Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “high performance space 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 1,063 records · Page 59

The Mira-Titan Universe. II. Matter Power Spectrum Emulation

We introduce a new cosmic emulator for the matter power spectrum covering eight cosmological parameters. Targeted at optical surveys, the emulator provides accurate predictions out to a wavenumber k ~ 5Mpc -1 and redshift z ≤ 2. Besides covering the standard set of CDM parameters, massive neutrinos and a dynamical dark energy of state are included. The emulator is built on a sample set of 36 cosmological models, carefully chosen to provide accurate predictions over the wide and large parameter space. For each model, we have performed a high-resolution simulation, augmented with sixteen medium-resolution simulations and TimeRG perturbation theory results to provide accurate coverage of a wide k-range; the dataset generated as part of this project is more than 1.2Pbyte. With the current set of simulated models, we achieve an accuracy of approximately 4%. Because the sampling approach used here has established convergence and error-control properties, follow-on results with more than a hundred cosmological models will soon achieve ~1% accuracy. We compare our approach with other prediction schemes that are based on halo model ideas and remapping approaches. The new emulator code is publicly available.

97 MATHEMATICS AND COMPUTING↗

Uncertainty-driven dynamics for active learning of interatomic potentials

Machine learning (ML) models, if trained to data sets of high-fidelity quantum simulations, produce accurate and efficient interatomic potentials. Active learning (AL) is a powerful tool to iteratively generate diverse data sets. In this approach, the ML model provides an uncertainty estimate along with its prediction for each new atomic configuration. If the uncertainty estimate passes a certain threshold, then the configuration is included in the data set. Here we develop a strategy to more rapidly discover configurations that meaningfully augment the training data set. The approach, uncertainty-driven dynamics for active learning (UDD-AL), modifies the potential energy surface used in molecular dynamics simulations to favor regions of configuration space for which there is large model uncertainty. The performance of UDD-AL is demonstrated for two AL tasks: sampling the conformational space of glycine and sampling the promotion of proton transfer in acetylacetone. The method is shown to efficiently explore the chemically relevant configuration space, which may be inaccessible using regular dynamical sampling at target temperature conditions.

97 MATHEMATICS AND COMPUTING↗

The Effect of Geometry, Spin, and Orbital Optimization in Achieving Accurate, Correlated Results for Iron–Sulfur Cubanes

Iron–sulfur clusters comprise an important functional motif in the catalytic centers of biological systems, capable of enabling important chemical transformations at ambient conditions. This remarkable capability derives from a notoriously complex electronic structure that is characterized by a high density of states that is sensitive to geometric changes. The spectral sensitivity to subtle geometric changes has received little attention from correlated, large active space calculations, owing partly to the exceptional computational complexity for treating these large and correlated systems accurately. Here, to provide insight into this aspect, we report the first Complete Active Space Self Consistent Field (CASSCF) calculations for different geometries of the [Fe(II/III) 4 S 4 (SMe) 4 ] -2 clusters using two complementary, correlated solvers: spin-pure Adaptive Sampling Configuration Interaction (ASCI) and Density Matrix Renormalization Group (DMRG). We find that the previously established picture of a double-exchange driven magnetic structure, with minute energy gaps (<1 mHa) between consecutive spin states, has a weak dependence on the underlying geometry. However, the spin gap between the singlet and the spin state 2S + 1 = 19, corresponding to a maximal number of Fe-d electrons being unpaired and of parallel spin, is strongly geometry dependent, changing by a factor of 3 upon slight deformations that are still within biologically relevant parameters. The CASSCF orbital optimization procedure, using active spaces as large as 86 electrons in 52 orbitals, was found to reduce this gap compared to typical mean-field orbital approaches. Our results show the need for performing large active space calculations to unveil the challenging electronic structure of these complex catalytic centers and should serve as accurate starting points for fully correlated treatments upon inclusion of dynamical correlation outside the active space.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Transitions, Dynamics, and Driven States in Quantum Magnets (Final Report Revised)

