Search NASA⌕ Search

SEARCH · Search NASA

Results for “computational statistics”

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 73 records · Page 4
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

On the convergence of statistics in simulations of stationary incompressible turbulent flows

When reporting statistics from simulations of statistically stationary chaotic phenomenon, it is important to verify that the simulations are time-converged. This condition is connected with the statistical error or number of digits with which statistics can be reliably reported. In this work we consider numerical experiments of low Reynolds number incompressible homogeneous and isotropic turbulence as a model problem to investigate statistical convergence over finite simulation times. Specifically, we investigate the time integration requirements that allow meaningful reporting of the statistical error associated with finiteness of the temporal domain. We address two key questions: (1) How long should a simulation be performed in terms of large eddy time, and (2) How should the simulation time be divided among temporal windows over which a quantity of interest is estimated so that its statistical error could be reliably reported? We find that reliable reporting of statistical errors requires simulations on the order of 10 4 large eddy times, which is orders of magnitude longer than typically performed. Additionally, data post-processing should employ windows of at least ten times the large eddy time scale, with the most robust computation of statistical error of the mean requiring window sizes of an additional factor of ten. For practical simulations, we demonstrate that it is possible to estimate the statistical error within a factor of two under a less stringent condition in which a minimum of four windows with size at least ten large eddy times are used. In conclusion, our observations for homogeneous isotropic turbulence are also shown to hold in turbulent channel flow.

42 ENGINEERING↗
DOE OSTI2020

TuckerMPI: A Parallel C++/MPI Software Package for Large-scale Data Compression via the Tucker Tensor Decomposition

With this study, our goal is compression of massive-scale grid-structured data, such as the multi-terabyte output of a high-fidelity computational simulation. For such data sets, we have developed a new software package called TuckerMPI, a parallel C++/MPI software package for compressing distributed data. The approach is based on treating the data as a tensor, i.e., a multidimensional array, and computing its truncated Tucker decomposition, a higher-order analogue to the truncated singular value decomposition of a matrix. The result is a low-rank approximation of the original tensor-structured data. Compression efficiency is achieved by detecting latent global structure within the data, which we contrast to most compression methods that are focused on local structure. In this work, we describe TuckerMPI, our implementation of the truncated Tucker decomposition, including details of the data distribution and in-memory layouts, the parallel and serial implementations of the key kernels, and analysis of the storage, communication, and computational costs. We test the software on 4.5 and 6.7 terabyte data sets distributed across 100 s of nodes (1,000 s of MPI processes), achieving compression ratios between 100 and 200,000×, which equates to 99--99.999% compression (depending on the desired accuracy) in substantially less time than it would take to even read the same dataset from a parallel file system. Moreover, we show that our method also allows for reconstruction of partial or down-sampled data on a single node, without a parallel computer so long as the reconstructed portion is small enough to fit on a single machine, e.g., in the instance of reconstructing/visualizing a single down-sampled time step or computing summary statistics. The code is available at https://gitlab.com/tensors/TuckerMPI.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Revisiting ENSO and IOD Contributions to Australian Precipitation

Tropical modes of variability, such as El Niño–Southern Oscillation (ENSO) and the Indian Ocean Dipole (IOD), exert a strong influence on the interannual variability of Australian precipitation. Nevertheless, commonly used indices of ENSO and IOD variability display significant co-variability that prevents a robust quantification of the independent contribution of each mode to precipitation anomalies. This co-variability issue is often addressed by statistically removing ENSO or IOD variability from the precipitation field before calculating teleconnection patterns. However, by performing a suite of coupled and uncoupled modeling experiments in which either ENSO or IOD variability is physically removed, we show that ENSO-only-driven precipitation patterns computed by statistically removing the IOD influence significantly underestimate the impact of ENSO on Australian precipitation variability. Here, inspired by this, we propose a conceptual model that allows one to effectively separate the contribution of each mode to Australian precipitation variability.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

