• 대한수학회보
• /
• 제54권2호
• /
• pp.667-677
• /
• 2017

Let $F{\subseteq}{\mathbb{P}}^3$ be a smooth surface of degree $3{\leq}d{\leq}9$ whose equation can be expressed as either the determinant of a $d{\times}d$ matrix of linear forms, or the pfaffian of a $(2d){\times}(2d)$ matrix of linear forms. In this paper we show that F supports families of dimension p of pairwise non-isomorphic, indecomposable, Ulrich bundles for arbitrary large p.

• Journal of Magnetics
• /
• 제17권2호
• /
• pp.153-157
• /
• 2012

This study evaluated the ability of Diffusion-Weight Image (DWI), which is one of pulse sequences used in MRI based on the T2 weighted images, to detect samples placed within phantoms according to their size. Two identically sized phantoms, which could be inserted into the breast coil bilaterally, were prepared. Five samples with different sizes were placed in the phantoms, and the T2 weighted images and DWI were obtained. The Breast 2 channel coil of SIEMENS MAGNETOM Avanto 1.5 Tesla equipment was used for the experiments. 2D T2 weighted images were obtained using the following parameters: TR/TE = 6700/74 msec, Thickness/gap = 5/1 mm, Inversion Time (TI) = 130 ms, and matrix = $224{\times}448$. The parameters of DWI were that TR/TE = 8100/90 msec, Thickness/gap = 5/1 mm, matrix = $128{\times}128$, Inversion Time = 185 ms, and b-value = 0, 100, 300, 600, 1000 s/mm. The ratio of the sample volume on DWI compared to the T2 weighted images, which show excellent ability to detect lesions on MR images, was presented as the mean b-value. The measured b-value of the samples was obtained: 0.5${\times}$0.5 cm=0.33/0.34 square ${\times}$ cm (103%), 1${\times}$1 cm=1.28/1.25 square ${\times}$ cm (102.4%), 1.5${\times}$1.5 cm = 2.28/2.67 square ${\times}$ cm (85.39%), 2${\times}$2 cm=3.56/4.08 square ${\times}$ cm (87.25%), and 2.5${\times}$2.5 cm=7.53/8.77 square ${\times}$ cm (85.86%). In conclusion, the detection ability by the size of a sample was measured to be over 85% compared to T2 weighted image, but the detection ability of DWI was relatively lower than that of T2 weighted image.

• 대한임베디드공학회논문지
• /
• 제17권6호
• /
• pp.367-374
• /
• 2022

The acceleration of neural networks has become an important topic in the field of computer vision. An accelerator is absolutely necessary for accelerating the lightweight model. Most accelerator-supported operators focused on direct convolution operations. If the accelerator does not provide GEMM operation, it is mostly replaced by CPU operation. In this paper, we proposed an optimization technique for 2-stage tiling-based GEMM routines on VTA. We improved performance of the matrix multiplication routine by maximizing the reusability of the input matrix and optimizing the operation pipelining. In addition, we applied the proposed technique to the DarkNet framework to check the performance improvement of the matrix multiplication routine. The proposed GEMM method showed a performance improvement of more than 2.4 times compared to the non-optimized GEMM method. The inference performance of our DarkNet framework has also improved by at least 2.3 times.

• Journal of Radiation Protection and Research
• /
• 제22권2호
• /
• pp.127-133
• /
• 1997

2차원 방사선 측정시스템에서 측정된 감마선의 에너지 스펙트럼으로부터 실제 에너지 스펙트럼을 얻기 위하여 CsI(T1) 섬광계수기에 대한 EGS4 모의실험을 실시하였다. $0.1{\sim}2$ MeV의 감마선에 대하여 50 keV 간격으로 모의실험을 실시하여 $39{\times}39$ response matrix를 계산하였고 이에 대한 역행렬을 구하였다. 이를 이용하여 실제 측정된 Co-60 에너지 스펙트럼에 대하여 unfolding을 실시하였다. 그 결과 계산된 response matrix가 실제 스펙트럼 분석에 유용한 것으로 판단되었다.

• 한국컴퓨터정보학회논문지
• /
• 제17권9호
• /
• pp.9-16
• /
• 2012

본 논문에서는 유한요소법에서 희소행렬의 효율적인 저장을 위한 2차원 가변길이 벡터 저장구조를 제안한다. 제안한 저장구조는 유한요소 전체 방정식의 거대희소행렬 $N{\times}N$ 대신, 전체 행의 개수 N의 상삼각행렬에서 0이 아닌 실제 필요한 값들만 2차원 가변길이 벡터를 이용하여 저장하는 방법이다. 이 방법을 이용하면, 해석대상의 2차원 격자구조에서는 각 절점당 최소 1개에서 최대 5개까지의 저장 공간이 필요하게 되고, 3차원 격자구조에서는 각 절점당 최소 1개에서 최대 14개까지의 저장 공간이 필요하게 된다. 인덱스를 포함해도 2배 이상을 넘지 않는다. 본 논문의 실험 결과에 의해, 제안한 저장구조는 총 절점 개수가 많아질수록 기존의 최대칼럼 높이를 저장하는 스카이 라인 저장구조보다 메모리 공간을 효과적으로 줄일 수 있는 구조임을 알 수 있었다.

