Search

In this letter, we propose an optimal voltage feed-back control algorithm for wind farms where the voltage set-point of wind-farm-side voltage source converter (WFVSC) and reactive powers of wind turbines (WTs) are coordinated. The proposed control establishes on the gradient descent and operates in an online fashion. Each WT updates its reactive power based on the local instantaneous terminal voltage and therefore, the proposed control can be implemented in a decentralized manner. Besides, we also address its asynchronous counterpart to enhance the practical applicability in real world.

We report a new approach to search for structures of minimum energy conical intersection (MECIs) automatically. Gradient projection (GP) method and single component artificial force induced reaction (SC-AFIR) method were combined in the present approach. As case studies, MECIs of benzene and naphthalene between their ground and first excited singlet electronic states (S-0/S-1-MECIs) were explored. All S-0/S-1-MECIs reported previously were obtained automatically. Furthermore, the number of force calculations was reduced compared to the one required in the previous search. Improved convergence in a step in which various geometrical displacements are induced by SC-AFIR would contribute to the cost reduction. (C) 2017 Elsevier B.V. All rights reserved.

We report a new approach to search for structures of minimum energy conical intersection (MECIs) automatically. Gradient projection (GP) method and single component artificial force induced reaction (SC-AFIR) method were combined in the present approach. As case studies, MECIs of benzene and naphthalene between their ground and first excited singlet electronic states (S-0/S-1-MECIs) were explored. All S-0/S-1-MECIs reported previously were obtained automatically. Furthermore, the number of force calculations was reduced compared to the one required in the previous search. Improved convergence in a step in which various geometrical displacements are induced by SC-AFIR would contribute to the cost reduction. (C) 2017 Elsevier B.V. All rights reserved.

Phase field models are widely used to describe the two-phase system. The evolution of the phase field variables is usually driven by the gradient flow of a total free energy functional. The generalization of the approach to an N phase (N >= 3) system requires some extra consistency conditions on the free energy functional in order for the model to give physically relevant results. A projection approach is proposed for the derivation of a consistent free energy functional for the three-phase Cahn-Hilliard equations. The system is then coupled with the Navier-Stokes equations to describe the three-phase flow on solid surfaces with moving contact line. An energy stable scheme is developed for the three-phase flow system. The discrete energy law of the numerical scheme is proved which ensures the stability of the scheme. We also show some numerical results for the dynamics of triple junctions and four phase contact lines. (C) 2016 Elsevier Inc. All rights reserved.

Phase field models are widely used to describe the two-phase system. The evolution of the phase field variables is usually driven by the gradient flow of a total free energy functional. The generalization of the approach to an N phase (N >= 3) system requires some extra consistency conditions on the free energy functional in order for the model to give physically relevant results. A projection approach is proposed for the derivation of a consistent free energy functional for the three-phase Cahn-Hilliard equations. The system is then coupled with the Navier-Stokes equations to describe the three-phase flow on solid surfaces with moving contact line. An energy stable scheme is developed for the three-phase flow system. The discrete energy law of the numerical scheme is proved which ensures the stability of the scheme. We also show some numerical results for the dynamics of triple junctions and four phase contact lines. (C) 2016 Elsevier Inc. All rights reserved.

Phase field models are widely used to describe the two-phase system. The evolution of the phase field variables is usually driven by the gradient flow of a total free energy functional. The generalization of the approach to an N phase (N ≥ 3) system requires some extra consistency conditions on the free energy functional in order for the model to give physically relevant results. A projection approach is proposed for the derivation of a consistent free energy functional for the three-phase Cahn-Hilliard equations. The system is then coupled with the Navier-Stokes equations to describe the three-phase flow on solid surfaces with moving contact line. An energy stable scheme is developed for the three-phase flow system. The discrete energy law of the numerical scheme is proved which ensures the stability of the scheme. We also show some numerical results for the dynamics of triple junctions and four phase contact lines.

Implementation and experimentation of Total Variation Regularization based on Collaborative Norms for denoising of hyperspectral images.

Image restoration models based on total variation (TV) have become popular since their introduction by Rudin, Osher, and Fatemi (ROF) in 1992. The dual formulation of this model has a quadratic objective with separable constraints, making projections onto the feasible set easy to compute. This paper proposes application of gradient projection (GP) algorithms to the dual formulation.We test variants of GP with different step length selection and line search strategies, including techniques based on the Barzilai-Borwein method. Global convergence can in some cases be proved by appealing to existing theory. We also propose a sequential quadratic programming (SQP) approach that takes account of the curvature of the boundary of the dual feasible set. Computational experiments show that the proposed approaches perform well in a wide range of applications and that some are significantly faster than previously proposed methods, particularly when only modest accuracy in the solution is required. © Springer Science+Business Media, LLC 2008.

