非线性算子具误差的隐迭代程序的强收敛性

设K是一致凸Banach空间中的非空闭凸子集,T_i:K→K(i=1,2,…,N)是有限族完全渐近非扩张映象.对任意的x_0∈K,具误差的隐迭代序列{x_n}为:x_n=α_nx_n-1+β_nT_n~kx_n+γ_nu_n,n≥1,其中{α_n},{β_n},{γ_n}■[0,1]满足α_n+β_n+γ_n=1,{u_n}是K中的有界序列.在一定的条件下,该文建立了隐迭代序列{x_n}的强收敛性.得到隐迭代序列{x_n}强收敛于有限族完全渐近非扩张映象公共不动点的充要条件.所得结果改进和推广了Shahzad与Zegeye,Zhou与Chang,Chang,Tan,Lee与Chan等人的相应结果.     

Banach空间中极大单调算子零点的迭代收敛定理及应用

令E为实光滑、一致凸的Banach空间,E*为其对偶空间.令A E×E*为极大单调算子且A-10≠.假设{rn}(0,+∞)为实数列且满足rn→∞,n→∞,数列{αn}[0,1]满足∑∞n=1(1-αn)<+∞,对给定的向量xn∈E,寻找向量{x∧n}及{en}使之满足:αnJxn+(1-αn)Jen∈Jx∧n+rnAx∧n,其中{en}E为误差序列而且满足一定的限制条件.即而定义迭代序列{xn}n 1如下:xn+1=J-1[βnJx1+(1-βn)Jx∧n],n 1,其中数列{βn}[0,1]满足βn→0,n→∞且∑∞n=1βn=+∞,则{xn}强收敛于QA-10(x1),这里QA-10为从E到A-10上的广义投影算子.利用Lyapunov泛函,Qr算子与广义投影算子等新技巧,证明了引入的新迭代序列强收敛于极大单调算子A的零点,并讨论了此结论在求解一类凸泛函最小值上的应用.     

严格伪压缩映象的复合隐格式迭代序列的收敛率估计

设K是实Banach空间E的非空闭凸集,{Ti}iN=1:K→K是N个严格伪压缩映象且公共不动集F=∩Ni=1F(Ti)≠φ,其中F(Ti)={x∈K:Tix=x}.{αn}n∞=1,{βn}n∞=1[0,1]是实序列且满足条件:(i)sum from n=1 to ∞ (αn)(ii)lim(n→∞)αn=lim(n→∞)βn=0(iii)αnβnL2<1,n≥1其中L≥1是{Ti}iN=1的公共Lipschitz常数.对于任意的x0∈K,设{xn}n∞=1是由下列产生的复合隐格式迭代序列:xn=(1-αn)xn-1+αn Tnynyn=(1-βn)xn-1+βnTnxn其中Tn=Tn mod N,则{xn}强收敛到{Ti}iN=1的公共不动点.结果推广和改进了相关文献的结果,且主要定理的证明方法也是不同的.并且进一步给出了序列的收敛率估计.     

渐近非扩张非自映射的收敛定理(英文)

设E是实的一致凸Banach空间,K是E的一个非空闭凸集,P是E到K上的非扩张的保核收缩映射.设T1,T2,T3:K→E分别是具有数列{hn},{ln},{kn}[1,∞)的渐近非扩张非自映射,使得sum (hn-1) from n=1 to ∞<∞,sum ((ln-1)) from n=1 to ∞<∞及sum (n=1(kn-1) from n=1 to ∞<∞,且F=F(T1)∩F(T2)∩F(T3)={x∈K:T1x=T2x=T3x}≠Ф.定义迭代序列{xn}:x1∈K,xn+1=P((1-αn)xn+αnT1(PT1)n-1yn),yn=P((1-βn)xn+βnT2(PT2)n-1zn),zn=P((1-γn)xn+γnT3(PT3)n-1xn),其中{αn},{βn},{γn}[ε,1-ε],ε是大于零的实数.(i)如果T1,T2,T3中有一个是全连续的或者半紧的,则{xn}强收敛于某一点q∈F;(ii)如果E具有Frechet可微范数或者满足Opial’s条件或者E的对偶空间E~*具有Kadec-Klee性质,则{xn}弱收敛于某一点q∈F.     

