The Experts below are selected from a list of 37347 Experts worldwide ranked by ideXlab platform
Tomoaki Ohtsuki - One of the best experts on this subject based on the ideXlab platform.
-
damping factor learning of bp detection with node selection in massive mimo using neural network
Vehicular Technology Conference, 2020Co-Authors: Junta Tachibana, Tomoaki OhtsukiAbstract:In a massive multiple-input multiple-output (MIMO) system, belief propagation (BP) detection is known as a method to separate and detect received signals. In BP detection, a MIMO channel is represented by a factor graph and the transmitted symbols are estimated by message passing. However, the Convergence Property of BP deteriorates due to multiple loops included in the MIMO channel. As a method to improve the Convergence Property and the detection performance, the damped BP that averages two successive messages with a weighing factor (called damping factor) is known. To train the damping factors off-line for each antenna configuration, deep neural network-based damped BP (DNN-dBP) has been reported. The problem with DNN-dBP is that the detection performance deteriorates when there is a difference in the channel correlation between training and test. This is because the optimal damping factors vary with the channel correlation. In this paper, to solve this issue, we derive the damping factors of BP with the node selection (NS) method that selects nodes to be updated to lower spatial correlation using DNN-dBP. By applying the NS method, the channel correlation among the selected nodes in BP detection is lowered. Therefore, the proposed method can reduce the detection performance deterioration due to the mismatches of the channel correlations between training and test in DNN-dBP. In addition, the Convergence Property of BP is improved by applying the NS method. Therefore, the proposed method can improve the detection performance compared to the conventional DNN-dBP with the same computational complexity. By computer simulation, it is shown that the proposed method significantly reduces the bit error rate (BER) performance deterioration due to the mismatches of the channel correlations between training and test in DNN-dBP. The results also show that the proposed method has the better BER performance than the conventional DNN-dBP with the same computational complexity.
-
learning and analysis of damping factor in massive mimo detection using bp algorithm with node selection
IEEE Access, 2020Co-Authors: Junta Tachibana, Tomoaki OhtsukiAbstract:In a massive multiple-input multiple-output (MIMO) system, belief propagation (BP) detection is known as a method to separate and detect received signals. In BP detection, a MIMO channel is represented by a factor graph and the transmitted symbols are estimated by message passing. However, the Convergence Property of BP deteriorates due to multiple loops included in the MIMO channel. As a method to improve the Convergence Property and the detection performance, the damped BP that averages two successive messages with a weighing factor (called damping factor) is known. To train the damping factors off-line for each antenna configuration, deep neural network-based damped BP (DNN-dBP) has been reported. The problem with DNN-dBP is that the detection performance deteriorates when there is a difference of the channel correlation between training and test. This is because the optimal damping factors vary with the channel correlation. In this paper, to solve this issue, we derive the damping factors of BP with the node selection (NS) method that selects nodes to be updated to lower spatial correlation using DNN-dBP. By applying the NS method, the channel correlation among the selected nodes in BP detection is lowered. Therefore, the proposed method can improve the detection performance deterioration due to the mismatches of the channel correlations between training and test in DNN-dBP. In addition, the Convergence Property of BP is improved by applying the NS method. Therefore, the proposed method has the same detection performance with low computational complexity as the conventional DNN-dBP. By computer simulation, it is shown that the proposed method significantly improves the bit error rate (BER) performance deterioration due to the mismatches of the channel correlations between training and test in DNN-dBP. The results also show that the proposed method can show the same BER performance with low computational complexity as the conventional DNN-dBP. We also investigate the distribution of the trained damping factors and evaluate the tendency of that.
Thomas P Wihler - One of the best experts on this subject based on the ideXlab platform.
-
an adaptive newton method based on a dynamical systems approach
Communications in Nonlinear Science and Numerical Simulation, 2014Co-Authors: Mario Amrein, Thomas P WihlerAbstract:Abstract The traditional Newton method for solving nonlinear operator equations in Banach spaces is discussed within the context of the continuous Newton method. This setting makes it possible to interpret the Newton method as a discrete dynamical system and thereby to cast it in the framework of an adaptive step size control procedure. In so doing, our goal is to reduce the chaotic behavior of the original method without losing its quadratic Convergence Property close to the roots. The performance of the modified scheme is illustrated with various examples from algebraic and differential equations.
