resources that are required, the quantity required and the likely distribution. We address also the question of the quality of the resources and the ability to guarantee resources and negotiate intelligently. quality and quantity Resource needs: type Compute Resources In both the classroom environment and clearly in the computational researchers situation there is often the necessity for large compute resources. These might be for the simulation or for the analysing of the data. Distributed Data resources and File Systems The chemistry computations discussed above often result in multiple sets of data stored in different locations using different technology. This occurs due to the research teams being made up of a distributed set of individuals and also be the nature of distributed computing. The data tends to reside at a location local to the lastest point of computation. The computations can take several weeks of wallclock time to complete. Virtual Environments Recently, higher bandwidth communications channels have made collaborative science and engineering using shared virtual environments practical. Research efforts have focused on two main areas: software infrastructure and application development. Researchers concentrating on software infrastructure have developed software architectures, network protocols, and data-sharing algorithms to enable distributed, shared, virtual environments. Application developers have built specific applications to serve as proofs-of-concept for the effectiveness of distributed, shared virtual environments. To achieve real-time collaboration over long distances, two approaches have emerged: (1) rendering graphics locally on similar platforms and exchanging state data at high rates over a communications link and; (2) rendering graphics at a single facility and delivering those graphics, over communications channels, to remote users. We propose to work with both approaches. UH will develop high performance remote rendering software, based on work initially done at SGI and ongoing at the UH with DARPA support. This will support the remote "writing" of frame buffers so that a powerful graphics supercomputer at one site can effectively drive the displays of inexpensive workstations on the scientist's or engineer's desktop. In this effort, much of the focus will be on enhancing interaction and collaboration including use of acoustic and haptic interfaces. Another major use of visualization will be its application as a user interface to large-scale mathematical models and simulations. We will seek to make possible the more effective use of such models and simulations executing on supercomputers, locally or remotely. This will be accomplished by allowing scientists and engineers to construct and work directly with near-real-time meaningful representations of the physical systems that they are attempting to understand. Linking this technology with ARMS and TANGOsim (which supports such custom viewers) allows us to deliver this environment to multiple distributed users by creating an effective interface and giving the deterministic control of the scheduling of the multiple resources required. Network Resources Contemporary network technologies provide the ability to guarantee QOS at the time a virtual connection is initiated. However, little work has been done recently to assure that the necessary network resources will be available at a scheduled time in the future. It would be very undesirable to find out that the necessary network resources are unavailable when other valuable computational and human resources have been scheduled in advance. We intend to work closely with the NGI and Internet technology developers to determine how best to integrate ARMS with developing network technologies and to provide requirements based on our test application assessments. Classrooms Monitoring Equipment There are limited physical resources that are equipped as UofH with X-ray crystallography and NMR spectroscopy, the infrastructure we are planning to build will allow the remote use of these facilities and the reservation of them through ARMS. Video Servers where/how much Sweb is also being used in the development of an advanced set of distributed resource monitoring and scheduling tools that will allow users to query the status and availability of resources and submit jobs from the same interface no matter where the resources are located geographically. This is an important breakthrough in masking the complexity of a large scale distributed system from users who do not need detailed information on resource location. guarantee negotiability Within the scope of this proposal, we will develop an intelligent negotiator. This software will be an integrated piece of ARMS that will allow the user the flexibility of the system choosing a n equivalent resource within limits defined by the user. (this defines the job) Types of resources identified: compute, VR equip, Video servers, network resources, distributed data stores, file systems, classroom, 3. Approach Preposed architecture Our proposed collaborative computing environment is built on top of emerging pervasive Web technology which will provide a well supported distributed software base which naturally includes linkage to such critical services as databases, Corba Object Brokers and multimedia (video) servers. We have a three-layer model (which sometimes can be considered as two with the first and third elements combined as generalized services to be linked together. The client systems with users and their visualization and analysis systems. These can of course vary from high-end Caves to a student studying course material on an HTML page in a Web browser. Here we expect VRML2 and especially Java3D to allow remarkably powerful unification of a variety of visualization systems. The integrating middleware, which is considered as a network of Web Servers-- currently dominantly HTTP servers such as Apache but there is a growing interest in specialized or general Java based servers. In the latter case the Jeeves system from JavaSoft appears to be establishing the basic approach. A set of services that include