Positional restraint of floating structures is achieved using mooring systems anchored to the seafloor. Design and risk assessment of these systems is challenging due to variability in wave, wind loading, ground condi...
详细信息
ISBN:
(纸本)9780791885949
Positional restraint of floating structures is achieved using mooring systems anchored to the seafloor. Design and risk assessment of these systems is challenging due to variability in wave, wind loading, ground conditions leading to uncertainty in the load-displacement behaviour of mooring line foundations for design. Accordingly, new simulation tools shall be able to support risk-informed integrated analysis of the floating structure as a system of interacting components subjected to realistic dynamic excitation. Also, all components of the floating structure simulation model shall be supported by experimental validation. In response to this need, this paper presents a computational framework for simulating the dynamic response of floating structures up to two second-order hydrodynamic loading. The framework utilizes the finite-element method and supports both integrated analysis and hybrid testing with focus on mooring line foundations. computational efficiency is the driving criterion for tailoring the fidelity of the element library. Results of an ideal case study are used to test computational efficiency.
In this study, we present strategies to improve the computational efficiency of high-order space-time DG. The relation between DG in time and implicit Runge-Kutta methods is discussed further and efficient smoothing s...
详细信息
ISBN:
(数字)9781624106996
ISBN:
(纸本)9781624106996
In this study, we present strategies to improve the computational efficiency of high-order space-time DG. The relation between DG in time and implicit Runge-Kutta methods is discussed further and efficient smoothing strategies for solving the implicit systems at each time step are investigated. We suggest a simple IMEX multi-stage optimized RK scheme for pseudo-time stepping that treats the spatial residual explicitly and the temporal residual implicitly. An IMEX four-stage optimized RK scheme has been tested for the isentropic vortex problem at high spatial and temporal orders. For small physical time steps, the IMEX scheme provides an 8.7x speedup over the purely explicit pseudo-time stepping scheme. For large physical time steps, the IMEX scheme reduces to an explicit RK scheme optimized for stability in pseudo-time and provides 7.8x faster convergence rate than single-stage RK scheme in pseudo-time. The performance of the IMEX pseudo-time stepping scheme is examined for flow over a NACA0012 airfoil in 2D and 3D as a function of the physical time step size and variations in grid resolution, showing superior performance to the equivalent explicit RK scheme using either local or global minimum time stepping in pseudo time.
We consider a regulator willing to drive individual choices towards increasing social welfare by providing incentives to a large population of individuals. For that purpose, we formalize and solve the problem of findi...
详细信息
ISBN:
(数字)9781665468800
ISBN:
(纸本)9781665468800
We consider a regulator willing to drive individual choices towards increasing social welfare by providing incentives to a large population of individuals. For that purpose, we formalize and solve the problem of finding an optimal personalized-incentive policy: optimal in the sense that it maximizes social welfare under an incentive budget constraint, personalized in the sense that the incentives proposed depend on the alternatives available to each individual, as well as her preferences. We propose a polynomial time approximation algorithm that computes a policy within few seconds and we analytically prove that it is boundedly close to the optimum. We then extend the problem to efficiently calculate the Maximum Social Welfare Curve, which gives the maximum social welfare achievable for a range of incentive budgets (not just one value). This curve is a valuable practical tool for the regulator to determine the right incentive budget to invest. Finally, we simulate a large-scale application to mode choice in a French department (about 200 thousands individuals) and illustrate the effectiveness of the proposed personalizedincentive policy in reducing CO2 emissions.
Visual odometry is important for plenty of applications such as autonomous vehicles, and robot navigation. It is challenging to conduct visual odometry in textureless scenes or environments with sudden illumination ch...
详细信息
ISBN:
(纸本)9798350323658
Visual odometry is important for plenty of applications such as autonomous vehicles, and robot navigation. It is challenging to conduct visual odometry in textureless scenes or environments with sudden illumination changes where popular feature-based methods or direct methods cannot work well. To address this challenge, some edge-based methods have been proposed, but they usually struggle between the efficiency and accuracy. In this work, we propose a novel visual odometry approach called EdgeVO, which is accurate, efficient, and robust. By efficiently selecting a small set of edges with certain strategies, we significantly improve the computational efficiency without sacrificing the accuracy. Compared to existing edge-based method, our method can significantly reduce the computational complexity while maintaining similar accuracy or even achieving better accuracy. This is attributed to that our method removes useless or noisy edges. Experimental results on the TUM datasets indicate that EdgeVO significantly outperforms other methods in terms of efficiency, accuracy and robustness.
When training complex neural networks, memory usage can be an important bottleneck. The question of when to rematerialize, i.e., to recompute intermediate values rather than retaining them in memory, becomes critical ...
