A mean–variance optimization problem for discounted Markov decision processes

In this paper, we consider a mean–variance optimization problem for Markov decision processes (MDPs) over the set of (deterministic stationary) policies. Different from the usual formulation in MDPs, we aim to obtain the mean–variance optimal policy that minimizes the variance over a set of all policies with a given expected reward. For continuous-time MDPs with the discounted criterion and finite-state and action spaces, we prove that the mean–variance optimization problem can be transformed to an equivalent discounted optimization problem using the conditional expectation and Markov properties. Then, we show that a mean–variance optimal policy and the efficient frontier can be obtained by policy iteration methods with a finite number of iterations. We also address related issues such as a mutual fund theorem and illustrate our results with an example.     

Uniform strength design of beams with non-zero minimum flexural rigidity under multiple loadings

The analytical method for uniform strength design (U.S.D.) of statically indeterminate beams is extended to deal with a preassigned constraint of non-zero minimum flexural rigidity and multiple load cases. And a numerical method of finding the U.S.D. is presented. Therefore we have a uniform procedure of solution of U.S.D. to beams with arbitrary cross-sectional shape and acted upon by several loadings and subjected to the constraint of minimum flexural rigidity.     

Asymptotic and essentially singular solutions of the Feigenbaum equation

For suitably defined largeN, we express Feigenbaum's equation as a singular Schroder functional equation whose solution is obtained using a scaling ansatz. In the limit of infiniteN certain self-consistency conditions on the scaled Schroder solution lead to an essentially singular solution of Feigenbaum's equation with a length scale factor of 0.0333 and. a limiting feigenvalue of 30.50, in agreement with Eckmann and Wittwer's value of =0.0333831... and their conjectured estimate of 30.     

Homotopic residual correction processes

We present and analyze homotopic (continuation) residual correction algorithms for the computation of matrix inverses. For complex indefinite Hermitian input matrices, our homotopic methods substantially accelerate the known nonhomotopic algorithms. Unlike the nonhomotopic case our algorithms require no pre-estimation of the smallest singular value of an input matrix. Furthermore, we guarantee rapid convergence to the inverses of well-conditioned structured matrices even where no good initial approximation is available. In particular we yield the inverse of a well-conditioned matrix with a structure of Toeplitz/Hankel type in flops. For a large class of input matrices, our methods can be extended to computing numerically the generalized inverses. Our numerical experiments confirm the validity of our analysis and the efficiency of the presented algorithms for well-conditioned input matrices and furnished us with the proper values of the parameters that define our algorithms.

     

Successively iterative method for fractional differential equations with integral boundary conditions

In this paper, we study the boundary value problem of a fractional differential equation with nonlocal integral boundary conditions. Two successively iterative sequences are constructed, the conditions for the existence of the nontrivial sign-changing solutions to the differential equation are established.     

Convergence of Ishikawa Iterates for a Multi-Valued Mapping with a Fixed Point

Existence of fixed points of multivalued mappings that satisfy a certain contractive condition was proved by N. Mizoguchi and W. Takahashi. An alternative proof of this theorem was given by Peter Z. Daffer and H. Kaneko. In the present paper, we give a simple proof of that theorem. Also, we define Mann and Ishikawa iterates for a multivalued map T with a fixed point p and prove that these iterates converge to a fixed point q of T under certain conditions. This fixed point q may be different from p. To illustrate this phenomenon, an example is given. This work is supported by the U. G. C. Grant No. U4/4997/97-98. G. V. R. Babu thanks the University Grants Commission, India for the financial support.     

A SPLITTING ITERATION METHOD FOR DOUBLE X_0-BREAKING BIFURCATION POINTS

1.IntroductionConsiderthefollowingtwo-parameterdependentnonlinearproblemwhereX=Re,^,parerealpaxameters,fEC"(r23),D.fo(=D.f(xo,^o,Po))isanedholmmapwithindexzero.Oneofourmainassumptions,whichariseinmanyapplications[1'2,5--7],isthatfsatisfiesZZ--symmetry:the…     

Finding optimal memoryless policies of POMDPs under the expected average reward criterion

In this paper, partially observable Markov decision processes (POMDPs) with discrete state and action space under the average reward criterion are considered from a recent-developed sensitivity point of view. By analyzing the average-reward performance difference formula, we propose a policy iteration algorithm with step sizes to obtain an optimal or local optimal memoryless policy. This algorithm improves the policy along the same direction as the policy iteration does and suitable step sizes guarantee the convergence of the algorithm. Moreover, the algorithm can be used in Markov decision processes (MDPs) with correlated actions. Two numerical examples are provided to illustrate the applicability of the algorithm.     

Efficient high-order methods based on golden ratio for nonlinear systems

We derive new iterative methods with order of convergence four or higher, for solving nonlinear systems, by composing iteratively golden ratio methods with a modified Newton’s method. We use different efficiency indices in order to compare the new methods with other ones and present several numerical tests which confirm the theoretical results.     

线性规划中一个避免人工变元的方法的改进

有许多文献讨论了线性规划问题中单纯形方法的改进(如文献[1～5]等)。我们在文献[1]的基础上,突破了传统方法中要求单纯形表中的基变量始终非负的想法,给出了求解线性规划问题中一个新的避免人工变量的方法,使其计算量得到减少。