Boundary image matching supporting partial denoising using time-series matching techniques

Multimedia Tools and Applications - In this paper, we deal with the problem of boundary image matching which finds similar boundary images regardless of partial noise exploiting time-series...     

Numerical approximation of incompressible flows with net flux defective boundary conditions by means of penalty techniques

We consider incompressible flow problems with defective boundary conditions prescribing only the net flux on some inflow and outflow sections of the boundary. As a paradigm for such problems, we simply refer to Stokes flow. After a brief review of the problem and of its well posedness, we discretize the corresponding variational formulation by means of finite elements and looking at the boundary conditions as constraints, we exploit a penalty method to account for them. We perform the analysis of the method in terms of consistency, boundedness and stability of the discrete bilinear form and we show that the application of the penalty method does not affect the optimal convergence properties of the finite element discretization. Since the additional terms introduced to account for the defective boundary conditions are non-local, we also analyze the spectral properties of the equivalent algebraic formulation and we exploit the analysis to set up an efficient solution strategy. In contrast to alternative discretization methods based on Lagrange multipliers accounting for the constraints on the boundary, the present scheme is particularly effective because it only mildly affects the structure and the computational cost of the numerical approximation. Indeed, it does not require neither multipliers nor sub-iterations or additional adjoint problems with respect to the reference problem at hand.     

The RISC penalty

All the major players in the workstation and personal computer market are committed to RISC. The author looks at the view that RISC is not better, and does not hold the key to high speed computing. He claims that the only reason this foolishness has got so far is that the real differences between RISC and complex-instruction-set computing (CISC) are relatively small in the big performance picture. Hard work and good marketing more than compensate for the differences     

Generic parameter penalty architecture

A new architecture for Multidisciplinary Design Optimization, called Generic Parameter Penalty Architecture, is introduced. This architecture can be configured through three parameters that manage the distribution of objective functions and consistencies between main and sub-level optimizations. The parameters can adopt values between zero and one. Five different configurations of these parameters were tested on three problems (an analytical problem, Golinski’s speed reducer, and the combustion of propane). All three parameters of four of these configurations have extreme values (either zero or one), and the fifth one has intermediate values. These values make some of configurations manage the main level and sub-level optimizations in a similar manner to that of All at Once, Collaborative Optimization, and Analytical Target Cascading, which were studied for benchmarking purposes. The convergence and relative error of the solutions obtained by the new architecture were studied and compared to those of the previously stated widespread architectures. Results show that the performance of the new architecture depends on the values of its three parameters. It adopted behaviors similar to those of the reference architectures. Finally, its convergence and relative error, in contrast to the reference architectures, increased with the complexity of the problem.     

On penalty methods for interelement constraints

The problem of enforcing constraints across interelement boundaries can be treated directly in the construction of the basis or by using multiplier or penalty methods. We demonstrate that the multiplier approach can induce linear dependence in the constraints and lead to a singular system of equations if appropriate precautions are not exercised. The penalty method, on the other hand, is shown to be directly applicable to the overconstrained¹ problem. There are, however, some subtle points regarding loss of accuracy in the penalty approach. We present some new results and corroborating numerical experiments to support the arguments. In particular, we show that there is progressive erosion of accuracy in the approximate solution using finite precision arithmetic as the penalty factor is increased beyond an optimal value. This implies that a control must be applied on the size of the penalty factor. These ideas are quite broadly pertinent to computation of solutions to other constrained optimization problems in which multiplier and penalty methods are used as well as the specific class of problems we consider here.     

The penalty method for the Navier-Stokes equations

Summary The penalty finite element method as it applies to the Stokes and Navier-Stokes flow equations is reviewed. The main developments are discussed and selected but still extensive list of references is provided.     

Reducing the branch penalty in pipelined processors

A probabilistic model is developed to quantify the performance effects of the branch penalty in a typical pipeline. The branch penalty is analyzed as a function of the relative number of branch instructions executed and the probability that a branch is taken. The resulting model shows the fraction of maximum performance achievable under the given conditions. Techniques to reduce the branch penalty include static and dynamic branch prediction, the branch target buffer, the delayed branch, branch bypassing and multiple prefetching, branch folding, resolution of branch decision early in the pipeline, using multiple independent instruction streams in a shared pipeline, and the prepare-to-branch instruction     

关于惩罚函数中惩罚系数的讨论

对惩罚函数法中的惩罚系数进行了系统分析和讨论. 首先, 针对Deb 基于可行性规则对约束违反相同情况下的比较没有具体说明的现状, 提出一种改进的Deb 基于可行性规则. 在此基础上, 证明了惩罚系数过大或者过小均不会影响排序的结论, 并给出了惩罚系数影响排序的上下边界. 实例分析表明了所得结论的有效性, 为基于惩罚系数的算法设计提供了依据.

     

Second-order learning algorithm with squared penalty term

This article compares three penalty terms with respect to the efficiency of supervised learning, by using first- and second-order off-line learning algorithms and a first-order on-line algorithm. Our experiments showed that for a reasonably adequate penalty factor, the combination of the squared penalty term and the second-order learning algorithm drastically improves the convergence performance in comparison to the other combinations, at the same time bringing about excellent generalization performance. Moreover, in order to understand how differently each penalty term works, a function surface evaluation is described. Finally, we show how cross validation can be applied to find an optimal penalty factor.     

Backup agreements with penalty scheme under supply disruptions

This article considers a supply chain for a single product involving one retailer and two independent suppliers, when the main supplier might fail to supply the products, the backup supplier can always supply the products at a higher price. The retailer could use the backup supplier as a regular provider or a stand-by source by reserving some products at the supplier. A backup agreement with penalty scheme is constructed between the retailer and the backup supplier to mitigate the supply disruptions and the demand uncertainty. The expected profit functions and the optimal decisions of the two players are derived through a sequential optimisation process. Then, the sensitivity of two players’ expected profits to various input factors is examined through numerical examples. The impacts of the disruption probability and the demand uncertainty on the backup agreement are also investigated, which could provide guideline for how to use each sourcing method.     

Challenging the abstraction penalty in parallel patterns libraries

The Journal of Supercomputing - In the last years, pattern-based programming has been recognized as a good practice for efficiently exploiting parallel hardware resources. Following this approach,...     

A hybrid variable penalty method for nonlinear programming

The paper describes a hybrid variable penalty method (in which the penalty functions are finite on the boundary of the constrained set) for solving a general nonlinear programming problem. The method combines two types of variable penalty functions to obtain a hybrid formulation in such a way that the error in the approximation of the Hessian matrix resulting from using only the first derivatives of the constraints, are minimized both in the feasible and infeasible domains. The hybrid algorithm poses a sequence of unconstrained optimization problems with mechanism to control the quality of the approximation for the Hessian matrix. The unconstrained problems are solved using a modified Newton's algorithm. The hybrid method is found to exhibit the following characteristics: (i) Admit small values of the penalty weight r to start SUMT (i.e. r = 1 × 10⁻³ or smaller) with no apparent ill-conditioning; (ii) Stimulate faster rate of convergence for a single r; (iii) Permit feasible or infeasible initial starting points.

The numerical effectiveness of the algorithm is demonstrated on a relatively large set of test problems.