亚洲免费av电影一区二区三区,日韩爱爱视频,51精品视频一区二区三区,91视频爱爱,日韩欧美在线播放视频,中文字幕少妇AV,亚洲电影中文字幕,久久久久亚洲av成人网址,久久综合视频网站,国产在线不卡免费播放

        ?

        隨機矩陣特征值新蓋爾型包含集(英)

        2021-01-09 02:44:18trace
        工程數(shù)學(xué)學(xué)報 2020年6期
        關(guān)鍵詞:現(xiàn)代史面向社會蓋爾

        (A)=trace(A)+(n ?1)(A)?1,

        |μ?0.2255|≤1.6380.

        |μ+0.2008|≤1.3461.

        |μ|≤0.5846.

        1 Introduction

        As it is well known,stochastic matrices and the eigenvalue localization of stochastic matrices play the central role in many application fields such as birth and death processes, computer aided geometric design and Markov chain, see [1-6]. An nonnegative matrix A=(aij)∈Rn×nis called a row stochastic matrix(or simply stochastic matrix)if for each i ∈N ={1,2,··· ,n},that is,the sum of each row is 1. Since the row sum condition can be written as Ae=1e,we find that 1 is an eigenvalue of A with a corresponding eigenvector e=[1,1,··· ,1]T.It follows from the Perron-Frobenius theorem[7]that |λ| ≤1,λ ∈σ(A). In fact, 1 is a dominant eigenvalue for A. Furthermore, η is called a subdominant eigenvalue of a stochastic matrix if η is the second-largest modulus after 1[8,9].

        In 2011,Cvetkovi′c et al[10]discovered the following region including all eigenvalues different from 1 of a stochastic matrix A by refining the Gersgorin circle[11]of A.

        Theorem 1[10]Let A = (aij) ∈Rn×nbe a stochastic matrix, and let si, i = 1,2,··· ,n be the minimal element among the off-diagonal entries of the ith column of A.Taking γ(A)=maxi∈N(aii?si), if λ ?=1 is an eigenvalue of A, then

        However, although (1) of Theorem 1 provides a circle with the center γ(A) and radius r(A) to localize the eigenvalue λ different from 1, it is not effective when A is stochastic, and aii=si=0, for each i ∈N.

        Recently, in order to conquer this drawback, Li and Li[12]obtained the following modified region including all eigenvalues different from 1.

        Theorem 2[12]Let A = (aij) ∈Rn×nbe a stochastic matrix, and let Si, i = 1,2,··· ,n be the maximal element among the off-diagonal entries of the ith column of A. Taking(A)=maxi∈N(Si?aii), if λ ?=1 is an eigenvalue of A, then

        Note that (2) of Theorem 2 provides a circle with the center ?(A) and radius(A).

        In general, the circle of Theorem 1 and Theorem 2 can be large when n is large. It is very interesting how to provide a more sharper Gersgorin circle than those in[10,12].

        In this paper,we will continue to investigate the eigenvalue localization for stochastic matrices and present a new and simple Gersgorin circle set that consists of one disk.Moreover, an algorithm is obtained to estimate an upper bound for the modulus of subdominant eigenvalues of a positive stochastic matrix. Numerical examples are also given to show that our results are more effective than those in [10,12].

        2 A new eigenvalue localization for stochastic matrices

        Here matrices A with constant row or column sum are considered,that is,Ae=λe or ATe = λe for some λ ∈R. Obviously, λ is an eigenvalue of A, when λ = 1, it is stochastic.

        To obtain a new set including all eigenvalues different from 1 of a stochastic matrix,we start with the following propositions.

        Proposition 1[10]Let A=(aij)∈Rn×nbe such that ATe=λe, for any d=[d1,d2,··· ,dn]T∈Rn×n, let μ∈σ(A){λ} . Then ?μ is also an eigenvalue of the matrix

        Applying this result to AT, we have:

        Proposition 2[10]Let A = (aij) ∈Rn×nbe such that Ae = λe, for any d = [d1,d2,··· ,dn]T∈Rn×n, let μ∈σ(A){λ}. Then ?μ is also an eigenvalue of the matrix

        It is easy to see that the best choice of each diin Proposition 1 and Proposition 2 should minimize the radius of the ith Gersgorin row circle of C and B. In particular,we present the following choice for an nonnegative matrix

        which not only leads to a reduction of the radii of the Gersgorin circles,but also localizes all eigenvalues in one circle.