conventional relational databases linked by JDBC (the so-called Java Database Connectivity) and object brokers where we expect a growing adoption of CORBA by the Web community. There are now many Java based Object Brokers and for instance JavaIDL from JavaSoft has support for ORBlets which Netscape expects to bundle in their new browsers. Other capabilities needed by particular applications, will augment these general services. Here we put MPP compute and disk systems and other HPCC facilities that in our model are essentially hosted by a Web Server providing the middleware linking them to the community network. For our educational applications,video servers are a critical additional service. We use Web technology because it provides the base integration technology for distributed computing. We note that use of Web Software does not require us to implement our applications on the Internet, which obviously has insufficient performance in many cases. Rather this software will be used in production on appropriate Intranets such as the vBNS. However it is important that our architecture gives systems that can, in part or whole, run on the Internet. This is how we guarantee educational opportunities for everybody and more mundanely that computational scientists can continue their work started on high-end visualization systems on their home PC's. In this proposal, we intend to augment the broad capabilities inherited from the Web by special services designed to enable and improve performance of special educational and computational applications. We can illustrate this concept with an example. One can implement parallel computing with all message passing implemented by either HTTP or Java sockets at the Web client and server level. This is method used by elegant Javelin system from UCSB, our collaboration that factored RSA130 using the Web and is the default in our new AVS like WebFlow approach. (All these concepts are described in the proceedings of the Syracuse workshop on the use of Java in Science and Engineering, which is published in Concurrency: Practice and Experience June 1997). However most HPCC applications require higher performance and here one would replace the Web implementation with MPI or Globus systems. In this way, we view classic HPCC technology as high performance subsystems, which selectively replace Web technology when necessary. Key technologies in our proposal are ARMS and TANGOsim, which supplement the commercial Web distributed infrastructure in critical areas. ARMS supplies the necessary resource management and scheduling for the different services supported by the Object Broker and Web Server middleware. TANGOsim essentially provides the software integration between clients and servers so that one can view clients (people) and computers (instruments, Web services) on a common basis. A brief description of each system follows. The issues related to integrating the systems and the tools that will be developed to complete the system are then discussed. TANGOsim TANGOsim as an open, extensible system that provides a technological framework for building subsequent generations of collaborative systems. TANGOsim is aimed at creation of shareable information spaces: The basic paradigm of many traditional collaboratory systems is to connect identical instances of few applications and allow them to exchange data objects or streams. We consider this paradigm way too restrictive. There is no basic reason to restrict collaborating people and teams to look at the same information in the same way or using the same tools. Collaboratory system should support coordinated but independent views of related information. Different presentation of information may be needed for different reasons: a 2D GIS display on the command and control main screen may be sufficient while a mission planner may want to see a much higher resolution 3D representation of the same information. In other cases, different information views my be displayed on workstation with different performance or graphics capabilities. In TANGOsim, we allow different applications to exchange messages and act accordingly. The applications exchanging information can reside on either the same node or on different machines. This is an important feature, since, independently of supporting much richer collaboration model, it allows us to build complex applications from small, reusable modules. Consider, as an example, a tandem of a chat and a web browser. This tandem is replicated on multiple nodes. As the users chat, they may pass information about interesting URLs in the chat window. In TANGOsim system, this information will be automatically recognized and passed to the set of browsers working in the collaborative mode. Another example is a simulation system created from a customized TANGOsim environment: a simulation engine sends messages to both local and remote applications of various kinds, creating a war game or a crisis management center training module. In both examples the users of the system have full access to the enormous information resources of the entire Web. This information may be presented to them either via a standard, familiar web browser interface, or via a set of specialized filters that tailor information display to a specific task. TANGOsim is very tightly integrated with the Web: This statement applies both to TANGOsim as a functional model and to its implementation. In the functional sense, tight integration with Web implies access to the entire informational potential of the Web. This is critical for the construction of the shareable information space discussed above. Further, designing TANGOsim functionality we have opted for the self-distributing software model for collaboratory applications. While this decision created a number of difficult software implementation issues, it also makes the system very easy to install, use, and extend. TANGOsim provides support for all synchronous collaboratory functions: Built in a stateless Web environment, TANGOsim is a statefull system. Basic design supports all functions of a synchronous collaboratory system, including session management, data/event