Image restoration models based on total variation (TV) have become popular since their introduction by Rudin, Osher, and Fatemi (ROF) in 1992. The dual formulation of this model has a quadratic objective with separable constraints, making projections onto the feasible set easy to compute. This paper proposes application of gradient projection (GP) algorithms to the dual formulation.We test variants of GP with different step length selection and line search strategies, including techniques based on the Barzilai-Borwein method. Global convergence can in some cases be proved by appealing to existing theory. We also propose a sequential quadratic programming (SQP) approach that takes account of the curvature of the boundary of the dual feasible set. Computational experiments show that the proposed approaches perform well in a wide range of applications and that some are significantly faster than previously proposed methods, particularly when only modest accuracy in the solution is required. © Springer Science+Business Media, LLC 2008.

Image restoration models based on total variation (TV) have become popular since their introduction by Rudin, Osher, and Fatemi (ROF) in 1992. The dual formulation of this model has a quadratic objective with separable constraints, making projections onto the feasible set easy to compute. This paper proposes application of gradient projection (GP) algorithms to the dual formulation.We test variants of GP with different step length selection and line search strategies, including techniques based on the Barzilai-Borwein method. Global convergence can in some cases be proved by appealing to existing theory. We also propose a sequential quadratic programming (SQP) approach that takes account of the curvature of the boundary of the dual feasible set. Computational experiments show that the proposed approaches perform well in a wide range of applications and that some are significantly faster than previously proposed methods, particularly when only modest accuracy in the solution is required. © Springer Science+Business Media, LLC 2008.

This course will focus on fundamental subjects in (deterministic) optimization, connected through the themes of convexity, geometric multipliers, and duality. The aim is to develop the core analytical and computational issues of continuous optimization, duality, and saddle point theory using a handful of unifying principles that can be easily visualized and readily understood. The mathematical theory of convex sets and functions will be central, and will allow an intuitive, highly visual, geometrical approach to the subject. This theory will be developed in detail and in parallel with the optimization topics. The first part of the course develops the analytical issues of convexity and duality. The second part is devoted to convex optimization algorithms, and their applications to a variety of large-scale optimization problems from resource allocation, machine learning, engineering design, and other areas.

This course will focus on fundamental subjects in (deterministic) optimization, connected through the themes of convexity, geometric multipliers, and duality. The aim is to develop the core analytical and computational issues of continuous optimization, duality, and saddle point theory using a handful of unifying principles that can be easily visualized and readily understood. The mathematical theory of convex sets and functions will be central, and will allow an intuitive, highly visual, geometrical approach to the subject. This theory will be developed in detail and in parallel with the optimization topics. The first part of the course develops the analytical issues of convexity and duality. The second part is devoted to convex optimization algorithms, and their applications to a variety of large-scale optimization problems from resource allocation, machine learning, engineering design, and other areas.

The support vector machines (SVMs) have been very successful in many machine learning problems. However, they can be slow during testing because of the possibly large number of support vectors obtained. Recently, Wu et al. (2005) proposed a sparse formulation that restricts the SVM to use a small number of expansion vectors. In this paper, we further extend this idea by integrating with techniques from multiple-kernel learning (MKL). The kernel function in this sparse SVM formulation no longer needs to be fixed but can be automatically learned as a linear combination of kernels. Two formulations of such sparse multiple-kernel classifiers are proposed. The first one is based on a convex combination of the given base kernels, while the second one uses a convex combination of the so-called "equivalent" kernels. Empirically, the second formulation is particularly competitive. Experiments on a large number of toy and real-world data sets show that the resultant classifier is compact and accurate, and can also be easily trained by simply alternating linear program and standard SVM solver.

The support vector machines (SVMs) have been very successful in many machine learning problems. However, they can be slow during testing because of the possibly large number of support vectors obtained. Recently, Wu et al. (2005) proposed a sparse formulation that restricts the SVM to use a small number of expansion vectors. In this paper, we further extend this idea by integrating with techniques from multiple-kernel learning (MKL). The kernel function in this sparse SVM formulation no longer needs to be fixed but can be automatically learned as a linear combination of kernels. Two formulations of such sparse multiple-kernel classifiers are proposed. The first one is based on a convex combination of the given base kernels, while the second one uses a convex combination of the so-called "equivalent" kernels. Empirically, the second formulation is particularly competitive. Experiments on a large number of toy and real-world data sets show that the resultant classifier is compact and accurate, and can also be easily trained by simply alternating linear program and standard SVM solver.

The support vector machines (SVMs) have been very successful in many machine learning problems. However, they can be slow during testing because of the possibly large number of support vectors obtained. Recently, Wu et al. (2005) proposed a sparse formulation that restricts the SVM to use a small number of expansion vectors. In this paper, we further extend this idea by integrating with techniques from multiple-kernel learning (MKL). The kernel function in this sparse SVM formulation no longer needs to be fixed but can be automatically learned as a linear combination of kernels. Two formulations of such sparse multiple-kernel classifiers are proposed. The first one is based on a convex combination of the given base kernels, while the second one uses a convex combination of the so-called "equivalent" kernels. Empirically, the second formulation is particularly competitive. Experiments on a large number of toy and real-world data sets show that the resultant classifier is compact and accurate, and can also be easily trained by simply alternating linear program and standard SVM solver.

