Search NASA⌕ Search

SEARCH · Search NASA

Results for “Gaussian Process Models”

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 487 records · Page 27

Collaborative simulations and experiments for a novel yield model of coal devolatilization in oxy-coal combustion conditions

Oxy-coal combustion is an emerging low-cost “clean coal” technology for emissions reduction and Carbon Capture and Sequestration (CCS). The use of Computational Fluid Dynamics (CFD) tools is crucial for the development of cost-effective oxy-fuel technologies and the minimization of environmental concerns at industrial scale. The coupling of detailed chemistry models and CFD simulations is still challenging, especially for large-scale plants, because of the high computational efforts required. The development of scale-bridging models is therefore necessary, to find a good compromise between computational efforts and the physical-chemical modeling precision. This paper presents a procedure for scale-bridging modeling of coal devolatilization, in the presence of experimental error, that puts emphasis on the thermodynamic aspect of devolatilization, namely the final volatile yield of coal, rather than kinetics. The procedure consists of an engineering approach based on dataset consistency and Bayesian methodology including Gaussian-Process Regression (GPR). Experimental data from devolatilization tests carried out in an oxy-coal entrained flow reactor were considered and CFD simulations of the reactor were performed. Jointly evaluating experiments and simulations, a novel yield model was validated against the data via consistency analysis. In parallel, a Gaussian-Process Regression was performed, to improve the understanding of the uncertainty associated to the devolatilization, based on the experimental measurements. Potential model forms that could predict yield during devolatilization were obtained. The set of model forms obtained via GPR includes the yield model that was proven to be consistent with the data. Finally, the overall procedure has resulted in a novel yield model for coal devolatilization and in a valuable evaluation of uncertainty in the data, in the model form, and in the model parameters.

01 COAL, LIGNITE, AND PEAT↗

Unfolding of Proteins: Thermal and Mechanical Unfolding

We have employed a Hamiltonian model based on a self-consistent Gaussian appoximation to examine the unfolding process of proteins in external - both mechanical and thermal - force elds. The motivation was to investigate the unfolding pathways of proteins by including only the essence of the important interactions of the native-state topology. Furthermore, if such a model can indeed correctly predict the physics of protein unfolding, it can complement more computationally expensive simulations and theoretical work. The self-consistent Gaussian approximation by Micheletti et al. has been incorporated in our model to make the model mathematically tractable by signi cantly reducing the computational cost. All thermodynamic properties and pair contact probabilities are calculated by simply evaluating the values of a series of Incomplete Gamma functions in an iterative manner. We have compared our results to previous molecular dynamics simulation and experimental data for the mechanical unfolding of the giant muscle protein Titin (1TIT). Our model, especially in light of its simplicity and excellent agreement with experiment and simulation, demonstrates the basic physical elements necessary to capture the mechanism of protein unfolding in an external force field.

Hur, Joe S.↗

Lattice thermal conductivity of filled skutterudites: An anharmonicity perspective

We report a phenomenological model to calculate the high-temperature lattice thermal conductivity of filled skutterudite antimonides. The model needs no phonon resonant scattering terms. Instead, we assume that umklapp processes dominate the high-temperature phonon scattering. In order to represent the anharmonicity introduced by the filling atom, we introduce a Gaussian term into the relaxation time of the umklapp process. The developed model agrees remarkably well with the experimental results of RE{sub f}Co{sub 4}Sb{sub 12} and RE{sub f}Fe{sub 4}Sb{sub 12} (RE = Yb, Ba, and Ca) alloys. To further test the validity of our model, we calculate the lattice thermal conductivity of nanostructured or multi-filled skutterudites. The calculation results are also in good agreement with experiment, increasing our confidence in the developed anharmonicity model.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Robust Online Monitoring for Calibration Assessment of Transmitters and Instrumentation

