Banach embedding properties of non-commutative L^p-spaces

合集下载

Banach空间中渐近非扩张映射的强收敛定理

Banach空间中渐近非扩张映射的强收敛定理

Banach空间中渐近非扩张映射的强收敛定理不动点理论是泛函分析的一个重要的研究分支,它在微分方程、积分方程、数值分析、对策论、控制论以及最优化等学科中有广泛而深入的应用.不动点理论的研究起源于Banach,Banach给出了第一个不动点定理,即Banach压缩映射原理.Browder利用Banach压缩映射原理在Hilbert空间中证明了非扩张映射的不动点存在性定理.Browder定理被Reich推广至一致光滑的Banach空间中.Kirk 在具有一致正规结构的Banach空间中证明了非扩张映射的不动点存在性定理.Goebel和Kirk首先提出渐近非扩张映射,并证明了一致凸Banach空间中非空有界闭凸子集上的每个渐近非扩张映射都有不动点.Kim和Xu将该结果推广至空间具有一致正规结构的情形.2002年,Li和Sims证明了在具有一致正规结构的Banach空间中渐近非扩张型映射在适当条件下具有不动点:设E是一个具有一致正规结构的Banach空间, C是E的一个非空有界子集, T :C→C是渐近非扩张型映射且T在C上连续,若C存在非空闭凸子集K具有性质: ( )z∈K ?ωwz ? K,则T在K中具有不动点.在这些定理证明中,都是利用压缩映射的不动点直接逼近或迭代逼近非扩张映射的不动点.1998年,Shioji和Takahashi给出了Hilbert 空间中非扩张半群的隐式粘性平均迭代序列的强收敛定理.Shimizu和Takahashi在Hilbert空间中证明了非扩张半群的显式粘性平均迭代序列是强收敛的.2007年,Chen和Song研究了具有一致Gateaux可微范数的一致凸Banach 空间中的非扩张半群的隐式粘性平均迭代和显式粘性平均迭代的收敛性问题.本文主要利用Li和Sims的不动点存在性定理,研究了在具有一致Gateaux可微范数与一致正规结构的Banach空间中,渐近非扩张映射及渐近非扩张半群的粘性隐式迭代序列{ }z n和粘性显式迭代序列{ }xn的收敛性问题.在第二章中,本文研究了在具有一致Gateaux可微范数与一致正规结构的Banach空间中,由下式定义的粘性迭代序列{ }z n和{ }xn :其中f∈ΠK, K是E的非空闭凸子集, T :K →K是渐近非扩张映射且F (T )≠? ,都收敛于T的不动点p ,且p是变分不等式0的唯一解.在第三章中,本文研究了在具有一致Gateaux可微范数与一致正规结构的Banach空间中,由下式定义的粘性迭代序列{ }z n和{ }xn :其中f∈ΠK, K是E的非空闭凸子集, ?= {T (t ), t≥0}是K上渐近非扩张半群且证明了{ }z n和{ }xn都强收敛于?的公共不动点p ,且p是变分不等式本文的主要结果推广和改进了文[9,10]中的结果.。

Banach空间中有限族严格伪压缩映像隐迭代序列的收敛性问题

Banach空间中有限族严格伪压缩映像隐迭代序列的收敛性问题

定义 1 设 E为一个实 B nc 间 , . 2 aah空 是 的一个非 空 闭凸子集 , 自映像
实数 L 0 V YE > , , K有
K称为 L Lpci 的 , - i ht 如果存在一 s z l I .
公共不动点. 最近 . 0O ikt 他们 的结果从非扩 张映像扩展到严 M..sie l  ̄
【 e od] oepni ;tcy s dcnlte ap g; anir i oes i r r K y rsN nxas eSi e oot cv pi sM av p cs s t eo w v rt p u l Iim n a n t te r e w h rs e
0 引言和预 备知 识
T eC n e g n eo mp i tI r t eP o esf raF n t a l f tit s u o o ta t eMa pn si a a hS a e h o v r e c fI l i t ai c e v r c s o ii eF mi yo S rc yP e d c n r ci l v p i g nB n c p cs
要】 本文主要通过构造有限族严格伪压缩映像具有误差的 Man型迭代序列来研 究在 B nc n a ah空间框 架下的有 限族严格
伪 压 缩 映像 的 隐迭代 序 列的 收敛 性 问题 . 文结果 是 一 些作 者早 期 与最 近的 相应 结 果 的改进 与 推广 . 本
【 关键词】 非扩张映像 ; 严格伪 压缩映像 ; 具有误 差的 M n 迭代序列 an
a e c n i e e n B n c p c s b i ig t e Ma n i r t e p o e s swi ro s h e u t p e e td i h s p p r e t n n r o s rd i a a h s a e y g vn h n t a i r c s e t e r r.T e r s l rs n e n t i a e x e d a d d e v h s

Banach空间中广义平衡问题和一族相对非扩张映象的强收敛定理

Banach空间中广义平衡问题和一族相对非扩张映象的强收敛定理
题、 变分不 等式 问题 、 最大最小 问题等 [ 2 13 -. 我 们称 S: C— c 一非扩张映象 , 是 如果 _ 一勖 l I l s l I ≤ x一训 ,xY∈C. V,
20 0 7年 , a a和 T l hL i一 ] T k h si Td a a £h[ ( s 3 及 aa ah S和 T k h s i] 了 Hi let空间 中广义平 a a ah[得到 5 e br r 衡 问题 和相对非扩 张映象 的弱 收敛定理和强 收敛定理 . 特别地 , a a和 T n ah[ 用 N k j Td k h s i】 a ao和

族相对非扩张映象的强收敛定理
蘑金芳
宜宾 学 院 数 学 系, 宜宾 6 4 0 407
摘要:在 Ba a h空间中, nc 用收缩投影的方法证 明了广义平衡问题和一族相对非扩张映象的公共不动点的
强收敛定理.结论改进了最近一些人的研究结果 关键词:广义平衡问题; 一族相对非扩张映象; 不动点问题

∈ E
收稿 日期: 0 90-5 2 0—51 作者简介: 唐金芳 (9 9 , 四川省仁寿县人, - i j fnt7 ̄13cm 17一)女, E mal i a g.9 6 . :n o
20 6
应 用 泛 函 分 析 学 报
第 l 卷 2
定义广义投影I : — C : I E
nG ()
) ∈E , V
称映象 : — E C 是 O L 一逆强单调的, 如果存在常数 >0 使得对 V , , x Y∈C,z—Y A ( , x—A ) y≥
Ol x—A l. LA l yl 引理 21 一】 .[ 。设 巳光滑 的严格 凸的 白反的 B nc aah空间, 是 E的非 空闭凸子集 , 则对 V x∈

堆垛层错

堆垛层错

The stacking fault
FCC crystal structure, close-packed plane stacking sequence of ABCABC ... ... with "Δ" said AB, BC, CA ... ... order to use "▽" said the expressed a contrary order, that is, BA, CB, AC ... ..., The FCC stacking sequence of the normal ΔΔΔ ... ..., HCP for Δ ▽ Δ ▽ ... ... If taken out of the FCC layer of C, then the ABCAB ↓ ABCABC. ΔΔΔΔ ▽ ΔΔΔΔΔ Insert a layer of A, then A B C A B ↓ A ↓ C A B C ΔΔΔΔ ▽ ▽ △ △ △ At "↓" stacking sequence of localized disorder, FCC crystal stacking faults can be seen as embedded in a thin layer close-packed hexagonal structure.
What is the stacking fault ?
1 We produced a stacking fault because the stacking sequence ABCABCA..., has been changed to the faulty sequence ABCABABCA... 2 Stacking faults by themselves are simple two-dimensional defects. They carry a certain stacking fault energy g; very roughly around a few 100 mJ/m2.

CAS中ABA问题的解决

CAS中ABA问题的解决

CAS中ABA问题的解决转⾃()在运⽤CAS做Lock-Free操作中有⼀个经典的ABA问题:线程1准备⽤CAS将变量的值由A替换为B,在此之前,线程2将变量的值由A替换为C,⼜由C替换为A,然后线程1执⾏CAS时发现变量的值仍然为A,所以CAS成功。

但实际上这时的现场已经和最初不同了,尽管CAS成功,但可能存在潜藏的问题,例如下⾯的例⼦:现有⼀个⽤单向链表实现的堆栈,栈顶为A,这时线程T1已经知道A.next为B,然后希望⽤CAS将栈顶替换为B:pareAndSet(A,B);在T1执⾏上⾯这条指令之前,线程T2介⼊,将A、B出栈,再pushD、C、A,此时堆栈结构如下图,⽽对象B此时处于游离状态:此时轮到线程T1执⾏CAS操作,检测发现栈顶仍为A,所以CAS成功,栈顶变为B,但实际上B.next为null,所以此时的情况变为:其中堆栈中只有B⼀个元素,C和D组成的链表不再存在于堆栈中,平⽩⽆故就把C、D丢掉了。