￠—半压缩算子和￠—强增殖算子方程的迭代

设E是任意实Banach空间,K是E的非空闭凸子集,T：K→K是一致连续￠-半压缩映像且值域有界。设{an},{bn},{cn},{a'n},{b'n}和{c'n}是[0,1]中的序列且满足条件：Ⅰ）an bn cn=a'n b'n c'n=1,任意n≥0;Ⅱ）limbn=limb'n=limc'n=0;Ⅲ）∑n=0^∞bn=∞;Ⅳ）cn=o(bn).对任意给定的x0,u0,v0∈K,定义Ishikawa迭代{xn}如下：{xn 1=anxn bnTyn cnun,yn=a'nxn b'nTxn c'nvn(任意n≥0),其中{un}和{vn}是K中两个有界序列。则{xn}强收敛于T的唯一不动点。最后研究了￠-强增殖算子方程解的Ishikawa迭代收敛性。     

三步投影算法的收敛性及其在变分不等式组中的应用

First a general model for a three-step projection method is introduced, and second it has been applied to the approximation solvability of a system of nonlinear variational inequality problems in a Hilbert space setting. Let H be a real Hilbert space and K be a nonempty closed convex subset of H. For arbitrarily chosen initial points x0, y0, z0 ∈ K,compute sequences xn, yxn, zxn such that { xn+1=(1-αn-rn)xn+αxPk[yn-ρTyn]+rnun,yn=(1-β-δn)xn+βnPk[zn-ηTxn]+δnun,zn=(1-an-λn)xn+akPk[xn-γTxn]+λnwn.For η, ρ,γ>0 are constants,{αn}, {βn}, {an}, {rn}, {δn}, {λn} C [0,1], {un}, {vn}, {wn} are sequences in K, and 0≤n + rn ≤ 1,0 ≤βn + δn ≤ 1,0 ≤ an + λn ≤ 1,(A)n ≥ 0, where T : K → H is a nonlinear mapping onto K. At last three-step models are applied to some variational inequality problems.     

伪压缩映象的广义Halpern迭代

本文在具有一致G(a)teaux 可微范数的Banach空间中,利用一种广义Halpern迭代,研究了满足条件,C1)lim n→∞αn=0和C2)∞∑n=0αn=+∞的数列{αn)是可以充分保证伪压缩映象和B-P型严格伪压缩映象迭代序列有强收敛性.所得的结论改进和推广了相关的结果.     

伪压缩族公共不动点的复合隐格式迭代逼近问题

设H是一实Hillber空间,K是H之一非空间凸子集,设{Ti}Ni=1是N个Lipschitz伪压缩映象使得F=∩Ni=1F(Ti)≠0,其中F(Ti)={x∈K:Tix=x}并且{αn}n∞=1,{βn}∞n=1[0,1]是满足如下条件的实序列(i)∑∞n=1(1-αn)2= ∞;(ii)limn→∞(1-αn)=0;(iii)∑∞n=1(1-βn)< ∞;(iv)(1-αn)L2<1,n1;(v)αn(1-βn)2 αn[βn L(1-βn)]2<1,其中L1是{Ti}iN=1的公共Lipschitz常数,对于x0∈K,设{xn}n∞=1是由下列定义的复合隐格式迭代xn=αnxn-1 (1-αn)Tnyn,yn=βnxn (1-βn)Tnxn,其中Tn=TnmodN,则(i)limn→∞‖xn-p‖存在,对于所有的p∈F;(ii)limn→∞d(xn,f)存在,其中d(xn,F)=infp∈F‖xn-p‖;(iii)liminfn→∞‖xn-Tnxn‖=0.本文的结果推广并且改进H-K.Xu和R.G.Ori在2001年的结果和Osilike在2004年的结果,并且在这篇文章中,主要的证明方法也不同与H-K.Xu和Osilike的方法.     

求高阶常系数非齐次线性微分方程特解的新方法

求高阶常系数非齐次线性微分方程:y(n)+P1y(n-1)+…+Pny=f(x)(P1,P2,…,Pn是实数)的特解的一种新方法.首先将该方程降为n个一阶非齐次线性微分方程组:其中w1,w2,…,wn是对应的齐次方程的特征方程:tn+P1tn-1+…+Pn=0的n个根.然后得出了求原方程一个特解的迭代公式.     

高阶常系数非齐次线性微分方程的新解法

提出了高阶常系数非齐次线性微分方程y(n)+P1y(n-1)+…+Pny=f(x)(P1,P2,…,Pn是实数)的一种新解法.首先将该方程降为n个一阶非齐次线性微分方程组:y1′-w1y1=f(x),y2′-w2y2=y1,…………………yn′-wnyn=yn-1,其中w1,w2,…,wn是对应的齐次方程的特征方程tn+P1tn-1+…+Pn=0的n个根.然后求出它的通解y=yn,最后得出了求原方程一个特解的迭代公式.     