        Firstly, let us introduce some notations. For an nonnegative matrix A = (aij) ∈Rn×n, which is a stochastic matrix, and B = diag{d1,d2,··· ,dn}eeT?AT, where di=aii, define, for i=1,2,··· ,n,

        The main result of this paper is the following theorem.

        Theorem 3Let A=(aij)∈Rn×nbe a stochastic matrix,taking di=aii, i ∈N,if λ ?=1 is an eigenvalue of A, then:

        (I) If N1=N, then |λ|≤ˉr(A)=trace(A)?1;

        (II) If N2=N, then |λ|≤ˉr(A)=1 ?trace(A);

        (III) If N3?=?, then

        Proof Since di=aii, i ∈N,and B =diag{d1,d2,··· ,dn}eeT?AT,we have that for any i ∈N, and j ?=i,

        如自學(xué)習(xí)《鴉片戰(zhàn)爭》時,鴉片戰(zhàn)爭打開了我國的大門,為我國帶來了侵略、傷害。但同時進了我國自然經(jīng)濟的解題,讓我過從封閉天國轉(zhuǎn)變出來,開始面向社會。它也成為了我國現(xiàn)代史的開端,所以學(xué)生不能夠從單一的角度去認(rèn)識和學(xué)習(xí)它,而要從辯證的角度去看待它。

        Firstly, we consider the following two special cases (I) and (II). From (3), we have:(I) If N1=N, for each i ∈N, j ?=i, it has

        so, by (4), Proposition 2 and Gersgorin circle theorem, we have

        (II) If N2=N, similarly, for each i ∈N, j ?=i,

        then

        In general, that is, the following case (III), we have:

        (III) If N3?=?, so, by (4), Proposition 2 and Gersgorin circle theorem, we obtain

        The proof is completed.

        Remark 1Note that the special cases (5) and (6) of Theorem 3 provide circles with the center 0 and radius equal to trace(A)?1 and 1 ?trace(A), respectively. The two bounds are not sharper than (1) of Theorem 1 and (2) of Theorem 2. However,more generally, for case (III), we have

        Remark 21) Consider the following matrix, it illustrates that the bound provided by Theorem 3 is sharp. Take the matrix

        Then

        by Theorem 3, |λ|≤0. In fact, the eigenvalues of A different from 1 are 0.2) Furthermore, consider the following matrix

        By Theorem 1, for any μ ∈σ(A)1, we have |μ?0.3096| ≤1.1234. By Theorem 2,we have |μ+0.1273| ≤0.6242. By Theorem 3, we have μ| ≤0.3776. These regions are shown in Figure 1. It is easy to see that the region in Theorem 3 localizing all eigenvalues different from 1 of A is better than those in Theorem 1 and Theorem 2.

        Figure 1 The region |μ|≤0.3776 is represented by the innermost circle

        In order to further compare obtained results, we consider the following stochastic matrix generated by the Matlab code

        By Theorem 1, for any μ∈σ(A)1, we have

        |μ?0.2255|≤1.6380.

        By Theorem 2, we have

        |μ+0.2008|≤1.3461.

        By Theorem 3, we have

        |μ|≤0.5846.

        These regions are shown in Figure 2, obviously,our result is better than those got from Theorem 1 and Theorem 2 in some cases.

        Figure 2 The region |μ|≤0.5846 is represented by the innermost circle

        3 Comparison of the subdominant eigenvalue of stochastic matries

        Note that if A is stochastic, then Amis also stochastic for any positive integer m.Cvetkovi′c et al[10], applied Theorem 1 to obtain

        and proved that the sequences {γm(A)} and rm(A) all converge to 0, where the value γm(A)=γ(Am).