以上就是由于ABA问题带来的隐患,各种乐观锁的实现中通常都会⽤版本戳version来对记录或对象标记,避免并发操作带来的问题,在Java中,AtomicStampedReference<E>也实现了这个作⽤,它通过包装[E,Integer]的元组来对对象标记版本戳stamp,从⽽避免ABA问题,例如下⾯的代码分别⽤AtomicInteger和AtomicStampedReference来对初始值为100的原⼦整型变量进⾏更新,AtomicInteger会成功执⾏CAS操作,⽽加上版本戳的AtomicStampedReference对于ABA问题会执⾏CAS失败:public class Test {private static AtomicInteger atomicInt = new AtomicInteger(100);private static AtomicStampedReference atomicStampedRef = new AtomicStampedReference(100, 0);public static void main(String[] args) throws InterruptedException {Thread intT1 = new Thread(new Runnable() {@Overridepublic void run() {pareAndSet(100, 101);pareAndSet(101, 100);}});Thread intT2 = new Thread(new Runnable() {@Overridepublic void run() {try {TimeUnit.SECONDS.sleep(1);} catch (InterruptedException e) {}boolean c3 = pareAndSet(100, 101);System.out.println(c3); // true}});intT1.start();intT2.start();intT1.join();intT2.join();Thread refT1 = new Thread(new Runnable() {@Overridepublic void run()try {TimeUnit.SECONDS.sleep(1);} catch (InterruptedException e) {}pareAndSet(100, 101, atomicStampedRef.getStamp(), atomicStampedRef.getStamp() + 1); pareAndSet(101, 100, atomicStampedRef.getStamp(), atomicStampedRef.getStamp() + 1); }});Thread refT2 = new Thread(new Runnable() {@Overridepublic void run() {int stamp = atomicStampedRef.getStamp();try {TimeUnit.SECONDS.sleep(2);} catch (InterruptedException e) {}boolean c3 = pareAndSet(100, 101, stamp, stamp + 1);System.out.println(c3); // false}});refT1.start();refT2.start();}PS: AtomicStampedReference解决了CAS的ABA问题,其实是在出现ABA问题的时候执⾏不成功,⽽不是⾃动纠正这个问题.。

代数英语

代数英语

(0,2) 插值||(0,2) interpolation0#||zero-sharp; 读作零井或零开。

0+||zero-dagger; 读作零正。

1-因子||1-factor3-流形||3-manifold; 又称“三维流形”。

AIC准则||AIC criterion, Akaike information criterionAp 权||Ap-weightA稳定性||A-stability, absolute stabilityA最优设计||A-optimal designBCH 码||BCH code, Bose-Chaudhuri-Hocquenghem codeBIC准则||BIC criterion, Bayesian modification of the AICBMOA函数||analytic function of bounded mean oscillation; 全称“有界平均振动解析函数”。

BMO鞅||BMO martingaleBSD猜想||Birch and Swinnerton-Dyer conjecture; 全称“伯奇与斯温纳顿-戴尔猜想”。

B样条||B-splineC*代数||C*-algebra; 读作“C星代数”。

C0 类函数||function of class C0; 又称“连续函数类”。

CA T准则||CAT criterion, criterion for autoregressiveCM域||CM fieldCN 群||CN-groupCW 复形的同调||homology of CW complexCW复形||CW complexCW复形的同伦群||homotopy group of CW complexesCW剖分||CW decompositionCn 类函数||function of class Cn; 又称“n次连续可微函数类”。

Cp统计量||Cp-statisticC。

óòó×ùò× òDóòòèóT××òùòòDDòòY×ò ×óDDùòùDóòùòóò ìàáèìá

óòó×ùò× òDóòòèóT××òùòòDDòòY×ò ×óDDùòùDóòùòóò ìàáèìá

1 Introduction
During the doping process impurity atoms of higher or lower chemical valence as silicon are introduced into a silicon layer to in uence its electrical properties. Such dopants penetrate under high temperatures, usually around 10000C, with the so-called pair di usion mechanism into the (homogeneous) layer. A precise description of the process can be found in 1], 2], 6], 12] and in the literature cited therein. Usually, dopant atoms (A) occupy substitutional sites in the silicon crystal lattice, loosing (donors, such as Arsenic and Phosphorus) or gaining (acceptors, such as Boron) by this an electron. The dopants move by interacting with native point defects called interstitials (I) and vacancies (V). Interstitials are silicon atoms which are not placed on a lattice site and move through the crystal unconstrained, and vacancies are empty lattice sites. Both can form mobile pairs with dopant atoms(AI,AV), while the unpaired dopants are immobile. The formation and decay of such pairs as well as the recombination of defects cause a movement of the dopants. These interactions can be modelled in terms of chemical reactions. The resulting nonlinear model contains a set of reaction{drift{di usion equations for the point defects and pairs, reaction equations for the immobile dopants and a Poisson equation for the electrostatic potential, which may not be neglected if the doping concentrations are high, the situation we are concerned with here.

堆叠自动编码器的特征选择方法(八)

堆叠自动编码器的特征选择方法(八)