• 한국데이타베이스학회:학술대회논문집
• /
• 한국데이타베이스학회 2001년도 추계 컨퍼런스: 인터넷 비즈니스 환경에서의 디지털 컨텐츠 기술 발전 및 활용을 위한 컨퍼런스
• /
• pp.567-581
• /
• 2001

인터넷이 일반화된 지 불과 10년이 못되었지만, 기업뿐만 아니라 우리 사회 전반에 걸친 패러다임을 변화시키고 있다. 이와 같이 무선인터넷을 기반으로 한 모바일 비즈니스에 대한 관심은 높지만, 기존의 문헌들은 무선인터넷 서비스 수요, 사용자 만족도 등에 관한 연구가 주를 이루고 있다. 즉 모바일 비즈니스 모델을 통한 무선인터넷 서비스의 향후 진화방향에 관한 연구는 전무한 실정이다. 본 연구의 목적은 문헌고찰을 통하여 모바일 비즈니스의 모델들을 정리하고, 오재인(2000)의 The 2$\times$2 Matrix를 분석 툴을 활용하여 모바일 서비스별 현재 수요와 향후 수요를 분석함으로써 모바일 비즈니스의 진화방향을 실증적으로 규명하는 것이다. 1540명의 회수된 설문지 중 유효표본 786개를 대상으로 통계분석을 실시하여 분석된 주요 연구결과는 다음과 같다. 첫째, 모바일 서비스들을 The 2$\times$2 Matrix상에 매핑한 결과 예측에는 별다른 차이가 없는 것으로 나타났다. 둘째, 전체응답자, 사용자, 비사용자들의 모바일 서비스에 대한 순위는 The 2$\times$2 Matrix에서 Mobile Hub, Mobile Trade, Mobile Care, Mobile Support 순으로 나타났다. 셋째, 현재와 향후 각 서비스들 간에는 차이가 있고, 현재 모바일 서비스들 각각과 향후 모바일 서비스들 각각에 대하여 서비스들 순위에는 차이가 없는 것으로 나타났다. 넷째, 무선인터넷의 전반적인 만족도에 영향을 미치는 요인은 우수하고 독특한 컨텐츠, 패킷 과금으로 저렴한 서비스를 제공하는 Cost전략 등으로 나타났다.

• 대한수학회지
• /
• 제52권3호
• /
• pp.625-636
• /
• 2015

The Boolean rank of a nonzero $m{\times}n$ Boolean matrix A is the least integer k such that there are an $m{\times}k$ Boolean matrix B and a $k{\times}n$ Boolean matrix C with A = BC. We investigate the structure of linear transformations T : $\mathbb{M}_{m,n}{\rightarrow}\mathbb{M}_{p,q}$ which preserve Boolean rank. We also show that if a linear transformation preserves the set of Boolean rank 1 matrices and the set of Boolean rank k matrices for any k, $2{\leq}k{\leq}$ min{m, n} (or if T strongly preserves the set of Boolean rank 1 matrices), then T preserves all Boolean ranks.

• 정보통신설비학회논문지
• /
• 제10권1호
• /
• pp.7-13
• /
• 2011

Matrix multiplication is a fundamental operation of linear algebra and arises in many areas of science and engineering. This paper introduces an efficient parallel matrix multiplication scheme on N ${\times}$ N mesh-connected SIMD array processor, called multiple hierarchical SIMD architecture (HMSA). The architectural characteristic of HMSA is the hierarchically structured control units which consist of a global control unit, N local control units configured diagonally, and $N^2$ processing elements (PEs) arranged in an N ${\times}$ N array. PEs are communicating through local buses connecting four adjacent neighbor PEs in mesh-torus networks and global buses running across the rows and columns called horizontal buses and vertical buses, respectively. This architecture enables HMSA to have the features of diagonally indexed concurrent broadcast and the accessibility to either rows (row control mode) or columns (column control mode) of 2D array PEs alternately. An algorithmic mapping method is used for performance evaluation by mapping matrix multiplication on the proposed architecture. The asymptotic time complexities of them are evaluated and the result shows that paralle matrix multiplication on HMSA can provide significant performance improvement.

• 전기학회논문지P
• /
• 제58권1호
• /
• pp.85-89
• /
• 2009

We investigated the quench characteristics in accordance with increase of turns number of trigger coil and shunt resistance of matrix-type superconducting fault current limiter (SFCL) with $2{\times}3$ array. The matrix-type SFCL consists of the trigger part to apply magnetic field and the current-limiting part to limit fault current. The fault current limiting characteristics according to the increase of magnetic field and applied voltage were nearly same. This is because the application of magnetic field hasn't an affect on total impedance of the SFCL. When turns number of a reactor increased, the voltage difference between two superconducting units in the current-limiting part according was decreased. The resistance difference generated in two superconducting units was also decreased. Therefore, we confirmed that the differences of the critical behaviors between superconducting units were reduced by application of magnetic field. By this results, we could decide the optimum turns number of reactor to apply magnetic field.

• Journal of Biosystems Engineering
• /
• 제32권6호
• /
• pp.389-394
• /
• 2007

The power output of the stirling engine is influenced by the regenerator effectiveness. The regenerator effectiveness is influenced by heat transfer and flow friction loss of the regenerator matrix. In this paper, in order to provide basic data for the design of the regenerator matrix, characteristics of working fluid velocities were investigated by a packed method of matrix in the oscillating flow as the same condition of operation in a Stirling engine. As matrices, two different wire screens were used. The results are summarized as follows; 1. When a regenerator is not filled with any wire screen, working fluid velocity of the oscillating flow shows 1.3 times faster than that of one directional flow. 2. When a regenerator is filled with the wire screen of No.50, working fluid velocity of the oscillating flow reveals 2.5 times faster than that of one directional flow. 3. When a regenerator is filled with the wire screen of No. 100, working fluid velocity of the oscillating flow shows 2 times faster than that of one directional flow, regardless of the number of packed wire screens. 4. Working fluid velocity is decreased wire the increase in number of meshes and packed wire screens.