Dynamic Decomposition Method Research Articles

Comparison of aerodynamic performance of trains running on bridges under crosswinds using various motion modes

The high-speed railway bridge, accounting for over 50% of railway lines, plays an important role in high-speed railways. When the train runs at high speed on these bridges, the strong winds will directly pose a risk to the safety of train operation, even causing train overturning accidents. Therefore, it is significantly urgent to accurately simulate the relative motion of the train-bridge system in crosswind environment. An improved delayed detached-eddy simulation study based on the shear stress transfer k-omega (I-ω) turbulence model was used to analyze the aerodynamic characteristics of trains running on bridges under two relative motion modes, i.e., the static synthesis method (SSM) and the dynamic decomposition method (DDM), and the variation rule of train aerodynamic loads. The differences of aerodynamic forces, surface pressure distributions, and flow characteristics around the train body under two motions were discussed. The results show that the applied velocity-inlet boundaries in two relative train-bridge motion modes under crosswinds lead to a large difference in the flow characteristics around bridges. Compared with the SSM, the DDM generates more small-scale vortices in the leeward side of the train, causes more disturbances, and enhances the flow instability. Under the DDM, the side force coefficient and lift force coefficient of the train are decreased by 1.99%, 66.67% (Train on the windward side) and 24.47%, 48.24% (Train on the leeward side), respectively, as compared to the SSM. The achievements obtained in the current study can provide data support for the operation standard of high-speed trains traveling on bridges.

Stochastic Gradient Descent and Anomaly of Variance-Flatness Relation in Artificial Neural Networks

Stochastic gradient descent (SGD), a widely used algorithm in deep-learning neural networks, has attracted continuing research interests for the theoretical principles behind its success. A recent work reported an anomaly (inverse) relation between the variance of neural weights and the landscape flatness of the loss function driven under SGD [Feng Y and Tu Y Proc. Natl. Acad. Sci. USA 118 e2015617118 (2021)]. To investigate this seeming violation of statistical physics principle, the properties of SGD near fixed points are analyzed with a dynamic decomposition method. Our approach recovers the true “energy” function under which the universal Boltzmann distribution holds. It differs from the cost function in general and resolves the paradox raised by the the anomaly. The study bridges the gap between the classical statistical mechanics and the emerging discipline of artificial intelligence, with potential for better algorithms to the latter.

A Graph-Based Soft Actor Critic Approach in Multi-Agent Reinforcement Learning

Multi-Agent Reinforcement Learning (MARL) is widely used to solve various real-world problems. In MARL, the environment contains multiple agents. A good grasp of the environment can guide agents to learn cooperative strategies. In Centralized Training Decentralized Execution (CTDE), a centralized critic is used to guide cooperative strategies learning. However, having multiple agents in the environment leads to the curse of dimensionality and influence of other agents’ strategies, resulting in difficulties for centralized critics to learn good cooperative strategies. We propose a graph-based approach to overcome the above problems. It uses a graph neural network, which uses partial observations of agents as input, and information between agents is aggregated by graph methods to extract information about the whole environment. In this way, agents can improve their understanding of the overall state of the environment and other agents in the environment while avoiding dimensional explosion. Then we combine a dual critic dynamic decomposition method with soft actor-critic to train policy. The former uses individual and global rewards for learning, avoiding the influence of other agents’ strategies, and the latter help to learn an optional policy better. We call this approach Multi-Agent Graph-based soft Actor-Critic (MAGAC). We compare our proposed method with several classical MARL algorithms under the Multi-agent Particle Environment (MPE). The experimental results show that our method can achieve a faster learning speed while learning better policy.

Decomposition method of synthesis of autonomous control system of multidimensional dynamic object

The work deals with issues of synthesis of multidimensional system of automatic control of dynamic object, having interconnected adjustable variables. In order to solve the task, a dynamic decomposition method is proposed, which provides autonomous control of output variables of the object. The method allows analytically synthesizing a multidimensional regulator with minimal complexity and allows ensuring stability of the control system, required dynamic properties. The structure of the multi-dimensional regulator is determined on the basis of models of the multi-dimensional control object in state variables without introducing additional stabilization elements, which will significantly simplify the structure of the synthesized system. The effectiveness of the proposed method is shown in the example of synthesis of a multi-dimensional control system of a two-dimensional object.