A road map to cosmological parameter analysis with third-order shear statistics: III. Efficient estimation of third-order shear correlation functions and an application to the KiDS-1000 data

Context. Third-order lensing statistics contain a wealth of cosmological information that is not captured by second-order statistics. However, the computational effort it takes to estimate such statistics in forthcoming stage IV surveys is prohibitively expensive. Aims. We derive and validate an efficient estimation procedure for the three-point correlation function (3PCF) of polar fields such as weak lensing shear. We then use our approach to measure the shear 3PCF and the third-order aperture mass statistics on the KiDS-1000 survey. Methods We constructed an efficient estimator for third-order shear statistics that builds on the multipole decomposition of the 3PCF. We then validated our estimator on mock ellipticity catalogs obtained from N -body simulations. Finally, we applied our estimator to the KiDS-1000 data and presented a measurement of the third-order aperture statistics in a tomographic setup. Results. Our estimator provides a speedup of a factor of ∼100–1000 compared to the state-of-the-art estimation procedures. It is also able to provide accurate measurements for squeezed and folded triangle configurations without additional computational effort. We report a significant detection of tomographic third-order aperture mass statistics in the KiDS-1000 data (S/N = 6.69). Conclusions. Our estimator will make it computationally feasible to measure third-order shear statistics in forthcoming stage IV surveys. Furthermore, it can be used to construct empirical covariance matrices for such statistics.

Astronomy & Astrophysics↗
DOE OSTI2024

Divide and conquer: using RhizoVision Explorer to aggregate data from multiple root scans using image concatenation and statistical methods

Roots are important in agricultural and natural systems for determining plant productivity and soil carbon inputs. Sometimes, the amount of roots in a sample is too much to fit into a single scanned image, so the sample is divided among several scans, and there is no standard method to aggregate the data. Here, we describe and validate two methods for standardizing measurements across multiple scans: image concatenation and statistical aggregation. We developed a Python script that identifies which images belong to the same sample and returns a single, larger concatenated image. These concatenated images and the original images were processed with RhizoVision Explorer, a free and open-source software. An R script was developed, which identifies rows of data belonging to the same sample and applies correct statistical methods to return a single data row for each sample. These two methods were compared using example images from switchgrass, poplar, and various tree and ericaceous shrub species from a northern peatland and the Arctic. Most root measurements were nearly identical between the two methods except median diameter, which cannot be accurately computed by statistical aggregation. We believe the availability of these methods will be useful to the root biology community.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2020

Research Needs for Trusted Analytics in National Security Settings

As artificial intelligence, machine learning, and statistical modeling methods become commonplace in national security applications, the drive to create trusted analytics becomes increasingly important. The goal of this report is to identify areas of research that can provide the foundational understanding and technical prerequisites for the development and deployment of trusted analytics in national security settings. Our review of the literature covered several disjoint research communities, including computer science, statistics, human factors, and several branches of psychology and cognitive science, which tend not to interact with one another or cite each other's literatures. As a result, there exists no agreed-upon theoretical framework for understanding how various factors influence trust and no well-established empirical paradigm for studying these effects. This report therefore takes three steps. First, we define several key terms in an effort to provide a unifying language for trusted analytics and to manage the scope of the problem. Second, we outline an empirical perspective that identifies key independent, moderating, and dependent variables in assessing trusted analytics. Though not a substitute for a theoretical framework, the empirical perspective does support research and development of trusted analytics in the national security domain. Finally, we discuss several research gaps relevant to developing trusted analytics for the national security mission space.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2025

