Search NASA⌕ Search

SEARCH · Search NASA

Results for “Ensemble computing.”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 181 records · Page 10

Workflows for Science: A comprehensive guide for ensemble workflow tools usage with applications on OLCF systems

The growing demand for robust computational and workflow environments for scientific applications and user communities at the Oak Ridge Leadership Computing Facility (OLCF) has prompted collaboration with ensemble tools development teams and facility users to produce this technical paper. We connect science applications to the RADICAL-Pilot (RP) workflow tool to execute ensemble instantiations using the Frontier supercomputer. The documented installation, usage, and execution demonstrates how RP streamlines scientific workflows at OLCF. We outline the specific steps OLCF users can follow to integrate this tool with their applications and advance their research. This document stands as a comprehensive guide to OLCF users of ensemble workflow tools with examples on real applications using the Frontier supercomputer.

97 MATHEMATICS AND COMPUTING↗

Embedded ensemble propagation for improving performance, portability, and scalability of uncertainty quantification on emerging computational architectures

In this study, quantifying simulation uncertainties is a critical component of rigorous predictive simulation. A key component of this is forward propagation of uncertainties in simulation input data to output quantities of interest. Typical approaches involve repeated sampling of the simulation over the uncertain input data, and can require numerous samples when accurately propagating uncertainties from large numbers of sources. Often simulation processes from sample to sample are similar and much of the data generated from each sample evaluation could be reused. We explore a new method for implementing sampling methods that simultaneously propagates groups of samples together in an embedded fashion, which we call embedded ensemble propagation. We show how this approach takes advantage of properties of modern computer architectures to improve performance by enabling reuse between samples, reducing memory bandwidth requirements, improving memory access patterns, improving opportunities for fine-grained parallelization, and reducing communication costs. We describe a software technique for implementing embedded ensemble propagation based on the use of C++ templates and describe its integration with various scientific computing libraries within Trilinos. We demonstrate improved performance, portability and scalability for the approach applied to the simulation of partial differential equations on a variety of CPU, GPU, and accelerator architectures, including up to 131,072 cores on a Cray XK7 (Titan).

97 MATHEMATICS AND COMPUTING↗

Comparisons of some large scientific computers

In 1975, the National Aeronautics and Space Administration (NASA) began studies to assess the technical and economic feasibility of developing a computer having sustained computational speed of one billion floating point operations per second and a working memory of at least 240 million words. Such a powerful computer would allow computational aerodynamics to play a major role in aeronautical design and advanced fluid dynamics research. Based on favorable results from these studies, NASA proceeded with developmental plans. The computer was named the Numerical Aerodynamic Simulator (NAS). To help insure that the estimated cost, schedule, and technical scope were realistic, a brief study was made of past large scientific computers. Large discrepancies between inception and operation in scope, cost, or schedule were studied so that they could be minimized with NASA's proposed new compter. The main computers studied were the ILLIAC IV, STAR 100, Parallel Element Processor Ensemble (PEPE), and Shuttle Mission Simulator (SMS) computer. Comparison data on memory and speed were also obtained on the IBM 650, 704, 7090, 360-50, 360-67, 360-91, and 370-195; the CDC 6400, 6600, 7600, CYBER 203, and CYBER 205; CRAY 1; and the Advanced Scientific Computer (ASC). A few lessons learned conclude the report.

Credeur, K. R.↗

Numerical and experimental evaluations of the flow past nested chevrons

An effort is made to contribute to the development of CFD by relating the successful use of vortex dynamics in the computation of the pressure drop past a planar array of chevron-shaped obstructions. An ensemble of results was used to compute the loss coefficient k, stimulating an experimental program for the assessment of the measured loss coefficient for the same geometry. The most provocative result of this study has been the representation of kinetic energy production in terms of vorticity source terms.

Foss, J. F.↗

A Machine Learning Framework for Modeling Ensemble Properties of Atomically Disordered Materials

Atomic disorder can strongly influence material properties such as charge transport, optical response, and catalytic activity. However, efficiently modeling these disorder effects remains challenging for first-principles methods due to the cost of sampling large configurational spaces and computing complex physical quantities. Recent advances of machine learning techniques, particularly graph neural networks (GNNs), has enabled the efficient and accurate predictions of complex material properties, offering promising tools for studying disordered systems. In this work, we present a general machine-learning-assisted computational framework that integrates equivariant GNNs with Monte Carlo simulations to compute the thermodynamic and ensemble-averaged functional properties of disordered materials. Using the surface-termination-disordered MXene monolayer Ti 3 C 2 T 2–x as a representative system, we find that electrical conductivity exhibits an emergent peak near the order–disorder phase transition temperature due to the interplay between electron scattering and doping. In contrast, optical conductivity remains largely insensitive to local atomic disorder and reflects the global surface chemical composition. These results highlight the role of atomic disorder in affecting material properties and demonstrate the potential of our approach for statistically modeling disorder effects in a wide range of materials such as high-entropy alloys and spin liquids.