Transitions into unusual electronic, optical, and magnetic states at the absolute zero of temperature display special properties that teach us about fundamental quantum mechanics and also underlie important technologies. These so-called quantum phase transitions intertwine the static and dynamical responses of the materials changing state. They are extremely sensitive to the effects of disorder and etch in high relief the role of quantum fluctuations. There are ample questions remaining about the character of such transitions and the nature of the competing quantum states. There are also opportunities to drive quantum materials out of equilibrium, with the possibility of new types of correlated and coherent order, and new ways to access the dynamical evolution of ground and excited states. We have investigated how the order develops in time and space, and the nature of the final configurations. We are able to compare and contrast classical and quantum excitations in our experimental system of choice, and thereby can trace the relative speed by which the system settles into its lowest-energy ground state. This comparison of quantum and classical annealing protocols lies at the heart of strategies for harnessing the power of quantum computers dedicated to optimization problems. A combination of magnetic susceptibility measurements at finite frequency where the magnetic spins are queried by an external ac field, dc magnetometry measurements that characterize the macroscopic strength of magnetic order and its resistance to change, noise measurements that reveal the microscopic fluctuations of the spins as they prepare to change state, and microwave spectroscopy to look at the response of both the electronic and nuclear degrees of freedom, have been performed as functions of temperature, magnetic field, frequency, excitation amplitude, magnetic spin concentration, and disorder. This combination of probes addresses issues of stability over time, overlap between spin states, the ability of the spins to respond independently or coherently, and the promise of control on the nanometer length scale. The research reported here provides insights into fundamental quantum physics. It also addresses the question of how best to use complex systems, such as magnetic solids, to stably process quantum information.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

High-throughput screening of tribological properties of monolayer films using molecular dynamics and machine learning

Monolayer films have shown promise as a lubricating layer to reduce friction and wear of mechanical devices with separations on the nanoscale. These films have a vast design space with many tunable properties that can affect their tribological effectiveness. For example, terminal group chemistry, film composition, and backbone chemistry can all lead to films with significantly different tribological properties. This design space, however, is very difficult to explore without a combinatorial approach and an automatable, reproducible, and extensible workflow to screen for promising candidate films. Here, using the Molecular Simulation Design Framework (MoSDeF), a combinatorial screening study was performed to explore 9747 unique monolayer films (116 964 total simulations) and a machine learning (ML) model using a random forest regressor, an ensemble learning technique, to explore the role of terminal group chemistry and its effect on tribological effectiveness. The most promising films were found to contain small terminal groups such as cyano and ethylene. The ML model was subsequently applied to screen terminal group candidates identified from the ChEMBL small molecule library. Approximately 193 131 unique film candidates were screened with approximately a five order of magnitude speed-up in analysis compared to simulation alone. The ML model was thus able to be used as a predictive tool to greatly speed up the initial screening of promising candidate films for future simulation studies, suggesting that computational screening in combination with ML can greatly increase the throughput in combinatorial approaches to generate in silico data and then train ML models in a controlled, self-consistent fashion.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Research on the ATLAS Experiment at the University of Illinois: Final Report

The goal of particle physics is to understand the Universe at its most fundamental level, including the constituents of matter and their interactions, and the nature of space and time itself. The quest to understand the fundamental building blocks of nature, and their interactions, is one of the most ambitious and enduring of human endeavors. Recent developments in particle physics and cosmology raise the exciting prospect that we are on the threshold of a major step forward in this pursuit. Facilities such as CERN's Large Hadron Collider (LHC) provide a unique window into the subatomic world to pursue this understanding. Mark Neubauer and his research group at the University of Illinois were supported over a nine-month period (07/01/2017 through 03/31/2018) to perform high-energy particle physics research on the ATLAS experiment at the LHC. The primary focus of this research was to search for physics beyond the standard model through an analysis of high-energy proton-proton collision data from the ATLAS experiment, with an emphasis on the Higgs boson as a sensitive tool for this search, and development of fast particle tracking systems for use the ATLAS trigger system.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Reification of latent microstructures: On supervised unsupervised and semi-supervised deep learning applications for microstructures in materials informatics

Machine learning (ML), including deep learning (DL), has become increasingly popular in the last few years due to its continually outstanding performance. In this context, we apply machine learning techniques to "learn" the microstructure using both supervised and unsupervised DL techniques. In particular, we focus (1) on the localization problem bridging (micro)structure (localized) property using supervised DL and (2) on the microstructure reconstruction problem in latent space using unsupervised DL. The goal of supervised and semi-supervised DL is to replace crystal plasticity finite element model (CPFEM) that maps from (micro)structure (localized) property, and implicitly the (micro)structure (homogenized) property relationships, while the goal of unsupervised DL is (1) to represent high-dimensional microstructure images in a non-linear low-dimensional manifold, and (2) to discover a way to interpolate microstructures via latent space associating with latent microstructure variables. At the heart of this report is the applications of several common DL architectures, including convolutional neural networks (CNN), autoencoder (AE), and generative adversarial network (GAN), to multiple microstructure datasets, and the quest of neural architecture search for optimal DL architectures.