Basic Research Needs for Inverse Methods for Complex Systems under Uncertainty

Inverse problems, which aim to infer unknown properties of a system using experimental and observational data, are central to addressing many of the U.S. Department of Energy’s (DOE) most critical scientific and engineering challenges. Accurate, computationally efficient, and data-efficient solutions to inverse problems are essential for advancing DOE mission-critical science drivers, including analyzing data from large-scale experimental facilities, optimizing fusion reactor performance, accelerating materials discovery, enhancing geophysical imaging, improving wildfire predictions, and enabling autonomous systems and digital twins. However, these problems are becoming increasingly complex, often involving nonlinear, highdimensional, and interconnected systems and models that span multiple physics and scales, while relying on data with varying quantity, quality, and information content. Compounding these challenges is the uncertainty inherent in DOE-relevant systems, where errors in inputs, noise in data, incompleteness of data, and discrepancies between models and reality constrain the accuracy and precision of solutions. At the same time, the convergence of recent scientific computing trends—scientific machine learning, artificial intelligence, and computing advances such as exascale computing—is creating unprecedented opportunities for tackling these challenges. The cross-cutting nature of inverse problems, combined with their growing complexity and rapidly evolving data and algorithmic demands, strongly motivates the formulation of a prioritized research agenda to maximize their capabilities and impact. In response to this need, DOE’s Advanced Scientific Computing Research (ASCR) program in the Office of Science convened the Workshop on Basic Research Needs for Inverse Problems for Complex Systems Under Uncertainty in June 2025. This workshop brought together experts across disciplines to identify grand challenges and major opportunities in the field. Through collaborative discussions, the workshop defined transformative research directions aimed at addressing the mathematical, statistical, and computational challenges posed by inverse problems under uncertainty. As a result of these efforts, four priority research directions (PRDs) were identified to guide future research and development in this area. These PRDs, summarized below, represent a roadmap for advancing the foundational science and mathematics of inverse problems, enabling robust, scalable, and uncertainty-aware solutions that are critical for DOE applications.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Mathematics: The Tao of Data Science

The two pieces, "Ten Research Challenge Areas in Data Science" by Jeannette M. Wing and “Challenges and Opportunities in Statistics and Data Science: Ten Research Areas” by Xuming He and Xihong Lin, provide an impressively complete list of data science challenges from luminaries in the field of data science. They have done an extraordinary job, so this response offers a complementary viewpoint from a mathematical perspective and evangelizes advanced mathematics as a key tool for meeting the challenges they have laid out. Notably, we pick up the themes of scientific understanding of machine learning and deep learning, computational considerations such as cloud computing and scalability, balancing computational and statistical considerations, and inference with limited data. We propose that mathematics is an important key to establishing rigor in the field of data science and as such has an essential role to play in its future.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Ultrafast radiographic imaging and tracking: An overview of instruments, methods, data, and applications

Ultrafast radiographic imaging and tracking (U-RadIT) use state-of-the-art ionizing particle and light sources to experimentally study sub-nanosecond transients or dynamic processes in physics, chemistry, biology, geology, materials science and other fields. These processes are fundamental to modern technologies and applications, such as nuclear fusion energy, advanced manufacturing, communication, and green transportation, which often involve one mole or more atoms and elementary particles, and thus are challenging to compute by using the first principles of quantum physics or other forward models. One of the central problems in U-RadIT is to optimize information yield through, e.g. high-luminosity X-ray and particle sources, efficient imaging and tracking detectors, novel methods to collect data, and large-bandwidth online and offline data processing, regulated by the underlying physics, statistics, and computing power. We review and highlight recent progress in: (a.) Detectors such as high-speed complementary metal-oxide semiconductor (CMOS) cameras, hybrid pixelated array detectors integrated with Timepix4 and other application-specific integrated circuits (ASICs), and digital photon detectors; (b.) U-RadIT modalities such as dynamic phase contrast imaging, dynamic diffractive imaging, and four-dimensional (4D) particle tracking; (c.) U-RadIT data and algorithms such as neural networks and machine learning, and (d.) Applications in ultrafast dynamic material science using XFELs, synchrotrons and laser-driven sources. Hardware-centric approaches to U-RadIT optimization are constrained by detector material properties, low signal-to-noise ratio, high cost and long development cycles of critical hardware components such as ASICs. Interpretation of experimental data, including comparisons with forward models, is frequently hindered by sparse measurements, model and measurement uncertainties, and noise. Alternatively, U-RadIT make increasing use of data science and machine learning algorithms, including experimental implementations of compressed sensing. Machine learning and artificial intelligence approaches, refined by physics and materials information, may also contribute significantly to data interpretation, uncertainty quantification and U-RadIT optimization.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2021