Geetanjali Panda - One of the best experts on this subject based on the ideXlab platform.
-
Two-phase quasi-Newton method for unconstrained optimization problem
Afrika Matematika, 2019Co-Authors: Suvra Kanti Chakraborty, Geetanjali PandaAbstract:In this paper, a two-phase quasi-Newton scheme is proposed for solving an unconstrained optimization problem. The global Convergence Property of the scheme is provided under mild assumptions. The superlinear Convergence rate of the scheme is also proved in the vicinity of the solution. The advantages of the proposed scheme over the traditional schemes are justified with numerical table and graphical illustrations.
-
Two-Phase-SQP Method with Higher-Order Convergence Property
Journal of the Operations Research Society of China, 2016Co-Authors: Suvra Kanti Chakraborty, Geetanjali PandaAbstract:We propose a two-phase-SQP (Sequential Quadratic Programming) algorithm for equality-constrained optimization problem. In this paper, an iteration process is developed, and at each iteration, two quadratic sub-problems are solved. It is proved that, under some suitable assumptions and without computing further higher-order derivatives, this iteration process achieves higher-order local Convergence Property in comparison to Newton-SQP scheme. Theoretical advantage and a note on $$l_{1}$$ l 1 merit function associated to the method are provided.
Junta Tachibana - One of the best experts on this subject based on the ideXlab platform.
-
damping factor learning of bp detection with node selection in massive mimo using neural network
Vehicular Technology Conference, 2020Co-Authors: Junta Tachibana, Tomoaki OhtsukiAbstract:In a massive multiple-input multiple-output (MIMO) system, belief propagation (BP) detection is known as a method to separate and detect received signals. In BP detection, a MIMO channel is represented by a factor graph and the transmitted symbols are estimated by message passing. However, the Convergence Property of BP deteriorates due to multiple loops included in the MIMO channel. As a method to improve the Convergence Property and the detection performance, the damped BP that averages two successive messages with a weighing factor (called damping factor) is known. To train the damping factors off-line for each antenna configuration, deep neural network-based damped BP (DNN-dBP) has been reported. The problem with DNN-dBP is that the detection performance deteriorates when there is a difference in the channel correlation between training and test. This is because the optimal damping factors vary with the channel correlation. In this paper, to solve this issue, we derive the damping factors of BP with the node selection (NS) method that selects nodes to be updated to lower spatial correlation using DNN-dBP. By applying the NS method, the channel correlation among the selected nodes in BP detection is lowered. Therefore, the proposed method can reduce the detection performance deterioration due to the mismatches of the channel correlations between training and test in DNN-dBP. In addition, the Convergence Property of BP is improved by applying the NS method. Therefore, the proposed method can improve the detection performance compared to the conventional DNN-dBP with the same computational complexity. By computer simulation, it is shown that the proposed method significantly reduces the bit error rate (BER) performance deterioration due to the mismatches of the channel correlations between training and test in DNN-dBP. The results also show that the proposed method has the better BER performance than the conventional DNN-dBP with the same computational complexity.
-
learning and analysis of damping factor in massive mimo detection using bp algorithm with node selection
IEEE Access, 2020Co-Authors: Junta Tachibana, Tomoaki OhtsukiAbstract:In a massive multiple-input multiple-output (MIMO) system, belief propagation (BP) detection is known as a method to separate and detect received signals. In BP detection, a MIMO channel is represented by a factor graph and the transmitted symbols are estimated by message passing. However, the Convergence Property of BP deteriorates due to multiple loops included in the MIMO channel. As a method to improve the Convergence Property and the detection performance, the damped BP that averages two successive messages with a weighing factor (called damping factor) is known. To train the damping factors off-line for each antenna configuration, deep neural network-based damped BP (DNN-dBP) has been reported. The problem with DNN-dBP is that the detection performance deteriorates when there is a difference of the channel correlation between training and test. This is because the optimal damping factors vary with the channel correlation. In this paper, to solve this issue, we derive the damping factors of BP with the node selection (NS) method that selects nodes to be updated to lower spatial correlation using DNN-dBP. By applying the NS method, the channel correlation among the selected nodes in BP detection is lowered. Therefore, the proposed method can improve the detection performance deterioration due to the mismatches of the channel correlations between training and test in DNN-dBP. In addition, the Convergence Property of BP is improved by applying the NS method. Therefore, the proposed method has the same detection performance with low computational complexity as the conventional DNN-dBP. By computer simulation, it is shown that the proposed method significantly improves the bit error rate (BER) performance deterioration due to the mismatches of the channel correlations between training and test in DNN-dBP. The results also show that the proposed method can show the same BER performance with low computational complexity as the conventional DNN-dBP. We also investigate the distribution of the trained damping factors and evaluate the tendency of that.