distribution, flexible floor control, and multiple, configurable security levels. The system does not impose any restrictions on a number of concurrent sessions, users, or collaborative applications. Asynchronous collaboration is supported via database back-end: Session record and playback capability has been designed into the system as its fundamental component. This capability applies to both events and data streams and can be used review collaborative sessions in asynchronous fashion. Playback/review capability is provided also for real-time continuous data streams such as audio and video. Distributed visualization and manipulation of multimedia information streams TANGOsim provides scaleable multimedia support. Recognizing different performance requirements of continuous multimedia streams, TANGOsim provides mechanisms for decentralized distribution of the such streams under TANGOsim session control. Language independence: While almost entire TANGOsim runtime and most of the applications are written in Java, there is no restriction on the language used to implement collaborative applications compatible with TANGOsim. This design decision appears to violate the aforementioned requirement of application downloadability. Actually, methodologies for automatic distribution of applications written in languages other than Java exist and we intend to incorporate them into our implementation. We have decided however that, at least at present, restricting ourselves to Java would prevent us from extending our systems to such important domains as 3D visualization, high quality video,streaming, and videoconferencing (especially multimedia encoding).In addition, TANGOsim provides a truly unique support for applications written in JavaScript. Our ability to do so stems from the earlier decision of very tight integration of TANGOsim with Web. Extensibility: As stated above, we firmly believe that ultimate success or failure,of a collaboratory system depends on its application set. Video, audio, and whiteboard,conferencing is just a seed of a collaboratory. Domain specific applications must be either implemented from scratch or ported from their stand-alone versions for a collaboratory to become truly useful. For this reason, TANGOsim APIs for Java, C/C++ and Javascript have been written and well documented. TANGOsim is currently implemented using Netscape's LiveConnect but as JavaSoft's collaboration architecture (JSDA) matures we can expect the core implementation to change. In this proposal, we will not develop this core capability but use it our chosen applications and integrate it with the resource management of ARMS. (Architecture Proposed in this Section: figure) (Example of WebFlow used to link HPCC Simulation with visualization Filter: figure) Recent Web technology developments open the way for new collaborative computing environments that integrate traditional video conferencing, distributed simulation as developed by the DMSO office (SIMNET), and emerging ideas of computational steering. In this proposal we will link the ARMS system with TANGOsim developed at NPAC. TANGOsim uses Java Servers to manage Web collaboration in the traditional fashion already seen in Habanero (NCSA) and Shaking Hands (IBM but originating in NPAC). The Java Server supplies session management with multicasting allowing shared applications between multiple users. TANGOsim was built to support command and control with scripted environments allowing real and virtual members in the command teams. This is implemented as an event driven simulator replacing the simple session manager used in the other Web collaboration systems. Since ARMS and TANGOsim are built on a modern Web backbone we can integrate them so that ARMS provides the asynchronous scheduling and TANGOsim the synchronous collaborative computational steering, linking in each case distributed computers, databases, visualization and multiple collaborating computational scientists. TANGOsim has been developed with other funding for command and control and distance education. Here we propose funding for its integration with ARMS and the special features needed for distributed computational steering. TANGOsim is implemented as a Java Server linking a net of Java Applets with a special downloadable plug-in providing the critical connectivity. It already supports client applications written in Java, JavaScript, and C++ (and hence other traditional languages). We have demonstrated a prototype visualization client for a 3D GIS (Geographical Information System) with a custom VRML browser written in C++. Other major capabilities of TANGOsim include a video teleconferencing module and integration with a state-of-the-art digital video server that supports efficient multicast services. Further over the next few months, we will add a database backend (using the new JDBC Java Database Connectivity) which will allow one to log the complete multimedia collaborative session. TANGOsim provides traditional basic services such as chatboard, shared whiteboard and multimedia mail. Further we have ported several interesting educational applications, the WebWisdom dissemination systems and some interesting Physics education applets , to collaborative mode. This proposal involves the integration of ARMS and TANGOsim with a lower level TANGOsim integrating the components of a synchronous computational steering environment. The details of this integration will depend on the evolution of Web technology but we expect to continue the basic architecture built around Netscape's LiveConnect on the client and Java Servers. Note that TANGOsim can be viewed as "just integration technology" since collaboration systems only provide integration between the involved clients. existing tools Advanced Resource Management System (ARMS) The Advanced Resource Management System, ARMS is a collaborative effort to develop a secure, scalable, distributed resource management system. The ARMS project is led by the Cornell Theory Center working with Lawrence Livermore National Laboratory, Pennsylvania State University, Syracuse University, and University of Houston. The key features of ARMS are heterogeneous