Statistical emulation of a perturbed basal melt ensemble of an ice sheet model to better quantify Antarctic sea level rise uncertainties

Abstract. Antarctic ice shelves are vulnerable to warming ocean temperatures, and some have already begun thinning in response to increased basal melt rates. Sea level is therefore expected to rise due to Antarctic contributions, but uncertainties in its amount and timing remain largely unquantified. In particular, there is substantial uncertainty in future basal melt rates arising from multi-model differences in thermal forcing and how melt rates depend on that thermal forcing. To facilitate uncertainty quantification in sea level rise projections, we build, validate, and demonstrate projections from a computationally efficient statistical emulator of a high-resolution (4 km) Antarctic ice sheet model, the Community Ice Sheet Model version 2.1. The emulator is trained to a large (500-member) ensemble of 200-year-long 4 km resolution transient ice sheet simulations, whereby regional basal melt rates are perturbed by idealized (yet physically informed) trajectories. The main advantage of our emulation approach is that by sampling a wide range of possible basal melt trajectories, the emulator can be used to (1) produce probabilistic sea level rise projections over much larger Monte Carlo ensembles than are possible by direct numerical simulation alone, thereby providing better statistical characterization of uncertainties, and (2) predict the simulated ice sheet response under differing assumptions about basal melt characteristics as new oceanographic studies are published, without having to run additional numerical ice sheet simulations. As a proof of concept, we propagate uncertainties about future basal melt rate trajectories, derived from regional ocean models, to generate probabilistic sea level rise estimates for 100 and 200 years into the future.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

FORESTR: Finding, Organizing, Representing, Explaining, Summarizing, and Thinning Random forests

Random forests have become popular models used for data driven predictions. As a result, random forests are currently used or being considered for high-consequence mission applications in national security, such as the prediction of yield from optical signals and malware detection. While random forests may provide accurate predictions, the complexity of the algorithm causes a lack of interpretability. Random forests are an ensemble of regression or decision trees. Individual regression and decision trees are interpretable, but ensembles are inherently difficult to interpret due to the compilation of many models. We aim to increase the interpretability of random forests by finding patterns in the ensemble of trees that can be used to “thin” (or remove) trees. As a starting point, in this report, we develop a new distance metric for quantifying the similarity between trees based on their topologies (i.e., shapes). We base the metric on a novel distance metric for graphs that is a proper mathematical distance, is invariant to transformations, has registration between graphs, and computes topological evolutions between graphs. We use the tree distance metric to compute tree statistics such as a “mean tree” and to identify clusters of trees. We apply the developed methodology to a toy dataset and a mission relevant product inspection dataset to demonstrate how the metric can provide insight into random forests. Furthermore, we discuss the limitations of the approach and ideas for future research into how the metric could be used as a thinning tool to develop less complex models.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Uncertainties in Density and Simulation Parameters for Radiographic Reconstructions Using Machine Learning

We will develop new and improved methods that will enable a fuller and more extensive use of experimental data, such as from radiographic imaging, towards better characterizing and reducing uncertainties in predictive modeling of weapons performance. We will achieve this by leveraging recent developments in the areas of computational imaging, statistical and machine learning, and reduced order modeling. Key new aspects of our approach include (a) a coupling of deep learning based density reconstruction with fast hydrodynamics simulators that will permit the development of certain guarantees on extrapolatability and bounds on certain other hydrodynamics-related para- metric uncertainties and (b) a dual approach towards better characterizing uncertainties in density reconstruction using deep learning based surrogates to accelerate Bayesian reconstructions on the one hand and using variational inference and Bayesian machine learning on the other hand.