堆叠自动编码器的特征选择方法自动编码器是一种无监督学习算法,可以用于特征提取和数据降维。

堆叠自动编码器是由多个自动编码器组成的深度神经网络,通过层层训练来学习数据的高级抽象特征。

在实际应用中,特征选择是非常重要的,可以帮助我们减少数据维度,提高模型效率和预测准确率。

本文将探讨堆叠自动编码器的特征选择方法。

一、特征选择的意义特征选择是指从原始数据中选择出最具代表性的特征,排除冗余和噪声,以提高模型的泛化能力和预测性能。

在实际应用中,原始数据往往包含大量特征,而其中只有一部分特征对模型的训练和预测起到关键作用。

因此,通过特征选择可以减少数据维度,提高模型训练效率,降低过拟合的风险。

二、堆叠自动编码器的特征提取堆叠自动编码器是一种深度神经网络,可以用于学习数据的高级抽象特征。

在训练过程中,每一层自动编码器都能够学习数据的不同层次的特征表示,从而实现数据的逐层抽象和提取。

通过堆叠多个自动编码器,可以逐渐提取出数据的深层次特征,这些特征对于区分不同类别的数据具有很强的区分能力。

三、基于重构误差的特征选择方法在堆叠自动编码器中,每个自动编码器的训练都是通过最小化重构误差来实现的。

重构误差指的是输入数据与自动编码器的重构输出之间的差异,通过最小化重构误差,可以有效地学习数据的抽象特征表示。

因此,可以基于重构误差来进行特征选择,具体方法是通过分析每个特征对于重构误差的贡献程度,选择对重构误差影响较大的特征作为最终的特征集合。

四、基于梯度下降的特征选择方法除了基于重构误差的方法,还可以利用梯度下降的方法来进行特征选择。

在堆叠自动编码器的训练过程中,可以通过计算每个特征对于损失函数的梯度,来评估每个特征对于模型训练的重要性。

通过梯度下降的方法,可以筛选出对于损失函数梯度影响较大的特征,从而实现特征选择的目的。

五、正则化方法的特征选择在堆叠自动编码器的训练过程中,可以通过正则化方法来进行特征选择。

正则化方法可以通过添加惩罚项来约束模型的复杂度,从而实现对特征的选择和筛选。

Joint_Slot_Scheduling_and_Power_Allocation_in_Clus

Joint_Slot_Scheduling_and_Power_Allocation_in_Clus

LetterJoint Slot Scheduling and Power Allocation in ClusteredUnderwater Acoustic Sensor NetworksZhi-Xin Liu, Xiao-Cao Jin, Yuan-Ai Xie, and Yi YangDear Editor,This letter deals with the joint slot scheduling and power alloca-tion in clustered underwater acoustic sensor networks (UASNs),based on the known clustering and routing information, to maximize the network’s energy efficiency (EE). Based on the block coordi-nated decent (BCD) method, the formulated mixed-integer non-con-vex problem is alternatively optimized by leveraging the Kuhn-Munkres algorithm, the Dinkelbach’s method and the successive con-vex approximation (SCA) technique. Numerical results show that the proposed scheme has a better performance in maximizing EE com-pared to the separate optimization methods.Recently, the interest in the research and development of underwa-ter medium access control (MAC) protocol is growing due to its potentially large impact on the network throughput. However, the focus of many previous works is at the MAC layer only, which may lead to inefficiency in utilizing the network resources [1]. To obtain a better network performance, the approach of cross-layer design has been considered. In [1], Shi and Fapojuwo proposed a cross-layer optimization scheme to the scheduling problem in clustered UASNs.However, power allocation and slot scheduling were separately designed in [1], which cannot guarantee a global optimum solution.In [2], a power control strategy was introduced to achieve the mini-mum-frame-length slot scheduling. However, EE, as a non-negligi-ble aspect of network performance, is not being considered in [2].In this letter, we formulate a joint slot scheduling and power allo-cation optimization problem to maximize the network’s EE in clus-tered UASNs. The formulated problem with coupled variables is non-convex and mixed-integer, which is challenging to be solved.We propose an efficient iterative algorithm to solve it. Numerical results demonstrate the effectiveness of our proposed algorithm.N ≜{1,2,...,N }K ≜{1,2,...,K }M ≜{1,2,...,M }Problem statement: A clustered UASN with N sensor nodes grouped into K clusters is considered in this article, with the sets and . Sensor nodes’ operation time in a frame consists of M equal and length-fixed time slots with the index set . The sensor nodes send carriers at the same frequency. The half-duplex (HD) mode and the decode-and-for-ward (DF) mode are adopted for data relaying. The data packet length is assumed equal to the length of the time slot. Since packet collisions occur at the receiver but not the sender, we optimize the slot scheduling from the perspective of signal arrival time. As shown in Fig. 1, packets are scheduled to reach the destination at specific time slots. To avoid collisions, the arriving packets cannot overlap with each other as shown in the example of the packets at the sink from CH1, CH2 and CH3 in Fig. 1.z n =(M ,[t ,1])∈R M ×1M −1We use the sparse vector (which means the t -th element is 1 and the rest elements are 0) to represent the scheduling indicator, i.e., the t -th time slot is assigned to node n to Z ≜{z 1,z 2,...,z N }p n P ≜{p 1,p 2,...,p N }B log 2(1+γn )γn deliver data. Then the slot scheduling of the overall network can beexpressed by the set . The allocated transmission power of node n is denoted as , with the corresponding set of the overall network . By Shannon’s law, the achiev-able link rate of node n to its receiver can be given by R n =, where B is the bandwidth, and is the signal-to-interference-and-noise ratio (SINR) at the receiver of node n , which can be written asg nn n N 0(f )where is the link’s channel gain from node to the receiver of node n , is the power spectral density (p.s.d.) ofthe ambient noises at the receiver (refer to [3]), and binary variablen n n ∈N tionships between node n and node , where .We regard the links connecting to the sink (i.e., sea surface buoy node) directly as the bottleneck links, then the EE maximization˙KK Q k ≜{1,2,...,Q k }C (z k )k ∈˙Kγth NR th T p C 1C 2C 3C 4M min M max C 5C 6z k i ,j where is the set of the links connecting to the sinkdirectly, is the remaining part of after removing the cluster con-taining the sink, is the set of cluster members (CMs) in k -th cluster, represents the set of time slots occupiedby the cluster head (CH) to transmit data, is the required SINR threshold for each link, is the required threshold of net-work rate for the entire network, and is a constant integer. is the transmission power constraint. is the SINR constraint to ensure that the signals can be correctly demoduled as shown in the example of CH3 in Fig. 1. indicates that the output link rate of CH k is restrained by the rate of its subnetwork, which ensures that the links connecting to the sink directly are the bottleneck links. is the integer constraint of M ranging from to . is the required minimum network rate constraint. denotes is a binary variable, which is set as 1 when the j -th time slot is occupied by theCorresponding author: Zhi-Xin Liu.Citation: Z.-X. Liu, X.-C. Jin, Y.-A. Xie, and Y. Yang, “Joint slot scheduling and power allocation in clustered underwater acoustic sensor networks,” IEEE/CAA J. Autom. Sinica , vol. 10, no. 6, pp. 1501–1503, Jun.2023.The authors are with the School of Electrical Engineering, Yanshan University, Qinhuangdao 066004, China (e-mail:Color versions of one or more of the figures in this paper are available online at .Digital Object Identifier 10.1109/JAS.2022.106031CH1CH2CH2CH1CH3CH3Slot1Signal packetInterference packetSlot2Slot3Slot4Slot5SinkFig. 1. Receiver-synchronized slot scheduling table.C 7C 8C 9C 10i -th CM of the k -th cluster. denotes that each time slot accommo-dates at most one node in a cluster, which is given to avoid packetcollisions. denotes that each node is assigned one and only onetime slot to deliver data in a frame. denotes that the HD mode is adopted, thus CHs could not transmit and receive data simultane-ously as shown in the example of CH1 in Fig. 1. is the con-straint for CHs to ensure that frames will not affect each other.P Z C 2C 3C 5P Z Problem solution: The optimization problem (3) is a non-convex and mixed-integer optimization problem, which cannot be solved directly due to the challenge that the variables M , and are always coupled with each other in , , and the objective function. To tackle the coupled variables, firstly, the exhaustive search method is adopted to solve the variable M , then a BCD-based alternating optimization method is utilized to decouple and .P Z∗Given the , and M , sensors’ optimal slot scheduling solution Q k M Q k M each node is assigned one and only one slot to deliver data and each slot accommodates at most one node in a cluster, the slot scheduling problem in a cluster can be modeled as a weighted matching prob-lem for a bipartite graph, in which the CMs in the k -th cluster and the M time slots can be partitioned into two independent and disjoint sets and such that every edge connects a node in to a time slot in . The weight of the edge is defined as the network rate. Then problem (4) can be solved by the Kuhn-Munkres algorithm proposed in [4]. The process, that optimizing the slot scheduling of a cluster while keeping the other clusters unchangeable, continues until all the clusters are optimized. After a round of optimization, if the network rate is improved, another optimization round will be performed until the network rate no longer increases.C 2Although any two nodes in the same cluster have no mutual inter-ference, it still should be noted that a node in other clusters may be interfered by multi nodes in the optimization cluster. That means the optimal matching obtained by the Kuhn-Munkres algorithm may unsatisfy . For solving this problem, Criterion 1 is proposed to search for the eligible slot scheduling scheme.B B B Criterion 1: Supposing node A is the node unsatisfying the SINR constraint, firstly, we find its interference nodes (called set ) who belong to the optimization cluster. Then, we sort set in descending order in terms of the interference intensity to node A to find the node having the largest interference to node A (called node C). If nodeC has more than one available time slot, the previous assigned slot is forbidden to be assigned to node C. Otherwise, other nodes’ time slot will be checked and forbidden in same fashion unless there are no more time slot that can be forbidden in .Z P∗Given the and , sensors’ optimal transmission power solution can be optimized by solving the following problem:C 3C 5P C 3R k ∀k ∈˙K R k =B log 2A k −B log 2H k A k =p k g kk +∑∀k k ∈N δk (z k )p k g kk +N 0(f )B H k =∑∀k k ∈N δk (z k )p k g kk +N 0(f )B ∑i v i ≥∏i (v i /θi )θiv i ≥0θi >0∑i θi =1∑v objective function and the constraints and with respect to .To obtain a convex upper bound of the left-hand side (LHS) of ,we note that , , can be rewritten as , where , and. Making use of the deformation of arithmetic-geometric mean inequality, which states thatwith , and (the equality happensRk =B log 2A k −B f (P )R k ≤R k ,∀k ∈K ˜pn =ln p n ∀p n ∈P Letting , we have . And the equality happens when (7) holds. Letting , , it isC 3To obtain a concave lower bound of the right-hand side (RHS) of , the logarithmic approximation method used in [5] is adopted.ˇRl ,∀l ∈L ,R l ,∀l ∈L ,C 5˜pn =ln p n ∀p n ∈P of in the objective function and . Substitut-ing the undesired terms in (5) with the upper or lower boundsobtained above, and letting , , problem (5) can be12N ˜Ptional function with a concave numerator and a convex denominatorin terms of the transmission power , and the constraints are all con-vex. Therefore, we can exploit the Dinkelbach’s method [6] to trans-form it into the equivalent convex problemP The optimal solution of problem (5) can be obtained by solving the equivalent convex problem (13) iteratively, which can be tackled with existing optimization tools like CVX. The pseudocode of the optimization process in terms of sensors’ transmission power is shown in Algorithm 1.The pseudocode of the BCD-based alternating optimization algo-rithm is shown in Algorithm 2, in which two variable blocks are opti-mized alteratively corresponding to the two optimization subprob-lems (i.e., the slot scheduling subproblem and the power allocation subproblem) in each iteration of the alternating optimization process.f =10B =2P max =Simulation results: We consider a 10 km × 10 km × 200 m area,where N = 30 underwater sensor nodes deployed randomly at differ-ent sea depths are divided K clusters. We assume that the sensors are stationary, and the data in each sensor’s buffer is always sufficient.We take the carrier frequency kHz, kHz and 2 W.P 0Z 0For assessing the performance of the proposed alternating-opti-mization-based joint slot scheduling and power allocation algorithm (denoted as AO), we present three other schemes as contrasts, which include two kinds of separate optimization methods and the power allocation scheme and the slot scheduling scheme obtained by the proposed CMS-MAC algorithm in [2]. The two separate opti-mization methods are summarized as follows:Algorithm 1 Power Control Algorithm Based on the SCA Technique andthe Dinkelbach’s Methodτεt ←−0˜P{t }←−{ln p 1,ln p 2,...,ln p N }P 1: Set the maximum number of iterations and the maximum tolerance .Initialize iteration index and , where is the input powers;2: repeath ←−0˜P {h }temp←−˜P {t }3: Initialize iteration index and ;η{t }EE ˜P{t }4: Compute with given ;5: repeat αl =γl (˜P {h }temp )1+γl (˜P {h }temp )βl =ln1+γl (˜P {h }temp )γαl l(˜P {h }temp ) ∀l ∈L αq =γq (˜P {h }temp )1+γq (˜P {h }temp )βq =ln 1+γq (˜P {h }temp )γαq q (˜P {h }temp) ∀q ∈Q k ∀k ∈˙K θk ∀k k ∈N θN ∀k ∈˙K˜P {h }temp 6: Compute , , , , , , , and compute ,, , by (7) with given ;η{t }EE Z ˜P {h +1}temp7: Solve (13) with the given and , and obtain the optimal ;h ←−h +18: Update ;˜P h temp ˜P ∗temp 9: until converge to the optimal solution ;˜P{t +1}←−˜P ∗temp 10: ;t ←−t +111: Update ; f (η{t −1}EE )<ε,or t ≥τ12: until ;P ∗={e ˜p{t }1,e ˜p {t }2,...,e ˜p {t }N }13: Obtain the optimal solution ;Algorithm 2 Alternating-Optimization-Based Joint Time Slot Scheduling and Power Allocation AlgorithmM min M max P 0Z 0M min 1: Obtain the low bound and the upper bound of M , and the power allocation solution and the slot scheduling scheme under by the algorithm proposed in [2];ε2: Set the maximum tolerance ;M =M min ;M ≤M max ;M ++3: for do l ←04: Initialize iteration index ;Z {l }M ←Z 0P {l }M ←P 05: Initialize , and ;6: repeatP {l }M Z {l }M Z {l +1}M 7: Solve (4) with the given and by the Kuhn-Munkres algo-rithm, and obtain the optimal slot scheduling ;P {l }M Z {l +1}M P {l +1}M 8: Solve (5) with the given and by Algorithm (1), and obtainthe optimal power allocation ;l ←−l +19: Update ;10: until the increment of ηEE is smaller than ε;η∗M Z {l }M P {l }M 11: Obtain the optimal network EE , and the optimal solution and ;12: endη∗M ∗=Max {η∗M min ,η∗M min +1,...,η∗M max }13: Let ;M ∗Z {l }M ∗P {l }M ∗14: Return the optimal solution , and ;Z 01) Optimal power allocation with fixed slot scheduling (denoted as OPA_FSS): With the fixed slot scheduling scheme , the transmis-sion powers are optimized by Algorithm 1.P 02) Optimal slot scheduling with fixed power allocation (denoted as OSS_FPA): With the fixed power allocation scheme , the slotscheduling of all of sensors are optimized by the slot schedulingalgorithm proposed above.The corresponding comparison results are shown in Fig. 2. It can be observed that the proposed AO shows the best performance. The reason is that slot scheduling and power allocation may be influ-enced by each other, thus it is unreasonable to fix one of them and then solve another. For the proposed AO, slot scheduling and power allocation could be solved in an alternating way, which leads to bet-ter solutions. Furthermore, it can be found that AO achieves signifi-cant EE gains compared to CMS-MAC algorithm.(a)(b)K SINR (dB)2500E E (b i t s /H z /J )E E (b i t s /H z /J )200015001000500γth Fig. 2. Comparisons of EE. (a) for different clustering numbers with = 10dB; (b) for different SINR constraints with K = 7.Conclusion: In this letter, an EE maximization problem withcross-layer design is considered in clustered UASNs. To tackle the non-convex and mixed-integer optimization problem, a BCD-based iterative algorithm is proposed. Numerical results show that the pro-posed joint optimization scheme achieves significant EE gains com-pared to the separate optimization methods.Acknowledgment: This work was supported by the National Natu-ral Science Foundation of China (62273298, 61873223), the Natural Science Foundation of Hebei Province (F2019203095), and Provin-cial Key Laboratory Performance Subsidy Project (22567612H).ReferencesL. Shi and A. O. Fapojuwo, “TDMA scheduling with optimized energy efficiency and minimum delay in clustered wireless sensor networks,”IEEE. Trans. Mob. Comput., vol. 9, no. 7, pp. 927–940, 2010.[1]W. Bai, H. Wang, X. Shen, and R. Zhao, “Link scheduling method for underwater acoustic sensor networks based on correlation matrix,” IEEE Sens. J., vol. 16, no. 11, pp. 4015–4022, 2016.[2]M. Stojanovic, “On the relationship between capacity and distance in an underwater acoustic communication channel,” SIGMOBILE put. Commun. Rev., vol. 11, no. 4, pp. 34–43, 2007.[3]F. Xing, H. Yin, Z. Shen, and V. C. M. Leung, “Joint relay assignment and power allocation for multiuser multirelay networks over underwater wireless optical channels,” IEEE Internet Things J., vol. 7, no. 10, pp. 9688–9701, 2020.[4]Z. Liu, Y. Xie, Y. Yuan, K. Ma, K. Y. Chan, and X. Guan, “Robust power control for clustering-based vehicle-to-vehicle communication,”IEEE Syst. J., vol. 14, no. 2, pp. 2557–2568, 2020.[5]J.-P. Crouzeix and J. A. Ferland, “Algorithms for generalized fractional programming,” Mathematical Programming , vol. 52, no. 1, pp. 191–207,1991.[6]LIU et al .: JOINT SLOT SCHEDULING AND POWER ALLOCATION IN CLUSTERED UASNS 1503。