36 MATERIALS SCIENCE↗

Characterization of the Martinsburg Formation using Computed Tomography and Geophysical Logging Techniques

The computed tomography (CT) facilities and the Multi-Sensor Core Logger (MSCL) at the National Energy Technology Laboratory (NETL) Morgantown site were used to characterize 9 different core sections of Upper Ordovician black shale from the Eastern Panhandle of West Virginia. The samples are representative of the upper Ordovician Martinsburg Formation, a Utica Shale equivalent, which is a carbonaceous black shale in the Appalachian Basin. The primary impetus of this work was to develop a standardized methodology for the rapid and detailed characterization of chemical composition, physical characteristics, and internal micro structure of lithologic samples with the NETL core characterization equipment. Having a standardized methodology combining these innovative techniques allows large, rapidly produced data sets to be available for multiple researchers and allows collaborative studies. The resultant data sets are presented as part of this report as well and can be accessed from NETL's Energy Data eXchange (EDX) online system (https://edx.netl.doe.gov) using the following link: https://edx.netl.doe.gov/dataset/martinsburg-formation-ct-data. All equipment and techniques used were non-destructive, enabling future examinations to be performed on these cores with the additional information gathered through these tests. None of the equipment used was suitable for direct visualization of the shale pore space, though fractures and discontinuities were visible and detectable with all methods tested. High resolution CT imagery with the NETL industrial CT scanner was a powerful and insightful way to examine the details of fractures, discontinuities, minerals, and large crystals within the shale, but was time consuming both in image capture and analysis. As such, only a small percentage of the core was scanned at high resolution. Low resolution CT imagery with the NETL medical CT scanner was performed on the entire core. Qualitative analysis of the medical CT images, coupled with the measurements from the multi-sensor core logger (including x-ray fluorescence, XRF, measurements) were very useful in identifying zones of interest for more detailed analysis and fractured zones. The ability to quickly identify key areas for more detailed study with higher resolution will save time and resources in future studies. The combination of all methods used provided a multi-scale analysis of this core and provided both a macro and micro description of the core that is relevant for many subsurface energy related examinations of core that have traditionally been performed at NETL.

47 OTHER INSTRUMENTATION↗

Data Summarization and Inference at Scale

This is the final report for the DOE ASCR grant SC-0022260, Data Summarization and Inference at Scale, PI: Alex Pothen, Purdue University. The goal of the project was to solve data-intensive and compute-intensive problems in the physical sciences, engineering, information science, data science, etc. by designing and implementing new algorithms that could work with a subset of the data. The four subgoals were: (a) The solution of problems where the data is too large to be stored in the memory of a computer. In this streaming model of computation, the data arrives as a stream of elements to the computer, each element is processed as it arrives, and a decision is made to discard the data or to store it; only a small subset of the data proportional to the size of the output solution is stored, and when all the data has been streamed, a solution to the problem is computed from the stored subset. (b) The use of machine learning methods to compute solutions to data-intensive problems. The use of GPUs is critical to obtain high performance on machine learning tasks, but their memory sizes are smaller relative to that of CPUs. For large-scale problems, the data is sampled many times, and small samples are used with repetition, for robustness, to compute solutions to inference tasks. This sampling reduces the memory required to solve the problem, but attention is needed to avoid slow convergence to the solutions, and reduced accuracy of inference. We propose submodular optimization, Large Language Models, and physics-informed neural networks to enable GPU computations here. (c) Modeling and visualization of high-dimensional data using interpretable features. Clinical proteomic data sets from immunology for the detection of cancer and other diseases are temporal and high-dimensional, and algorithms for visualizing these data sets using clinically interpretable features are lacking. We propose methods that compute distances based on the optimal transportation problem and graph edit distances to address this problem. We also propose the use of optimal transport-based distances, spatial statistics, and network structure to classify image data sets, We apply these algorithms to electron micrographs of the peripheral nervous system in the digestive tract. (d) The design of data-intensive algorithms on emerging architectures, specifically, noisy, intermediate-scale quantum (NISQ) devices. Quantum computers offer the possibility of exploring large solution spaces due to the principle of superposition, but current quantum computers are limited by few qubits, short coherence times due to noise, poor interconections among the qubits, etc. We propose the use of the divide and conquer paradigm to solve large-scale problems, wherein collections of small subproblems are solved on the quantum devices, and the solutions to the subproblems are integrated into a solution for the original problem on a classical computer.