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗
DOE OSTI2020

Hydrodynamic and Radiographic Toolbox (HART)

With a multi-lab and university team, we propose to develop new methods for a Hydrodynamic and Radiographic Toolbox (HART) that will enable a fuller and more extensive use of experimental radiographic data towards better characterizing and reducing uncertainties in predictive modeling of weapons performance. We will achieve this by leveraging recent developments in areas of computational imaging, statistical and machine learning, and reduced order modeling of hydrodynamics. In terms of software practices, by partnering with XCP (RISTRA Project) we will conform to recent XCP standards that are inline with modern software practices and standards and ensure compatibility and ease of inter-operability with existing codes. Three new activities under this proposal include (a) the development and use of deep learning-based surrogates to accelerate reconstruction and variational inference of density fields from radiographs of hydrotests, (b) a model-data fusion strategy that couples deep learning-based density reconstructions with fast hydrodynamics simulators to better constrain the reconstruction, and (c) a method for treating asymmetries using techniques adopted from limited view tomography. All three new activities will be based on improved treatment of scatter, noise, beam spot movement, detector blur, and flat fielding in the forward model, and a use of sophisticated priors to aid in the re construction. The improvements to the forward model and improved algorithmic design of the reconstruction when complete will be contained in the iterative reconstruction code SHIVA—a code project that we have recently initiated. The many ways in which machine learning can be used in the reconstruction work will be contained in a code HERMES that has been initiated with DTRA support. For example, the significant levels of acceleration that will likely be achieved by the use of machine learning techniques will permit us (and are required) to quantify uncertainties in density retrievals. Next, the two-way coupling between density reconstruction and model-based simulation of the hydrodynamics will be contained in code EREBUS, and will permit a fuller realization of the potential of the data to constrain the hydrodynamic model and better address issues related to asymmetries in the problem. Finally, we anticipate that the better consistency with physics achieved in our reconstructions will allow them to be used by X-Division more so than today.

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗
DOE OSTI2024

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↗
DOE OSTI2022

Effects of random forest modeling decisions on biogeochemical time series predictions

Abstract Random forests (RF) are an increasingly popular machine learning approach used to model biogeochemical processes in the Earth system. While RF models are robust to many assumptions that complicate deterministic models, there are several important parameterization decisions for appropriate use and optimal model fit. We explored the role that parameter decisions, including training/testing data splitting strategies, variable selection, and hyperparameters play on RF goodness‐of‐fit by constructing models using 1296 unique parameter combinations to predict concentrations of nitrate, a key nutrient for biogeochemical cycling in aquatic ecosystems. Models were built on long‐term, publicly available water quality and meteorology time series collected by the National Estuarine Research Reserve monitoring network for two contrasting ecosystems representing freshwater and brackish estuaries. We found that accounting for temporal dependence when splitting data into training and testing subsets was key for avoiding over‐estimation of model predictive power. In addition, variable selection, the ratio of training to testing data, and to a lesser degree, variables per split and number of trees, were significant parameters for optimizing RF goodness‐of‐fit. We also explored how model parameter decisions influenced interpretation of the relative importance of predictors to the model, and model predictor‐dependent variable relationships, with results suggesting that both data structure and model parameterization influence these factors. Because much of the current RF literature is written for the computational and statistical science communities, the primary goal of this study is to provide guidelines for aquatic scientists new to machine learning to apply RF techniques appropriately to aquatic biogeochemical datasets.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2022

Experimental measurements of fluid flow in an 84-pin hexagonal rod bundle with spacer grid for a gas-cooled fast modular reactor