        Algorithm 1Given a positive stochastic matrix A=(aij)∈Rn×nand a positive integer T, for t=1, do the following:

        1) Set m=2t?1;

        2) Compute Bm=diag{d1,d2,··· ,dn}eeT?(Am)T;

        3) If N1=N, computem(A)=trace(Am)?1;

        4) If N2=N, computem(A)=1 ?trace(Am);

        5) If N3?=?, computem(A)=maxi∈NCi(Bm);

        6) Set A=A×A and t=t+1. If t>T, output r, stop, otherwise, go to 1).

        Example 1Stochastic matrices A1, A2are the same as in [10,12].

        Using Matlab, we can compute vm(A),m(A) andm(A), see Table 1 and Table 2,respectively.

        Table 1 The value for A1 when m=2t, t=0,1,2

        By Matlab computations, the eigenvalues of A1are 0.1634,0.2094±0.1109i, and 0.1732.

        Table 2 The values for A2 when m=2t, t=0,1,2,3,4

        By Matlab computations, the eigenvalues of A2are 0.7934,?0.3683±0.0088i, and 0.1936. We can observe that Theorem 3 is better than those in [10,12].

        Example 2

        Matrix A3is the example in [10] with xi= i/10, i ∈{1,2,3,4}. By computations,vm(A3),m(A3) andm(A3) are shown in Table 3.

        Table 3 The values of vm(A3), m(A3) and m(A3)

        Table 3 The values of vm(A3), m(A3) and m(A3)

        m vm(A1) ~vm(A1) rm(A1)4 0.9505 0.8604 0.9481 8 0.8951 0.8395 0.8218 16 0.8310 0.8062 0.7906 32 0.7757 0.7705 0.7662 64 0.7682 0.7660 0.7591

        By Matlab computations, the second-largest eigenvalue of A3is 0.7513. It is not difficult to see from this example that our bound performs better than that in [10,12].

        猜你喜歡
        現(xiàn)代史面向社會蓋爾
        走丟的車輪奶酪
        陜西省檔案館面向社會公開征集防控新冠肺炎疫情檔案
        陜西檔案(2020年1期)2020-04-14 06:07:16
        迷霧中的蓋爾瑪
        暴力之旅——伊恩·麥克蓋爾訪談錄
        蓋爾瑪 百年老企轉(zhuǎn)戰(zhàn)直銷
        關(guān)于我國高校圖書館面向社會開放現(xiàn)狀調(diào)查
        文體求索與寓意取向
        雪蓮(2015年11期)2015-11-25 08:17:59
        試論高等體育院校圖書館面向社會開放的特殊意義
        體育師友(2011年4期)2011-03-20 15:29:43
        略談世界現(xiàn)代史的開端問題
        《中國現(xiàn)代史地圖集》評介
        軍事歷史(2000年5期)2000-08-21 05:38:14
        国产精品白丝喷水在线观看| 亚洲视频免费在线观看| 四虎成人精品在永久免费| 无码午夜成人1000部免费视频| 人妻丰满多毛熟妇免费区| 久久99热精品免费观看麻豆| 开心久久综合婷婷九月| 在线成人爽a毛片免费软件 | 91免费在线| 日韩极品免费在线观看| 麻豆精品一区二区av白丝在线| 午夜性无码专区| 亚洲成av人片天堂网九九| 国产自产在线视频一区| 精品国产午夜肉伦伦影院| 青青草97国产精品免费观看| 亚洲欧洲久久久精品| 精品日韩在线观看视频| 午夜成人理论福利片| 国产裸体歌舞一区二区| 欧美在线Aⅴ性色| 亚洲av成熟国产一区二区| 久久99精品国产麻豆不卡| 久久精品无码一区二区三区不| 玖玖资源网站最新网站| 成人欧美一区二区三区黑人| 国产av无码专区亚洲av琪琪| 国产日韩欧美视频成人| 日本va中文字幕亚洲久伊人| 男人的天堂无码动漫av| 9999精品视频| 久久久国产精品首页免费| 亚洲另类无码专区首页| 国产人澡人澡澡澡人碰视频| 国产免费一区二区av| 国产91精品高潮白浆喷水| 亚洲av无码av男人的天堂| 日本香蕉久久一区二区视频| 视频女同久久久一区二区| 久久久久亚洲av成人无码| 精品亚洲日韩国产一二三区亚洲|