97 MATHEMATICS AND COMPUTING↗

Feature Tracking Using Reeb Graphs

Tracking features and exploring their temporal dynamics can aid scientists in identifying interesting time intervals in a simulation and serve as basis for performing quantitative analyses of temporal phenomena. In this paper, we develop a novel approach for tracking subsets of isosurfaces, such as burning regions in simulated flames, which are defined as areas of high fuel consumption on a temperature isosurface. Tracking such regions as they merge and split over time can provide important insights into the impact of turbulence on the combustion process. However, the convoluted nature of the temperature isosurface and its rapid movement make this analysis particularly challenging. Our approach tracks burning regions by extracting a temperature isovolume from the four-dimensional space-time temperature field. It then obtains isosurfaces for the original simulation time steps and labels individual connected 'burning' regions based on the local fuel consumption value. Based on this information, a boundary surface between burning and non-burning regions is constructed. The Reeb graph of this boundary surface is the tracking graph for burning regions.

97 MATHEMATICS AND COMPUTING↗

Reducing Disk Storage of Full-3D Seismic Waveform Tomography (F3DT) Through Lossy Online Compression

Full-3D seismic waveform tomography (F3DT) is the latest seismic tomography technique that can assimilate broadband, multi-component seismic waveform observations into high-resolution 3D subsurface seismic structure models. The main drawback in the current F3DT implementation, in particular the scattering-integral implementation (F3DT-SI), is the high disk storage cost and the associated I/O overhead of archiving the 4D space-time wavefields of the receiver- or source-side strain tensors. The strain tensor fields are needed for computing the data sensitivity kernels, which are used for constructing the Jacobian matrix in the Gauss-Newton optimization algorithm. In this study, we have successfully integrated a lossy compression algorithm into our F3DT SI workflow to significantly reduce the disk space for storing the strain tensor fields. The compressor supports a user-specified tolerance for bounding the error, and can be integrated into our finite-difference wave-propagation simulation code used for computing the strain fields. The decompressor can be integrated into the kernel calculation code that reads the strain fields from the disk and compute the data sensitivity kernels. During the wave-propagation simulations, we compress the strain fields before writing them to the disk. To compute the data sensitivity kernels, we read the compressed strain fields from the disk and decompress them before using them in kernel calculations. Experiments using a realistic dataset in our California statewide F3DT project have shown that we can reduce the strain-field disk storage by at least an order of magnitude with acceptable loss, and also improve the overall I/O performance of the entire F3DT-SI workflow significantly. The integration of the lossy online compressor may potentially open up the possibilities of the wide adoption of F3DT-SI in routine seismic tomography practices in the near future.

97 MATHEMATICS AND COMPUTING↗

SANE: strategic autonomous non-smooth exploration for multiple optima discovery in multi-modal and non-differentiable black-box functions

Both computational and experimental material discovery bring forth the challenge of exploring multidimensional and multimodal parameter spaces, such as phase diagrams of Hamiltonians with multiple interactions, composition spaces of combinatorial libraries, material structure image spaces, and molecular embedding spaces. Often these systems are black-boxes and time-consuming to evaluate, which resulted in strong interest towards active learning methods such as Bayesian optimization (BO). However, these systems are often noisy which make the black box function severely multi-modal and non-differentiable, where a vanilla BO can get overly focused near a single or faux optimum, deviating from the broader goal of scientific discovery. To address these limitations, here we developed Strategic Autonomous Non-Smooth Exploration (SANE) to facilitate an intelligent Bayesian optimized navigation with a proposed cost-driven probabilistic acquisition function to find multiple global and local optimal regions, avoiding the tendency to becoming trapped in a single optimum. To distinguish between a true and false optimal region due to noisy experimental measurements, a human (domain) knowledge driven dynamic surrogate gate is integrated with SANE. We implemented the gate-SANE into pre-acquired piezoresponse spectroscopy data of a ferroelectric combinatorial library with high noise levels in specific regions, and piezoresponse force microscopy (PFM) hyperspectral data. SANE demonstrated better performance than classical BO to facilitate the exploration of multiple optimal regions and thereby prioritized learning with higher coverage of scientific values in autonomous experiments. Our work showcases the potential application of this method to real-world experiments, where such combined strategic and human intervening approaches can be critical to unlocking new discoveries in autonomous research.

