A Study of Massively Parallel Molecular Computing - Creating IntBrconnection'Free Computers -

大规模并行分子计算的研究 - 创建 IntBrconnectionFree 计算机 -

基本信息

批准号：
13680385
负责人：
AOKI Takafumi
金额：
$ 2.75万
依托单位：
Tohoku University
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (C)
财政年份：
2001
资助国家：
日本
起止时间：
2001 至 2002
项目状态：
已结题

项目摘要

The exponential growth of VLSI circuit complexity causes interconnection problems to become particularly severe. This research project is to investigate a possibility of interconnectionfree molecular computing from the viewpoint of integrated circuit technlogy as well as fundametal computer science. Listed below are major results of this project:1. We proposed a basic model of artificial catalyst devices for molecular computing, and investigated some possible approaches to device implementation. We investigated a possibility of realizing massively parallel computation using reaction-diffusion dynamics induced by a collective behavior of artificial catalyst devices, which are fabricated on a 2D substrate. As a first step for demonstrating the proposed concept, we developed an experimental model of a redox microarray consisting of a two-dimensional array of platinum microelectrodes, each of which works as an artificial catalyst device. We confirmed that the redox microarray could control reversible redox reactions in a thin liquid layer on the surface of the array, and could realize artificial excitable reacton diffusion-dynamics. We successfully observed active chemical-wave propagation within the liquid layer.2. We investigated a systematic way of designing artificial reactiondiffusion dynamics for performong specific image/pattern processing. We employed the framework of Digital Rraction Diffusion Systems (DRDSs) --- a special class of multidimensional digital filters having pattern/texture formation capability --- for theoretical discussion. Wedveloped a new restoration algorithm for texture images, such as fingerprint images, on the basis of DRDS. We also developed a shortest path search algorithm for arbitrary 2D/3D maps using excitable DRDSs.Further investigations on a systematic way of designing molecular computing systems are deing left as future research subjects.

VLSI电路复杂性呈指数级增长，导致互连问题变得尤为严重。本研究项目旨在从集成电路技术和基础计算机科学的角度研究无互连分子计算的可能性。本项目的主要成果如下： 1．我们提出了用于分子计算的人工催化剂装置的基本模型，并研究了一些可能的装置实现方法。我们研究了利用由人造催化剂装置的集体行为引起的反应扩散动力学实现大规模并行计算的可能性，这些催化剂装置是在二维基板上制造的。作为演示所提出概念的第一步，我们开发了一个氧化还原微阵列的实验模型，该模型由二维铂微电极阵列组成，每个电极都充当人工催化剂装置。我们证实氧化还原微阵列可以控制阵列表面薄液层中的可逆氧化还原反应，并可以实现人工可激发反应扩散动力学。我们成功地观测到了液层内活跃的化学波传播。2．我们研究了一种设计人工反应扩散动力学以执行特定图像/图案处理的系统方法。我们采用数字牵引扩散系统（DRDS）的框架（一种特殊类型的具有图案/纹理形成能力的多维数字滤波器）进行理论讨论。在DRDS的基础上，开发了一种新的指纹图像等纹理图像的恢复算法。我们还使用可激励的 DRDS 开发了一种用于任意 2D/3D 地图的最短路径搜索算法。对设计分子计算系统的系统方法的进一步研究将作为未来的研究课题。