resource management and monitoring, cross-realm authentication using DCE, and an intelligent heterogeneous distributed job scheduler that can interface to any localized scheduling system. ARMS is the underlying infrastructure for developing seamless, cohesive, distributed compuing environments and layered collaborativetools. ARMS provides secure, reliable job scheduling on computational,synthetic environment, software, data, and network resources. ARMS uses Ensemble, developed at Cornell University, for inter-site communication which guarantees the secure, ordered arrival of scheduling negotiation messages. ARMS relies on cross-realm authentication and file access through the use of DCE/DFS. This ensures that users will not have to remember account information for the wide range of resources that ARMS will have access to, and that systems administrators will not have to coordinate user and group IDs across all ARMS systems. Users will be able to authenticate once to their local system and their credentials will be transfered as necessary by ARMS to resources they are allocated time on. Figure 1 shows the basic ARMS architecture. Underlying R&D: EASY Deterministic Scheduling Algorithm Fair Scheduling for Large Parallel Jobs EASY-LL Benefits of EASY Scheduling Reliable Resource Management Scalable Job Starting and Stopping EASY-LSF EASY-NT ARMS will perform distributed backfill. ARMS Participation Requirements: Must Provide resource management API routines. ARMS API consortia to ``standardize'' API. Must support advanced resource reservation. ARMS Network Resource Scheduling: ARMS network traffic characteristics Node information: 30 kbytes/512 nodes Queue information: 40 kbytes/512 jobs Information for 15 sites: 1 mbyte Information for 1000 sites: 70 mbytes Low latency is necessary for optimal ARMS operation. Bandwidth requirements are influenced by application requirements Advantages of ARMS: Sites maintain their own scheduling systems and usage policies. Efficient use of resources by distributed backfilling. Use of ARMS will provide real data on distributed system limitations and usage trends. ARMS will allow users to access geographically distributed resources through the scheduling interface they are currently using. The Cornell Theory Center (CTC) has been developing an advanced resource management and scheduling system for distributed, heterogeneous computational, visualization, network, and data acquisition resources. This system, called ARMS, deterministically schedules applications on these resources so that network connectivity and quality of service are ensured. ARMS is a technology enabler, in that it provides access for applications to these distributed resources in a seamless, cohesive fashion. This will allow researchers from all areas to focus on their areas of expertise without being concerned with the logistics of running their applications or experiments on geographically distributed heterogeneous resources. ARMS includes a two-level intelligent scheduling system that uses an Informix distributed database. The database maintains status and availability information, which is then used to schedule jobs with specific computing, synthetic environment, software, data, and networking resource requirements. The foundation for ARMS is based on key components from the EASY-IBM LoadLeveler application programming interface (API) project . Through a minimal set of API calls, all the necessary information about local and remote computational resources and user job requirements can be queried reliably. This API is being extended to work with other major scheduling systems such as Platform Computing's LSF. ARMS provides secure, survivable resource management and job scheduling. Local resource monitoring agents will be developed that can relay status and availability information to the ARMS database using the Ensemble system developed at Cornell University. Ensemble, is a reliable networking protocol that guarantees the arrival of information and also its ordered arrival. ARMS relies on DCE/DFS for cross-realm authentication where we are working closely with Lawrence Livermore National Laboratory. The growing complexity of distributed environments suggests that distributed system monitoring tools and navigational tools will be extremely important for both systems administrators and users. We propose the development of Web-based Java tools for ARMS which will allow location of resources and monitoring of their status and availability. ARMS is being tested in several testbeds. Development work is being done primarily at CTC on its 512 node IBM SP system and SGI Power Onyx systems. As code matures it is deployed to a campus-based distributed system with ATM network connectivity. In addition, mature ARMS code has been tested on a national testbed including CTC, Pennsylvania State University, the University of California at Los Angeles, and the University of Maryland. The geographical distance between these sites has proven to be an excellent test of ARMS' capabilities. The reliable and deterministic service that ARMS will be able to guarantee will make it an essential part of the NGI. It will provide a mechanism to predict and control network traffic - a feature that is missing in the current Internet. Our working closely with key distributed application developers will ensure the ultimate usability of the ARMS system To test the usability of the ARMS system for the development of distributed programming environments, we have been working on several prototype applications and collaborative tools layered on top of ARMS. CTC and Syracuse have collaborated on the development of a secure Web-based scheduler interface that provides ARMS users with a common interface for submitting jobs to the ARMS scheduler and for browsing and editing DFS filespace, no matter where the jobs reside. "Smart Make" facility, another interface under development, will allow users to submit source code to the ARMS scheduler, which in turn will locate the necessary software and computing resources for the code, build the correct executable, and store metadata on the executable pertaining to its