关于2~2~k值n元逻辑函数的几何图象模型与图象分析初探

本文建立了由2值向2~(2~k)值(k∈{0,1,2…})拓广的多值n维逻辑空间几何图象模型及其图象分析基础。主要内容有:(1)2~(2~k)值n维逻辑空间的结构与几何图象模型;(2)2~(2~k)值n元逻辑函数、范式及0—1方程的图象分析;(3)用图象分析法求2~(2~k)值n元逻辑方程的再生解。     

单体型组装问题MEC/GI模型的参数化算法

根据DNA测序片段数据的特点,提出了一个时间复杂度为0（nk2^2k2＋mlogm＋mk1）的单体型组装问题MEC／GI模型的参数化算法,其中m为片段数,n为单体型的SNP位点数,k1为一个片段覆盖的最大SNP位点数（通常小于10）,k2为覆盖同一SNP位点的片段的最大数（通常不大于10）。对于实际DNA测序中的片段数据,即使m和n都相当大,该算法也可以在较短的时间得到MEC／GI模型的精确解,具有良好的可扩展性和较高的实用价值。     

基于参数化最小割方法的SAR图像区域分割

由于SAR图像中存在大量的相干斑噪声,最小割准则(Minimum Cut Criterion)倾向于分割出小的孤立点集,因此最小割准则难以直接应用于SAR图像分割中。本文给出了一种可以根据需要来控制分割区域大小的参数化最小割准则(Parametric Minimum Cut Criterion),由此减少小的孤立点集的产生。该准则通过构建参数化割树(Parametric Cut Tree)得到理论最优解。为了参数化最小割准则更有效地应用于SAR图像分割,结合了区域分割方法中的四叉树分裂合并技术,给出了一种基于图论的SAR图像区域分割算法。在利用四叉树技术分裂图像时,根据SAR图像的统计特性,给出了一种新的区域一致性度量,有效地抑制了相干斑噪声的影响。实验结果表明了本文方法的有效性。     

一种新型快速收敛的LMS谐波检测算法

针对传统基于固定步长LMS算法在电网谐波检测中存在收敛速度与稳态精度需折中选择的问题,提出了一种新型具有较高收敛速度的改进变步长LMS算法。该算法以同一相位下相邻两时刻的误差信号e(n)、e(n－1)的自相关估计调节步长更新,并且采用归一化处理方式,以误差信号在总电流信号中的比例k(n)、k(n－1)作为新反馈量,同时对新的反馈量进行相干平均估计来调整步长迭代;在权值迭代公式中引入相邻时刻估计误差绝对值之差的扰动量来加快自适应滤波器权矢量的迭代速度。MATLAB/Simulink仿真和实验证明,该方法相对于传统固定步长LMS谐波检测算法在收敛速度和稳态精度上有了进一步的提高,尤其在负载发生突变时的跟踪能力。     

Cesàro平均的带权强求和

设f(x)∈LP(Ωn),1≤P≤2,δ＞(n-1)(1)/(P)-(1)/(2)),σδN(f)(x)表示f(x)在n维球面Ωn上的Cesàro平均.证明了 limN→∞(1)/(N+1) ΣN)/(k=0|σδk(f)(x)-f(x)| 2ak=0 a.e.x∈Ωn 其中权系数ak＞0,满足 1≤(1)/(N+1)ΣN)/(k=0ak≤A (A是一个绝对常数).     

逻辑空间的结构及图象——兼论逻辑函数化简

本文是关于“逻辑空间图象”理论研究的一个小结.主要内容:1.从解析角度分析了n维逻辑空间及n元逻辑函数的结构性质,论证了二者之间的一一对应关系.其中,得到了m维子空间的集合的基数|α~m|=[n!/((n-m)!m!)]2~(n-m),n维逻辑空间的示性数X(P)=sum from K=0 to n(-1)~k|α~k|=1,以及|α~0|=|γ~0|=2~n,|α~m|=|γ~m|=[n!/((n-m)!m!)]2~(n-m)|α|=sum from m=0 to n|α~m|=|γ|=sum from m=0 to n|γ~m|=3~n等定理.2.根据等价原理,采用“k阶a维子图的m维集合”及“a维子图的k阶m维集合”两种方法,得到了简明的多维逻辑空间图象。3.分析了逻辑函数的化简,得到了二个结果.从而提出了一个化简逻辑函数的普遍程序:“由低维到高维,按次序地提取f中的不重复孤立子空间,直到提完所有的点为止”。     