详细信息
When training complex neural networks, memory usage can be an important bottleneck. The question of when to rematerialize, i.e., to recompute intermediate values rather than retaining them in memory, becomes critical to achieving the best time and space efficiency. In this work we consider the rematerialization problem and devise efficient algorithms that use structural characterizations of computation graphs-treewidth and pathwidth-to obtain provably efficient rematerialization schedules. Our experiments demonstrate the performance of these algorithms on many common deep learning models.
In this paper a simulation-based approach to finding optimal defender strategy in multi-act Security Games (SG) played on a graph is proposed. The method employs the Upper Confidence Bounds applied to Trees (UCT) algo...
详细信息
ISBN:
(纸本)9781614996729;9781614996712
In this paper a simulation-based approach to finding optimal defender strategy in multi-act Security Games (SG) played on a graph is proposed. The method employs the Upper Confidence Bounds applied to Trees (UCT) algorithm which relies on massive simulations of possible game scenarios. Three different variants of the algorithm are presented and compared with each other as well as against the Mixed Integer Linear Program (MILP) exact solution in terms of computational efficiency and memory requirements. Experimental evaluation shows that the method has a few times lower memory demands and is faster than MILP approach in majority of test cases while preserving quality of the resulting mixed strategies.
Understanding unsteady flow phenomena in compressor stages often requires the use of time-accurate CFD simulations. Due to the inherent differences in blade pitch between adjacent blade rows, the flow conditions at an...
详细信息
ISBN:
(纸本)9780791854679
Understanding unsteady flow phenomena in compressor stages often requires the use of time-accurate CFD simulations. Due to the inherent differences in blade pitch between adjacent blade rows, the flow conditions at any given instant in adjacent blade rows differ. Simplified computation of the stage represented by a single blade in each row and simple periodic boundary conditions is therefore not possible. Depending on the blade counts, it may be necessary to model the entire annulus of the stage;however, this requires considerable computational time and memory resources. Several methods for modeling the transient flow in turbomachinery stages which require a minimal number of blade passages per row, and therefore reduced computational demands, have been presented in the literature. Recently, some of these methods have become available in commercial CFD solvers. This paper provides a brief description of the methods used, and how they are applied to a transonic compressor stage. The methods are evaluated and compared in terms of computational efficiency and storage requirements, and comparison is made to steady stage simulations. Comparisons to overall performance data and two-dimensional LDV measurements are used to assess the predictive capabilities of the methods. Computed flow features are examined, and compared with reported measurements.
High-performance object detection relies on expensive convolutional networks to compute features, often leading to significant challenges in applications, e.g. those that require detecting objects from video streams i...
详细信息
ISBN:
(纸本)9781538664209
High-performance object detection relies on expensive convolutional networks to compute features, often leading to significant challenges in applications, e.g. those that require detecting objects from video streams in real time. The key to this problem is to trade accuracy for efficiency in an effective way, i.e. reducing the computing cost while maintaining competitive performance. To seek a good balance, previous efforts usually focus on optimizing the model architectures. This paper explores an alternative approach, that is, to reallocate the computation over a scale-time space. The basic idea is to perform expensive detection sparsely and propagate the results across both scales and time with substantially cheaper networks, by exploiting the strong correlations among them. Specifically, we present a unified framework that integrates detection, temporal propagation, and across-scale refinement on a Scale-Time Lattice. On this framework, one can explore various strategies to balance performance and cost. Taking advantage of this flexibility, we further develop an adaptive scheme with the detector invoked on demand and thus obtain improved tradeoff. On ImageNet VID dataset, the proposed method can achieve a competitive mAP 79.6% at 20 fps, or 79.0% at 62 fps as a performance/speed tradeoff. (1)
Galerkin-Bubnov Integral Boundary Element Method (GB-lBEM) is used to model transient phenomena on thin wire structures directly in the time domain. One of the most prominent limitations of the method is the inability...
详细信息
ISBN:
(纸本)9789536037650
Galerkin-Bubnov Integral Boundary Element Method (GB-lBEM) is used to model transient phenomena on thin wire structures directly in the time domain. One of the most prominent limitations of the method is the inability to deal with the structures above finally conducting half plane, due to the high computational inefficiency when ground losses are accounted for in the numerical model. In this paper, this problem is tackled via various modifications, in order to optimize GB-IBEM with respect to the computational speed. These optimizations of the original method result in several orders of magnitude improvement in the overall calculation time, allowing GB-IBEM to be used with geometries above lossy ground.
A Multigrid Alternating Direction Implicit scheme has been developed to solve the two-dimensional Thin Layer Navier-Stokes equations for compressible turbulent flows, incorporating a simple algebraic turbulence model....
详细信息
暂无评论