resource requirements. work to be done First draft: gcf second draft: dave 4. Evaluation and Verification: Integration and Assessment We propose to deploy a distributed collaborative computing environment that focuses on high level user services. Little experience exists on either the useful features of such a system or on the needed infrastructure capabilities in areas of network, parallel multimedia database for session logs, and computational services including visualization. Thus our proposal will be structured as a set of application experiments described below where we will carefully assess both the value of particular system features and needed performance and functionality of underlying infrastructure. This assessment will use traditional computer and network performance tools augmented by enhancements in TANGOsim to log message traffic through the session manger. Here we will follow the strategy exploited in our Virtual Programming Laboratory of using a web implementation of Illinois's Pablo system with ,performance data logged in TANGOsim's backend database using the SDDF data format. (a) applications (writers as above) Need something here with respect to the chemistry applications. Geoffrey do you want to put in words on the HBCU project. (b) international testbed since this wasn't in the final preproposal should it be in the proposal 5. Statement of Work Integration of ARMS and Tango Resource needs: NPAC ?? CTC ?? Extensions of ARMS: Resource needs Resource reservation/QoS: NPAC ???? CTC ???? UH 90k Resource management: NPAC ???? CTC ???? UH 0? .......... Extensions to Tango: NPAC ??? CTC ??? UH ??? Evaluation of the Distr Env. Application/Evaluation Software NPAC CTC UH 90k Experiments and Analysis NPAC CTC UH 40k .......... The timeframe for this proposals is expected to be 3 years. The figures shown for each institution are the yearly requested amounts. Cornell Theory Center: Staff including overhead $200,000 Distributed management Servers $ 30,000 Software licenses $ 5000 Travel $ 5000 Total $240,000 Syracuse University: Staff including overhead $140,000 Distributed management server $ 20,000 Travel $ 5000 Software licenses $ 5000 Total $170,000 University of Houston: Staff including overhead $140,000 Distributed management server $ 20,000 Travel $ 5000 Software licenses $ 5000 Total $170,000 Total yearly request $580,000 Biographical Sketches David A. Lifka Senior Systems Programmer Cornell Theory Center Cornell University Ithaca, NY 14853 (607) 254-8621 lifka@tc.cornell.edu Education Ph.D. in Computer Science, Illinois Institute of Technology, expected completion Fall 1996. Dissertation topic: "Job Scheduling for MPP Architectures supervised by Dr. Ewing Lusk, Dr. James Kenevan, and Dr. Martha Evens. M.S. in Computer Science, North Central College, 1991. Thesis topic: "The Computational Assessment of Algorithms for Solving the Laplace Equation on Vector and Parallel Architectures". B.S. in Computer Science, Illinois Benedictine College, 1988. Employment Background 1995 - Senior Systems Programmer, Cornell Theory Center, Cornell University, Ithaca NY 14853 1993 - 1995 Senior Technical Support Analyst, Mathematics and Computer Science Division, Argonne National Laboratory, Argonne IL 60439 1992 - 1993 Associate Computer Scientist, Petabyte Access Storage Solutions project, High Energy Physics Division, Argonne National Laboratory Argonne IL 60439 1991 - 1992 Associate Computer Scientist, Computing and Telecommunications Division, Argonne National Laboratory, Argonne IL 60439 1988 - 1991 Assistant Computer Scientist, Computing and Telecommunications Division, Argonne National Laboratory, Argonne IL 60439 Recent Papers David A. Lifka, Joseph Skovira, Waiman Chan, Honbo Zhou "The EASY - LoadLeveler API Project" IPPS 1996 Workshop on Job Scheduling Strategies for Parallel Processing April 1996 David A. Lifka "The ANL/IBM SP Scheduling System" IPPS 1995 Workshop on Job Scheduling Strategies for Parallel Processing April 1995 David A. Lifka, Mark W. Henderson, and Karen Rayl. ANL TM-201, "User's Guide to the Argonne SP Scheduling System" May 1, 1995 Professional Activities Association for Computing Machinery Computer Society of The Institute for Electrical and Electronics Engineers, Inc. Geoffrey Charles Fox gcf@nova.npac.syr.edu , http://www.npac.syr.edu, Phone: (315) 443-2163, Fax: (315) 443-4741 Citizen Status: Permanent Resident Alien; Citizen of United Kingdom Education: B.A. in Mathematics from Cambridge Univ., Cambridge, England (1961-1964) Ph.D. in Theoretical Physics from Cambridge University (1964-1967) M.A. from Cambridge University (1968) Professional Experience: 1990- Professor of Computer Science, Syracuse University 1990- Professor of Physics, Syracuse University 1990- Director of Northeast Parallel Architectures Center 1979-1990 Professor of Physics, California Inst. of Tech. 1986-1988 Associate Provost for Computing, California Inst. of Tech. 1983-1985 Dean for Educational Computing, California Inst. of Tech. 1981-1983 Executive Officer of Physics, California Inst. of Tech. 1974-1979 Associate Professor of Physics, California Inst. of Tech. 1971-1974 Assistant Professor of Physics, California Inst. of Tech. 1970-1971 Millikan Research Fellow in Theoretical Physics, Caltech 1970 Visiting Scientist (April-May), Brookhaven National Laboratory 1969-1970 Research Fellow at Peterhouse College, Cavendish Lab.,Cambridge 1968-1969 Research Scientist, Lawrence Berkeley Lab., Berkeley, Calif. 1967-1968 Member of School of Natural Science, Inst. for Advanced Study, Princeton, New Jersey Awards and Honors Senior Wrangler, Part III Mathematics, Cambridge (1964) Alfred P. Sloan Foundation Fellowship (1973-75) Fellow of the American Physical Society (1990) Journal Editorships Principal: Concurrency: Practice and Experience (John Wiley, Inc.) Physics and Computers (International Journal of Modern Physics C - World Scientific) Associate: Journal of Supercomputing, Selected List of Publications - Geoffrey C. Fox 1. Fox, G.C., Johnson, M.A., Lyzenga, G.A., Otto, S.W., Salmon, J.K., Walker, D.W., Solving Problems on Concurrent Processors, Vol. 1, Prentice-Hall, Inc. 1988; Vol. 2, 1990. 