扩展巴科斯范式(转自维基)

扩展巴科斯范式(转自维基)

扩展巴科斯范式(转⾃维基)https:///wiki/%E6%89%A9%E5%B1%95%E5%B7%B4%E7%A7%91%E6%96%AF%E8%8C%83%E5%BC%8F扩展巴科斯范式[]维基百科,⾃由的百科全书扩展巴科斯-瑙尔范式(EBNF, Extended Backus–Naur Form)是表达作为描述计算机和的正规⽅式的的(metalanguage)符号表⽰法。

它是基本(BNF)元语法符号表⽰法的⼀种扩展。

它最初由开发,最常⽤的 EBNF 变体由标准,特别是 ISO-14977 所定义。

⽬录[隐藏]基本[],如由即可视字符、数字、标点符号、空⽩字符等组成的的。

EBNF 定义了把各符号序列分别指派到的:digit excluding zero = "1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9" ;digit = "0" | digit excluding zero ;这个产⽣规则定义了在这个指派的左端的⾮终结符digit。

竖杠表⽰可供选择,⽽终结符被引号包围,最后跟着分号作为终⽌字符。

所以digit是⼀个 "0"或可以是 "1或2或3直到9的⼀个digit excluding zero"。

产⽣规则还可以包括由逗号分隔的⼀序列终结符或⾮终结符:twelve = "1" , "2" ;two hundred one = "2" , "0" , "1" ;three hundred twelve = "3" , twelve ;twelve thousand two hundred one = twelve , two hundred one ;可以省略或重复的表达式可以通过花括号 { ... } 表⽰:natural number = digit excluding zero , { digit } ;在这种情况下,字符串1, 2, ...,10,...,12345,... 都是正确的表达式。

stable diffusion embedding 训练 -回复

stable diffusion embedding 训练 -回复

stable diffusion embedding 训练-回复稳定扩散嵌入(stable diffusion embedding)是一种用于数据降维和可视化的机器学习技术。

它可以将高维数据投射到低维空间中,同时保留原始数据的结构和相似性。

本文将逐步解释stable diffusion embedding 的训练过程。

首先,为了理解stable diffusion embedding的训练过程,我们需要先了解一些基本概念。

稳定扩散嵌入是基于数据的相似性矩阵进行计算的。

相似性矩阵是一个对数据点之间的相似度进行量化的矩阵,通常使用欧氏距离或相关性来度量。

第一步是构建相似性矩阵。

对于给定的数据集,我们首先需要计算每对数据点之间的相似度。

这可以通过计算欧氏距离、相关性或其他相似性度量来完成。

相似性矩阵是一个对称矩阵,其中每个元素表示相应数据点之间的相似度。

接下来,我们需要对相似性矩阵进行标准化。

标准化可以帮助处理不同度量之间的差异,并确保嵌入结果的稳定性。

一种常用的标准化方法是通过行归一化,即将相似性矩阵的每一行除以该行元素之和。

然后,我们使用标准化后的相似性矩阵构建转移概率矩阵。

转移概率矩阵描述了数据点之间的转移概率,即从一个数据点转移到另一个数据点的概率。

转移概率可以通过将标准化后的相似性矩阵除以每一行元素之和来计算。

接下来,我们引入扩散核(diffusion kernel)。

扩散核是用于建模转移概率的核函数,通常采用高斯核函数。

通过将扩散核应用于转移概率矩阵,我们可以计算得到扩散矩阵。

现在,我们可以开始计算稳定扩散嵌入了。

稳定扩散嵌入的计算过程可以通过迭代计算扩散矩阵的特征向量来完成。

特征向量表示了数据在嵌入空间中的位置,其中最大的特征值对应于最主要的轴,最小的特征值对应于最次要的轴。

根据特征值和特征向量的数量,我们可以选择保留最重要的特征向量,将数据投射到低维空间。

最后,我们可以通过使用保留的特征向量将高维数据投射到低维空间。

Banach空间中隐补问题解的存在性定理

Banach空间中隐补问题解的存在性定理

第 3期
沈琴琴 等 : a a h空间 中隐补 问题 解 的存在 性 定理 B nc
¨¨ ・
2 3
() 1 对任 意 网{ cP 当{ } , )
时 , ( — )g ‰) ( ) 有f x) , ( 且

() 2 存在 两 个非 空弱 紧集 , MCP V P , Y∈ 使得 ( , ∈ 有 M 则存 在 ∈P, 得 T() 且 ( ()gx )0 使 fx ∈ r x ,()= 。 f
相 补 问题 理论 的研究 始 于 2 0世 纪 6 0年代 L m e C te和 D m i g的工作 l I 一理 论具 有广 泛 的实 e k 、o l t a ti z l 。这 _ 2 际背景 。在一 定 的 条件 下 , 经济 和 交通 的平 衡 问题 、 电路 联结 问题 、 学 规划 问题 以及流 体 渗透 空 隙介 质 问 数
设 Y, - , E【,] ∈S1 ) O 0 1。注意 到 g是 仿射 算子 , Y ( / 则有
( ^ , ( _ 0 I1 a)2 = 刁 )g x -a ( + 1 )(2]= 刁( g ) - - y] ( ,( ) [gy) (一 g y)) ) - ( ) (
在 定理 1中 , 令 可得 如下 结果 :
推 论 2 设 E为 B n c aa h空 间 , CE为 闭凸锥 。 P 设 : P , 非线 性算 子 , : P为 正齐 次满 的仿 射算 为 g P 子 ,满 足条件 :
W W II II -
( ) 任 意 M{ } 1对 x CP, } 时 , g x) ( 且 当{ — 有 ( 一g )
( )g 一 ( )> 刁 ,( ) gy ) 0
于是 ∈ () Sy 。这表明