MXene↗

Combining the ensemble and Franck-Condon approaches for calculating spectral shapes of molecules in solution

The correct treatment of vibronic effects is vital for the modeling of absorption spectra of many solvated dyes. Vibronic spectra for small dyes in solution can be easily computed within the Franck-Condon approximation using an implicit solvent model. However, implicit solvent models neglect specific solute-solvent interactions on the electronic excited state. On the other hand, a straightforward way to account for solute-solvent interactions and temperature-dependent broadening is by computing vertical excitation energies obtained from an ensemble of solute-solvent conformations. Ensemble approaches usually do not account for vibronic transitions and thus often produce spectral shapes in poor agreement with experiment. Here we address these shortcomings by combining zero-temperature vibronic fine structure with vertical excitations computed for a room-temperature ensemble of solute-solvent configurations. In this combined approach, all temperature-dependent broadening is treated classically through the sampling of configurations and quantum mechanical vibronic contributions are included as a zero-temperature correction to each vertical transition. In our calculation of the vertical excitations, significant regions of the solvent environment are treated fully quantum mechanically to account for solute-solvent polarization and charge-transfer. For the Franck-Condon calculations, a small amount of frozen explicit solvent is considered in order to capture solvent effects on the vibronic shape function. We test the proposed method by comparing calculated and experimental absorption spectra of Nile red and the green fluorescent protein chromophore in polar and non-polar solvents. For systems with strong solute-solvent interactions, the combined approach yields significant improvements over the ensemble approach. For systems with weak to moderate solute-solvent interactions, both the high-energy vibronic tail and the width of the spectra are in excellent agreement with experiments

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Computationally efficient implementation of combustion chemistry in parallel PDF calculations

In parallel calculations of combustion processes with realistic chemistry, the serial in situ adaptive tabulation (ISAT) algorithm [S.B. Pope, Computationally efficient implementation of combustion chemistry using in situ adaptive tabulation, Combustion Theory and Modelling, 1 (1997) 41-63; L. Lu, S.B. Pope, An improved algorithm for in situ adaptive tabulation, Journal of Computational Physics 228 (2009) 361-386] substantially speeds up the chemistry calculations on each processor. To improve the parallel efficiency of large ensembles of such calculations in parallel computations, in this work, the ISAT algorithm is extended to the multi-processor environment, with the aim of minimizing the wall clock time required for the whole ensemble. Parallel ISAT strategies are developed by combining the existing serial ISAT algorithm with different distribution strategies, namely purely local processing (PLP), uniformly random distribution (URAN), and preferential distribution (PREF). The distribution strategies enable the queued load redistribution of chemistry calculations among processors using message passing. They are implemented in the software x2f{sub m}pi, which is a Fortran 95 library for facilitating many parallel evaluations of a general vector function. The relative performance of the parallel ISAT strategies is investigated in different computational regimes via the PDF calculations of multiple partially stirred reactors burning methane/air mixtures. The results show that the performance of ISAT with a fixed distribution strategy strongly depends on certain computational regimes, based on how much memory is available and how much overlap exists between tabulated information on different processors. No one fixed strategy consistently achieves good performance in all the regimes. Therefore, an adaptive distribution strategy, which blends PLP, URAN and PREF, is devised and implemented. It yields consistently good performance in all regimes. In the adaptive parallel ISAT strategy, the type and extent of redistribution is determined 'on the fly' based on the prediction of future simulation time. Compared to the PLP/ISAT strategy where chemistry calculations are essentially serial, a speed-up factor of up to 30 is achieved. The study also demonstrates that the adaptive strategy has acceptable parallel scalability.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Calculating adsorption isotherms using the two-phase thermodynamic method and molecular dynamics simulations