2. Fox G. and Mills K., "Information Processing and Opportunities for HPCN Use in Industry", pages 1-14. Number 796 in Lecture Notes in Computer Science, Springer- Verlag, New York, April 1994. Proceedings of "High Performance Computing and Networking(HPCN) Conference, Munich, 3. Fox, G.C., Copty, N.,Ranka, S.,Shankar, R. "Solving the region growing problem on the Connection Machine," in Proceedings of the 22nd International Conference on Parallel Processing, volume 3, pages 1993. 4. Fox, G. C., Messina, P., Williams, R., Parallel Computing Works!, Morgan Kaufmann, San Mateo Ca, 1994. 5. Fox G.C., Mills K., "InfoMall: an Innovative strategy for high-performance computing and communications application development", Internet Research, 4:31- 45, 1994. 6. Fox, G.C., Hipes, P., Salmon, J., "Practical Parallel Supercomputing: Examples from Chemistry and Physics", Proceedings of Supercomputing '89, pp 58-70. ACM Press, November 1989. IEEE Computer Society and ACM SIGARCH, Reno, Nevada. C3P- 818. 7. Fox, G.C., Hiranadani, S., Kennedy, K., Koelbel, C., Kremer, U., Tseng, C.W., Wu, M.Y., "FortranD Language Specifications", Rice COMP TR90079, December 1990, Revised, April 1991. 8. Fox, G. C. "Approaches to Physical Optimization," in Proceedings of 5th SIAM Conference on Parallel Processes for Scientific Computation, pp 153-162, March 25-27, 1991, Houston, TX, J. Dongarra, Kennedy, P. Messina, D. Sorensen, R. Voigt, editors, SIAM, 1992. C3P-959, CRPC-TR91124 9. Fox G.C., Mansour N., "Parallel Physical Optimization Algorithms for allocating data to multicomputer nodes", Journal of Supercomputing, 8:53-80,1994. 10. Fox, G. C. "Parallel Computing and Education," Daedalus, Journal of the American Academy of Arts and Sciences, Vol. 121, No. 1, pps 111-118, Winter 1992. C3P-958, CRPC-TR91123. Summary of Interests Fox is an internationally recognized expert in the use of parallel architectures and the development of concurrent algorithms. He leads a major project to develop prototype high performance Fortran compilers and their runtime support. He is also a leading proponent for the development of computational science as an academic discipline and a scientific method. He has established at Syracuse University both graduate and undergraduate programs which cover both simulation and information technologies. All course have been made available on the Web and his research includes HPCC technology to support education at both K-12 and University level. His research on parallel computing has focused on development and use of this technology to solve large scale computational problems. Fox directs InfoMall, which is focused on accelerating the introduction of high speed communications and parallel computing into New York State industry and developing the corresponding software and systems industry. Much of this activity is centered on NYNET with ISDN and ATM connectivity throughout the state including schools where Fox is leading developments of new K-12 curricula material built using VRML, Java and other new technology. PhD Advisor: Dr. Richard Eden Cambridge University, England Collaboraters: Fox has participated in HPCC since 1981 and in that time has collaborated in various fashions with many scientists. In particular current work involves scientists in CRPC (Argonne,Caltech,Los Alamos,Rice,Tennessee), and Grand Challenge scientists at Texas and NASA Goddard. Anne E. Trefethen aet@tc.cornell.edu , http://www.tc.cornell.edu/~anne Phone: (607) 254-4462, Fax: (607) 254-8888 Citizen Status: Permanent Resident Alien, US; Citizen of United Kingdom Education BSc. (Hons), Mathematics, First Class, Coventry University, UK 1983. Ph.D., Mathematics, Royal Military College of Science (RMCS), Cranfield University, UK, 1987. Experience 1995- present : Associate Director for Scientific Computational Support, Cornell Theory Center, Cornell University 1992-95 Research Scientist, Cornell Theory Center, Cornell University 1994 : Visiting Research Fellow at ETH Zurich, IPS (Interdisciplinary Project Center for Supercomputing), May-Aug. 1990 : Visiting Research Fellow at School of Mathematics, University of New South Wales, Sept-Dec. 1988-92: Research Scientist, Thinking Machines Corporation 1987-88: Permanent Research Fellow, Computational Mathematics Group, RMCS. 1983-87: Research Associate, Computational Mathematics Group, RMCS. 1981-82: Industrial Trainee, Meteorological Office, Bracknell, UK. As the Associate Director for Scientific Computational Support, Anne Trefethen is responsible for the groups within the Cornell Theory Center who provide support for the users, including software support, consulting, education and training, strategic applications support and outreach programs including the Smart Node Program. She is the principle investigator on the MultiMATLAB project, a project to develop a toolbox to enable the use of MATLAB in parallel. Before taking the position as Associate Director she was a research scientist in the performance and algorithms group at the Theory Center where she worked on algorithms and application implementation on the IBM SP1/2 and the KSR. Prior to joining the Theory Center she worked for three years at Thinking Machines Corporation where she was the project leader for the first release of the Connection Machine Scientific Software Library (CMSSL); as such she was involved in all aspects of the project from the overall design of the library and interfaces, to the underlying algorithms and implementation. During her four years at RMCS she was active in a number of research contracts combining scientific computing and numerical analysis. Throughout her career she has been involved in the development of education programs or materials. Most recently this includes design and development of materials for a Virtual Workshop. Present Interests: Large scale scientific computing; education in high performance computing, particularly development of computer-aided education. Societies Member of the Society for Industrial and Applied Mathematics. Selected Publications Electronic Communications - Education via a Virtual Workshop, R. Leibensperger, S Mehringer, A.E.Trefethen