线性时变离散系统渐近稳定性的充分条件

本文讨论线性时变离散系统 x(k+1)=A(k)x(k),k=0,1,…的零解渐近稳定性,这里x(k)∈C_n,A(k)∈C_(n×n)。 定义 称系统(1)的零解渐近稳定,如果它的任一解x(k)→0(k→∞)。 显然,(1)的零解渐近稳定当且仅当     

一类反应扩散方程组的奇摄动

考虑一类反应扩散方程组定解问题:(j=1,2),其中ε为正的小参数,x=(X_1,X_2,…x_n),Ω为n维欧氏空间中的有界域,为Ω的边界的外法向导数,而     

穿孔区域中非线性波动方程解的存在性与齐次化

本文研究如下非线性波动方程的齐次化问题 uε'div(AεΔuε) g(uε') = fε in Ωε× (0,T),在Ωε× (0,T) 内,这里Ωε为一周期穿孔区域,g为一非降标量函数,我们首先证明该问题解的存在性与唯一性,然后论证该解收敛于其齐次化问题的解。     

随机权网络稀疏正则化算法及其应用

随机权网络是一种有效的前馈神经网络(FNNs),尤其是内权和偏置值的随机选取极大地提高了网络的学习速率,并克服了其他学习算法的一些不足.但是,其在计算外权的过程中也存在着一些不足.我们就此不足提出了一个新的算法——稀疏正则化算法,并结合梯度投影算法给出了一种迭代解,进而提出了相应的参数选择方法和算法终止准则.实验证明所提出的算法的优势,尤其是当隐层神经元数较多和训练样本较大时,所给出的算法具有明显的优势.     

Comparison on Mathematical Models for Description of Flow Curves of Stable Austenitic Steels

The flow curves were measured for the stable austenitic steels 304L and 304LN by meas of tensile test at room temperature,which are described by the models σ=K1ε^n1 exp(K2 n2ε),σ=Kε^n1 n2lnε and σ=σ0 Kε^n(where,K1,K2,n1 and n2:K,n1 and n2;σ0,K and n are constant ).The comparison of the maximum deviations and the consideration of the variation of the work hardening rate with true strain show that the flow curves for the austenitic steels 304L and 304LN can be described by the model σ=Kε^n1 n2lnε at higher precision .The derivatives of the models σ=K1ε^n1 exp(K2 n2ε)and σ=Kε^n1 n2lnε with respect to true strain ,exhibit the extreme at low true strain,This inherent character indicates that both models are unsuitable to describe the part of the work hardening rate curve at low true strain.     

Inconel 625合金高速热变形动态再结晶的临界条件

通过等温热压缩试验获得Inconel625合金在变形温度为1000～1200℃,应变速率为1～80S^-1条件下的真应力-应变曲线,利用加工硬化率,结合lnθ-ε曲线上的拐点判据及-δ（1nθ）／δε-ε曲线上的最小值,来研究Inconel625合金动态再结晶的临界条件。结果表明,在该实验条件下,Inconel625合金的lnθε曲线均出现拐点特征,对应的-δ（lnθ）／δε-ε曲线出现最小值,该最小值处对应的应变即为临界应变;临界应变随应变速率的增大和变形温度的降低而增加,并且临界应变和峰值应变之间有一定的关系,即εc=0．69εp;动态再结晶时临界应变的预测模型可以表示为εc=4．41×10^-4Z^0.14261。     

