405905 research outputs found
Sort by
Hydro Gym-GPU: From 2D to 3D Benchmark Environments for Reinforcement Learning in Fluid Flows
No full textFluid flow modeling and control is a significant modern challenge with potential impacts across science, technology, and industry. Improved flow control could enhance drag reduction, mixing, and noise reduction in areas like transportation, energy, and medicine. However, progress in flow control is currently hindered by the lack of systematically standardized benchmarks and the high computational cost of fluid simulations. While two-dimensional problems have been extensivelystudied, three-dimensional simulations with larger meshes are rarely considered due to the need forhighly parallelized and specialized solvers. As a result, the engineering burden of encapsulating thesesimulations in benchmark environments has proven to be a significant barrier. In this paper, a GPU-based extension of the HydroGym platform coupling the multiphysics solver framework m-AIA witha state-of-the-art reinforcement learning platform is presented for fluid flow control problems. Basedon the highly-parallelized lattice Boltzmann solver, which is part of m-AIA, a new set of three-dimensional, non-differentiable fluid flow environments is added that extend existing flow controlchallenges to a new level of physical and computational complexity
Streamlining Performance Analysis Workflows Using Compiler-Assisted Instrumentation Selection
No full textEfficient performance analysis is crucial for optimizing scientific applications. The traditional instrumentation workflow may require multiple manual refinements of the instrumentation configuration to minimize measurement perturbations, which can be time-consuming. This paper presents two approaches to streamline these workflows, guided by a compiler-generated static whole-program call graph of the target application. The Compiler-assisted Performance Instrumentation tool (CaPI) enables the user to build modular selection pipelines, which can be tailored to the application and measurement objective. Performance Instrumentation Refinement Automation (PIRA) combines static call graph analysis with dynamic profiling for iterative refinement. Ongoing work aims to integrate these approaches into a unified production tool, emphasizing workflow efficiency
Computational Investigation of the Atmospheric Boundary Layer in the GABLS Benchmark Problem Using the Spectral Element Code NekRS
No full textThis paper investigates the stable atmospheric boundary layer (ABL) using the spectral element code NekRS. Simulations were performed with various grid resolutions to study grid independence and performance. The results show that grid independence is achieved at higher grid resolutions, with minor differences observed between the finest grids. Performance tests on two high-performance computing (HPC) systems, Summit and JUWELS-Booster, reveal that newer generation GPUs offer significant performance improvements. The study underscores the critical role of advanced HPC systems in enabling detailed simulations of complex ABL phenomena and highlights the potential of exascale computing and hybrid CPU-GPU architectures for future research
Bounding the rotating wave approximation for coupled harmonic oscillators
No full textIn this work we study the validity of the rotating wave approximation of an ideal system composed of two harmonic oscillators evolving with a quadratic Hamiltonian and arbitrarily strong interaction. We prove its validity for arbitrary states by bounding the error introduced. We then restrict ourselves to the dynamics of Gaussian states and are able to fully quantify the deviation of arbitrary pure Gaussian states that evolve through different dynamics from a common quantum state. We show that this distance is fully determined by the first and second moments of the statistical distribution of the number of excitations created from the vacuum during an appropriate effective time-evolution. We use these results to completely control the dynamics for this class of states, therefore providing a toolbox to be used in quantum optics and quantum information. Applications and potential physical implementations are also discussed
Remote sensing of the light-obscuring smoke properties in real-scale fires using a photometric measurement method
No full text
