|
|
|
|
Redfish Lab: Keywords: Artificial
Life, Self-Organization, Distributed Objects, Redfish is working on tools to facilitate the gathering and delivery of Chinese information. The foundation for these tools is a network of software objects poised to self-organize and reproduce. Networks compete for limited network processing time and memory. Objects are developed in Java and Macromedia Director/Shockwave. Objects with fit behavior reproduce. Unfit ones don't. Cooperative objects persist and extend their relations with other objects. Behaviors are governed by neural networks with topologies and connectivity weights indirectly determined by objects' genomes, developmental processes and interaction with the environment. Objects evolve to higher levels of fitness via the mechanisms of natural selection and self-organization. Nodes of the neural networks interact in a simulated chemical matrix. The network is an open design running on UNIX, Windows and Macintosh. The objects make up the business rule layer of an n-tier information system. The presentation layer is made up of traditional tools including web browsers, java apps, Microsoft Office apps and database forms. The data services layer allows for the necessary persistence and query of objects. This layer can include an enterprise's existing legacy system , off-the-shelf relational databases and client/server systems. The network is flexible enough to include inputs from many sources, including bar code readers, video cameras, inspection microscopes and analog sensors on the factory floor. As the business objects prove themselves in real-time environments, Redfish can take samples from successful "cultures" of objects and transplant them into secure areas of additional corporate and personal LANs. The evolved business objects can be transplanted independent of the data specific to a corporate client thus ensuring a corporations privacy. This type of information system is is not suitable
for all types of enterprises. It is primarily designed for those enterprises
competing in extremely dynamic market environments where business rules
are in constant flux and more traditional information management techniques
prove too rigid. An example of a good candidate company would be one
that relies on the Internet as both a primary data source and as a primary
marketing medium. |
|
|
Below is a screen shot from Redfish's Chaos projector developed in Macromedia Director (requires Shockwave). The software explores the dynamic behavior lurking within the all too familiar logistic map. This feedback equation can be found in many academic disciplines particularly economics, biology, chemistry and physics. We've used the tool to approximate fluctuations in populations of software objects as reproduction rates and lifespans are manipulated. The equation NextPopulation=GrowthRate x CurrentPopulation x ( 1 - CurrentPopulation), captures the fact that a culture's next population count is a dependent on its current count and growth rate. The culture can be a colony of ants, a petri dish of bacteria or, as in our case, a society of software objects. Constraints to growth, such as food, space, predators, memory or disk space are captured by the (1-CurrentPopulation) term. In the example below, a population is started at an arbitrary level (N seed=.2820), a reproduction rate of 1 (Start A=1.000) and represented by a blue dot on the graph. The next generation is calculated via the equation N2=A*N1*(1-N1 ) and plotted with a new color. The equation is repeated 12 times (Generations = 12) with the resulting population. The reproduction rate is then increased to 1.6 (A step = 0.6000). This iterative process continues until the reproduction rate is 4.2 (End A = 4.200). Below, the first and second generations are not graphed (Only graph generations>2). This graph is distinct to other graphs we've seen as we are tracking populations for a given "generation" as the term "a" changes. Restated, like colors represent a population level after the same number of generations.
Depending on the parameters of the equation (shown in green numbers) , the population will settle to single point attactors, periodic attractors, complex or strange attractors or chaos. Through experimentation with the parameters of the population function, the software user can get a better intuitive feel for the edge-of-chaos. It is this region in dynamical systems where the system dances between periodicities and deterministic chaos. Dynamical systems can often use a single variable, referred to as a control parameter to adjust the system to regions of stasis, periodicities, chaos and edge of chaos. The software projector was designed for in-house experimental use and during client education on the basics of chaos and dynamical systems theory. If you have difficulty running the projector in your browser, you can download the windows EXE. The uncompiled director file is also available.
Stephen Guerin started researching chaotic systems as they applied to economic systems and business cycles in 1989. He founded Redfish in 1991 to provide computer solutions for desktop publishing and video editing. In 1993, Redfish Group delivered several award-winning multimedia and computer animation projects. Operations were shifted to Beijing in 1994 to develop a series of Chinese language learning tools. Mark Haviland joined Redfish Group at this time and was instrumental in making RedfishGroup the first foreign company to establish a host on China's nascent Internet backbone. We went on to provide Internet services and consulting to multinational and Chinese firms with presences in Beijing and Shanghai. Our web-based Chinese-English dictionary sans interface is now available. |