Robust online monitoring (OLM) technologies are expected to enable the extension or elimination of periodic sensor calibration intervals in operating and new reactors. These advances in OLM technologies will improve the safety and reliability of current and planned nuclear power systems through improved accuracy and increased reliability of sensors used to monitor key parameters. In this article, we discuss an overview of research being performed within the Nuclear Energy Enabling Technologies (NEET)/Advanced Sensors and Instrumentation (ASI) program, for the development of OLM algorithms to use sensor outputs and, in combination with other available information, 1) determine whether one or more sensors are out of calibration or failing and 2) replace a failing sensor with reliable, accurate sensor outputs. Algorithm development is focused on the following OLM functions: • Signal validation • Virtual sensing • Sensor response-time assessment These algorithms incorporate, at their base, a Gaussian Process-based uncertainty quantification (UQ) method. Various plant models (using kernel regression, GP, or hierarchical models) may be used to predict sensor responses under various plant conditions. These predicted responses can then be applied in fault detection (sensor output and response time) and in computing the correct value (virtual sensing) of a failing physical sensor. The methods being evaluated in this work can compute confidence levels along with the predicted sensor responses, and as a result, may have the potential for compensating for sensor drift in real-time (online recalibration). Evaluation was conducted using data from multiple sources (laboratory flow loops and plant data). Ongoing research in this project is focused on further evaluation of the algorithms, optimization for accuracy and computational efficiency, and integration into a suite of tools for robust OLM that are applicable to monitoring sensor calibration state in nuclear power plants.

fault detection↗

Guided search for desired functional responses via Bayesian optimization of generative model: Hysteresis loop shape engineering in ferroelectrics

Advances in theoretical modeling across multiple disciplines have yielded generative models capable of high veracity in predicting macroscopic functional responses of materials emerging as a result of complex non-local interactions. Correspondingly, of interest is the inverse problem of finding the model parameter that will yield desired macroscopic responses, such as stress–strain curves, ferroelectric hysteresis loops, etc. Here, we suggest and implement Gaussian process based methods that allow to effectively sample the degenerate parameter space of a complex non-local model to output regions of parameter space which yield desired functionalities. In this work, we discuss the specific adaptation of the acquisition function and sampling function to make the process efficient and balance the efficient exploration of parameter space for multiple possible minima and exploitation to densely sample the regions of interest where target behaviors are optimized. Additionally, this approach is illustrated via the hysteresis loop engineering in ferroelectric materials but can be adapted to other functionalities and generative models.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Using a coupled dispersion model to estimate depletion of a tritium oxide plume by a forest

Tritium processing facilities may release tritium oxide (HTO) to the atmosphere which poses potential health risks to exposed co-located workers and to offsite individuals. Most radiological consequence analyses determine HTO dose by applying Gaussian plume models to simulate the transport of HTO. Within these models, deposition velocity is used to assess the sum of all deposition processes acting on the plume. While this may account for vegetative and soil uptake or respiration processes, it may currently lack inclusion of the complex interactions within heterogeneous forested environments. In this complex morphology, dispersion patterns are significantly altered by changing flow regimes above and below the forest canopy and by the transfer of plume material across the canopy boundary. To determine the effects of a heterogeneous forest canopy on an airborne HTO plume, a Gaussian plume model coupled with an advection-diffusion plume model was applied to estimate transport in the free atmosphere above the forest and within the forest canopy and understory. During 2012, wind speed and wind direction measurements taken at 5 heights, ranging from 2-m to 28-m, on an instrumented meteorological tower located in a loblolly pine forest at the Department of Energy (DOE) Savannah River Site (SRS), near Aiken, SC. From these measurements, model predictions were made over a full spectrum of meteorological conditions. Deposition and resuspension velocities were calculated based on the model-predicted flux of plume material across the top of the forest canopy. Additionally, net deposition velocity of the plume material was calculated as the difference between the deposition and resuspension velocities. The 1st and 5th percentile net deposition velocities were estimated to be 0.7 cm s-1 and 1.2 cm s-1, respectively.

63 RADIATION, THERMAL, AND OTHER ENVIRON. POLLUTAN↗

A Novel Approach for Determining Source–Receptor Relationships in Model Simulations: A Case Study of Black Carbon Transport in Northern Hemisphere Winter

A Gaussian process (GP) emulator is applied to quantify the contribution of local and remote emissions of black carbon (BC) on the BC concentrations in different regions using a Latin Hypercube sampling strategy for emission perturbations in the offline version of the Community Atmosphere Model Version 5.1 (CAM5) simulations. The source-receptor relationships are computed based on simulations constrained by a standard free-running CAM5 simulation and the ERA-Interim reanalysis product. The analysis demonstrates that the emulator is capable of retrieving the source-receptor relationships based on a small number of CAM5 simulations. Most regions are found susceptible to their local emissions. The emulator also finds that the source-receptor relationships retrieved from the model-driven and the reanalysis-driven simulations are very similar, suggesting that the simulated circulation in CAM5 resembles the assimilated meteorology in ERA-Interim. The robustness of the results provides confidence for applying the emulator to detect dose-response signals in the climate system.

