http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
경험적 정보를 이용한 결합자 감축기의 세대별 가베지 수집기 개선
김성배(Sung-Bae Kim),김홍읍(Hong-Eub Kim),이갑래(Gab-Rae Lee),주형석(Hyung-Seok Joo),유원희(Won-Hee Yoo) 한국정보과학회 1995 한국정보과학회 학술발표논문집 Vol.22 No.2B
세대별 가베지 수집기를 이용함으로써 일반적인 복사 방법에 의한 가베지 수집 보다는 실행 프로그램이 중단되는 시간을 줄일 수 있는 장점을 가지지만, 최적의 ‘복사공간’ 크기를 알 수 없어서 ‘복사공간’을 필요 이상으로 크게 운영하므로써 기억장소의 이용 효율이 떨어지는 문제점을 지니고 있다. 본 논문에서는 세대별 가베지 수집기의 이러한 문제점을 개선하여 ‘할당공간’의 크기를 가능한 한 크게 함으로써 기억장소 이용 효율을 향상시킨 가베지 수집 방법을 제안한다. 이와같이 하기 위해서 ‘할당공간’과 ‘복사공간’을 고정된 같은 크기로 분할하지 않고, 할당된 노드와 가베지 수집 후 회수된 기억장소로부터 얻어지는 경험적 정보를 이용하여 다음 단계의 ‘할당공간’과 ‘복사공간’의 크기를 가변적으로 운영하는 알고리즘과 실험 결과를 보인다.
주형석(Joo Hyung Seok),김홍읍(Kim Hong Eub),유원희(Yoo Weon Hee) 한국정보처리학회 1997 정보처리학회논문지 Vol.4 No.3
Despite elegant semantics and a lot of features, pure functional programming language do not provide an efficient way of representing states. Many researches have been done to resolve the problem, however, another problem arises that it is hard to implement because of the complex type system and reduction rule. Therefore, the scheme which simplifies the reduction rule and maintains states efficiently needed to have the implementation effective. This paper proposes st-calculus, the execution model of a functional language with states, and proves that the proposed model satisfies the Church-Rosser theorem. It has simple reduction rules and the ability of representing states without compromizing the properties of pure functional languages. The expressiveness can be increased through this model, and the difficulties with implementation may be reduced by simplifying the reduction rules.
비평가인자 함수언어의 다중스레드 코드 생성을 위한 생성기의 설계
김홍읍,이형렬 김천과학대학 1998 김천과학대학 논문집 Vol.24 No.-
Non-strict functional language with implicit parallelism needs dynamic scheduling and synchronization, because of non-strict's mechanism. Multithreaded model is efficient parallel processing model to solve latency time, dynamic scheduling, synchronization problemby combination of dataflow mode and von Neumann model. This model's efficiency is determined by construcion of thread and source language. This paper's generator translates kernel language, intermediate code, from non-strict functional language with implicit parallelism. Now and then generator translates multithreaded code from kernel language through dataflow graph.
지연어의 고계함수 프로그램의 병렬성 향상을 위한 스트릭트성 분석
김홍읍 김천과학대학 2003 김천과학대학 논문집 Vol.29 No.-
In this work, we propose the method of strictness analysis for the lazy evaluation, which is performed during the compilation time in order for the parallel analysis of high-order function. To analyze the strictness of high-order function, the application of the proper function as well as the definition of function should be characterized. Therefore, the accurate and strict information, which contain the definition of first-order function, high-order function, and recursive function and its characteristics, can be obtained by analyzing the strictness as a function of characteristic of each subsidiary function, affecting the strictness of the arbitrary function.
수퍼컴비네이터 수행모델을 위한 리덕션 규칙 및 리덕션 시스템의 모형
김홍읍 김천과학대학 2006 김천과학대학 논문집 Vol.32 No.-
SK combinator's generation and reduction rule is simple. But combinator code to be created from functional expression is large and complex. Therefore, it is proposed that supercombinator's target code is smaller than SK combinator. This paper design a prototype for execution model on supercombinator code to be generated from Lambda expressions.
함수형 언어의 다중스레드 모델에서의 프르세서 통신 메카니즘의 개선
김홍읍 김천과학대학 1997 김천과학대학 논문집 Vol.23 No.-
Multi-threaded architecture solves data flow architecture's problems using thread to have a advantage of data flow architecture. Multi-threaded is executed by message passing mechanism and uses the thread scheduling based on frame. In order to solve the problem using frame and message processing in traditional multi-threaded architecture, this paper proposes the method to separate message into inner/outter message and the method to store thread scheduling information into thread scheduling table instead of frame memory.
김홍읍 김천과학대학 2006 김천과학대학 논문집 Vol.32 No.-
Despite elegant semantics and a lot of features, pure functional programming languages do not provide an efficient way of representing states. Many researches have been done to resolve the problem however, another problem arises that it is hard to implement because of the complex type system and reduction rule. Threrefore, the scheme which simplifies the reduction rule and maintains states efficiently is needed to have the implementation effective. This paper proposes A^-calculus, the execution model of a functional language with states, and proves that the proposed model satisfies the Church-Rosser theorem. It has simple reduction rules and the ability of representing states without compromizing the properties of pure functional languages. The expressiveness can be increased through this model, and the difficulties with implementation may be reduced by simplifying the reduction rules.
이형렬,김홍읍 김천과학대학 2002 김천과학대학 논문집 Vol.28 No.-
Information representing the thickness of the original lines from the thinning results can be applied efficiently in order to implement the automated vectorizing system. This paper propose a preprocessing scheme of thinning which can show the information of the original lines' thickness of the thinning result. In the proposed scheme, the depth of each pixel constructing the lines of map was calculated, which was represented by the number of layers composed of neighboring pixels surrounding the original pixel. And then the original lines' thickness could be recognized through the depth information of the skeleton from the thinning results. We implemented the proposed scheme and experimented on a contour map. Using the depth information of the skeleton, we could easily distinguish each line of the contour either an intermediate or an index contour.