UW’s Meritorious Applications

The University of Wyoming is committed to advancing the concepts and research atmosphere of the Next Generation Internet [1]. UW has several research applications where our lack of high quality connectivity is presently constraining the advancement of researchers in solving problems and improving our knowledge base. In addition, dialogue with numerous UW faculty suggests that their research directions and pursuits are limited by their lack of high performance connectivity. The illustrative applications discussed in this section are examples of current research that is being limited by our poor connectivity. In addition to these examples, UW could and would use improved connectivity to enhance its distance education and telemedicine initiatives.

The University of Wyoming recently demonstrated its commitment to large data set research by using funding from NSF-EPSCoR, the State of Wyoming and the private sector, to create the Spatial Data and Visualization Center (sdvc.uwyo.edu). The SDVC houses large (multiple terabyte) size data sets which can be served and accessed by researchers outside UW. In addition, the SDVC performs computationally intensive modeling and visualization tasks. A current example is a beta version of the GeDET being developed jointly with Environmental Systems Research, Inc. (ESRI) which requires real time access to large data sets. The full potential of the SDVC cannot be realized without enhanced connectivity. Other specific and current UW applications examples follow.

The UW Real Time Collaboration Network (Meta-Amphion and Amphion)
Recent advances in networking software have made many new network-based applications possible. There are numerous advantages of such applications. Two of these are of interest here. First, applications can be built that enable real time collaboration between multiple remote locations, greatly reducing the time required to collaborate on scientific (and nonscientific) projects. Second, applications can be made available over the network, removing the requirement for porting software between platforms. Unfortunately, while networked applications have these advantages, they often require real time communication of large amounts of information and can be bandwidth limited. The Computer Science Department at the University of Wyoming is building two such applications for which our current Internet connection is severely limiting.

The Computer Science Department, in collaboration with NASA Ames Research Center, is also involved in a more generalized project, called Meta-Amphion, to develop a tool that increases the access non-experts have to complex subroutine libraries such as the space opportunity analysis library. This increased access to complex subroutine libraries is realized through automated deductive synthesis. First, an abstraction is constructed for the very complex subroutine library. This abstraction is in a language familiar to users and hides programming details. The abstraction language is normally graphical, allowing users to develop graphical specifications of their desired end products without attention to programming issues. These graphical specifications are used to automatically construct programs that call library subroutines to produce the desired result.

This project is using new technology that we have developed in automated deductive synthesis. Most existing automated deductive synthesis technology requires automated software engineering experts to use. As a result, very few usable automated deductive synthesis systems exist. Meta-Amphion is a notable exception to this. The technology we are developing enables the construction of specialized deductive synthesis systems by individuals that are not experts in automated software engineering.

We believe that Meta-Amphion will lead to unprecedented access to complex subroutine libraries that are currently accessible to only a few. By intregrating into Meta-Amphion the ability to place such applications on the WorldWide Web, accessibility to these libraries will be further increased. In addition, we intend to host Meta-Amphion over the WorldWide Web to make it widely accessible. This will require high bandwidth network access to computers at the University of Wyoming.

Another more specialized application, called Amphion, assists space scientists in the analysis of astronomical observation opportunities. For instance, an experimental version of Amphion has already been used by scientists at the Jet Propulsion Laboratory to analyze opportunities to observe Saturn’s faintest rings and smallest satellites when the Sun and Earth passed through Saturn’s ring plane. The method of interaction with Amphion is to use a graphical editor to create a high-level specification of the observation in which one is interested. For example, space scientists used the animation created for the Saturn rings observation to understand what observations could usefully be made when Saturn’s rings were on edge. From such a specification, Amphion automatically generates a Fortran program that assists in analyzing the observation opportunity. This program consists of calls to a subroutine library developed at JPL that performs geometric analysis with light-time correction using accurate ephemeris data. Amphion also contains an animator that renders a real time interactive animation of the opportunity, which is very useful for visual analysis of the opportunity.