Biswas, Arpan [University of Tennessee, Knoxville,↗

Learning genetic perturbation effects with variational causal inference

Advances in sequencing technologies have enhanced the understanding of gene regulation in cells. In particular, Perturb-seq has enabled high-resolution profiling of the transcriptomic response to genetic perturbations at the single-cell level. This understanding has implications in functional genomics and potentially for identifying therapeutic targets. Various computational models have been developed to predict perturbational effects. While deep learning models excel at interpolating observed perturbational data, they tend to overfit in the lack of enough data and may not generalize well to unseen perturbations. In contrast, mechanistic models, such as linear causal models based on gene regulatory networks, hold greater potential for extrapolation, as they encapsulate regulatory information that can predict responses to unseen perturbations. However, their application has been limited to small studies due to overly simplistic assumptions, making them less effective in handling noisy, large-scale single-cell data. We propose a hybrid approach that combines a mechanistic causal model with variational deep learning, termed Single Cell Causal Variational Autoencoder (SCCVAE). The mechanistic model employs a learned regulatory network to represent perturbational changes as shift interventions that propagate through the learned network. SCCVAE integrates this mechanistic causal model into a variational autoencoder, generating rich, comprehensive transcriptomic responses. Our results indicate that SCCVAE exhibits superior performance over current state-of-the-art baselines for extrapolating to predict unseen perturbational responses. Additionally, for the observed perturbations, the latent space learned by SCCVAE allows for the identification of functional perturbation modules and simulation of single-gene knockdown experiments of varying penetrance, presenting a robust tool for interpreting and interpolating perturbational responses at the single-cell level.

59 BASIC BIOLOGICAL SCIENCES↗

Modeling Material Interfaces with the Six-Equation Model in FLEXO

FLEXO is a multiphysics code developed at Sandia National Laboratories for predictive simulation of pulsed power target physics with extended magnetohydrodynamics modeling. Given this application space, FLEXO must be able to perform high-fidelity simulations of pulsed-power systems that include multiple materials. This work presents the augmentation of FLEXO with a six-equation model for multimaterial flow. We design and analyze discretizations of the six-equation model, emphasizing desirable properties for robust multimaterial simulation. We also present solution limiting, bounds preservation, and pressure equilibration/relaxation techniques that enhance the robustness of the overall multimaterial scheme. We present a suite of test problems to document these new capabilities of the FLEXO code.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Rapid sampling of stochastic displacements in Brownian dynamics simulations

Here, we present a new method for sampling stochastic displacements in Brownian Dynamics (BD) simulations of colloidal scale particles. The method relies on a new formulation for Ewald summation of the Rotne-Prager-Yamakawa (RPY) tensor, which guarantees that the real-space and wave-space contributions to the tensor are independently symmetric and positive-definite for all possible particle configurations. Brownian displacements are drawn from a superposition of two independent samples: a wave-space (far-field or long-ranged) contribution, computed using techniques from fluctuating hydrodynamics and non-uniform fast Fourier transforms; and a real-space (near-field or short-ranged) correction, computed using a Krylov subspace method. The combined computational complexity of drawing these two independent samples scales linearly with the number of particles. The proposed method circumvents the super-linear scaling exhibited by all known iterative sampling methods applied directly to the RPY tensor that results from the power law growth of the condition number of tensor with the number of particles. For geometrically dense microstructures (fractal dimension equal three), the performance is independent of volume fraction, while for tenuous microstructures (fractal dimension less than three), such as gels and polymer solutions, the performance improves with decreasing volume fraction. This is in stark contrast with other related linear-scaling methods such as the force coupling method and the fluctuating immersed boundary method, for which performance degrades with decreasing volume fraction. Calculations for hard sphere dispersions and colloidal gels are illustrated and used to explore the role of microstructure on performance of the algorithm. In practice, the logarithmic part of the predicted scaling is not observed and the algorithm scales linearly for up to 4×10 6 particles, obtaining speed ups of over an order of magnitude over existing iterative methods, and making the cost of computing Brownian displacements comparable to the cost of computing deterministic displacements in BD simulations. Finally, a high-performance implementation employing non-uniform fast Fourier transforms implemented on graphics processing units and integrated with the software package HOOMD-blue is used for benchmarking.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Speed vs Accuracy: Effect on Ligand Pose Accuracy of Varying Box Size and Exhaustiveness in AutoDock Vina

Abstract Structure‐based virtual high‐throughput screening involves docking chemical libraries to targets of interest. A parameter pertinent to the accuracy of the resulting pose is the root mean square deviation (RMSD) from a known crystallographic structure, i. e., the ‘docking power’. Here, using a popular algorithm, Autodock Vina, as a model program, we evaluate the effects of varying two common docking parameters: the box size (the size of docking search space) and the exhaustiveness of the global search (the number of independent runs starting from random ligand conformations) on the RMSD from the PDBbind v2017 refined dataset of experimental protein‐ligand complexes. Although it is clear that exhaustiveness is an important parameter, there is wide variation in the values used, with variation between 1 and >100. We, therefore, evaluated a combination of cubic boxes of different sizes and five exhaustiveness values (1, 8, 25, 50, 75, 100) within the range of those commonly adopted. The results show that the default exhaustiveness value of 8 performs well overall for most box sizes. In contrast, for all box sizes, but particularly for large boxes, an exhaustiveness value of 1 led to significantly higher median RMSD (mRMSD) values. The docking power was slightly improved with an exhaustiveness of 25, but the mRMSD changes little with values higher than 25. Therefore, although low exhaustiveness is computationally faster, the results are more likely to be far from reality, and, conversely, values >25 led to little improvement at the expense of computational resources. Overall, we recommend users to use at least the default exhaustiveness value of 8 for virtual screening calculations.

Agarwal, Rupesh↗

A Validation Study of the Compressible Rayleigh–Taylor Instability Comparing the Ares and Miranda Codes

In this paper, the compressible Rayleigh–Taylor (RT) instability is studied by performing a suite of large eddy simulations (LES) using the Miranda and Ares codes. A grid convergence study is carried out for each of these computational methods, and the convergence properties of integral mixing diagnostics and late-time spectra are established. A comparison between the methods is made using the data from the highest resolution simulations in order to validate the Ares hydro scheme. We find that the integral mixing measures, which capture the global properties of the RT instability, show good agreement between the two codes at this resolution. The late-time turbulent kinetic energy and mass fraction spectra roughly follow a Kolmogorov spectrum, and drop off as k approaches the Nyquist wave number of each simulation. The spectra from the highest resolution Miranda simulation follow a Kolmogorov spectrum for longer than the corresponding spectra from the Ares simulation, and have a more abrupt drop off at high wave numbers. The growth rate is determined to be between around 0.03 and 0.05 at late times; however, it has not fully converged by the end of the simulation. Finally, we study the transition from direct numerical simulation (DNS) to LES. The highest resolution simulations become LES at around t/τ ≃ 1.5. Finally, to have a fully resolved DNS through the end of our simulations, the grid spacing must be 3.6 (3.1) times finer than our highest resolution mesh when using Miranda (Ares).

42 ENGINEERING↗

Accelerated Dimension-Independent Adaptive Metropolis

This work describes improvements from algorithmic and architectural means to black-box Bayesian inference over high-dimensional parameter spaces. The well-known adaptive Metropolis (AM) algorithm [33] is extended herein to scale asymptotically uniformly with respect to the underlying parameter dimension for Gaussian targets, by respecting the variance of the target. The resulting algorithm, referred to as the dimension-independent adaptive Metropolis (DIAM) algorithm, also shows improved performance with respect to adaptive Metropolis on non-Gaussian targets. This algorithm is further improved, and the possibility of probing high-dimensional (with dimension d 1000) targets is enabled, via GPU-accelerated numerical libraries and periodically synchronized concurrent chains (justi ed a posteriori). Asymptotically in dimension, this GPU implementation exhibits a factor of four improvement versus a competitive CPU-based Intel MKL parallel version alone. Strong scaling to concurrent chains is exhibited, through a combination of longer time per sample batch (weak scaling) and yet fewer necessary samples to convergence. The algorithm performance is illustrated on several Gaussian and non-Gaussian target examples, in which the dimension may be in excess of one thousand.

97 MATHEMATICS AND COMPUTING↗