We describe the calculation of adsorption isotherms from molecular dynamics simulations based on the two-phase thermodynamic (2PT) model. The 2PT model developed for bulk fluid phases treats the gas-like components as hard spheres (HSs), which correctly recovers the limiting behaviors of unconfined fluids. We showed that this treatment, however, does not always lead to the correct zero-loading behavior in strongly confining systems. For methane adsorption into zeolite MFI, the HS reference state underestimates entropy by up to 20% at low loadings and leads to an order-of-magnitude increase in the adsorption onset pressure. To fix these issues, we propose the use of ideal adsorbed gas (IAG) as the gas reference model, the properties of which can be computed using the Widom insertion method on an empty adsorbent. We further describe three routes to compute adsorption isotherms from the Helmholtz free energy at different loadings. Comparing against established Monte Carlo (MC) methods, we found that the adsorption isotherms obtained using the IAG reference state agrees to within 40%, which corresponds to deviations of <5% in adsorption free energy. The isotherms calculated using the HS reference state underestimate the adsorption uptake at low to medium loadings in strongly confining systems, but its accuracy improves at higher loadings and as the pore size increases relative to the sorbate diameter. The methods described here provide an alternative approach for computing adsorption isotherms when MC simulations in an open ensemble are undesirable and enable a direct comparison of computed adsorption thermodynamics with experiments.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Analysis of Filesystem Utilization by the ?Ensemble of Models? Approach (U)

In order to execute Uncertainty Quantification (UQ) studies, the number of reads placed on the filesystem will increase. This document works through the file I/O for a climatology UQ study. The utilization of the filesystem for the above discussed will be similar for ICF and Stockpile Stewardship applications. The current state of the art for the quantification of uncertainty of a multi-physics simulation code is the utilization of an ensemble of models approach. As an overview of the ensemble of models approach, a set of uncertain input parameters is identified along with an identified set of observational and output parameters. The model is computed n number of times with each ensemble simulation using a unique set of parametric combinations of input parameters thereby creating an ensemble of simulations. Response surface models (also known as statistical emulator models, surrogate models, or meta models) are trained using the ensemble results. The response models are then convolved with observational data to further constrain input parameters and to create uncertainty bounds on the model outputs. Using the Community Climate System Model (CCSM) specifically the atmospheric component of CCSM, the Community Atmospheric Model (CAM), as the model of interest, this document provides a rough model of the demands on a filesystem that is needed to execute a UQ study on CAM. Each ensemble simulation consists of 12 simulation years and uses 384 processors on the Atlas machine. The LLNL UQ Pipeline is LLNL's standard tool to execute UQ studies. The UQ Pipeline possesses the capability to execute the ensemble simulations on LLNL's diverse set of HPC environments, produce response models, generate uncertainty bounds, and analyze the results. The process executing the LLNL UQ Pipeline is run on a different compute node from the set of concurrent, executing ensemble simulations.

99 GENERAL AND MISCELLANEOUS↗

Reducing the complexity of finite-temperature auxiliary-field quantum Monte Carlo

The auxiliary-field quantum Monte Carlo (AFMC) method is a powerful and widely used technique for ground-state and finite-temperature simulations of quantum many-body systems. Here we introduce several algorithmic improvements for finite-temperature AFMC calculations of dilute fermionic systems that reduce the computational complexity of most parts of the algorithm. This is principally achieved by reducing the number of single-particle states that contribute at each configuration of the auxiliary fields to a number that is of the order of the number of fermions. Our methods are applicable for both the canonical and grand-canonical ensembles. We demonstrate the reduced computational complexity of the methods for the homogeneous unitary Fermi gas.

67.85.Lm↗

Portable, heterogeneous ensemble workflows at scale using libEnsemble

libEnsemble is a Python-based toolkit for running dynamic ensembles, developed as part of the DOE Exascale Computing Project. The toolkit utilizes a unique generator–simulator–allocator paradigm, where generators produce input for simulators, simulators evaluate those inputs, and allocators decide whether and when a simulator or generator should be called. The generator steers the ensemble based on simulation results. Generators may, for example, apply methods for numerical optimization, machine learning, or statistical calibration. libEnsemble communicates between a manager and workers. Flexibility is provided through multiple manager–worker communication substrates each of which has different benefits. These include Python’s multiprocessing, mpi4py, and TCP. Multisite ensembles are supported using Balsam or Globus Compute. We overview the unique characteristics of libEnsemble as well as current and potential interoperability with other packages in the workflow ecosystem. We highlight libEnsemble’s dynamic resource features: libEnsemble can detect system resources, such as available nodes, cores, and GPUs, and assign these in a portable way. These features allow users to specify the number of processors and GPUs required for each simulation; and resources will be automatically assigned on a wide range of systems, including Frontier, Aurora, and Perlmutter. Such ensembles can include multiple simulation types, some using GPUs and others using only CPUs, sharing nodes for maximum efficiency. We also describe the benefits of libEnsemble’s generator–simulator coupling, which easily exposes to the user the ability to cancel, and portably kill, running simulations based on models that are updated with intermediate simulation output. We demonstrate libEnsemble’s capabilities, scalability, and scientific impact via a Gaussian process surrogate training problem for the longitudinal density profile at the exit of a plasma accelerator stage. In conclusion, the study uses gpCAM for the surrogate model and employs either Wake-T or WarpX simulations, highlighting efficient use of resources that can easily extend to exascale.