字母不重复的子串-概述说明以及解释

字母不重复的子串-概述说明以及解释

字母不重复的子串-概述说明以及解释1.引言1.1 概述字母不重复的子串是指在一个字符串中,找出不包含重复字母的最长子串。

也就是说,子串中的每个字符在该子串中只出现一次。

这个问题在字符串处理和算法设计方面有着广泛的应用。

在实际生活中,我们经常需要处理包含字母的字符串,例如英文单词、句子甚至是DNA序列。

而找出其中不重复的子串,能够帮助我们理解字符串的特征和模式,从而进行更深入的分析和处理。

字母不重复的子串问题在实际应用中有着很广泛的应用。

以文本信息处理为例,比如在自然语言处理中,我们需要通过分析文本中的子串来提取关键词、识别命名实体等。

另外,在数据压缩和加密中,也经常需要使用字母不重复的子串来提高存储和传输的效率。

本文将深入探讨字母不重复的子串问题的意义和应用,并通过具体的算法设计与实例分析,展示如何高效地找出字母不重复的子串。

最后,我们将对问题进行总结,并展望未来该领域可能的发展方向。

让我们一起来探索吧!1.2 文章结构本文将分为三个主要部分进行讨论。

第一部分是引言部分。

在这部分中,我们将概述本文的主要内容,包括字母不重复的子串的概念和意义,以及本文的结构和目的。

第二部分是正文部分。

在这部分中,我们将详细讨论字母不重复的子串的意义和应用。

首先,我们将探讨字母不重复的子串在计算机科学领域的重要性,并介绍一些相关的算法和数据结构。

然后,我们将探讨字母不重复的子串在其他领域的应用,如自然语言处理、图像处理等。

通过这些案例研究,我们将阐述字母不重复的子串在现实生活中的实际应用和潜在价值。

第三部分是结论部分。

在这部分中,我们将对前文进行总结,并提出一些展望。

我们将回顾字母不重复的子串对于解决实际问题的作用,并讨论可能的未来研究方向。

我们还将提供一些实用建议,以帮助读者更好地理解和应用字母不重复的子串的概念。

通过以上结构,本文旨在全面介绍字母不重复的子串的意义和应用,并为读者提供深入理解和有效运用该概念的指导。

Parallel and Distributed Computing and Systems

Parallel and Distributed Computing and Systems

Proceedings of the IASTED International ConferenceParallel and Distributed Computing and SystemsNovember3-6,1999,MIT,Boston,USAParallel Refinement of Unstructured MeshesJos´e G.Casta˜n os and John E.SavageDepartment of Computer ScienceBrown UniversityE-mail:jgc,jes@AbstractIn this paper we describe a parallel-refinement al-gorithm for unstructuredfinite element meshes based on the longest-edge bisection of triangles and tetrahedrons. This algorithm is implemented in P ARED,a system that supports the parallel adaptive solution of PDEs.We dis-cuss the design of such an algorithm for distributed mem-ory machines including the problem of propagating refine-ment across processor boundaries to obtain meshes that are conforming and non-degenerate.We also demonstrate that the meshes obtained by this algorithm are equivalent to the ones obtained using the serial longest-edge refine-ment method.Wefinally report on the performance of this refinement algorithm on a network of workstations.Keywords:mesh refinement,unstructured meshes,finite element methods,adaptation.1.IntroductionThefinite element method(FEM)is a powerful and successful technique for the numerical solution of partial differential equations.When applied to problems that ex-hibit highly localized or moving physical phenomena,such as occurs on the study of turbulence influidflows,it is de-sirable to compute their solutions adaptively.In such cases, adaptive computation has the potential to significantly im-prove the quality of the numerical simulations by focusing the available computational resources on regions of high relative error.Unfortunately,the complexity of algorithms and soft-ware for mesh adaptation in a parallel or distributed en-vironment is significantly greater than that it is for non-adaptive computations.Because a portion of the given mesh and its corresponding equations and unknowns is as-signed to each processor,the refinement(coarsening)of a mesh element might cause the refinement(coarsening)of adjacent elements some of which might be in neighboring processors.To maintain approximately the same number of elements and vertices on every processor a mesh must be dynamically repartitioned after it is refined and portions of the mesh migrated between processors to balance the work.In this paper we discuss a method for the paral-lel refinement of two-and three-dimensional unstructured meshes.Our refinement method is based on Rivara’s serial bisection algorithm[1,2,3]in which a triangle or tetrahe-dron is bisected by its longest edge.Alternative efforts to parallelize this algorithm for two-dimensional meshes by Jones and Plassman[4]use randomized heuristics to refine adjacent elements located in different processors.The parallel mesh refinement algorithm discussed in this paper has been implemented as part of P ARED[5,6,7], an object oriented system for the parallel adaptive solu-tion of partial differential equations that we have devel-oped.P ARED provides a variety of solvers,handles selec-tive mesh refinement and coarsening,mesh repartitioning for load balancing,and interprocessor mesh migration.2.Adaptive Mesh RefinementIn thefinite element method a given domain is di-vided into a set of non-overlapping elements such as tri-angles or quadrilaterals in2D and tetrahedrons or hexahe-drons in3D.The set of elements and its as-sociated vertices form a mesh.With theaddition of boundary conditions,a set of linear equations is then constructed and solved.In this paper we concentrate on the refinement of conforming unstructured meshes com-posed of triangles or tetrahedrons.On unstructured meshes, a vertex can have a varying number of elements adjacent to it.Unstructured meshes are well suited to modeling do-mains that have complex geometry.A mesh is said to be conforming if the triangles and tetrahedrons intersect only at their shared vertices,edges or faces.The FEM can also be applied to non-conforming meshes,but conformality is a property that greatly simplifies the method.It is also as-sumed to be a requirement in this paper.The rate of convergence and quality of the solutions provided by the FEM depends heavily on the number,size and shape of the mesh elements.The condition number(a)(b)(c)Figure1:The refinement of the mesh in using a nested refinement algorithm creates a forest of trees as shown in and.The dotted lines identify the leaf triangles.of the matrices used in the FEM and the approximation error are related to the minimum and maximum angle of all the elements in the mesh[8].In three dimensions,the solid angle of all tetrahedrons and their ratio of the radius of the circumsphere to the inscribed sphere(which implies a bounded minimum angle)are usually used as measures of the quality of the mesh[9,10].A mesh is non-degenerate if its interior angles are never too small or too large.For a given shape,the approximation error increases with ele-ment size(),which is usually measured by the length of the longest edge of an element.The goal of adaptive computation is to optimize the computational resources used in the simulation.This goal can be achieved by refining a mesh to increase its resolution on regions of high relative error in static problems or by re-fining and coarsening the mesh to follow physical anoma-lies in transient problems[11].The adaptation of the mesh can be performed by changing the order of the polynomi-als used in the approximation(-refinement),by modifying the structure of the mesh(-refinement),or a combination of both(-refinement).Although it is possible to replace an old mesh with a new one with smaller elements,most -refinement algorithms divide each element in a selected set of elements from the current mesh into two or more nested subelements.In P ARED,when an element is refined,it does not get destroyed.Instead,the refined element inserts itself into a tree,where the root of each tree is an element in the initial mesh and the leaves of the trees are the unrefined elements as illustrated in Figure1.Therefore,the refined mesh forms a forest of refinement trees.These trees are used in many of our algorithms.Error estimates are used to determine regions where adaptation is necessary.These estimates are obtained from previously computed solutions of the system of equations. After adaptation imbalances may result in the work as-signed to processors in a parallel or distributed environ-ment.Efficient use of resources may require that elements and vertices be reassigned to processors at runtime.There-fore,any such system for the parallel adaptive solution of PDEs must integrate subsystems for solving equations,adapting a mesh,finding a good assignment of work to processors,migrating portions of a mesh according to anew assignment,and handling interprocessor communica-tion efficiently.3.P ARED:An OverviewP ARED is a system of the kind described in the lastparagraph.It provides a number of standard iterativesolvers such as Conjugate Gradient and GMRES and pre-conditioned versions thereof.It also provides both-and -refinement of meshes,algorithms for adaptation,graph repartitioning using standard techniques[12]and our ownParallel Nested Repartitioning(PNR)[7,13],and work mi-gration.P ARED runs on distributed memory parallel comput-ers such as the IBM SP-2and networks of workstations.These machines consist of coarse-grained nodes connectedthrough a high to moderate latency network.Each nodecannot directly address a memory location in another node. In P ARED nodes exchange messages using MPI(Message Passing Interface)[14,15,16].Because each message has a high startup cost,efficient message passing algorithms must minimize the number of messages delivered.Thus, it is better to send a few large messages rather than many small ones.This is a very important constraint and has a significant impact on the design of message passing algo-rithms.P ARED can be run interactively(so that the user canvisualize the changes in the mesh that results from meshadaptation,partitioning and migration)or without directintervention from the user.The user controls the systemthrough a GUI in a distinguished node called the coordina-tor,.This node collects information from all the other processors(such as its elements and vertices).This tool uses OpenGL[17]to permit the user to view3D meshes from different angles.Through the coordinator,the user can also give instructions to all processors such as specify-ing when and how to adapt the mesh or which strategy to use when repartitioning the mesh.In our computation,we assume that an initial coarse mesh is given and that it is loaded into the coordinator.The initial mesh can then be partitioned using one of a num-ber of serial graph partitioning algorithms and distributed between the processors.P ARED then starts the simulation. Based on some adaptation criterion[18],P ARED adapts the mesh using the algorithms explained in Section5.Af-ter the adaptation phase,P ARED determines if a workload imbalance exists due to increases and decreases in the num-ber of mesh elements on individual processors.If so,it invokes a procedure to decide how to repartition mesh el-ements between processors;and then moves the elements and vertices.We have found that PNR gives partitions with a quality comparable to those provided by standard meth-ods such as Recursive Spectral Bisection[19]but which(b)(a)Figure2:Mesh representation in a distributed memory ma-chine using remote references.handles much larger problems than can be handled by stan-dard methods.3.1.Object-Oriented Mesh RepresentationsIn P ARED every element of the mesh is assigned to a unique processor.V ertices are shared between two or more processors if they lie on a boundary between parti-tions.Each of these processors has a copy of the shared vertices and vertices refer to each other using remote ref-erences,a concept used in object-oriented programming. This is illustrated in Figure2on which the remote refer-ences(marked with dashed arrows)are used to maintain the consistency of multiple copies of the same vertex in differ-ent processors.Remote references are functionally similar to standard C pointers but they address objects in a different address space.A processor can use remote references to invoke meth-ods on objects located in a different processor.In this case, the method invocations and arguments destined to remote processors are marshalled into messages that contain the memory addresses of the remote objects.In the destina-tion processors these addresses are converted to pointers to objects of the corresponding type through which the meth-ods are invoked.Because the different nodes are inher-ently trusted and MPI guarantees reliable communication, P ARED does not incur the overhead traditionally associated with distributed object systems.Another idea commonly found in object oriented pro-gramming and which is used in P ARED is that of smart pointers.An object can be destroyed when there are no more references to it.In P ARED vertices are shared be-tween several elements and each vertex counts the number of elements referring to it.When an element is created, the reference count of its vertices is incremented.Simi-larly,when the element is destroyed,the reference count of its vertices is decremented.When the reference count of a vertex reaches zero,the vertex is no longer attached to any element located in the processor and can be destroyed.If a vertex is shared,then some other processor might have a re-mote reference to it.In that case,before a copy of a shared vertex is destroyed,it informs the copies in other processors to delete their references to itself.This procedure insures that the shared vertex can then be safely destroyed without leaving dangerous dangling pointers referring to it in other processors.Smart pointers and remote references provide a simple replication mechanism that is tightly integrated with our mesh data structures.In adaptive computation,the struc-ture of the mesh evolves during the computation.During the adaptation phase,elements and vertices are created and destroyed.They may also be assigned to a different pro-cessor to rebalance the work.As explained above,remote references and smart pointers greatly simplify the task of creating dynamic meshes.4.Adaptation Using the Longest Edge Bisec-tion AlgorithmMany-refinement techniques[20,21,22]have been proposed to serially refine triangular and tetrahedral meshes.One widely used method is the longest-edge bisec-tion algorithm proposed by Rivara[1,2].This is a recursive procedure(see Figure3)that in two dimensions splits each triangle from a selected set of triangles by adding an edge between the midpoint of its longest side to the opposite vertex.In the case that makes a neighboring triangle,,non-conforming,then is refined using the same algorithm.This may cause the refinement to prop-agate throughout the mesh.Nevertheless,this procedure is guaranteed to terminate because the edges it bisects in-crease in length.Building on the work of Rosenberg and Stenger[23]on bisection of triangles,Rivara[1,2]shows that this refinement procedure provably produces two di-mensional meshes in which the smallest angle of the re-fined mesh is no less than half of the smallest angle of the original mesh.The longest-edge bisection algorithm can be general-ized to three