DYNAMISM FOR MATERIALS DESIGN AFTER INFORMATION FUSION. Shuichi Iwata, RACE, The University of Tokyo 4-6-1 Komaba, Meguro-ku, Tokyo 153 Japan

Trials and errors to design materials for structural, opto-electric and electro- magnetic applications are reported so as to develop a set of software tools for materials research and developments. Fusion of relevant information is the first step and three types of information, namely, (1)calculated data by different models with explicit semantics on the basis of underlying principles and mechanisms, (2)a set of structure-related data of phase diagrams, crystallographic data, basic intrinsic properties, diffraction data and microstructural images, and (3)engineering data for different applications, are tried to be integrated on a workbench to find out solutions. In general problem solutions do not follow one deterministic way and trials and errors are the normal case. Therefore the software tools need to have an adaptability to any changes in the process of materials design so that we are designing the functions of the software tools by using the following five aspects.

(1)Establish Intentions : De fact driven approaches by taking advantage of VRML and other browsers are proposed on the basis of the following considerations. How to get a new idea emerged and to discover multi-variate patterns through the mining of data;browsing of continuous, heterogeneous information via filtering/focusing/erasing/selecting/reasoning to extract useful information, with fluctuating/changing viewpoints.

(2)Establish View : How to define, reduce dimension of, and represent design space to visualize solutions using indices, structure/properties/functional/process/etc. maps based upon focus attributes. Parameters describing structural stability, band structures and microstructural images are the keys for the above applications.

(3)Establish Concepts and Categories : How to enable in situ process discovery, classify/organize design and/or control space into sub-spaces and/or sub- processes searching for feasible solutions and/or regularities to include problem decomposition based upon multi-variate criteria/guidelines/heuristics.

(4)Establish Associatively and Causality : How to create design 'windows' for unique linear combinations of material properties, e.g., strength, temperature, stress, creep rate, corrosion, etc. for applications involving high temperature.

(5)Establish a Paradigm : How to implement and combine geometric reasoning and analytic models with discrete optimization methods in available computing resources for the simultaneous design of material, shape and process.

These latter three aspects are mixed up to form a "virtual production line" to verify each candidate material until we get a suitable material in the reality by replacing all parts of the "virtual production line" by a "real" line, tactics and strategies of which are proposed in the presentation.