Dynamic ensembles↗

GPS Monitor Station Upgrade Program at the Naval Research Laboratory

One of the measurements made by the Global Positioning System (GPS) monitor stations is to measure the continuous pseudo-range of all the passing GPS satellites. The pseudo-range contains GPS and monitor station clock errors as well as GPS satellite navigation errors. Currently the time at the GPS monitor station is obtained from the GPS constellation and has an inherent inaccuracy as a result. Improved timing accuracy at the GPS monitoring stations will improve GPS performance. The US Naval Research Laboratory (NRL) is developing hardware and software for the GPS monitor station upgrade program to improve the monitor station clock accuracy. This upgrade will allow a method independent of the GPS satellite constellation of measuring and correcting monitor station time to US Naval Observatory (USNO) time. THe hardware consists of a high performance atomic cesium frequency standard (CFS) and a computer which is used to ensemble the CFS with the two CFS's currently located at the monitor station by use of a dual-mixer system. The dual-mixer system achieves phase measurements between the high-performance CFS and the existing monitor station CFS's to within 400 femtoseconds. Time transfer between USNO and a given monitor station is achieved via a two way satellite time transfer modem. The computer at the monitor station disciplines the CFS based on a comparison of one pulse per second sent from the master site at USNO. The monitor station computer is also used to perform housekeeping functions, as well as recording the health status of all three CFS's. This information is sent to the USNO through the time transfer modem. Laboratory time synchronization results in the sub nanosecond range have been observed and the ability to maintain the monitor station CFS frequency to within 3.0 x 10 (sup minus 14) of the master site at USNO.

Galysh, Ivan J.↗

Estimating Uncertainty in Simulated ENSO Statistics

Abstract Large ensembles of model simulations are frequently used to reduce the impact of internal variability when evaluating climate models and assessing climate change induced trends. However, the optimal number of ensemble members required to distinguish model biases and climate change signals from internal variability varies across models and metrics. Here we analyze the mean, variance and skewness of precipitation and sea surface temperature in the eastern equatorial Pacific region often used to describe the El Niño–Southern Oscillation (ENSO), obtained from large ensembles of Coupled model intercomparison project phase 6 climate simulations. Leveraging established statistical theory, we develop and assess equations to estimate, a priori, the ensemble size or simulation length required to limit sampling‐based uncertainties in ENSO statistics to within a desired tolerance. Our results confirm that the uncertainty of these statistics decreases with the square root of the time series length and/or ensemble size. Moreover, we demonstrate that uncertainties of these statistics are generally comparable when computed using either pre‐industrial control or historical runs. This suggests that pre‐industrial runs can sometimes be used to estimate the expected uncertainty of statistics computed from an existing historical member or ensemble, and the number of simulation years (run duration and/or ensemble size) required to adequately characterize the statistic. This advance allows us to use existing simulations (e.g., control runs that are performed during model development) to design ensembles that can sufficiently limit diagnostic uncertainties arising from simulated internal variability. These results may well be applicable to variables and regions beyond ENSO.

54 ENVIRONMENTAL SCIENCES↗

Evaluation of LRC Spectral Analyses of Simulated Gust Velocity Data

The NASA Langley Research Center (LRC) has been computing power spectra of simulated atmospheric turbulence data by various techniques, in preparation for the B-57 gust velocity data analysis program. The results of these studies have confirmed earlier suspicions that conventional first-difference pre-whitening of gust velocity data can produce serious distortions in the spectral densities at very low frequencies (below the frequency of the gust velocity spectral "knee"). The results also indicate that the frequency averaging procedures have certain merits over ensemble averaging procedures in the computation of spectra by direct Fourier transform operations. Finally, the results do not reveal any significant difference in the spectral estimates obtained using Hann versus Parzen smoothing procedures.