Cooperative Coevolution with Two-Stage Decomposition for Large-Scale Global Optimization Problems

Cooperative coevolution (CC) is an effective framework for solving large-scale global optimization (LSGO) problems. However, CC with static decomposition method is ineffective for fully nonseparable problems, and CC with dynamic decomposition method to decompose problems is computationally costly. Therefore, a two-stage decomposition (TSD) method is proposed in this paper to decompose LSGO problems using as few computational resources as possible. In the first stage, to decompose problems using low computational resources, a hybrid-pool differential grouping (HPDG) method is proposed, which contains a hybrid-pool-based detection structure (HPDS) and a unit vector-based perturbation (UVP) strategy. In the second stage, to decompose the fully nonseparable problems, a known information-based dynamic decomposition (KIDD) method is proposed. Analytical methods are used to demonstrate that HPDG has lower decomposition complexity compared to state-of-the-art static decomposition methods. Experiments show that CC with TSD is a competitive algorithm for solving LSGO problems.

The Interplay between Structural Stability and Plasticity Determines Mutation Profiles and Chaperone Dependence in Protein Kinases.

We present a novel comparative analysis of representative protein kinases to characterize the main dynamic and energetic determinants of functional regulation shared among different families. The relationships between stability and plasticity are also used to rationalize kinase tendencies to interact with the molecular chaperone Hsp90. These questions are tackled through newly developed molecular-dynamics-based methods of analysis of internal energy and dynamics applied to a total of 37 different systems, which represent wild-type and mutated proteins, including active and inactive states. Energetic decomposition analysis is coupled to multiple structural alignments and dynamic decomposition methods and identifies, across different families, common elements that underlie fold stabilization and conformational regulation. This analysis also exposes which substructures play a key role in determining chaperone dependence. Overall, the results highlight common interaction networks that underpin kinase stabilization, are modulated by mutations (even if located at a distance), and underlie their tendencies to act as clients or nonclients of Hsp90.

Low-frequency variability of the shallow meridional overturning circulation in the South China Sea

The low-frequency variability of the shallow meridional overturning circulation (MOC) in the South China Sea (SCS) is investigated using a Simple Ocean Data Assimilation (SODA) product for the period of 1900–2010. A dynamical decomposition method is used in which the MOC is decomposed into the Ekman, external mode, and vertical shear components. Results show that all the three dynamical components contribute to the formation of the seasonal and annual mean shallow MOC in the SCS. The shallow MOC in the SCS consists of two cells: a clockwise cell in the south and an anticlockwise cell in the north; the former is controlled by the Ekman flow and the latter is dominated by the external barotropic flow, with the contribution of the vertical shear being to reduce the magnitude of both cells. In addition, the strength of the MOC in the south is found to have a falling trend over the past century, due mainly to a weakening of the Luzon Strait transport (LST) that reduces the transport of the external component. Further analysis suggests that the weakening of the LST is closely related to a weakening of the westerly wind anomalies over the equatorial Pacific, which leads to a southward shift of the North Equatorial Current (NEC) bifurcation and thus a stronger transport of the Kuroshio east of Luzon.

Analytical Design of Multivariable Control Systems by Dynamical Decomposition Method

Design of multivariable control systems for complex technical plants with several interrelated channels and controlled variables is difficult problem. Usually here it is necessary to provide either independent (autonomous, unrelated) or coherent control of output variables. This design problem can be solved by using the analytical method implying exact, dynamical decomposition of the multivariable plant to the set of an independent single input–single output channels. The dynamic decomposition method is based on decomposing property of the adjunct matrix and provides realized of control action as function of output variables, reference input and measured disturbances (control on output and inputs) of the system. The efficiency of this analytical method is shown by the numerical example of control system design for the turbojet engine of the gas-pumping station with three interrelated channels.

High-purity Cu nanocrystal synthesis by a dynamic decomposition method.