Amphion is a large system that is difficult to port to new sites and platforms. We have converted Amphion to a web-based tool to allow space scientists to access it remotely. While this tool is currently running at the University of Wyoming, it is not available to many potential users because of our limited Internet accessibility. From the early experiences of space scientists with Amphion, it appears that this most benefit real time collaborative tool, which allows space scientists at remote locations to develop specifications and view animations collaboratively, would realize significant use and further development if we had enhanced connectivity.

Both the graphical editor and specific bandwidth animator require real time (low-latency) interaction to be used effectively. This need requires highspeed specified bandwidth network connectivity. This is particularly true for the animator. Our current connection to the Internet has severely limited both the fidelity and flexibility of the animations that can be delivered to a remote site in a timely fashion and has caused us to place on hold plans for a collaborative version of our system.

Collaborative Computational Astrophysics
Astronomy has made numerous advances recently through the use of high-speed processors and multi-dimensional visualization. At the University of Wyoming, faculty in the department of Physics and Astronomy are currently involved in two large scale computational modeling problems. One is the calculation of evolutionary sequences of interacting binary stars from formation along the asymptotic giant branch (AGB) to demise as faint, weakly interacting degenerate star-brown dwarf pair. Development of the EVOLve model, in collaboration with other astronomers at MIT, Arizona State University, and University College London, has recently made major advances in the field via the addition of new equation of state calculations added to the code and advanced computational algorithms for the beginning evolutionary sequences. The development of the secular evolution code for this model includes the latest brown dwarf interior models and was funded by NSF and NASA.

Further involvement of this collaboration is moving towards the next logical step, that of real time interaction with the code. Model formation and evolution involves many complex steps with multiple possible branches for the outcome. Working with the various models in such a manner as to be able to effect their outcome along the proper lines is essential. Otherwise, numerous uninteresting, time consuming and computationally intensive paths get explored, making the progress for the specific project slow. One method of dealing with this issue would be to allow each collaborator to view each progressing model and each total ensemble, as they progress in time. This would involve not only fast processors but multi-dimensional graphics distributed across the net to our collaborators. The University of Wyoming, which leads this collaborative effort, cannot currently proceed in this direction because of our limited bandwidth and high latency situation.

The second computationally intensive project involves numerical investigations of the formation and evolution of accretion disks. These disks of material which form encircling numerous astronomical objects such as quasars, cataclysmic variables, and proto-planetary disks, are the result of gravitational in-fall of material from a source surrounding a gravitation potential such as a star or black hole. Disks are dynamic objects and can not be correctly modeled in 1-D or as time independent entities. They must involve temporal constraints and at least 2-dimensional structures. At times, accretion disks can undergo outbursts during which time local hydrodynamic variables change over orders of magnitude and within very short dynamic times. These changes are critical to understand in order to pass information from one model step in the calculations to another. As before, without real time interaction and complex visualization by remotely located investigators, the keys to allowing efficient solution of these problems are unavailable.

Tools such as those discussed above are also ideal for use as educational mechanisms. Allowing students, not only at the University of Wyoming, but throughout the state and country, to benefit from the ability to "see" models built in real time, to "see" the power of computational physics, and to "see" how they themselves can interact with models and explore the parameter space of the physical world, is one of the greatest learning tools potentially available. We currently have an in-house wire-frame interacting binary model which allows a few physical parameters to be varies, but the "realness" of the model is limited.

Current bandwidth limitations and latency are the biggest hurdles to the advancing research in both of the above situations. Numerous sites exist for fast computation, advanced graphics, three-dimensional model visualization, etc. but our usage of these in a real time, interactive manner is limited today. Wire-frame graphics, low-resolution imagery, and limited color tables are also the current norm. Usage of these already developed codes and/or already existing complex, expensive peripherals would save additional costs to these projects, but these devices often only exist at remote sites, thereby requiring efficient connectivity. In addition, already developed resources at Wyoming could be used in a likewise manner from other sites. For example, porting codes, like Java scripts, which tie all the above together, across the net from Wyoming to collaborators machines, would be an ideal solution to many of these issues.

next