[About the Foresight Institute] [Index of Foresight Conferences]

[Index of Abstracts]

Computational neural networks: a general purpose tool for nanotechnology

S.A. Meyer*(a), M. Morgenstern(a), J.A. Darsey(b), D.W. Noid(c), B.G. Sumpter(c)

This is an abstract for a poster to be presented at the Fifth Foresight Conference on Molecular Nanotechnology. There will be a link from here to the full article when it is available on the web.

A computational scheme which utilizes neural networks was developed to predict properties of nano-structured materials and optimization and control of nano-devices. Using a set of simple algorithms to encode the structure and composition of the material directly into numerical vectors neural network modules can correlate these numeric inputs with a set of desired properties. Calculated results for a series of hydrocarbons, fluorohydrocarbons, amines, and crown ethers demonstrate average accuracies of 0.2-8.1% with maximum deviations of 16-20% for a broad range of thermodynamic, physical, biological (toxicity: human and environmental) and physical-chemical characteristics (heat capacity, enthalpy, heat of evaporation, boiling point, density, refractive index, stability constants, etc.). A molecular design tool based on the neural network capabilities of formulating accurate quantitative structure-property relationships is described. This technique, called computational synthesis, is capable of formulating the structure and composition of materials which will give a set of specified properties. In other applications, this technique has been proven useful in the reverse engineering of nano-fluidics and nano-motors.

Research sponsored by the Division of Materials Sciences, Office of Basic Energy Sciences, U.S. Department of Energy under contract DE-AC05-96OR22464 with Lockheed-Martin Energy Research Corp.