A 50 MW e helium-cooled fast modular reactor (FMR) is under development, and as part of the Department of Energy Integrated Research Project (IRP), Texas A&M University is conducting the thermal-hydraulic characterization of its baseline core configuration. Here, we experimentally investigated the axial and cross flow fields characteristics in the hexagonal fuel rod bundle composed of 84 rods, a central rod, and spacer grids at a Reynolds number of 12,000. A fully transparent experimental facility resembling of one unit of the fuel assembly was constructed. Time-resolved particle image velocimetry (TR-PIV) measurements were performed to characterize hydraulic behavior downstream the spacer grid. Velocity measurements were conducted in the axial and radial direction of the rod bundle. From the PIV velocity vector fields, the full-field flow statistics were computed for the mean velocity, vorticity, and Reynolds stresses. An analysis of the energy decay downstream of the spacer grid was conducted with calculations of secondary-flow intensities, turbulent kinetic energy, power spectrum, and spatial-temporal velocity cross correlations. The vorticity field was obtained and pairs of counter-rotating vortexes were identified using a 3D reconstruction of several measurement planes. The data from this experimental campaign will be used to validate numerical models based on Computational Fluid Dynamics and other codes.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2020

Assessing Historical Variability of South Asian Monsoon Lows and Depressions With an Optimized Tracking Algorithm

Abstract Cyclonic low‐pressure systems (LPS) produce abundant rainfall in South Asia, where they are traditionally categorized as monsoon lows, monsoon depressions, and more intense cyclonic storms. The India Meteorological Department (IMD) has tracked monsoon depressions for over a century, finding a large decline in their number in recent decades, but their methods have changed over time and do not include monsoon lows. This study presents a fast, objective algorithm for identifying monsoon LPS and uses it to assess interannual variability and trends in reanalyses. Variables and thresholds used in the algorithm are selected to best match a subjectively analyzed LPS data set while minimizing disagreement between four reanalyses in a training period. The stream function of 850 hPa horizontal wind is found to be optimal in this sense; it is less noisy than vorticity and represents the complete nondivergent wind, even when flow is not geostrophic. Using this algorithm, LPS statistics are computed for five reanalyses, and none show a detectable trend in monsoon depression counts since 1979. Both the Japanese 55‐year Reanalysis (JRA‐55) and the IMD data set show a step‐like reduction in depression counts when they began using geostationary satellite data, in 1979 and 1982, respectively; the 1958–2018 linear trend in JRA‐55, however, is smaller than in the IMD data set, and its error bar includes 0. There are more LPS in seasons with above‐average monsoon rainfall and in La Niña years, but few other large‐scale modes of interannual variability are found to modulate LPS counts, lifetimes, or track length consistently across reanalyses.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2023

Exact Gaussian processes for massive datasets via non-stationary sparsity-discovering kernels

Abstract A Gaussian Process (GP) is a prominent mathematical framework for stochastic function approximation in science and engineering applications. Its success is largely attributed to the GP’s analytical tractability, robustness, and natural inclusion of uncertainty quantification. Unfortunately, the use of exact GPs is prohibitively expensive for large datasets due to their unfavorable numerical complexity of $$O(N^3)$$ O ( N 3 ) in computation and $$O(N^2)$$ O ( N 2 ) in storage. All existing methods addressing this issue utilize some form of approximation—usually considering subsets of the full dataset or finding representative pseudo-points that render the covariance matrix well-structured and sparse. These approximate methods can lead to inaccuracies in function approximations and often limit the user’s flexibility in designing expressive kernels. Instead of inducing sparsity via data-point geometry and structure, we propose to take advantage of naturally-occurring sparsity by allowing the kernel to discover—instead of induce—sparse structure. The premise of this paper is that the data sets and physical processes modeled by GPs often exhibit natural or implicit sparsities, but commonly-used kernels do not allow us to exploit such sparsity. The core concept of exact, and at the same time sparse GPs relies on kernel definitions that provide enough flexibility to learn and encode not only non-zero but also zero covariances. This principle of ultra-flexible, compactly-supported, and non-stationary kernels, combined with HPC and constrained optimization, lets us scale exact GPs well beyond 5 million data points.

97 MATHEMATICS AND COMPUTING↗