This paper presents a trust region algorithm with null space technique fornonlinear equality constrained optimization. Considering in the null space methods that,the convergent rate of range space step is faster than ...This paper presents a trust region algorithm with null space technique fornonlinear equality constrained optimization. Considering in the null space methods that,the convergent rate of range space step is faster than the null space step for the most cases,the proposed algorithm computes null steps more often than range space step. Moreover,the new algorithm is based on the reduced Hessian SQP method. Global convergence ofthe proposed algorithm is proved. The effectiveness of the method is demonstrated bysome numerical examples.展开更多
In this paper, a trust region method for equality constrained optimizationbased on nondifferentiable exact penalty is proposed. In this algorithm, the trail step ischaracterized by computation of its normal component ...In this paper, a trust region method for equality constrained optimizationbased on nondifferentiable exact penalty is proposed. In this algorithm, the trail step ischaracterized by computation of its normal component being separated from computation of itstangential component, i.e., only the tangential component of the trail step is constrained by trustradius while the normal component and trail step itself have no constraints. The other maincharacteristic of the algorithm is the decision of trust region radius. Here, the decision of trustregion radius uses the information of the gradient of objective function and reduced Hessian.However, Maratos effect will occur when we use the nondifferentiable exact penalty function as themerit function. In order to obtain the superlinear convergence of the algorithm, we use the twiceorder correction technique. Because of the speciality of the adaptive trust region method, we usetwice order correction when p = 0 (the definition is as in Section 2) and this is different from thetraditional trust region methods for equality constrained optimization. So the computation of thealgorithm in this paper is reduced. What is more, we can prove that the algorithm is globally andsuperlinearly convergent.展开更多
In recent years,numerous recurrent neural network(RNN)models have been reported for solving time-dependent nonlinear optimization problems.However,few existing RNN models simultaneously involve nonlinear equality cons...In recent years,numerous recurrent neural network(RNN)models have been reported for solving time-dependent nonlinear optimization problems.However,few existing RNN models simultaneously involve nonlinear equality constraints,direct discretization,and noise suppression.This limitation presents challenges when existing models are applied to practical engineering problems.Additionally,most current discrete-time RNN models are derived from continuous-time models,which may not perform well for solving essentially discrete problems.To handle these issues,a robust direct-discretized RNN(RDD-RNN)model is proposed to efficiently realize time-dependent optimization constrained by nonlinear equalities(TDOCNE)in the presence of various time-dependent noises.Theoretical analyses are provided to reveal that the proposed RDD-RNN model possesses excellent convergence and noise-suppressing capability.Furthermore,numerical experiments and manipulator control instances are conducted and analyzed to validate the superior robustness of the proposed RDD-RNN model under various time-dependent noises,particularly quadratic polynomial noise.Eventually,small target detection experiments further demonstrate the practicality of the RDD-RNN model in image processing applications.展开更多
基金This research is partly supported by the Hunan Provincial Natural Science Foundtion of China and Hunan Provincial Education Foundation of China 02B021
文摘This paper presents a trust region algorithm with null space technique fornonlinear equality constrained optimization. Considering in the null space methods that,the convergent rate of range space step is faster than the null space step for the most cases,the proposed algorithm computes null steps more often than range space step. Moreover,the new algorithm is based on the reduced Hessian SQP method. Global convergence ofthe proposed algorithm is proved. The effectiveness of the method is demonstrated bysome numerical examples.
基金This research is supported in part by the National Natural Science Foundation of China(Grant No. 39830070,10171055)and China Postdoctoral Science Foundation
文摘In this paper, a trust region method for equality constrained optimizationbased on nondifferentiable exact penalty is proposed. In this algorithm, the trail step ischaracterized by computation of its normal component being separated from computation of itstangential component, i.e., only the tangential component of the trail step is constrained by trustradius while the normal component and trail step itself have no constraints. The other maincharacteristic of the algorithm is the decision of trust region radius. Here, the decision of trustregion radius uses the information of the gradient of objective function and reduced Hessian.However, Maratos effect will occur when we use the nondifferentiable exact penalty function as themerit function. In order to obtain the superlinear convergence of the algorithm, we use the twiceorder correction technique. Because of the speciality of the adaptive trust region method, we usetwice order correction when p = 0 (the definition is as in Section 2) and this is different from thetraditional trust region methods for equality constrained optimization. So the computation of thealgorithm in this paper is reduced. What is more, we can prove that the algorithm is globally andsuperlinearly convergent.
基金supported in part by the National Key Research and Development Program of China(2023YFC3011100)the National Natural Science Foundation of China(62476294)+1 种基金the Science and Technology Planning Project of Guangdong Province,China(2021B1212040017)the Guangdong Basic and Applied Basic Research Foundation(2025A1515010377,2023A1515110697).
文摘In recent years,numerous recurrent neural network(RNN)models have been reported for solving time-dependent nonlinear optimization problems.However,few existing RNN models simultaneously involve nonlinear equality constraints,direct discretization,and noise suppression.This limitation presents challenges when existing models are applied to practical engineering problems.Additionally,most current discrete-time RNN models are derived from continuous-time models,which may not perform well for solving essentially discrete problems.To handle these issues,a robust direct-discretized RNN(RDD-RNN)model is proposed to efficiently realize time-dependent optimization constrained by nonlinear equalities(TDOCNE)in the presence of various time-dependent noises.Theoretical analyses are provided to reveal that the proposed RDD-RNN model possesses excellent convergence and noise-suppressing capability.Furthermore,numerical experiments and manipulator control instances are conducted and analyzed to validate the superior robustness of the proposed RDD-RNN model under various time-dependent noises,particularly quadratic polynomial noise.Eventually,small target detection experiments further demonstrate the practicality of the RDD-RNN model in image processing applications.