dimensions[3]where a tetrahedron is bisected into two tetrahedrons by inserting a triangle between the midpoint of its longest edge and the two vertices not in-cluded in this edge.The refinement propagates to neigh-boring tetrahedrons in a similar way.This procedure is also guaranteed to terminate,but unlike the two dimensional case,there is no known bound on the size of the small-est angle.Nevertheless,experiments conducted by Rivara [3]suggest that this method does not produce degenerate meshes.In two dimensions there are several variations on the algorithm.For example a triangle can initially be bisected by the longest edge,but then its children are bisected by the non-conforming edge,even if it is that is not their longest edge[1].In three dimensions,the bisection is always per-formed by the longest edge so that matching faces in neigh-boring tetrahedrons are always bisected by the same com-mon edge.Bisect()let,and be vertices of the trianglelet be the longest side of and let be the midpoint ofbisect by the edge,generating two new triangles andwhile is a non-conforming vertex dofind the non-conforming triangle adjacent to the edgeBisect()end whileFigure3:Longest edge(Rivara)bisection algorithm for triangular meshes.Because in P ARED refined elements are not destroyed in the refinement tree,the mesh can be coarsened by replac-ing all the children of an element by their parent.If a parent element is selected for coarsening,it is important that all the elements that are adjacent to the longest edge of are also selected for coarsening.If neighbors are located in different processors then only a simple message exchange is necessary.This algorithm generates conforming meshes: a vertex is removed only if all the elements that contain that vertex are all coarsened.It does not propagate like the re-finement algorithm and it is much simpler to implement in parallel.For this reason,in the rest of the paper we will focus on the refinement of meshes.5.Parallel Longest-Edge RefinementThe longest-edge bisection algorithm and many other mesh refinement algorithms that propagate the refinement to guarantee conformality of the mesh are not local.The refinement of one particular triangle or tetrahedron can propagate through the mesh and potentially cause changes in regions far removed from.If neighboring elements are located in different processors,it is necessary to prop-agate this refinement across processor boundaries to main-tain the conformality of the mesh.In our parallel longest edge bisection algorithm each processor iterates between a serial phase,in which there is no communication,and a parallel phase,in which each processor sends and receives messages from other proces-sors.In the serial phase,processor selects a setof its elements for refinement and refines them using the serial longest edge bisection algorithms outlined earlier. The refinement often creates shared vertices in the bound-ary between adjacent processors.To minimize the number of messages exchanged between and,delays the propagation of refinement to until has refined all the elements in.The serial phase terminates when has no more elements to refine.A processor informs an adjacent processor that some of its elements need to be refined by sending a mes-sage from to containing the non-conforming edges and the vertices to be inserted at their midpoint.Each edge is identified by its endpoints and and its remote ref-erences(see Figure4).If and are sharedvertices,(a)(c)(b)Figure4:In the parallel longest edge bisection algo-rithm some elements(shaded)are initially selected for re-finement.If the refinement creates a new(black)ver-tex on a processor boundary,the refinement propagates to neighbors.Finally the references are updated accord-ingly.then has a remote reference to copies of and lo-cated in processor.These references are included in the message,so that can identify the non-conforming edge and insert the new vertex.A similar strategy can be used when the edge is refined several times during the re-finement phase,but in this case,the vertex is not located at the midpoint of.Different processors can be in different phases during the refinement.For example,at any given time a processor can be refining some of its elements(serial phase)while neighboring processors have refined all their elements and are waiting for propagation messages(parallel phase)from adjacent processors.waits until it has no elements to refine before receiving a message from.For every non-conforming edge included in a message to,creates its shared copy of the midpoint(unless it already exists) and inserts the new non-conforming elements adjacent to into a new set of elements to be refined.The copy of in must also have a remote reference to the copy of in.For this reason,when propagates the refine-ment to it also includes in the message a reference to its copies of shared vertices.These steps are illustrated in Figure4.then enters the serial phase again,where the elements in are refined.(c)(b)(a)Figure5:Both processors select(shaded)mesh el-ements for refinement.The refinement propagates to a neighboring processor resulting in more elements be-ing refined.5.1.The Challenge of Refining in ParallelThe description of the parallel refinement algorithm is not complete because refinement propagation across pro-cessor boundaries can create two synchronization prob-lems.Thefirst problem,adaptation collision,occurs when two(or more)processors decide to refine adjacent elements (one in each processor)during the serial phase,creating two(or more)vertex copies over a shared edge,one in each processor.It is important that all copies refer to the same logical vertex because in a numerical simulation each ver-tex must include the contribution of all the elements around it(see Figure5).The second problem that arises,termination detection, is the determination that a refinement phase is complete. The serial refinement algorithm terminates when the pro-cessor has no more elements to refine.In the parallel ver-sion termination is a global decision that cannot be deter-mined by an individual processor and requires a collabora-tive effort of all the processors involved in the refinement. Although a processor may have adapted all of its mesh elements in,it cannot determine whether this condition holds for all other processors.For example,at any given time,no processor might have any more elements to re-fine.Nevertheless,the refinement cannot terminate because there might be some propagation messages in transit.The algorithm for detecting the termination of parallel refinement is based on Dijkstra’s general distributed termi-nation algorithm[24,25].A global termination condition is reached when no element is selected for refinement.Hence if is the set of all elements in the mesh currently marked for refinement,then the algorithmfinishes when.The termination detection procedure uses message ac-knowledgments.For every propagation message that receives,it maintains the identity of its source()and to which processors it propagated refinements.Each prop-agation message is acknowledged.acknowledges to after it has refined all the non-conforming elements created by’s message and has also received acknowledgments from all the processors to which it propagated refinements.A processor can be in two states:an inactive state is one in which has no elements to refine(it cannot send new propagation messages to other processors)but can re-ceive messages.If receives a propagation message from a neighboring processor,it moves from an inactive state to an active state,selects the elements for refinement as spec-ified in the message and proceeds to refine them.Let be the set of elements in needing refinement.A processor becomes inactive when:has received an acknowledgment for every propa-gation message it has sent.has acknowledged every propagation message it has received..Using this definition,a processor might have no more elements to refine()but it might still be in an active state waiting for acknowledgments from adjacent processors.When a processor becomes inactive,sends an acknowledgment to the processors whose propagation message caused to move from an inactive state to an active state.We assume that the refinement is started by the coordi-nator processor,.At this stage,is in the active state while all the processors are in the inactive state.ini-tiates the refinement by sending the appropriate messages to other processors.This message also specifies the adapta-tion criterion to use to select the elements for refinement in.When a processor receives a message from,it changes to an active state,selects some elements for refine-ment either explicitly or by using the specified adaptation criterion,and then refines them using the serial bisection algorithm,keeping track of the vertices created over shared edges as described earlier.When itfinishes refining its ele-ments,sends a message to each processor on whose shared edges created a shared vertex.then listens for messages.Only when has refined all the elements specified by and is not waiting for any acknowledgment message from other processors does it sends an acknowledgment to .Global termination is detected when the coordinator becomes inactive.When receives an acknowledgment from every processor this implies that no processor is re-fining an element and that no processor is waiting for an acknowledgment.Hence it is safe to terminate the refine-ment.then broadcasts this fact to all the other proces-sors.6.Properties of Meshes Refined in ParallelOur parallel refinement algorithm is guaranteed to ter-minate.In every serial phase the longest edge bisectionLet be a set of elements to be refinedwhile there is an element dobisect by its longest edgeinsert any non-conforming element intoend whileFigure6:General longest-edge bisection(GLB)algorithm.algorithm is used.In this algorithm the refinement prop-agates towards progressively longer edges and will even-tually reach the longest edge in each processor.Between processors the refinement also propagates towards longer edges.Global termination is detected by using the global termination detection procedure described in the previous section.The resulting mesh is conforming.Every time a new vertex is created over a shared edge,the refinement propagates to adjacent processors.Because every element is always bisected by its longest edge,for triangular meshes the results by Rosenberg and Stenger on the size of the min-imum angle of two-dimensional meshes also hold.It is not immediately obvious if the resulting meshes obtained by the serial and parallel longest edge bisection al-gorithms are the same or if different partitions of the mesh generate the same refined mesh.As we mentioned earlier, messages can arrive from different sources in different or-ders and elements may be selected for refinement in differ-ent sequences.We now show that the meshes that result from refining a set of elements from a given mesh using the serial and parallel algorithms described in Sections4and5,re-spectively,are the same.In this proof we use the general longest-edge bisection(GLB)algorithm outlined in Figure 6where the order in which elements are refined is not spec-ified.In a parallel environment,this order depends on the partition of the mesh between processors.After showing that the resulting refined mesh is independent of the order in which the elements are refined using the serial GLB al-gorithm,we show that every possible distribution of ele-ments between processors and every order of parallel re-finement yields the same mesh as would be produced by the serial algorithm.Theorem6.1The mesh that results from the refinement of a selected set of elements of a given mesh using the GLB algorithm is independent of the order in which the elements are refined.Proof:An element is refined using the GLBalgorithm if it is in the initial set or refinementpropagates to it.An element is refinedif one of its neighbors creates a non-conformingvertex at the midpoint of one of its edges.Therefinement of by its longest edge divides theelement into two nested subelements andcalled the children of.These children are inturn refined by their longest edge if one of their edges is non-conforming.The refinement proce-dure creates a forest of trees of nested elements where the root of each tree is an element in theinitial mesh and the leaves are unrefined ele-ments.For every element,let be the refinement tree of nested elements rooted atwhen the refinement procedure terminates. Using the GLB procedure elements can be se-lected for refinement in different orders,creating possible different refinement histories.To show that this cannot happen we assume the converse, namely,that two refinement histories and generate different refined meshes,and establish a contradiction.Thus,assume that there is an ele-ment such that the refinement trees and,associated with the refinement histories and of respectively,are different.Be-cause the root of and is the same in both refinement histories,there is a place where both treesfirst differ.That is,starting at the root,there is an element that is common to both trees but for some reason,its children are different.Be-cause is always bisected by the longest edge, the children of are different only when is refined in one refinement history and it is not re-fined in the other.In other words,in only one of the histories does have children.Because is refined in only one refinement his-tory,then,the initial set of elements to refine.This implies that must have been refined because one of its edges became non-conforming during one of the refinement histo-ries.Let be the set of elements that are present in both refinement histories,but are re-fined in and not in.We define in a similar way.For each refinement history,every time an ele-ment is refined,it is assigned an increasing num-ber.Select an element from either or that has the lowest number.Assume that we choose from so that is refined in but not in.In,is refined because a neigh-boring element created a non-conforming ver-tex at the midpoint of their shared edge.There-fore is refined in but not in because otherwise it would cause to be refined in both sequences.This implies that is also in and has a lower refinement number than con-。