Akio Yamamoto - One of the best experts on this subject based on the ideXlab platform.
-
generalized coarse mesh rebalance method for acceleration of neutron transport calculations
Nuclear Science and Engineering, 2005Co-Authors: Akio YamamotoAbstract:This paper proposes a new acceleration method for neutron transport calculations: the generalized coarse-mesh rebalance (GCMR) method. The GCMR method is a unified scheme of the traditional coarse-mesh rebalance (CMR) and the coarse-mesh finite difference (CMFD) acceleration methods. Namely, by using an appropriate acceleration factor, formulation of the GCMR method becomes identical to that of the CMR or CMFD method. This also indicates that the Convergence Property of the GCMR method can be controlled by the acceleration factor since the Convergence properties of the CMR and CMFD methods are generally different. In order to evaluate the Convergence Property of the GCMR method, a linearized Fourier analysis was carried out for a one-group homogeneous medium, and the results clarified the relationship between the acceleration factor and the spectral radius. It was also shown that the spectral radius of the GCMR method is smaller than those of the CMR and CMFD methods. Furthermore, the Fourier analysis showed that when an appropriate acceleration factor was used, the spectral radius of the GCMR method did not exceed unity in this study, which was in contrast to the results of the CMR or the CMFD method. Application of the GCMR method to practical calculations willmore » be easy when the CMFD acceleration is already adopted in a transport code. By multiplying a suitable acceleration factor to a coefficient (D{sup FD}) of a finite difference formulation, one can improve the numerical instability of the CMFD acceleration method.« less
-
Convergence Property of response matrix method for various finite difference formulations used in the nonlinear acceleration method
Nuclear Science and Engineering, 2005Co-Authors: Akio YamamotoAbstract:Convergence properties were investigated for the response matrix method with various finite-difference formulations that can be utilized in the nonlinear acceleration method. The nonlinear acceleration method is commonly used for the diffusion calculation with the advanced nodal method or the transport calculation with the method of characteristics. Efficiency of the nonlinear acceleration method depends on Convergences on two different levels, i.e., those of the finite-difference calculation and the correction factor. This paper focuses on the former topic, i.e., the Convergence Property of finite-difference calculations using the response matrix method. Though various finite-difference formulations can be used in the nonlinear acceleration method, systematic analysis of the Convergence Property for the finite-difference calculation has not been carried out so far. The spectral radius of iteration matrixes was estimated for the various finite-difference calculations assuming the response matrix method with the red-black sweep. From the calculation results, numerical stability of the various finite-difference formulations was clarified, and a favorable form of the finite-difference formulation for the nonlinear iteration was recommended. The result of this paper will be useful for implementation of the nonlinear acceleration scheme with the response matrix method.
-
Convergence improvement of coarse mesh rebalance method for neutron transport calculations
Journal of Nuclear Science and Technology, 2004Co-Authors: Akio Yamamoto, Yasunori Kitamura, Tadashi Ushio, Naoki SugimuraAbstract:The coarse mesh rebalance (CMR) is a simple acceleration method that is commonly used for transport calculations though it is conditionally stable, i.e. acceleration failed under certain calculation conditions. In this paper, a new acceleration scheme, i.e. the generalized coarse mesh rebalance (GCMR) method, is proposed and applied to improve Convergence Property of the CMR method. Definitions of partial neutron currents used in CMR are modified in the present method and Convergence Property of CMR is improved by the modifications. The proposed method was applied to transport calculations in slab and light water reactor assembly geometries. The calculation results were compared with those by the CMR and the coarse mesh finite-difference (CMFD) acceleration methods, and it was revealed that the present method significantly improves the Convergence Property of the traditional CMR method. Since the present method can be easily applied to existing transport codes using the CMR method, it is considered as a p...