Inverse maximum capacity problems

Let be a finite set, be a family of subsets of and be a capacity vector for all elements of . For each , define the capacity of as the minimum capacity occurring in . The problem which we discuss in this paper is how to change the vector as little as possible so that a given has the maximum capacity. This model contains inverse maximum capacity spanning tree problem, inverse maximum capacity path problem and etc. as its special cases. We transform the problem into the minimum weight cut set problem and show that this problem can be solved efficiently if an efficient algorithm for finding minimum weight cut set of is available. Received: 19 July 1996 / Accepted: 12 November 1997     

Complete set of material properties of [011](c) poled 0.24Pb(In(1/2)Nb(1/2))O(3)-0.46Pb(Mg(1/3)Nb(2/3))O(3)-0.30PbTiO(3) single crystal

A complete set of elastic, piezoelectric, and dielectric constants of [011](c) poled multidomain 0.24Pb(In(1/2)Nb(1/2))O(3)-0.46Pb(Mg(1/3)Nb(2/3))O(3)-0.30PbTiO(3) ternary single crystal has been determined using resonance and ultrasonic methods and the temperature dependence of the dielectric permittivity has been measured at 3 different frequencies. The experimental results revealed that this [011](c) poled ternary single crystal has very large transverse piezoelectric coefficient d(32) = -1693 pC/N, transverse dielectric constant ε(11)/ε(0) ~ 7400 and a high electromechanical coupling factor k(32) ~ 90%. In addition, its coercive field is 2 times of that of the corresponding binary 0.7Pb(Mg(1/3)Nb(2/3))O(3)-0.30PbTiO(3) single system with much better temperature stability. Therefore, the crystal is an excellent candidate for transverse mode electromechanical devices.     

k-supplier问题的贪婪近似算法

给定无向完全图G=(V,E)和正整数k,图G的顶点集V被划分为子集F和子集D=V-F.k-supplier问题主要研究如何寻找F中顶点数不多于k的子集S,使得S中的顶点到D中顶点的最大距离最小.研究了k-supplier问题,得到了一个近似比为3的多项式时间贪婪近似算法,并通过实例验证了该算法的有效性.     

A simple MC-based algorithm for evaluating reliability of stochastic-flow network with unreliable nodes

A MP/minimal cutset (MC) is a path/cut set such that if any edge is removed from this path/cut set, then the remaining set is no longer a path/cut set. An intuitive method is proposed to evaluate the reliability in terms of MCs in a stochastic-flow network subject to both edge and node failures under the condition that all of the MCs are given in advance. This is an extension of the best of known algorithms for solving the d-MC (a special MC but formatted in a system-state vector, where d is the lower bound points of the system capacity level) problem from the stochastic-flow network without unreliable nodes to with unreliable nodes by introducing some simple concepts. These concepts were first developed in the literature to implement the proposed algorithm to reduce the number of d-MC candidates. This method is more efficient than the best of known existing algorithms regardless if the network has or does not have unreliable nodes. Two examples are illustrated to show how the reliability is determined using the proposed algorithm in the network with or without unreliable nodes. The computational complexity of the proposed algorithm is analyzed and compared with the existing methods.     

BATCH SEQUENCING

Consider the single machine scheduling problem where there are a number of part types to be processed. A part type is defined as follows: Two parts are of the same part type if the machine does not require a setup in between the processing of these parts. The problem investigated in this paper is to find a sequence of batches of parts (if there are any) where all the requirements for parts are met. A heuristic and an exact algorithm are developed, and computational analysis is performed to measure the performance of the heuristic. The time complexity function of the heuristic is O(n²), and the exact algorithm runs in polynomial time given a fixed upper bound on the number of setups.     

Construction of roadmaps in semi-algebraic sets

In this paper we describe an algorithm of construction of a roadmap for a compact semi-algebraic setS Rⁿ, which is similar to the algorithm of [3], but which is simpler in the sense that it does not need the use of Whitney stratifications, and more general because it accepts as input any compact semi-algebraic set. The complexity of this algorithm is measured in terms of the numberk of polynomials definingS, their maximum degreed, and the number of variablesn. With respect to those measures, our algorithm runs in time . The goal of this paper is not a strengthening of the previous results (similar bounds of complexity are obtained in [5] and [13], even for a non-compact set), but more a new approach, and we hope the efficiency of the algorithm.