Banach空间有限个λ-严格伪压缩映射的强收敛定理

Banach空间有限个λ-严格伪压缩映射的强收敛定理
f V ∈c
? 艿 +( 一 y= 1 s)

() 3
+’ , , ≥O V
+= l&
)+
我们 可 以证 明 由 ( ) 3 定义 的 { } 是强 收敛到 一般不 动点 的.
1 预 备知 识
层是 B e 间 , E 的范数 是 G t u  ̄ah空 称  ̄e x可微 的 , V , ∈s E) E :I l } a 若 x Y ( ={ l l:1
L J
如 果
( ) , 称 B nc 间 E是一 致光 滑 的. g>1 若存 在 常数 >o使 得 p () £ 则 aah空 设 , £
非线性函数的不动点是个非 常重要 的领域. 特别通过迭代算法寻找非扩张映射的不动点. 对有限 个的映射的迭代方法 已经研究得非常广泛了 , 可以参见文献[ ~ ] 1 5. 在这 篇 文章 中我 们介 绍 一种 新 的 映射 :
0=I


1 =t

+( ) 1一t , c +( ‘ ) , ¨ 1一 啦
l +,I< I I l,I I )f l l+ I l ) 设 :0 1 一[ ,)是 层的光滑模 , [ ,) 0 1 定义它为
P( =u{ I + I I一 ) 1 ∈ () I 1 ‘ E ) s 寺(I )f yI一 : se , )l } P , +} f I ≤ ,
l 一 1




() 2
1= t



2 1一 一) +( t 1
2 l
Ⅳ = t n
Ⅳ l 1一 一) 一 +( t 1
收 稿 日期 :O 0 l— l 2 1一 l 2

banach不动点定理的证明

banach不动点定理的证明

Banach不动点定理是一个非常重要的结果,它描述了以下情况:给定一个赋范线性空间,如果一个连续线性算子在这个空间上有一个不动点,那么这个不动点就是唯一的。

换句话说,Banach不动点定理表明,如果一个函数在某个空间上的定义域内有一个不动点,那么这个不动点就是该函数在该空间上的唯一驻点。

让我们来看看这个定理的证明。

假设X是一个赋范线性空间,T是X上的一个线性算子。

设P是T的不动点。

我们首先需要证明P是唯一的。

为此,我们需要构造一个等价关系(或者说是有序关系)π(x) = π(y)当且仅当x-y = ε时与P有关的等价关系。

为了实现这一点,我们需要使用线性映射的极限性质。

假设T的限制TT(x)和T的限制TT(y)都存在。

由于T是连续的,我们可以得出x-y属于T的定义域,即存在ε> 0使得T(x-y) = ε。

由于T是线性的,我们可以得出TT(x-y) = T(ε) = 0。

因此,如果π(x) = π(y),那么x-y = ε成立。

因此,我们得到了一个等价关系π(x) = π(y)当且仅当x-y = ε,这与我们的定义相符。

现在假设存在另一个点Q属于T的定义域,并且Q与P不等价。

这意味着存在ε> 0使得Q-P = ε成立。

这意味着存在两个不同的点x和y满足x-y = ε。

这意味着存在ε/2 > 0使得x-y的补集与π(x)的补集与π(y)的补集都不相等。

根据我们的假设T的定义域的定义和π的定义,我们有Tx -Ty = ε/2,这意味着x-y=ε/2并不成立,这显然是矛盾的。

因此Q不能属于T的定义域,这证明了唯一性P和Q不唯一π的实例点定义集合σπ表示所有的实例点的集合它构成π的一度划分所以所有P与T都重合不含有异类的其他成员σπ对每个pi也这样根据前一个论证显然这已经说明了我们的第一步骤的所有关键要素——X的一个赋范线性子空间S=XT且该子空间对π是第一度划分π对S的所有实例点构成σπ并且所有实例点都属于S这就是Banach不动点定理的证明过程。

Banach空间及其相关定理

Banach空间及其相关定理

课程论文课程现代分析基础学生姓名学号院系专业指导教师二O一五年十二月四日目录1 绪论 (1)2 Banach空间基本概念 (1)2.1拟范数定义及例子 (1)2.2 Banach空间 (2)2.3 Banach空间中线性变换及其性质 (3)3 一致有界定理及其推论 (4)3.1问题 (4)3.2基本概念 (4)3.3一致有界定理及其推论 (5)3.4一致有界性定理及其推论的应用 (6)4 Hahn-Banach定理与凸集分离定理 (7)4.1实线性空间上的Hahn-Banach定理 (7)4.2复线性空间上的Hahn-Banach定理 (8)4.3赋范线性空间上的Hahn-Banach定理 (8)4.4有关Hahn-Banach定理的一些推论 (9)4.5 Hahn-Banach定理的几何形式:凸集分离定理 (9)5 Banach空间中开映射、闭图像定理以及逆算子定理 (9)5.1开映射定理 (10)5.2逆算子定理 (11)5.3闭图像定理 (12)6 总结 (14)参考文献 (16)Banach空间及其相关定理南京理工大学自动化学院,江苏南京摘要:本文的主要是介绍了Banach空间以及其相关定理。

