Trimmed minimax estimator of a covariance matrix

Summary In the problem of estimating the covariance matrix of a multivariate normal population, James and Stein (Proc. Fourth Berkeley Symp. Math. Statist. Prob.,1, 361–380, Univ. of California Press) obtained a minimax estimator under a scale invariant loss. In this paper we propose an orthogonally invariant trimmed estimator by solving certain differential inequality involving the eigenvalues of the sample covariance matrix. The estimator obtained, truncates the extreme eigenvalues first and then shrinks the larger and expands the smaller sample eigenvalues. Adaptive version of the trimmed estimator is also discussed. Finally some numerical studies are performed using Monte Carlo simulation method and it is observed that the trimmed estimate shows a substantial improvement over the minimax estimator. The second author's research was supported by NSF Grant Number MCS 82-12968.     

A minimax optimal control problem

A control system x=f(t,x,u) is considered, and a cost functional ess supT ₀tT ₁ G(t, x(t),u(t)) is to be minimized. Necessary conditions for optimality (maximum principle and transversality conditions) are derived. It is also shown that an optimal control is optimal for the corresponding problem on a subinterval of [T ₀,T ₁], if a certain controllability condition is satisfied.     

Some alternative principles and mixed minimax theorems involving two functions

Some alternative principles and quantitative-topological minimax theorems involving two functions are given. Consequently, Geraghty–Lin's minimax theorem and Terkelsen's minimax theorem are generalized. This revised version was published online in June 2006 with corrections to the Cover Date.     

Local asymptotic normality and asymptotical minimax efficiency of the MLE under random censorship

Here we study the problems of local asymptotic normality of the parametric family of distributions and asymptotic minimax efficient estimators when the observations are subject to right censoring. Local asymptotic normality will be established under some mild regularity conditions. A lower bound for local asymptotic minimax risk is given with respect to a bowl-shaped loss function, and furthermore a necessary and sufficient condition is given in order to achieve this lower bound. Finally, we show that this lower bound can be attained by the maximum likelihood estimator in the censored case and hence it is local asymptotic minimax efficient.     

The credibility premiums for exponential principle

In the classical credibility theory, the credibility premium is derived on the basis of pure premium. However, the insurance practice demands that the premium must be charged under some adaptable premium principle and serves the purpose for insurance business. In this paper, the balanced credibility models have been built under exponential principle, and the credibility estimator of individual exponential premium is derived. This result is also extended to the versions of multitude contracts, and the estimation of the structure parameters is investigated. Finally, the simulations have been introduced to show the consistency of the credibility estimator and its differences from the classical one.     

Algorithms for the solution of stochastic dynamic minimax problems

In this paper, we present algorithms for the solution of the dynamic minimax problem in stochastic programs. This dynamic minimax approach is suggested for the analysis of multi-stage stochastic decision problems when there is only partial knowledge on the joint probability distribution of the random data. The algorithms proposed in this paper are based on projected sub-gradient and bundle methods.

---

Résumé Dans cet article, nous proposons des algorithmes pour la solution du problème du minimax dynamique stochastique. Ce problème se présente par exemple lorsque, dans un problème de décision dynamique stochastique, l'information disponible au sujet des distributions de probabilité des paramètres est incomplète. Les algorithmes proposés sont fondés sur la méthode de sous-gradient projeté et la méthode des faisceaux.

     

Maximum principle and second-order conditions for minimax problems of optimal control

In this paper, we study the optimal control problem of minimizing the functionalJ(x, u)=max_t1tt2(x(t),t). We formulate and prove necessary optimality conditions for this problem. We establish the equivalence between the initial minimax problem and a problem involving a terminal functional and phase constraints.     

Minimax and admissible minimax estimators of the mean of a multivariate normal distribution for unknown covariance matrix

Let X be a p-variate (p ≥ 3) vector normally distributed with mean μ and covariance Σ, and let A be a p × p random matrix distributed independent of X, according to the Wishart distribution W(n, Σ). For estimating μ, we consider estimators of the form δ = δ(X, A). We obtain families of Bayes, minimax and admissible minimax estimators with respect to the quadratic loss function (δ ? μ)′ Σ^?1(δ ? μ) where Σ is unknown. This paper extends previous results of the author [1], given for the case in which the covariance matrix of the distribution is of the form σ²I, where σ is known.     

The effective minimax value of asynchronously repeated games

We study the effect of asynchronous choice structure on the possibility of cooperation in repeated strategic situations. We model the strategic situations as asynchronously repeated games, and define two notions of effective minimax value. We show that the order of players moves generally affects the effective minimax value of the asynchronously repeated game in significant ways, but the order of moves becomes irrelevant when the stage game satisfies the non-equivalent utilities (NEU) condition. We then prove the Folk Theorem that a payoff vector can be supported as a subgame perfect equilibrium outcome with correlation device if and only if it dominates the effective minimax value. These results, in particular, imply both Lagunoff and Matsuis (1997) result and Yoon (2001)s result on asynchronously repeated games.I thank three anonymous referees as well as an associate editor for many helpful comments and suggestions. This research was supported by the Suam Foundation Research Grants.Received: October 2001     

Ekeland's variational principle, minimax theorems and existence of nonconvex equilibria in complete metric spaces

In this paper, we introduce the concept of τ-function which generalizes the concept of w-distance studied in the literature. We establish a generalized Ekeland's variational principle in the setting of lower semicontinuous from above and τ-functions. As applications of our Ekeland's variational principle, we derive generalized Caristi's (common) fixed point theorems, a generalized Takahashi's nonconvex minimization theorem, a nonconvex minimax theorem, a nonconvex equilibrium theorem and a generalized flower petal theorem for lower semicontinuous from above functions or lower semicontinuous functions in the complete metric spaces. We also prove that these theorems also imply our Ekeland's variational principle.     

