We study dynamics of nearly elastic particles constrained to move on a line with energy input from the boundaries. We find that for typical initial conditions, the system evolves to an ``extraordinary'' state with particles separated to two groups: The majority of the particles get clamped into a small region of space and move with very slow velocities; a few remaining particles travel between the boundaries at much higher speeds. Such a state clearly violates equipartition of energy. The simplest hydrodynamic approach fails to give a correct description of the system.
In this paper, a respiratory training system geared to speech training for hearing-impaired children is researched, which is mainly used to solve breath-related issues occurring during speech training process. The system comprises of a translucent respiratory system model with interactive function and a respiratory training measurement system based on multi-sensor. Hearing-impaired children can use respiratory training model to imitate pronunciation process. Meanwhile, the system can collect and quantify data about their breathes and then provide feedback, based on which it can guide hearing-impaired children in respiratory trainings and lays a foundation for children making clear, accurate and loud sounds after rehabilitation trainings.
This paper studies two typical ways to use multigrid methods in CFD simulations. One is implementing multigrid methods in CFD softwares start from scratch, and the other is calling multigrid methods from existing libraries. The experimental results on two benchmark cases, the lid-driven cavity flow and the flow around a cylinder, show that the solving performance and parallel scalability of calling multigrid methods from third-party libraries are much better than that of using the multigrid methods implemented in CFD softwares. The result not only provides some guidances for users who try to simulate with multigrid methods, but also gives a possible recommendation for the development and application of softwares for exascale systems. It is a more clever and efficient way to develop softwares on massively parallel systems based on some existing packages and libraries, which reduces the degree of coupling between different functional modules and improves the development efficiency.
When there are several application running on Chip-Multiprocessors (CMPs), it is a problem to allocate the on-chip cache capacities between these competing applications. Cache partitioning is commonly used to solve this problem. Existing cache partitioning schemes either dedicate to the shared design or partition the last level cache depending on limited memory information. This paper presents Private Cache Partitioning, a low-overhead, runtime mechanism that partitions all of the private low level caches which are organized as a large shared cache by a distributed directory. The experiment results show that PCP reduces the overall missrate of competing applications and improves the throughput as well as the weighted speedup.
Implementation of an opposing jet in design of a hypersonic blunt body significantly modifies the external flowfield and yields a considerable reduction in the aerodynamic drag. This study aims to investigate the effects of flowfield modeling parameters of injection and freestream on the flow structure and aerodynamics of a blunt body with an opposing jet in hypersonic flow. Reynolds-Averaged Navier-Stokes (RANS) equations with a Shear Stress Transport (SST) turbulence model are employed to simulate the intricate jet flow interaction. Through utilizing a Non-Intrusive Polynomial Chaos (NIPC) method to construct surrogates, a functional relation is established between input modeling parameters and output flowfield and aerodynamic quantities in concern. Sobol indices in sensitivity analysis are introduced to represent the relative contribution of each parameter. It is found that variations in modeling parameters produce large variations in the flow structure and aerodynamics. The jet-to-freestream total-pressure ratio, jet Mach number, and freestream Mach number are the major contributors to variation in surface pressure, demonstrating an evident location-dependent behavior. The penetration length of injection, reattachment angle of the shear layer, and aerodynamic drag are also most sensitive to the three crucial parameters above. In comparison, the contributions of freestream temperature, freestream density, and jet total temperature are nearly negligible.
Accuracy and performance are key issues for CFD simulation. How to meet the specific accuracy requirements, as well as the optimal simulation performance, is always the research hotspot. This paper presents a general theory of Mesh-Order Independence that is used to guide the configuration of two of the most critical control parameters in a concrete CFD simulation process: grid spacing and discretization order. A concept of optimal mesh-order independent pair which can meet both accuracy and performance requirements at the same time is proposed and analyzed. To find the optimal Mesh-order independent pair, the Mesh-Order Independence is applied to high order FEM simulation, and the specific process and key technologies are detailed. Test and results of two benchmark cases, the Laplace equation and the Helmholtz equation, show that the Mesh-order theory proposed in this paper provides an important guidance for the grid spacing selection and discretization order configuration in practical simulation, especially in the case of high precision requirements. Specifically, only 6 pre-runs with low discretization orders and coarse meshes are needed for the both cases to have a prediction accuracy of more than 70%.
This paper focuses on mesh-partitioning metrics in large-scale parallel computational fluid dynamics (CFD) simulations. Mesh partitioning has a significant influence on the efficiency of parallel preconditioned conjugated gradient (PCG) solving procedure, which is the most representative and time-consuming part in parallel CFD. As the efficiency of parallel PCG depends on load balancing, communication overhead and iterative convergence rate comprehensively, we present a detailed review of mesh-partitioning metrics on these three aspects respectively. Three typical large-scale CFD applications are built to numerically testify the validity of all those metrics.
The Segment Anything Model (SAM) is a recently developed all-range foundation model for image segmentation. It can use sparse manual prompts such as bounding boxes to generate pixel-level segmentation in natural images but struggles in medical images such as low-contrast, noisy ultrasound images. We propose a refined test-phase prompt augmentation technique designed to improve SAM's performance in medical image segmentation. The method couples multi-box prompt augmentation and an aleatoric uncertainty-based false-negative (FN) and false-positive (FP) correction (FNPC) strategy. We evaluate the method on two ultrasound datasets and show improvement in SAM's performance and robustness to inaccurate prompts, without the necessity for further training or tuning. Moreover, we present the Single-Slice-to-Volume (SS2V) method, enabling 3D pixel-level segmentation using only the bounding box annotation from a single 2D slice. Our results allow efficient use of SAM in even noisy, low-contrast medical images. The source code will be released soon.
This paper preliminarily discusses how to develop software towards exascale computing. Two typical development models are studied, one is "all-round contract" which means all the functional modules are implemented in a single framework, and the other is "co-design" which using the existing packages to realize functional modules. As a widely used CFD software whose numerical solving module is implemented in a typical "all-round contract" manner, OpenFOAM is concerned in this work. After redesigning and reimplementing the solving module of OpenFOAM in a "co-design" way by inserting PETSc, the source lines of the code decrease dramatically, which improves the development efficiency. Tests of two CFD benchmark cases and a practical large-scale case on a 128-node cluster show that the newly implemented numerical module also has a higher solving efficiency, compared with the original numerical module in OpenFOAM. Therefore, it is recommended to develop software in a "co-design" manner towards exascale computing, and softwares already implemented in "all-round contract" way should also be reconsidered.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