and M. Kalos, Proceedings of Sumposium on International Science and Engineering Education, June '96. MultiMATLAB : Matlab on Multiple Processors, A.E. Trefethen, V. Menon, C. Chang, G. Czaijkowski, C. Myers, L.N. Trefethen, CTC Tech Report CTC96TR239, 5/96. Submitted to Supercomputing '96. The Cornell MultiMATLAB Project, C. Chang, G. Czaijkowski, X. Liu, V. Menon, C. Myers, A.E. Trefethen, L.N. Trefethen , POOMA 96, Conference Proceedings, http://www.acl.lanl.gov/Pooma96, 1996. Implementing the NAS CG benchmark on the SP1/SP2, A.E.Trefethen and T. Zhang, CTC tech rep. CTC95TR208, 1995. Spectra and Pseudospectra for Pipe Poiseuille Flow, A.E.Trefethen, L.N.Trefethen and P.J.Schmid, ETH IPS Research Rep. No 94-16, 1994 Hydrodynamic Stability without Eigenvalues, L.N.Trefethen, A.E.Trefethen, S.C.Reddy and T.A.Driscoll, Science, Vol 261, July 1993. Constrained Complex Approximation Algorithms for Communication Engineering, A.E.Daman (maiden name), J.C Mason and S.J. Wilde, in Algorithms for Approximation 2 , Chapman and Hall, 1989. Advisory Expert System for Curve and Surface Fitting, A.E.Daman, in Scientific Software Systems, Ed. John Mason, Chapman and Hall, 1989. A generalized cross-validation method for meteorological data with gaps, in Algorithms for Approximation, A.E.Daman and J.C. Mason, Clarendon Press, 1987. On the integration of isoparametric quadratic elements for curved boundaries, A.E.Daman, R.N.L. Smith and S.Ellis, in BETECH '86, Ed. J.R. Connors and C. A. Brebbia, Computational Mechanics Publications, 1986. PhD Advisor: Professor John Mason, Huddersfield University, England S. Lennart Johnsson Department of Computer Science University of Houston, Houston, TX 77204-3475 Phone: 713-743-3371 e-mail: Johnsson@cs.uh.edu http://www.cs.uh.edu/~johnsson Education: Ph.D Control Engineering, Chalmers Institute of Technology, Gothenburg, Sweden, 1970. Previous affiliations: Director of Computational Sciences, Thinking Machines Corporation, Cambridge, 1986 -- 1995; Associate Professor of Computer Science and Electrical Engineering, Yale University, New Haven, 1983 -- 1990; Visiting Professor, Department of Computer Science, Uppsala University, Uppsala, Sweden, June 1987; Visiting Professor, Department of Computer Science, Uppsala University, Uppsala, Sweden, May -- June 1986; Visiting Scientist, Mathematics and Computer Science Division, Argonne National Laboratories, Argonne, IL, July 1985; Visiting Professor, Departments of Mathematics and Computer Science, Linkoeping University, Linkoeping, Sweden, May -- June 1985; Visiting Scientist, The Institute for Mathematics and its Applications, The Academy of Engineering Sciences, Stockholm, Sweden, August -- September 1983; Senior Research Associate, 1981 -- 1983, Research Associate, 1980 -- 1981, Visiting Assistant Professor, 1979 -- 1980, Computer Science, California Institute of Technology, Pasadena, CA; Manager, Systems Engineering, Electrical Systems, 1974 -- 1980, Systems Engineer, Electrical Systems Office, 1970 -- 1974, Central Research and Development, ASEA AB, Vaesteras, Sweden; Postdoctoral Scholar, Systems Science Department, School of Engineering and Applied Science, UCLA, Los Angeles, CA, 1970 -- 1971; Research Scholar, Swedish Board for Technical Development at the Control Engineering Department, Chalmers Institute of Technology, Gothenburg, Sweden. 1967 -- 1970. Current affiliations: Hugh Roy and Lillie Cranz Cullen Distinguished Professor of Computer Science, Mathematics and Electrical Engineering, University of Houston; Adjunct Professor Department of Computer Science and Center for Research in Parallel Computation, Rice University, Houston; Chair, Executive Committee, Houston Area Computational Science Consortium; Member, Keck Center for Computational Biology, Houston; Chair, Scientific Board, The National High Performance Computing Center, The Royal Institute of Technology, S--100 44 Stockholm, Sweden; Gordon McKay Professor of the Practice of Computer Science, Harvard University, Cambridge. Awards: John Ericson Medal, Best Paper Award at the 1986 International Conference on Parallel Processing, Nomination for Best Paper Award at the 1983 Design Automation Conference (2nd place). Boards: USRA Science Councils for NASA Langley (1990 - 1992) and CESDIS (1990 - 1995), CRA Board member, 1992 - 1994, Industrial Advisory Board, West Virginia Experimental Program to Stimulate Competitive Research (EPSCoR) (1994 - ), Steering Committee, DIMACS Parallel Implementation Challenge (1992 - 1994), Steering Committee, Conference series on Massively Parallel Processing Using Optical Interconnections (1993 - ). Member of editorial boards for International Journal of Supercomputer Applications, Journal of Scientific Programming, Journal for Numerical Linear Algebra with Applications, Journal on Concurrency: Practice and Experience, International Journal on High Speed Computing, Journal of Parallel and Distributed Computing. Five relevant publications: ---------------------------- ``Block Cyclic Dense Linear Algebra'', (with Woody Lichtenstein), SIAM J. of Sci. Comp., vol. 14, no. 6, pp. 1257 -- 1286, 1993. "All--to--All Communication on the Connection Machine system CM--200", (with Kapil K. Mathur), the Journal of Scientific Programming, vol. 4, no. 4, pp. 251 -- 273, 1995. "Scalability of Finite Element Applications on Distributed--Memory Parallel Computers", (with Zdenek Johan and Kapil K. Mathur and S. Lennart Johnsson and Thomas J..R. Hughes), in Computer Methods in Applied Mechanics and Engineering, vol. 119, nos. 1 -- 2, pp. 61 -- 72, November 1994. ``Implementing $O(N)$ N--body algorithms efficiently in data parallel languages'', (with Yu Hu), to appear in the Journal of Scientific Programming, vol. , no. , pp. , 1996. ``Optimal Communication Channel Utilization for Matrix Transposition and Related Permutations on Boolean Cubes'', (with Ching--Tien Ho) in the Journal of Discrete Applied Mathematics, vol. 53, pp. 251 -- 274, September 1994. Other relevant publications: ---------------------------- ``Local Basic Linear Algebra Subroutines (LBLAS) for the CM--5/5E'', (with David Kramer and Yu Hu), to appear in