Parametric Borwein-Preiss variational principle and applications

A parametric version of the Borwein-Preiss smooth variational principle is presented, which states that under suitable assumptions on a given convex function depending on a parameter, the minimum point of a smooth convex perturbation of it depends continuously on the parameter. Some applications are given: existence of a Nash equilibrium and a solution of a variational inequality for a system of partially convex functions, perturbed by arbitrarily small smooth convex perturbations when one of the functions has a non-compact domain; a parametric version of the Kuhn-Tucker theorem which contains a parametric smooth variational principle with constraints; existence of a continuous selection of a subdifferential mapping depending on a parameter.

The tool for proving this parametric smooth variational principle is a useful lemma about continuous -minimizers of quasi-convex functions depending on a parameter, which has independent interest since it allows direct proofs of Ky Fan's minimax inequality, minimax equalities for quasi-convex functions, Sion's minimax theorem, etc.

     

指数保费原理下的经验厘定

在经典的信度理论中,信度保费是在净保费原理下得到的. 但是, 保险商业中, 保险公司要求制定的保费必须适用于某合适的保费原理以适应具体的保险商业的需要. 本文建立了指数保费原理下的完全经验厘定模型, 得到了风险保费的信度估计和经验Bayes 信度估计, 并讨论了结构参数的估计及其性质. 最后证明了多合同模型的经验Bayes 信度估计的渐近最优性     

Asymptotic minimax properties of M-estimators of scale

We ask whether or not the saddlepoint property holds, for robust M-estimation of scale, in gross-errors and Kolmogorov neighbourhoods of certain distributions. This is of interest since the saddlepoint property implies the minimax property — that the supremum of the asymptotic variance of an M-estimator is minimized by the maximum likelihood estimator for that member of the distributional class with minimum Fisher information. Our findings are exclusively negative — the saddlepoint property fails in all cases investigated.     

非凸极小极大问题的优化算法与复杂度分析

非凸极小极大问题是近期国际上优化与机器学习、信号处理等交叉领域的一个重要研究前沿和热点,包括对抗学习、强化学习、分布式非凸优化等前沿研究方向的一些关键科学问题都归结为该类问题。国际上凸-凹极小极大问题的研究已取得很好的成果,但非凸极小极大问题不同于凸-凹极小极大问题,是有其自身结构的非凸非光滑优化问题,理论研究和求解难度都更具挑战性,一般都是NP-难的。重点介绍非凸极小极大问题的优化算法和复杂度分析方面的最新进展。     

The minimax approach for a class of variable order fractional differential equation

This paper introduces an approximate solution for Liouville‐Caputo variable order fractional differential equations with order 0 < α(t) ≤ 1 . The solution is adapted using a family of fractional‐order Chebyshev functions with unknown coefficients. These coefficients have been obtained by using an optimization approach based on minimax technique and the least pth optimization function. Several linear and nonlinear fractional‐order differential equations are discussed using the proposed technique for fixed and variable order fractional‐order derivatives. Moreover, the response of RC charging circuit with variable order fractional capacitor is studied for different cases. Several comparisons with related published techniques have been added to illustrate the accuracy of the proposed approach.     

方差分量模型中回归系数的线性Minimax估计

考虑方差分量(混合线性)模型y=Xβ+U1ξ1+U2ξ2+…+Ukξk,这里Xn×p,Ui,n×ti为已知设计矩阵,βp×1是固定效应,iξ是ti×1随机效应向量,满足E(iξ)=0,cov(iξ)=σ2iIti,iξ都不相关.往往Uk=In,ξk=ek,即最后一项为随机误差,热β∈RP和i2σ>0(i=1,2,…,k)为未知参数.我们考虑β的可估函数Sβ,选取二次损失函数L(d,Sβ)=(d-Sβ)′(d-Sβ)∑ki=1ciσi2+β′X′Vk-1Xβ,然后在线性估计类中给出Sβ的惟一的mini max估计.     

Note on André's reflection principle

This note generalizes André's reflection principle to give a new combinatorial proof of a formula for the number of lattice paths lying within certain trapezoids.     

On the invariance principle for U-statistics

Let T_n be a U-statistic and S_n its projection (in the sense of Hájek). Limit theory for U-statistics is usually considered in two disjoint cases, termed degenerate and nondegenerate. The traditional method is to treat the cases separately, using different techniques in each to obtain a solution. Here we present a unified treatment based on a joint invariance principle for the vector (T_n, T_n – S_n), from which the invariance principles in both the degenerate and nondegenerate cases follow as easy corollaries.