A new algorithm for solving smooth large-scale minimization problems with bound constraints is introduced. The way of dealing with active constraints is similar to the one used in some recently introduced quadratic solvers. A limited-memory multipoint symmetric secant method for approximating the Hessian is presented. Positive-definiteness of the Hessian approximation is not enforced. A combination of trust-region and conjugate-gradient approaches is used to explore a useful negative curvature information. Global convergence is proved for a general model algorithm. Results of numerical experiments are presented.

A new algorithm for solving smooth large-scale minimization problems with bound constraints is introduced. The way of dealing with active constraints is similar to the one used in some recently introduced quadratic solvers. A limited-memory multipoint symmetric secant method for approximating the Hessian is presented. Positive-definiteness of the Hessian approximation is not enforced. A combination of trust-region and conjugate-gradient approaches is used to explore a useful negative curvature information. Global convergence is proved for a general model algorithm. Results of numerical experiments are presented.

A new algorithm for solving smooth large-scale minimization problems with bound constraints is introduced. The way of dealing with active constraints is similar to the one used in some recently introduced quadratic solvers. A limited-memory multipoint symmetric secant method for approximating the Hessian is presented. Positive-definiteness of the Hessian approximation is not enforced. A combination of trust-region and conjugate-gradient approaches is used to explore a useful negative curvature information. Global convergence is proved for a general model algorithm. Results of numerical experiments are presented.

A new algorithm for solving smooth large-scale minimization problems with bound constraints is introduced. The way of dealing with active constraints is similar to the one used in some recently introduced quadratic solvers. A limited-memory multipoint symmetric secant method for approximating the Hessian is presented. Positive-definiteness of the Hessian approximation is not enforced. A combination of trust-region and conjugate-gradient approaches is used to explore a useful negative curvature information. Global convergence is proved for a general model algorithm. Results of numerical experiments are presented.

A new algorithm for solving smooth large-scale minimization problems with bound constraints is introduced. The way of dealing with active constraints is similar to the one used in some recently introduced quadratic solvers. A limited-memory multipoint symmetric secant method for approximating the Hessian is presented. Positive-definiteness of the Hessian approximation is not enforced. A combination of trust-region and conjugate-gradient approaches is used to explore a useful negative curvature information. Global convergence is proved for a general model algorithm. Results of numerical experiments are presented.

High computational demand in solving the optimization problems associated with the model predictive control (MPC) schemes is a major obstacle when applying the methods to large-scale or fast-sampling systems. In this paper, we propose a new structured neural network approach to solving the quadratic programming problem in the constrained MPC. This new approach has the advantage of solving large-scale quadratic programming problems in a massively parallel fashion. The structured neural network consists of a projection network and a network for implementing the gradient projection algorithm. where the projection network is constructed from specially structured linear neurons with a special training algorithm. We prove that the training algorithm converges to the optimal solution. Finally, we test the method on a simplified paper machine model. (C) 2001 Elsevier Science Ltd. All rights reserved.

High computational demand in solving the optimization problems associated with the model predictive control (MPC) schemes is a major obstacle when applying the methods to large-scale or fast-sampling systems. In this paper, we propose a new structured neural network approach to solving the quadratic programming problem in the constrained MPC. This new approach has the advantage of solving large-scale quadratic programming problems in a massively parallel fashion. The structured neural network consists of a projection network and a network for implementing the gradient projection algorithm. where the projection network is constructed from specially structured linear neurons with a special training algorithm. We prove that the training algorithm converges to the optimal solution. Finally, we test the method on a simplified paper machine model. (C) 2001 Elsevier Science Ltd. All rights reserved.

High computational demand in solving the optimization problems associated with the model predictive control (MPC) schemes is a major obstacle when applying the methods to large-scale or fast-sampling systems. In this paper, we propose a new structured neural network approach to solving the quadratic programming problem in the constrained MPC. This new approach has the advantage of solving large-scale quadratic programming problems in a massively parallel fashion. The structured neural network consists of a projection network and a network for implementing the gradient projection algorithm. where the projection network is constructed from specially structured linear neurons with a special training algorithm. We prove that the training algorithm converges to the optimal solution. Finally, we test the method on a simplified paper machine model. (C) 2001 Elsevier Science Ltd. All rights reserved.

Nonlinear programming methods are used to solve saddle-point problems subject to inequality constraints on the variables; in particular, the type of saddle-point problem arising in pursuit-evasion differential games is considered. The methods investigated fall into two groups: solution of the nonlinear simultaneous equations obtained from the Kuhn-Tucker conditions, and solution of a sequence of constrained optimization problems by the gradient projection algorithm. These methods are applicable to any real-valued function f(x,y) which is convex in x, concave in y, and has continuous and bounded second partial derivatives. Several examples are given which illustrate the characteristics of the numerical procedures ., http://archive.org/details/thepplicationofn1094511674, Lieutenant, United States Navy, Approved for public release; distribution is unlimited.