首先,本文讲了Banach空间产生的背景以及应用领域。

然后本文介绍了Banach空间的基本概念及其相关性质。

最后本文开始从一致有界定理开始,将Banach空间中Hahn-Banach定理、开映射、闭图像以及逆算子定理这几个重要定理逐一做出介绍并给出相应定理的证明。

关键词:Banach空间;一致有界定理;Hahn-Banach定理;开映射、闭图像、逆算子定理1 绪论巴拿赫空间(Banach space)是一种赋有“长度”的线性空间,泛函分析研究的基本对象之一。

数学分析各个分支的发展为巴拿赫空间理论的诞生提供了许多丰富而生动的素材。

从魏尔斯特拉斯,K.(T.W.)以来,人们久已十分关心闭区间[a,b]上的连续函数以及它们的一致收敛性。

霍恩子句 规则

霍恩子句 规则

在数理逻辑中,霍恩子句是带有最多一个肯定文字的子句(文字的析取)。

霍恩子句的合取是合取范式,也叫做霍恩公式。

霍恩子句在逻辑编程中扮演基本角色并且在构造性逻辑中很重要。

霍恩子句对定理证明的实用性是一阶归结提供的,两个霍恩子句的归结是一个霍恩子句。

在自动定理证明中,这能导致子句的在计算机上表示得更加高效。

实际上,Prolog就是完全在霍恩子句上构造的编程语言。

霍恩子句在计算复杂性中也是关键的,在这里找到一组变量指派使霍恩子句的合取的为真的问题是一个P-完全问题,有时叫做HORNSAT。

这是布尔可满足性问题的P 的变体,它是一个中心的NP-完全问题。

Banach空间中非线性算子半群的遍历理论的开题报告

Banach空间中非线性算子半群的遍历理论的开题报告

Banach空间中非线性算子半群的遍历理论的开题报

本文将介绍Banach空间中非线性算子半群的遍历理论的开题报告。

遍历理论是研究动力学系统的重要分支,其研究对象是非线性算子半群的性质和行为。

这里我们主要关注Banach空间中的情况。

首先,我们将介绍非线性算子半群和Banach空间的基本概念。

非线性算子半群是指一类由非线性算子组成的函数族,它们在时间上是连续的、单调的、递增的,并且在幂次下满足半群性质。

Banach空间是一种完备的、有范数的向量空间,通常用来描述无限维空间中的线性映射。

接着,我们将介绍遍历理论的基本思想。

遍历是指非线性算子半群中一种重要的动态行为,它表示系统在长时间内不断地游走于整个空间中,而不停留在某个有限的子集中。

遍历理论主要研究如何刻画系统的遍历行为以及如何构造遍历解。

最后,我们将介绍最新的研究成果和研究方向。

当前的研究工作主要集中在Banach空间中非线性算子半群的遍历性质和遍历解的存在性问题上。

流行的方法包括刻画遍历性质的充分条件、构造遍历解的方法以及研究遍历解的稳定性等。

总之,本文将介绍Banach空间中非线性算子半群的遍历理论的基本思想和最新研究成果,旨在为进一步研究非线性算子半群在动力学系统和应用数学中的应用提供基础和参考。

  1. 1、下载文档前请自行甄别文档内容的完整性,平台不提供额外的编辑、内容补充、找答案等附加服务。
  2. 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
  3. 3、如文档侵犯您的权益,请联系客服反馈,我们会尽快为您处理(人工客服工作时间:9:00-18:30)。

(1.1)
i=1
αi xi ∼
C
n
αi yi
i=1
for all n and scalars α1 , . . . , αn .
(Equivalently, there exists an invertible linear map T : [xi ] → [yi ] with T , T −1 ≤ C , where [xi ] denotes the closed linear span of (xi ).) (xj ) is called unconditional if there is a constant u so that for n any n and scalars c1 , . . . , cn and ε1 , . . . , εn with |εi | = 1 for all i, ci xi (then i=1 εi ci xi ≤ u p one says (xj ) is u-unconditional). The usual ℓ -basis refers to the unit vector basis (ej ) of ℓp , where ej (i) = δji for all i and j . Our main result goes as follows. Theorem 1.1. Let N be a finite von Neumann algebra, 1 ≤ p < 2, and let (xij ) be an infinite matrix in Lp (τ ) where τ is a fixed faithful normal tracial state on N . Assume for some C ≥ 1 that every row and column of (xij ) is C -equivalent to the usual ℓ2 -basis and that (xik ,jk )∞ k=1 is unconditional, whenever i1 < i2 < · · · and j1 < j2 < · · · . Then there exist i1 < i2 < · · · and j1 < j2 < · · · so that setting yk = xik ,jk for all k , then
BANACH EMBEDDING PROPERTIES OF NON-COMMUTATIVE Lp -SPACES
U. HAAGERUP, H.P. ROSENTHAL AND F.A. SUKOCHEV
arXiv:math/0005150v1 [math.FA] 15 May 2000
Abstract. Let N and M be von Neumann algebras. It is proved that Lp (N ) does not Banach embed in Lp (M) for N infinite, M finite, 1 ≤ p < 2. The following considerably stronger result is obtained (which implies this, since the Schatten p-class Cp embeds in Lp (N ) for N infinite). Theorem. Let 1 ≤ p < 2 and let X be a Banach space with a spanning set (xij ) so that for some C ≥ 1, (i) any row or column is C -equivalent to the usual ℓ2 -basis, (ii) (xik ,jk ) is C -equivalent to the usual ℓp -basis, for any i1 < i2 < · · · and j1 < j2 < · · · . Then X is not isomorphic to a subspace of Lp (M), for M finite. Complements on the Banach space structure of non-commutative Lp -spaces are obtained, such as the p-Banach-Saks property and characterizations of subspaces of Lp (M) containing ℓp isomorphically. The spaces Lp (N ) are classified up to Banach isomorphism, for N infinite-dimensional, hyperfinite and semifinite, 1 ≤ p < ∞, p = 2. It is proved that there are exactly thirteen isomorphism types; the corresponding embedding properties are determined for p < 2 via an eight level Hasse diagram. It is also proved for all 1 ≤ p < ∞ that Lp (N ) is completely isomorphic to Lp (M) if N and M are the algebras associated to free groups, or if N and M are injective factors of type IIIλ and IIIλ′ for 0 < λ, λ′ ≤ 1.
Contents §1. §2. §3. §4. §5. §6. §7. Introduction. The modulus of uniform integrability and weak compactness in L1 (N ). Proof of the Main Theorem. Improvements to the Main Theorem. Complements on the Banach/operator space structure of Lp (N )-spaces. The Banach isomorphic classification of the spaces Lp (N ) for N hyperfinite semi-finite. Lp (N )-isomorphism results for N type III hyperfinite or a free group von Neumann algebra. References 1. Introduction Let N be a finite von Neumann algebra and 1 ≤ p < 2. Our main theorem yields that Cp is not linearly isomorphic to a subspace of Lp (N ) (where Cp denotes the Schatten p-class). It follows immediately that for any infinite von Neumann algebra M, Lp (M) is not isomorphic to a subspace of Lp (N ), since Cp is then isomorphic to a subspace of Lp (M). (It is proved in [S1] that also Cp does not embed in Lp (N ) for any 2 < p < ∞.) For N a semi-finite von-Neumann algebra and τ a faithful normal semi-finite trace on N , Lp (τ ) denotes the non-commutative Lp space associated with (N , τ ) (see e.g., [FK]). The particular choice of trace τ is unimportant, for if β is another such trace, Lp (β ) is isometric to Lp (τ ). We also denote this (isometrically unique) Banach space by Lp (N ).
n
(1.2)
′ for all subsequences (yk ) of (yk ).
n→∞
lim n−1/p
i=1
′ yi
L p (τ )
=0
Corollary 1.2. Let p and N be as in 1.1. Let X be a Banach space spanned by an infinite matrix of elements (xij ) so that for some λ ≥ 1, (i) every row and column of (xij ) is λ-equivalent to the usual ℓ2 basis p (ii) (xin ,jn )∞ n=1 is λ-equivalent to the usual ℓ -basis, for all i1 < i2 < · · · and j1 < j2 < · · · . Then X is not Banach isomorphic to a subspace of Lp (τ ). In particular, Cp does not embed in Lp (τ ). The Corollary yields its final statement since the standard matrix units (xij ) for Cp satisfy (i) and (ii) with λ = 1. To see why 1.1 =⇒ 1.2, suppose to the contrary that T : X → X ′ ⊂ Lp (τ ) were an isomorphic embedding, where X is as in 1.2. Then (T xij ) satisfies the hypotheses of 1.1 with C = λ T T −1 . However if (ik ), (jk ) satisfies the conclusion of Theorem 1.1, (T xik ,jk ) and hence (xik ,jk ) cannot be equivalent to the usual ℓp -basis, a contradiction. Let Rad Cp denote the “Rademacher unconditionalized version” of Cp (1 ≤ p < ∞). That is, letting (rij ) be an independent matrix of {1, −1}-valued random variables with P (rij = 1) = P (rij = −1) = 1 2 for all i, j , and letting (cij ) be a matrix of scalars with only finitely many non-zero terms, then (1.3) (cij )
相关文档
最新文档