Piersol, Allan G.↗

Short-wavelength plasma turbulence and temperature anisotropy instabilities: Recent computational progress

Plasma turbulence consists of an ensemble of enhanced, broadband electromagnetic fluctuations, typically driven by multi-wave interactions which transfer energy in wavevector space via non- linear cascade processes. In addition, temperature anisotropy instabilities in collisionless plasmas are driven by quasi-linear wave–particle interactions which transfer particle kinetic energy to field fluctuation energy; the resulting enhanced fluctuations are typically narrowband in wavevector magnitude and direction. Whatever their sources, short-wavelength fluctuations are those at which charged particle kinetic, that is, velocity-space, properties are important; these are generally wavelengths of the order of or shorter than the ion inertial length or the thermal ion gyroradius. The purpose of this review is to summarize and interpret recent computational results concerning short-wavelength plasma turbulence, short-wavelength temperature anisotropy instabilities and relationships between the two phenomena.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Accurate predictions from small boxes: variance suppression via the Zel'dovich approximation

Simulations have become an indispensable tool for accurate modelling of observables measured in galaxy surveys, but can be expensive if very large dynamic range in scale is required. We describe how to combine Lagrangian perturbation theory models with N-body simulations to reduce the effects of finite computational volume in the prediction of ensemble average properties in the simulations within the context of control variates. In particular we use the fact that Zel'dovich displacements, computed during initial condition generation for any simulation, correlate strongly with the final density field. Since all the correlators of biased tracers can be computed with arbitrary precision for these displacements, pairing the Zel'dovich `simulation' with the N-body realization allows hundredfold reductions in sample variance for power spectrum or correlation function estimation. Additionally, Zel'dovich control variates can accurately extend matter or tracer field emulators to larger scales than previously possible, as well as improving measurements of statistics in simulations which are inherently limited to small volumes, such as hydrodynamical simulations of galaxy formation and reionization.

79 ASTRONOMY AND ASTROPHYSICS↗

Orchestrating Distributed Resource Ensembles for Petascale Science

Distributed, data-intensive computational science applications of interest to DOE scientific com- munities move large amounts of data for experiment data management, distributed analysis steps, remote visualization, and accessing scientific instruments. These applications need to orchestrate ensembles of resources from multiple resource pools and interconnect them with high-capacity multi- layered networks across multiple domains. It is highly desirable that mechanisms are designed that provide this type of resource provisioning capability to a broad class of applications. It is also important to have coherent monitoring capabilities for such complex distributed environments. In this project, we addressed these problems by designing an abstract API, enabled by novel semantic resource descriptions, for provisioning complex and heterogeneous resources from multiple providers using their native provisioning mechanisms and control planes: computational, storage, and multi-layered high-speed network domains. We used an extensible resource representation based on semantic web technologies to afford maximum flexibility to applications in specifying their needs. We evaluated the effectiveness of provisioning using representative data-intensive ap- plications. We also developed mechanisms for providing feedback about resource performance to the application, to enable closed-loop feedback control and dynamic adjustments to resource allo- cations (elasticity). This was enabled through development of a novel persistent query framework that consumes disparate sources of monitoring data, including perfSONAR, and provides scalable distribution of asynchronous notifications.

97 MATHEMATICS AND COMPUTING↗

Reconstruction of the 1997/1998 El Nino from TOPEX/POSEIDON and TOGA/TAO Data Using a Massively Parallel Pacific-Ocean Model and Ensemble Kalman Filter

Two massively parallel data assimilation systems in which the model forecast-error covariances are estimated from the distribution of an ensemble of model integrations are applied to the assimilation of 97-98 TOPEX/POSEIDON altimetry and TOGA/TAO temperature data into a Pacific basin version the NASA Seasonal to Interannual Prediction Project (NSIPP)ls quasi-isopycnal ocean general circulation model. in the first system, ensemble of model runs forced by an ensemble of atmospheric model simulations is used to calculate asymptotic error statistics. The data assimilation then occurs in the reduced phase space spanned by the corresponding leading empirical orthogonal functions. The second system is an ensemble Kalman filter in which new error statistics are computed during each assimilation cycle from the time-dependent ensemble distribution. The data assimilation experiments are conducted on NSIPP's 512-processor CRAY T3E. The two data assimilation systems are validated by withholding part of the data and quantifying the extent to which the withheld information can be inferred from the assimilation of the remaining data. The pros and cons of each system are discussed.

Keppenne, C. L.↗