Iteration xn+1=αn+1f(xn) + (1 - αn+1)Tn+1xn for an Infinite Family of Nonexpansive Maps {Tn}∞n=1

Under the framework of uniformly smooth Banach spaces, Chang[1] proved in 2006 that the sequence {xn} generated by the iteration xn+1 = αn+1f(xn) + (1 - αn+1)Tn+1xn converges strongly to a common fixed point of a finite family of nonexpansive maps {Tn}, where f : C → C is a contraction. However, in this paper, the author considers the iteration in more general case that {Tn} is an infinite family of nonexpansive maps, and proves that Chang's result holds still in the setting of reflexive Banach spaces with the weakly sequentially continuous duality mapping.     

关于非扩张映象的不动点逼近的Ishikawa迭代程序

设E是一致凸Banach空间,满足Opial条件或具有Frechet可微范数.又设C是E的有界闭凸子集.若T:C→C是非扩张映象,则对任给的初始数据x₀∈C,由Ishikawa迭代程序x_n+1=t_nT(s_nTx_n+(1-s_n)x_n)+(1-t_n)x_n,n≥0,定义的序列{x_n}弱收敛到T的     

具误差隐格式迭代逼近严格伪压缩映像族公共不动点

设K是实Banach空间E中非空闭凸集, {Ti}i=1N是N个具公共不动点集F的严格伪压缩映像, {an}(?)[0,1]是实数列, {un}(?)K是序列,且满足下面条件设X0∈K,{xn}由下式定义xn=αnxn-1 (1-αn)Tnxn-un-1,n≥1其中Tn=TnmodN,则有下面结论(i)limn→∞‖xn-p‖存在,对所有P∈F; (ii)limn→∞d(xn,F)存在,当d(xn,F)=infp∈F‖xn-p‖; (iii)liminfn→∞‖xn-Tnxn‖=0．文中另一个结果是,如果{xn}(?){1-2-n,1},则{xn}收敛．文中结果改进与扩展了Osilike(2004)最近的结果,证明方法也不同．     

指数索赔情形下一类相依两险种风险模型的破产概率

考虑一类稀疏过程下索赔相依的两险种风险模型:U(t)=u+ct-∑i=1N2(t)X_i-∑i=1N2(t)Y_(i),其中{N_1(t),t≥0}、{N_2(t),t≥0}分别表示两个险种的索赔次数,它们按下述方式相关:N_1(t)N_(11)(t)+N_(12)(t),N_2(t)=N_(22)(t)+N'_(12)(t),{N'_(12)(t),t≥0}是{N_(12)(t),t≥0}的一个p-稀疏.考虑下列两种情形:(Ⅰ){N_(11)(t),t≥0}、{N_(12)(t),t≥0}、{N_(22)(t),t≥0}均为Poisson过程;(Ⅱ){N_(11)(t),t≥0}、{N_(22)(t),t≥0}为Poisson过程,{N_(12)(t),t≥0}为Erlang(2)过程.在上述两种情形下,当两险种的单次索赔额均服从指数分布时,通过建立并求解生存概率所满足的微分方程,给出其破产概率的表达式.     

