• Journal of the Korean Mathematical Society
• /
• v.54 no.2
• /
• pp.461-477
• /
• 2017

A dual substructuring method with a penalty term was introduced in the previous works by the authors, which is a variant of the FETI-DP method. The proposed method imposes the continuity not only by using Lagrange multipliers but also by adding a penalty term which consists of a positive penalty parameter ${\eta}$ and a measure of the jump across the interface. Due to the penalty term, the proposed iterative method has a better convergence property than the standard FETI-DP method in the sense that the condition number of the resulting dual problem is bounded by a constant independent of the subdomain size and the mesh size. In this paper, a further study for a dual iterative substructuring method with a penalty term is discussed in terms of its convergence analysis. We provide an improved estimate of the condition number which shows the relationship between the condition number and ${\eta}$ as well as a close spectral connection of the proposed method with the FETI-DP method. As a result, a choice of a moderately small penalty parameter is guaranteed.

• Proceedings of the KSME Conference
• /
• 2004.11a
• /
• pp.916-921
• /
• 2004

The penalty method has been widely applied to analyses of incompressible fluid flow. However, we have not yet found any prior studies that employed penalty method to analyze compressible fluid flow. In this study, with an eye on the apparent similarity between the slight compressible formulation and the penalty formulation, we have proposed a new approximate approach that can analyze compressible packing process using the penalty parameter l. Based on the assumption of the isothermal flow, a set of reference solutions was obtained to verify the validity of the proposed scheme. Furthermore, we have applied the proposed scheme to the analysis of the packing process of different cases.

• Communications for Statistical Applications and Methods
• /
• v.22 no.1
• /
• pp.41-54
• /
• 2015

In this article, we propose nonconcave penalties on a reduced-rank regression model to select variables and estimate coefficients simultaneously. We apply HARD (hard thresholding) and SCAD (smoothly clipped absolute deviation) symmetric penalty functions with singularities at the origin, and bounded by a constant to reduce bias. In our simulation study and real data analysis, the new method is compared with an existing variable selection method using $L_1$ penalty that exhibits competitive performance in prediction and variable selection. Instead of using only one type of penalty function, we use two or three penalty functions simultaneously and take advantages of various types of penalty functions together to select relevant predictors and estimation to improve the overall performance of model fitting.

• Proceedings of the KIEE Conference
• /
• 2000.07a
• /
• pp.49-51
• /
• 2000

ln this paper, a method by which penalty factors of all generators including slack bus can be directly derived is presented. With a simple re-assignment of angle reference bus to a bus where no generation exists, penalty factors for slack bus is obtained without any physical assumption. While previous Jacobian-based techniques for generator penalty factor calculation have been derived with basis upon reference bus, proposed method are not dependent on reference bus and calculated penalty factors can be substituted directly into the general ELD equation to compute the economic dispatch. Equations for system loss sensitivity, penalty factors and optimal generation allocation are solved simultaneously in normal power flow computation.

• Korean Journal of Mathematics
• /
• v.25 no.2
• /
• pp.147-162
• /
• 2017

In this research works, we consider the general regularized penalty method for non-stationary set valued equilibrium problem in a Banach space. We define weak coercivity conditions and show that the weak and strong convergence problems of the regularized penalty method.

• Journal of Mechanical Science and Technology
• /
• v.17 no.1
• /
• pp.64-76
• /
• 2003

In this work, a new penalty formulation is proposed for the analysis of Mindlin-Reissner plates by using the element-free Galerkin method. A penalized weak form for the Mindlin-Reissner Plates is constructed through the exterior penalty method to enforce the essential boundary conditions of rotations as well as transverse displacements. In the numerical examples, some typical problems of Mindlin-Reissner plates are analyzed, and parametric studies on the order of integration and the size of influence domain are also carried out. The effect of the types of background cells on the accuracy of numerical solutions is observed and a proper type of background cell for obtaining optimal accuracy is suggested. Further, optimal order of integration and basis order of Moving Least Squares approximation are suggested to efficiently handle the irregularly distributed nodes through the triangular type of background cells. From the numerical tests, it is identified that unlike the finite element method, the proposed element-free Galerkin method with penalty technique gives highly accurate solution without shear locking in dealing with Mindlin-Reissner plates.

• Journal of applied mathematics & informatics
• /
• v.29 no.5_6
• /
• pp.1533-1539
• /
• 2011

In this paper, a class of nonlinear bilevel programs, i.e. the lower level problem is linear programs, is considered. Aiming at this special structure, we append the duality gap of the lower level problem to the upper level objective with a penalty and obtain a penalized problem. Using the penalty method, we give an existence theorem of solution and propose an algorithm. Then, a numerical example is given to illustrate the algorithm.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2005.10a
• /
• pp.105-110
• /
• 2005

The penalty method has been widely applied to analyses of incompressible fluid flow. However, we have not yet found any prior studies that employed penalty method to analyze compressible fluid flow. In this study, with an eye on the apparent similarity between the slight compressible formulation and the penalty formulation, we have proposed a modified approximate approach that can analyze compressible packing process using the penalty parameter, which is an improvement on an earlier formulation (KSME, 2004B). Based on the assumption of the isothermal flow, a set of reference solutions was obtained to verify the validity of the proposed scheme. Furthermore, we have applied the proposed scheme to the analysis of the packing process of different cases.

• Journal of Mechanical Science and Technology
• /
• v.14 no.10
• /
• pp.1122-1130
• /
• 2000

Based on the value of the Lagrange multiplier and the degree of constraint activeness, a new update rule is proposed for penalty parameters of the ALM method. The theoretical exposition of this suggested update rule is presented by using the algorithmic interpretation and the geometric interpretation of the augmented Lagrangian. This interpretation shows that the penalty parameters can effect the performance of the ALM method. Also, it offers a lower limit on the penalty parameters that makes the augmented Lagrangian to be bounded. This lower limit forms the backbone of the proposed update rule. To investigate the numerical performance of the update rule, it is embedded in our ALM based dynamic response optimizer, and the optimizer is applied to solve six typical dynamic response optimization problems. Our optimization results are compared with those obtained by employing three conventional update rules used in the literature, which shows that the suggested update rule is more efficient and more stable than the conventional ones.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.09a
• /
• pp.166-169
• /
• 2003

Reinforcement learning if a kind of machine learning. It aims to adapt an agent to a given environment with a clue to a reward and a penalty. Q-learning [8] that is a representative reinforcement learning system treats a reward and a penalty at the same time. There is a problem how to decide an appropriate reward and penalty values. We know the Penalty Avoiding Rational Policy Making algorithm (PARP) [4] and the Penalty Avoiding Profit Sharing (PAPS) [2] as reinforcement learning systems to treat a reward and a penalty independently. though PAPS is a descendant algorithm of PARP, both PARP and PAPS tend to learn a local optimal policy. To overcome it, ion this paper, we propose the Multi Best method (MB) that is PAPS with the multi-start method[5]. MB selects the best policy in several policies that are learned by PAPS agents. By applying PS, PAPS and MB to a soccer game environment based on the SoccerBots[9], we show that MB is the best solution for the soccer game environment.