the International Journal of Supercomputer Applications, vol. , no. , pp. , 1996. ``An Efficient Communication Strategy for Finite Element Methods on the Connection Machine CM--5 System", (with Zdenek Johan, Kapil K Mathur, and Thomas J.R. Hughes), in Computer Methods in Applied Mechanics and Engineering, vol. 113, pages 363 -- 387, 1994. ``Experience with the Conjugate Gradient Method for Stress Analysis on a Data Parallel Supercomputer'', (with Kapil K. Mathur) International Journal on Numerical Methods in Engineering, vol. 27, no. 3, pp. 523 -- 546, December 1989. ``Index Transformation Algorithms in a Linear Algebra Framework'', (with Alan Edelman and Steve Heller), in {\em Transactions on Parallel and Distributed Systems}, vol. 5, no. 12, pp. 1302 -- 1309, 1994. ``Spanning Graphs for Optimum Broadcasting and Personalized Communication in Hypercubes", (with Ching--Tien Ho), IEEE Trans. Computers, Vol. 38, No. 9, pp. 1249 -- 1268, September, 1989. Collaborators in addition to the co-authors above: -------------------------------------------------- Collaborators on this proposal: Dan Davison, Roland Glowinski, Olin Johnson, Bowen Loftin, Monte Pettitt, Marek Rusinkiewicz, Ridgeway Scott, all at the University of Houston. Other collaborators during the last 48 months: Ralph Brickner, Los Alamos National Laboratories, Roland Sweet, University of Colorado, Jean-Philippe Brunet, Pablo Tamayo and Jacek Mykowski, Thinking Machines Corporation, Steve Daly and John Richardson, Silicon Graphics Inc., Kirk Jordan and Jill Mesirov, IBM Corp. and Marina Chen, Boston University. Graduate students: Ching-Tien Ho, IBM Alamaden, Abhiram Ranade, UC Brekeley, Ted Nesson, Advanced Video Communications, Inc., Nadia Shalaby, Harvard University, Yu Hu, Harvard University, Dimitris Kehagias, Harvard University, Peggy Li, Jet Propulsion Laboratory. Tak-Kwong Ng, IBM. Graduate Advisor: Birger Quarnstroem, Chalmers Institute of Technology, Gothenburgh, Sweden, Lars-Erik Zachrison, The Royal Institute of Technology, Sweden. |HHк(џсџтљFG(ќHHи(d'`аЯрЁБс;џў џџџўџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџSummaryInformation(џџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџTheory Center Staff'@цМеtМ@zXэtМ@ШTєtМЏ˜@Microsoft Word 6.0.15;џў џџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџMќЅЎѓ [K}F-№-Ь2ю274|4T55172737G7х?і?‘BВBЛBМBІCЗCНCтCЉEэEъFЪGxJyJЩJЪJюKяKјKљKеNиN7R9RђWћWГXДXЙnDo{{{C{[…•…k•‹•}™•ŸуЁЂъ•ъќќњјќјјјјјјјјјјјјјјјјјјјјјјјјјјіјјієucUcUcF‚ƒ„…†‡ˆ‰Š‹ŒŽ‘’“”ўџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџLMXn’ЁЖЗИХэ1G[qŠ‹ŒŸОеь:;<RqЂЙафљњћќўўўїїїўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўф§Š§Іџ,ќ| } FGЃЄЅЏъыьэюя№ёђѓ  ./uс!D[\]IJK}~HI !‰Šo p ўўћћћћћўўўўўўўўўўўўўўўўўўўўўўўўўћћўўўћўћћћћћћ-p 1$2$‡%ˆ%ч)ш)ѕ)і)ї)ј)љ)њ)ћ)-,.,F-G-˜-Ї-Й-№-..Щ2Ъ2Ы2Ь2ъ2ю28494O4i4j4|4T5U552737H7I7у:ф:є<§§§§§§§§§§§§§§ћ§ћћћћћ§§Р!№§Р!№ћР!№ћР!№ћР!№ћР!№ћР!№§Р!№ћР!№ћР!№ћР!№ћР!№ћР!№§Р!№ћР!№ћР!№ћР!№ћР!№ћР!№ћР!№§ Р!№§Р!№§Р!№-є<ѕ<Ы>Ь>т?у?ф?х?ї?ј?B‘BœBBВBЇCЈCЖCЗCИCЙCКCЛCМCНCрCсCтCЉEЊEЩEьEэEщFъFыFGLG•G–GЃGЩGЪGŒIIўћўћўўўўўћўўўўўўўўўўўўўўўўўћўўўўўўўўўўўўўўћў,IЩJЪJљKњKзNиN8R9RDWEW№WёWђWћWˆX‰Xу`ф`QcRcee{fЌf­fьgоНо›о™–™––––––––––––––––––! h˜ў 4џџh. h˜ў 4џџh.! h˜ў 4џџh.ьgэgžkŸkymzmЛnМnьnэnAoBoCoDoEo­rЎrьxэxюzяz{{{C{џ}~78HMp˜ Мйќ‚ ‚‚§§§§§§ћћћћћћћ§§ћ§§§ћћћћћ§§§§§ййййййййй§! h˜ў 4џhЗ'‚6‚7‚X‚‡‚В‚о‚п‚ƒƒ'ƒNƒtƒ–ƒМƒ3„4„H„ˆ„П„…–…—…˜…"‹#‹nŽoŽё‘ђ‘••.•/•0•1•J•e•f•g•§§§лл맧§§ллл맧§ллл맧§й§§§й§§йййййййй! h˜ў 4џhЗ'g•h•i•j•‹•Œ•Ј•Љ•{™|™}™Љ™Њ™ъ™ы™%š&šHšIš‘š’š“šЈšЉšЧšШšђš››8››П›џ›?œ@œ€œРœџœ;<{ўўўўўўўћўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўў-{Кљњћ ž!žWžŒžЬžЭžŸ8ŸxŸyŸ”Ÿ•ŸнŸ  2 [ „ Ѓ В У Ф й Ё+Ё:ЁYЁjЁkЁ‚ЁЋЁдЁуЁЂЂЂ5Ђ6Ђ7Ђ8Ђ9Ђўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўў-9ЂOЂPЂQЂ`ЂzЂЂЃЂДЂУЂиЂйЂуЂ)ЃkЃВЃРЃСЃ ЄVЄzЄ{ЄЙЄКЄаЄђЄ ЅЅ-Ѕ.ЅuЅЄЅЕЅЖЅјЅ)ІEІVІWІЂІЩІкІлІ&ЇMЇ^Їўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўў-^Ї_Ї`ЇnЇВЇћЇЈЈgЈЄЈЅЈьЈ#Љ$Љ%Љ=ЉaЉbЉЃЉГЉДЉЕЉЖЉЗЉИЉЭЉЮЉЊ*Њ+ЊoЊpЊ{ЊФЊ Ћ-Ћ.ЋGЋ‚ЋДЋ№Ћ0Ќ{ЌХЌ ­W­ўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўў-W­Ё­ъ­1ЎЎЮЎЏ3Џ4ЏFЏŽЏжЏнЏоЏђЏ:АƒААКАЛАБGБ‘БšБхБ+ВqВИВГGГГкГъГ'ДoД’ДлД$ЕoЕ‡ЕвЕЖ=ЖˆЖЭЖЗўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўў-ЗaЗЂЗцЗєЗ<И…ИЉИЊИПИЙIЙ“ЙнЙ!КeКЈКэК5ЛyЛУЛМRМœМуМ,НsН‰НŠНЦНЧНОXО–ОпОшОщОъОыОьОэОюОяОППDПўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўў-DПbПcПЋПЌПЖПћПCРYРZРeРЎРиР#СlС‘СйСєС8ТƒТЧТУУУHУ‘УмУ Ф`ФЋФяФ6ХХИХЙХЦGЦ‘ЦжЦЧfЧЉЧГЧДЧіЧ>Шўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўў->ШoШpШƒШЮШЩЩЩMЩNЩOЩPЩfЩgЩЈЩяЩ7Ъ8Ъ‚ЪУЪіЪїЪ=Ы„ЫИЫЙЫЬ'Ь(ЬkЬЎЬЏЬљЬ2Э3ЭrЭЕЭюЭяЭ3ЮxЮyЮТЮЯЯЯўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўў-ЯWЯ ЯчЯшЯщЯъЯ.а/а0а1а2а3а4аHаIаJаiа€а—аЋаЦахацача.бHбIббгбвfвАвљв@гƒгЫгд^дІдщд,еoеИеџеGжўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўў-GжŠжЪжз&з'зlзГзћзFи‘или&йpй­йЎйЏйій@кoкpкБкњк=л~лХл мOмšмом&нUнVнWнXнtн‘н’ндноо[оІоПоРопўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўў-пNппйпппрп&рhрŽррирс`сpсqсŽсЋсЌсїс:тoтpтИтуDуsуtуНуфIф]ф^фЇфьф*х+хuхРхцхчхцMцNццецчўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўўў-чfчЌчяч:ш^ш_шŸшуш.щuщzщ{щ|щФщъъўўўўўўўўўўўўўўўўK@ёџNormala c"A@ђџЁ"Default Paragraph Fontч џџџџ ъ!џџ!џџ!џџ џџ џџ џџ џџ џџ џџ џџ  џџ џџ  џџ џџ џџ џџ џџ џџ џџ џџ џџ џџjx+ф7ъAžNф]эk”yП‘П˜ŸВЄоЌRЙ6ТxЫ&дpочDЙПšmT k W  @ !IJI•ъvќp є<Iьg‚g•{9Ђ^ЇW­ЗDП>ШЯGжпчъwxyz{|}~€‚ƒ„…†‡ˆ‰›Theory Center Staff2Macintosh HD:Temporary Items:Word Work File A 3006Theory Center Staff2Macintosh HD:Temporary Items:Word Work File A 3006Theory Center Staff2Macintosh HD:Temporary Items:Word Work File A 3006Theory Center Staff2Macintosh HD:Temporary Items:Word Work File A 3006Theory Center Staff$Macintosh HD:proposal folder:CCISE.2Theory Center Staff1Macintosh HD:Temporary Items:Word Work File A 950Theory Center Staff1Macintosh HD:Temporary Items:Word Work File A 950Theory Center Staff1Macintosh HD:Temporary Items:Word Work File A 950Theory Center Staff$Macintosh HD:proposal folder:CCISE.2Theory Center Staff$Macintosh HD:proposal folder:CCISE.2џ@€D*D*D*C*djOMTimes New RomanTimes Symbol"MArialHelveticaj MTimes"1аhEK&,L&іK&nЏ˜MƒDБGInfrastructure for distributed Heterogeneous Collaborative Computation.Theory Center StaffTheory Center StaffаЯрЁБс;џў џџџў}џџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџџ