A Data-driven Approach to Harvesting Latent Reduced Models to Precondition Lossy Compression for Scientific Data
Not provided.
SEARCH · Engineering Papers
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.
Not provided.
As parallel computing trends towards the exascale, scientific data produced by high-fidelity simulations are growing increasingly massive. For instance, a simulation on a three-dimensional spatial grid with 512 points per dimension that tracks 64 variables per grid point for 128 time steps yields 8 TB of data. By viewing the data as a dense five way tensor, we can compute a Tucker decomposition to find inherent low-dimensional multilinear structure, achieving compression ratios of up to 10000 on real-world data sets with negligible loss in accuracy. So that we can operate on such massive data, we present the first-ever distributed memory parallel implementation for the Tucker decomposition, whose key computations correspond to parallel linear algebra operations, albeit with nonstandard data layouts. Our approach specifies a data distribution for tensors that avoids any tensor data redistribution, either locally or in parallel. We provide accompanying analysis of the computation and communication costs of the algorithms. To demonstrate the compression and accuracy of the method, we apply our approach to real-world data sets from combustion science simulations. We also provide detailed performance results, including parallel performance in both weak and strong scaling experiments.
Machine learning interatomic potentials (MLIPs) balance high accuracy and lower costs compared to density functional theory calculations, but their performance often depends on the size and diversity of training datasets. Large datasets improve model accuracy and generalization but are computationally expensive to produce and train on, while smaller datasets risk discarding rare but important atomic environments and compromising MLIP accuracy/reliability. Here, we develop an information-theoretical framework to quantify the efficiency of dataset compression methods and propose an algorithm that maximizes this efficiency. By framing atomistic dataset compression as an instance of the minimum set cover (MSC) problem over atom-centered environments, our method identifies the smallest subset of structures that contains as much information as possible from the original dataset while pruning redundant information. The approach is extensively demonstrated on the GAP-20 and TM23 datasets and validated on 64 varied datasets from the ColabFit repository. Across all cases, MSC consistently retains outliers, preserves dataset diversity, and reproduces the long-tail distributions of forces even at high compression rates, outperforming other subsampling methods. Furthermore, MLIPs trained on MSC-compressed datasets exhibit reduced error for out-of-distribution data even in low-data regimes. We explain these results using an outlier analysis and show that such quantitative conclusions could not be achieved with conventional dimensionality reduction methods. The algorithm is implemented in the open-source QUESTS package and can be used for several tasks in atomistic modeling, from data subsampling, outlier detection, and training improved MLIPs at a lower cost.
Explore the source record for details and available documents.
Presented and discussed here is the implementation of a software solution that provides prompt X-ray diffraction data analysis during fast dynamic compression experiments conducted within the dynamic diamond anvil cell technique. It includes efficient data collection, streaming of data and metadata to a high-performance cluster (HPC), fast azimuthal data integration on the cluster, and tools for controlling the data processing steps and visualizing the data using the DIOPTAS software package. This data processing pipeline is invaluable for a great number of studies. The potential of the pipeline is illustrated with two examples of data collected on ammonia–water mixtures and multiphase mineral assemblies under high pressure. The pipeline is designed to be generic in nature and could be readily adapted to provide rapid feedback for many other X-ray diffraction techniques, e.g. large-volume press studies, in situ stress/strain studies, phase transformation studies, chemical reactions studied with high-resolution diffraction etc.
In this document, we provide the Hugoniot for sodium from two models: LEOS table L110 and Lynx table 110. We also provide the Hugoniot for lithium fluoride from LEOS (L2240) and Lynx (2240). The Hugoniot pressures are supplied for temperatures between 338.0 and 1.16×10 9 Kelvin and densities between 0.968 and 11.5 g/cc. These LEOS models were developed by the quotidian EOS methodology, which is a widely used and robust method for producing tabular EOS data. Tables list the model data for LEOS 110, Lynx 110, LEOS 2240, and Lynx 2240. The Lynx models follow the same methodology as the LEOS models; however, the Purgatorio average-atom DFT code was used to compute the electron thermal part of the EOS. The models for Lynx are only listed at high compression due to known issues with the Lynx library at lower pressures.
We report the multiscale complexity of modern problems in computational science and engineering can prohibit the use of traditional numerical methods in multi-dimensional simulations. Therefore, novel algorithms are required in these situations to solve partial differential equations (PDEs) with features evolving on a wide range of spatial and temporal scales. To meet these challenges, we present a multiresolution wavelet algorithm to solve PDEs with significant data compression and explicit error control. We discretize in space by projecting fields and spatial derivative operators onto wavelet basis functions. We provide error estimates for the wavelet representation of fields and their derivatives. Then, our estimates are used to construct a sparse multiresolution discretization which guarantees the prescribed accuracy. Additionally, we embed a predictor-corrector procedure within the temporal integration to dynamically adapt the computational grid and maintain the accuracy of the solution of the PDE as it evolves. We present examples to highlight the accuracy and adaptivity of our approach.
We present a new convolution layer for deep learning architectures which we call QuadConv — an approximation to continuous convolution via quadrature. Our operator is developed explicitly for use on non-uniform, mesh-based data, and accomplishes this by learning a continuous kernel that can be sampled at arbitrary locations. Moreover, the construction of our operator admits an efficient implementation which we detail and construct. As an experimental validation of our operator, we consider the task of compressing partial differential equation (PDE) simulation data from fixed meshes. Here, we show that QuadConv can match the performance of standard discrete convolutions on uniform grid data by comparing a QuadConv autoencoder (QCAE) to a standard convolutional autoencoder (CAE). Further, we show that the QCAE can maintain this accuracy even on non-uniform data. In both cases, QuadConv also outperforms alternative unstructured convolution methods such as graph convolution.
There has been a challenge for many decades to understand how heterogeneities influence the behavior of materials under shock loading, eventually leading to spall formation and failure. Experimental, analytical, and computational techniques have matured to the point where systematic studies of materials with complex microstructures under shock loading and the associated failure mechanisms are feasible. This is enabled by more accurate diagnostics as well as characterization methods. As interest in complex materials grows, understanding and predicting the role of heterogeneities in determining the dynamic behavior becomes crucial. Early computational studies, hydrocodes, in particular, historically preclude any irregularities in the form of defects and impurities in the material microstructure for the sake of simplification and to retain the hydrodynamic conservation equations. Contemporary computational methods, notably molecular dynamics simulations, can overcome this limitation by incorporating inhomogeneities albeit at a much lower length and time scale. This review discusses literature that has focused on investigating the role of various imperfections in the shock and spall behavior, emphasizing mainly heterogeneities such as second-phase particles, inclusions, and voids under both shock compression and release. Pre-existing defects are found in most engineering materials, ranging from thermodynamically necessary vacancies, to interstitial and dislocation, to microstructural features such as inclusions, second phase particles, voids, grain boundaries, and triple junctions. This literature review explores the interaction of these heterogeneities under shock loading during compression and release. Systematic characterization of material heterogeneities before and after shock loading, along with direct measurements of Hugoniot elastic limit and spall strength, allows for more generalized theories to be formulated. Further, continuous improvement toward time-resolved, in situ experimental data strengthens the ability to elucidate upon results gathered from simulations and analytical models, thus improving the overall ability to understand and predict how materials behave under dynamic loading.
Our primary use of the institutional computing clusters has been for Wikipedia-based disease surveillance and forecasting work. We've expanded breadth: the number of diseases and locations that can be surveilled using internet data. We've expanded forecasting: surprisingly little work towards forecasting disease spread with quantified uncertainty. We've expanded big data processing techniques: the Wikipedia access logs are 5TB of compressed text; we're working on efficient ways to process and store time series. Our work received significant media coverage and was highlighted as LANL's best science story of 2014.
The properties of silica (SiO 2 ) at extreme conditions have important applications for planetary processes and for high pressure research. Here, we report the results of 125 plate impact shock compression experiments on fused silica spanning 200–1100 GPa using the Z machine at Sandia National Laboratories. Additionally, we present a complementary set of density functional theory based molecular dynamics calculations based on an amorphous reference state that extend the Hugoniot to 2500 GPa. We find good agreement between the Z data, extant laser driven shock compression experiment data, and computational results over most of the pressure range. With these results, fused silica can be used as a new impedance matching standard for shock compression experiments.
As we continue toward exascale, scientific data volume is continuing to scale and becoming more burdensome to manage. In this paper, we lay out opportunities to enhance state of the art data management techniques. We emphasize well-principled data compression, and using it to achieve progressive refinement. This can both accelerate I/O and afford the user increased flexibility when she interacts with the data. The formulation naturally maps onto enabling partitioning of the progressively improving-quality representations of a data quantity into different media-type destinations, to keep the highest priority information as close as possible to the computation, and take advantage of deepening memory/storage hierarchies in ways not previously possible. Careful monitoring is requisite to our vision, not only to verify that compression has not eliminated salient features in the data, but also to better understand the performance of massively parallel scientific applications. Increased mathematical rigor would be ideal,to help bring compression on a better-understood theoretical footing, closer to the relevant scientific theory, more aware of constraints imposed by the science, and more tightly error-controlled. Throughout, we highlight pathfinding research we have begun exploring related these topics, and comment toward future work that will be needed.
Exploding dataset sizes from extreme-scale scientific simulations necessitates efficient data management and reduction schemes to mitigate I/O costs. With the discrepancy between I/O bandwidth and computational power, scientists are forced to capture data infrequently, thereby making data collection an inherently lossy process. Although data compression can be an effective solution, the random nature of real-valued scientific datasets renders lossless compression routines ineffective. These techniques also impose significant overhead during decompression, making them unsuitable for data analysis and visualization, which require repeated data access. To address this problem, we propose an effective method for In situ Sort-And-B-spline Error-bounded Lossy Abatement (ISABELA) of scientific data that is widely regarded as effectively incompressible. With ISABELA, we apply a pre-conditioner to seemingly random and noisy data along spatial resolution to achieve an accurate fitting model that guarantees a ≥0.99 correlation with the original data. We further take advantage of temporal patterns in scientific data to compress data b ≈ 85%, while introducing only a negligible overhead on simulations in terms of runtime. ISABELA significantly outperforms existing lossy compression methods, such as wavelet compression, in terms of data reduction and accuracy. We extend upon our previous paper by additionally building a communication-free, scalable parallel storage framework on top of ISABELA-compressed data that is ideally suited for extreme-scale analytical processing. The basis for our storage framework is an inherently local decompression method (it need not decode the entire data), which allows for random access decompression and low-overhead task division that can be exploited over heterogeneous architectures. Furthermore, analytical operations such as correlation and query processing run quickly and accurately over data in the compressed space. Copyright © 2012 John Wiley & Sons, Ltd.
In high-pressure isentropic compression experiments (ICE), the pressure is dominated by the cold curve. In order to obtain an accurate semi-empirical cold curve for Ta, we calculate the thermal pressure from ab initio phonon and electronic excitation spectra. The cold curve is then inferred from ultrasonic and shock data. Our empirical cold pressure is compared to density functional calculations and found to be closer to GGA results at low pressure and to approach LDA at high pressure.
As the memory and storage hierarchy get deeper and more complex, it is important to have new benchmarks and evaluation tools that allow us to explore the emerging middleware solutions to use this hierarchy. Skel is a tool aimed at automating and refining this process of studying HPC I/O performance. It works by generating application I/O kernel/benchmarks as determined by a domain-specific model. This paper provides some techniques for extending Skel to address new situations and to answer new research questions. For example, we document use cases as diverse as using Skel to troubleshoot I/O performance issues for remote users, refining an I/O system model, and facilitating the development and testing of a mechanism for runtime monitoring and performance analytics. We also discuss data oriented extensions to Skel to support the study of compression techniques for Exascale scientific data management.
Modern computing systems are capable of exascale calculations, which are revolutionizing the development and application of high-fidelity numerical models in computational science and engineering. While these systems continue to grow in processing power, the available system memory has not increased commensurately, and electrical power consumption continues to grow. A predominant approach to limit the memory usage in large-scale applications is to exploit the abundant processing power and continually recompute many low-level simulation quantities, rather than storing them. However, this approach can adversely impact the throughput of the simulation and diminish the benefits of modern computing architectures. We present three novel contributions to reduce the memory burden while maintaining, and sometimes improving, performance in simulations based on finite element discretizations. The first contribution develops dictionary-based data compression schemes that detect and exploit the structure of the discretization, due to redundancies across the finite element mesh. While these schemes are shown to reduce memory requirements by more than 99% on meshes with large numbers of identical mesh cells, there are applications where this structure does not exist. The second contribution leverages a recently developed augmented Lagrangian optimization algorithm to enable r-adaptivity for meshes with the goal of enhancing the redundancies in the mesh. The third contribution extends these methods to patch-based linear solvers and preconditioners by compressing local matrices. Numerical results demonstrate the effectiveness of the proposed methods to detect, enhance and exploit mesh structure on a suite of examples inspired by large-scale applications.
Large-scale data analytics workflows ingest massive input data into various data structures, including graphs and key-value datastores. These data structures undergo multiple transformations and computations and are typically reused in incremental and iterative analytics workflows. Persisting in-memory views of these data structures enables reusing them beyond the scope of a single program run while avoiding repetitive raw data ingestion overheads. Memory-mapped I/O enables persisting in-memory data structures without data serialization and deserialization overheads. However, memory-mapped I/O lacks the key feature of persisting consistent snapshots of these data structures for incremental ingestion and processing. The obstacles to efficient virtual memory snapshots using memory-mapped I/O include background writebacks outside the application’s control, and the significantly high storage footprint of such snapshots. To address these limitations, we present Privateer, a memory and storage management tool that enables storage-efficient virtual memory snapshotting while also optimizing snapshot I/O performance. Here, we integrated Privateer into Metall, a state-of-the-art persistent memory allocator for C++, and the Lightning Memory-Mapped Database (LMDB), a widely-used key-value datastore in data analytics and machine learning. Privateer optimized application performance by 1.22× when storing data structure snapshots to node-local storage, and up to 16.7× when storing snapshots to a parallel file system. Privateer also optimizes storage efficiency of incremental data structure snapshots by up to 11× using data deduplication and compression.