Cu nanocrystals are applied extensively in several fields, particularly in the microelectron, sensor, and catalysis. The catalytic behavior of Cu nanocrystals depends mainly on the structure and particle size. In this work, formation of high-purity Cu nanocrystals is studied using a common chemical vapor deposition precursor of cupric tartrate. This process is investigated through a combined experimental and computational approach. The decomposition kinetics is researched via differential scanning calorimetry and thermogravimetric analysis using Flynn-Wall-Ozawa, Kissinger, and Starink methods. The growth was found to be influenced by the factors of reaction temperature, protective gas, and time. And microstructural and thermal characterizations were performed by X-ray diffraction, scanning electron microscopy, transmission electron microscopy, and differential scanning calorimetry. Decomposition of cupric tartrate at different temperatures was simulated by density functional theory calculations under the generalized gradient approximation. High crystalline Cu nanocrystals without floccules were obtained from thermal decomposition of cupric tartrate at 271°C for 8 h under Ar. This general approach paves a way to controllable synthesis of Cu nanocrystals with high purity.

Ocean Eddy Dynamics in a Coupled Ocean–Atmosphere Model*

Abstract The role of mesoscale oceanic eddies is analyzed in a quasigeostrophic coupled ocean–atmosphere model operating at a large Reynolds number. The model dynamics are characterized by decadal variability that involves nonlinear adjustment of the ocean to coherent north–south shifts of the atmosphere. The oceanic eddy effects are diagnosed by the dynamical decomposition method adapted for nonstationary external forcing. The main effects of the eddies are an enhancement of the oceanic eastward jet separating the subpolar and subtropical gyres and a weakening of the gyres. The flow-enhancing effect is due to nonlinear rectification driven by fluctuations of the eddy forcing. This is a nonlocal process involving generation of the eddies by the flow instabilities in the western boundary current and the upstream part of the eastward jet. The eddies are advected by the mean current to the east, where they backscatter into the rectified enhancement of the eastward jet. The gyre-weakening effect, which is due to the time-mean buoyancy component of the eddy forcing, is a result of the baroclinic instability of the westward return currents. The diagnosed eddy forcing is parameterized in a non-eddy-resolving ocean model, as a nonstationary random process, in which the corresponding parameters are derived from the control coupled simulation. The key parameter of the random process—its variance—is related to the large-scale flow baroclinicity index. It is shown that the coupled model with the non-eddy-resolving ocean component and the parameterized eddies correctly simulates climatology and low-frequency variability of the control eddy-resolving coupled solution.

Croissance économique, pauvreté et inégalité des revenus au Cameroun

This study aims to construct a dynamic poverty profile for Cameroon in order to analyze the link between economy growth, poverty and inequality in that country, which has been under structural adjustment since more than 10 years. For poverty analysis, the costs of basic needs (CBN) method is used to estimate poverty lines, and the class of FGT indices developed by Foster, Greer, and Thorbecke (1984) is calculated to study, the incidence, intensity and severity of poverty. The analysis of the relationship between economic growth, poverty and inequality is based both on Kakwani's (1993) static decomposition approach which separately measures the impact of changes in average income and income inequality on poverty, and the dynamic decomposition methods by Datt and Ravaillon (1992) and Kakwani (1997) which break down changes in poverty into growth and redistribution components. The analysis shows that, between 1983/84 and 1996, poverty in Cameroon appears to be a significant phenomenon which affects rural areas more severely than urban areas. This result suggests that rural areas, where the majority of the poor reside, should be the primary beneficiaries of efforts for fighting against poverty. Moreover, the analysis of the link between growth, poverty and income inequality according to Kakwani's (1993) method indicates that in urban areas, the absolute values of expenditure elasticities are clearly higher than unity whatever the poverty measure. Furthermore, poverty elasticities relative to expenditures and to the Gini index are higher in urban than in rural areas thus indicating the higher social fragility of urban areas with respect to economic growth and changes in income inequality. Therefore, the dynamic decomposition results of changes in poverty in growth and redistribution effects confirm the fact that the increase in poverty over the 1983/84-1986 period was due the adverse effect of the contraction in economic activity on growth which was higher than the redistribution effect and thus was more favourable to the poor. The results of this study thus improve our understanding of the nature of poverty in Cameroon and may help in the design of appropriate policies to alleviate this phenomenon.

Dynamic decomposition method for linear programming problems with ceneralized upper bounds

Dynamic decomposition method for linear programming problems with ceneralized upper bounds