【作者】 洪龙；

【导师】 朱梧槚；

【作者基本信息】 南京航空航天大学 ， 计算机应用技术， 2006， 博士

【摘要】 基于中介数学,研究一种有别于模糊数学和粗集的数值化处理模糊现象的理论——中介真值程度的度量,并将其应用于计算机系统结构研究。全文的主要内容如下:在介绍模糊现象的研究历史及分析了处理模糊现象主要方法的特点后,简要说明课题的研究目的及意义,并以新思想、新概念为线索简述计算机系统结构的研究历史及进展。阐述了研究模糊否定词和真值程度词的数值化方法对于应用的必要性。在建立了谓词的标准度并将其用于描述一般应用的数值区域与其对应谓词的真值之间的关系后,提出相对于谓词的距离比率函数定义,并由此建立了一维情形下个体的真值程度函数。以此为基础,讨论了n维情形下个体的真值程度的度量。再以个体的真值程度度量为基础,讨论了离散型和连续区间群体的真值程度的度量。由于提出超态概念,理论上能把真值程度域从[0,1]拓宽到(-∞,+∞)。应用示例表明:真值程度函数具有合理性和有效性。在建立了中介标准度后,提出模糊程度概念、离散型中介熵定义和连续区间中介熵定义。以中介熵和传统的随机熵为基础,探讨了不确定度量。针对计算机技术中出现的硬件与软件的界限模糊不清的现象,提出兼容原定义的软件、硬件定义;根据中介思想提出中介件,并以中介真值程度度量讨论了中介件的主要性能,随之自然地赋予了计算机系统结构新的含义。提出评价设计质量的性能成本比的观点,并由此讨论了中介件的设计原则。针对反对对立概念强调“最大的差异”,提出弱化“两极”,突出“过渡”的观点。在对过渡与对立进行了一般性讨论后,建立了标准数值化映射和广义数值化映射概念,构造性地证明了过渡情形能转换为反对对立情形的充分条件,由此自然地引出广义中介件概念。针对传统的一维多层次结构存在的弱点,提出HCW(hierarchical- component-ware)三维结构。应用示例表明:HCW多层次结构适应计算机系统结构研究的新局面,有利于计算机技术的创新与发展。分子生物计算是当今世界科学前沿的研究热点。分子计算领域研究者的兴趣已从验证性的算法阶段开始走向实际应用,研究通用分子生物计算机是一件刻不容缓的事。设计了面向不同用户的分子生物计算机逻辑层次结构,并以本文提出的面向用户和面向专家的程序设计语言分别求解有向Hamilton路径为例,描述各层的作用和对各类用户的知识要求。在建立一种分子生物计算机物理组成模型之后,对分子生物计算机结构和组成进行了较详细的讨论。从设计角度用HCW结构描述了分子生物计算机系统。这些工作使我们在研究分子生物计算机时,有了不同的研究方向和途径。更多还原

【Abstract】 A new numerical quantification approach for processing vague phenomenon, measure of medium truth scale (MMTS), has been developed. The MMTS is based on medium mathematics, so it is different from fuzzy set and rough set. This approach has been further applied to computer architecture research. The main contents of this dissertation are as follows:After introducing the history of studying vague phenomenon and surveying the main features of several methods of processing it, the goal of our research is explained. The history and progress of computer architecture are also recounted by following the advancement of new ideas and concepts.We then demonstrate that a numerical quantification method for fuzzy negation and truth-value degree is essential for applications. After establishing the standard pointer of the predicate and describing the relation between truth-values of the predicate and numerical value areas of general application, the definition of the function of distance ratio is proposed. From this the individual truth grad function in one-dimensional and n-dimensional is found. Additionally, the MMTS of sets is discussed for both discrete and continuous based on MMTS of individual. With the concept of super-truth presented in this paper, the interval of truth scale can be extended from[0,1] to (-∞,+∞) in theory. Application examples are shown to demonstrate that MMTS is reasonable and effective.After establishing medium standard pointer, fuzzy scale and medium entropy are proposed, and the measure of uncertainty is explored on the basis of medium entropy and random entropy.Because of the indistinct boundary between hardware and software in computer technology, new concepts of software and hardware, that are compatible with existing definitions, are proposed. According to medium concept, medium-ware is defined and its primary performance is discussed with MMTS. The new significance of computer architecture is subsequently revealed. The ratio of performance and cost, which serves as a reference for design evaluation, is proposed. Finally, design principle of medium-ware is discussed.In accordance with de-emphasizing‘maximum difference’between concepts with opposite properties, the idea of weakening polarization and strengthening transition is presented. After a general discussion on transition and polarization, normal numerical quantification mapping and generalized numerical quantification mapping are constructed, and the sufficient condition for converting transition to opposing polarization is established. From this generalized medium-ware concept is defined. To overcome problems in traditional one-dimensional multi-layer structure, the HCW (hierarchical- component-ware) multi-layer structure that describes computer systems in three-dimensions is proposed. We demonstrate that HCW structure adapts to the new situation of computer architecture research, so that it is beneficial to the innovation and development of computer technology.Bio-molecular computation has already been the focus in cutting edge science research. As the interest in molecular computation has already advanced from validating algorithms to practical applications, it is urgent to develop general-purpose bio-molecular computers. A hierarchical structure of bio-molecular computers is designed; the rules of each hierarchical level and the specific requirements for users at different levels are described with the example of solving the Hamilton Path Problem by using the proposed user-oriented and expert-oriented programming languages. After establishing the physical model of bio-molecular computers, the detailed architecture and organization of these computers are discussed,and the bio-molecular computer system is described with HCW structure. The concept of hierarchical structure provides different directions and avenues to study the bio-molecular computer.更多还原