THROREMS OF DISTORTION FOR SCHLICHT FUNCTIONS

Let \[f(z) = z + \sum\limits_{n = 1}^\infty {{a_n}{z^n} \in S} {\kern 1pt} {\kern 1pt} {\kern 1pt} and{\kern 1pt} {\kern 1pt} {\kern 1pt} \log \frac{{f(z) - f(\xi )}}{{z - \xi }} - \frac{{z\xi }}{{f(z)f(\xi )}} = \sum\limits_{m,n = 1}^\infty {{d_{m,n}}{z^m}{\xi ^n},} \], we denote \[{f_v} = f({z_v})\] , \[\begin{array}{l} {\varphi _\varepsilon }({z_u}{z_v}) = {\left| {\frac{{{f_u} - {f_v}}}{{{z_u} - {z_v}}}} \right|^\varepsilon }\frac{1}{{(1 - {z_u}{{\bar z}_v})}},\g_m^\varepsilon (z) = - {F_m}(\frac{1}{{f(z)}}) + \frac{1}{{{z^m}}} + \varepsilon {{\bar z}^m}, \end{array}\], where \({F_m}(t)\) is a Faber polynomial of degree m. Theorem 1. If \[f(z) \in S{\kern 1pt} {\kern 1pt} {\kern 1pt} and{\kern 1pt} {\kern 1pt} {\kern 1pt} \sum\limits_{u,v = 1}^N {{A_{u,v}}{x_u}{{\bar x}_v} \ge 0} \] and then \[\begin{array}{l} \sum\limits_{u,v = 1}^N {{A_{u,v}}{\lambda _u}{{\bar \lambda }_v}} {\left| {\frac{{{f_u} - {f_v}}}{{{z_u} - {z_v}}}} \right|^\varepsilon }\exp \{ \alpha {F_l}({z_u},{z_v})\} \ \le \sum\limits_{u,v = 1}^N {{A_{u,v}}{\lambda _u}{{\bar \lambda }_v}} \varphi _\varepsilon ^\alpha ({z_u}{z_v})l = 1,2,3, \end{array}\], where \[\begin{array}{l} {F_1}({z_u},{z_v}) = \frac{1}{2}\sum\limits_{n = 1}^\infty {\frac{1}{n}} g_n^\varepsilon ({z_u})\bar g_n^\varepsilon ({z_v}),\{F_2}({z_u},{z_v}) = \frac{1}{{1 + {\varepsilon _n}R{d_{n,n}}}}Rg_n^\varepsilon ({z_u})Rg_n^\varepsilon ({z_v}),\{F_3}({z_u},{z_v}) = \frac{1}{{1 - {\varepsilon _n}R{d_{n,n}}}}Rg_n^\varepsilon ({z_u})Rg_n^\varepsilon ({z_v}). \end{array}\] The \[F({z_u},{z_v}) = \frac{1}{2}{g_1}({z_u}){{\bar g}_2}({z_v})\] is due to Kungsun. Theorem 2. If \(f(z) \in S\) ,then \[P(z) + \left| {\sum\limits_{u,v = 1}^N {{A_{u,v}}{\lambda _u}{{\bar \lambda }_v}} {{\left| {\frac{{{f_u} - {f_v}}}{{{z_u} - {z_v}}}\frac{{{z_u}{z_v}}}{{{f_u}{f_v}}}} \right|}^\varepsilon }} \right| \le \sum\limits_{u,v = 1}^N {{\lambda _u}{{\bar \lambda }_v}} \frac{1}{{1 - {z_u}{{\bar z}_v}}}\], where \[\begin{array}{l} P(z) = \frac{1}{2}\sum\limits_{n = 1}^\infty {\frac{1}{n}} {G_n}(z),\{G_n}(z) = {\left| {\left| {\sum\limits_{n = 1}^N {{\beta _u}({F_n}(\frac{1}{{f({z_u})}}) - \frac{1}{{z_u^n}})} } \right| - \left| {\sum\limits_{n = 1}^N {{\beta _u}z_u^n} } \right|} \right|^2}, \end{array}\], \(P(z) \equiv 0\) is due to Xia Daoxing.     

ON ADMISSffilLITY OF VARIANCE COMPONENTS ESTIMATES

Suppose that there is a variance components model $$\[\left\{ {\begin{array}{*{20}{c}} {E\mathop Y\limits_{n \times 1} = \mathop X\limits_{n \times p} \mathop \beta \limits_{p \times 1} }\{DY = \sigma _2^2{V_1} + \sigma _2^2{V_2}} \end{array}} \right.\]$$ where $\[\beta \]$,$\[\sigma _1^2\]$ and $\[\sigma _2^2\]$ are all unknown, $\[X,V > 0\]$ and $\[{V_2} > 0\]$ are all known, $\[r(X) < n\]$. The author estimates simultaneously $\[(\sigma _1^2,\sigma _2^2)\]$. Estimators are restricted to the class $\[D = \{ d({A_1}{A_2}) = ({Y^''}{A_1}Y,{Y^''}{A_2}Y),{A_1} \ge 0,{A_2} \ge 0\} \]$. Suppose that the loss function is $\[L(d({A_1},{A_2}),(\sigma _1^2,\sigma _2^2)) = \frac{1}{{\sigma _1^4}}({Y^''}{A_1}Y - \sigma _1^2) + \frac{1}{{\sigma _2^4}}{({Y^''}{A_2}Y - \sigma _2^2)^2}\]$. This paper gives a necessary and sufficient condition for $\[d({A_1},{A_2})\]$ to be an equivariant D-asmissible estimator under the restriction $\[{V_1} = {V_2}\]$, and a sufficient condition and a necessary condition for $\[d({A_1},{A_2})\]$ to equivariant D-asmissible without the restriction.     

On a hybrid of bilinear Hilbert transform and paraproduct

We introduce a class of tri-linear operators that combine features of the bilinear Hilbert transform and paraproduct. For two instances of these operators, we prove boundedness property in a large range D = {(p1,p2,p_3) ∈ R~3 : 1 p1,p2 ∞,1/(p1)+ 1/(p2)3/2,1 p3 ∞}.