black↗

Fast increased fidelity samplers for approximate Bayesian Gaussian process regression

Gaussian processes (GPs) are common components in Bayesian non-parametric models having a rich methodological literature and strong theoretical grounding. The use of exact GPs in Bayesian models is limited to problems containing several thousand observations due to their prohibitive computational demands. We develop a posterior sampling algorithm using H-matrix approximations that scales at O(n log 2 n). We show that this approximation’s Kullback-Leibler divergence to the true posterior can be made arbitrarily small. Though multidimensional GPs could be used with our algorithm, d-dimensional surfaces are modeled as tensor products of univariate GPs to minimize the cost of matrix construction and maximize computational efficiency. We illustrate the performance of this fast increased fidelity approximate GP, FIFA-GP, using both simulated and non-synthetic data sets

97 MATHEMATICS AND COMPUTING↗

A Hybrid Reduced-Order Model of Fine-Resolution Hydrologic Simulations at a Polygonal Tundra Site

High-resolution predictions of land surface hydrological dynamics are desirable for improved investigations of regional- and watershed-scale processes. Direct deterministic simulations of fine-resolution land surface variables present many challenges, including high computational cost. We therefore propose the use of reduced-order modeling techniques to facilitate emulation of fine-resolution simulations. We use an emulator, Gaussian process regression, to approximate fine-resolution four-dimensional soil moisture fields predicted using a three-dimensional surface-subsurface hydrological simulator (PFLOTRAN). A dimension-reduction technique known as “proper orthogonal decomposition” is further used to improve the efficiency of the resulting reduced-order model (ROM). The ROM reduces simulation computational demand to negligible levels compared to the underlying fine-resolution model. In addition, the ROM that we constructed is equipped with an uncertainty estimate, allowing modelers to construct a ROM consistent with uncertainty in the measured data. The ROM is also capable of constructing statistically equivalent analogs that can be used in uncertainty and sensitivity analyses. We apply the technique to four polygonal tundra sites near Barrow, Alaska that are part of the Department of Energy’s Next-Generation Ecosystem Experiments (NGEE)-Arctic project. The ROM is trained for each site using simulated soil moisture from 1998-2000 and validated using the simulated data for 2002 and 2006. The average relative RMSEs of the ROMs are under 1%.

58 GEOSCIENCES↗

Interpolating between small- and large-$\mathscr{g}$ expansions using Bayesian model mixing

Bayesian model mixing (BMM) is a statistical technique that can be used to combine models that are predictive in different input domains into a composite distribution that has improved predictive power over the entire input space. We explore the application of BMM to the mixing of two expansions of a function of a coupling constant $\mathscr{g}$ that are valid at small and large values of $\mathscr{g}$ respectively. This type of problem is quite common in nuclear physics, where physical properties are straightforwardly calculable in strong and weak interaction limits or at low and high densities or momentum transfers, but difficult to calculate in between. Interpolation between these limits is often accomplished by a suitable interpolating function, e.g., Padé approximants, but it is then unclear how to quantify the uncertainty of the interpolant. We address this problem in the simple context of the partition function of zero-dimensional Φ 4 theory, for which the (asymptotic) expansion at small $\mathscr{g}$ and the (convergent) expansion at large $\mathscr{g}$ are both known. We consider three mixing methods: linear mixture BMM, localized bivariate BMM, and localized multivariate BMM with Gaussian processes. Here, we find that employing a Gaussian process in the intermediate region between the two predictive models leads to the best results of the three methods. The methods and validation strategies we present here should be generalizable to other nuclear physics settings.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Estimation With Range Dependent Sensor Model

This paper focuses on the improvement of target tracking accuracy taking into account the sensor’s reading degradation as the relative range increases. The approach is based on the assumption that range measurement error depends on the actual range. Specifically, we model the measurement error as a proportional to the actual range term plus a zero-mean, Gaussian distributed, and uncorrelated process. We present two approaches to tracking a target with constant velocity model. The first approach uses Extended Kalman Filter (EKF) framework to estimate target's states and the unknown proportionality coefficient with linearization on the observation model on the predicted states at each time step. The second one uses a measurement conversion method to estimate the target's scaled by the unknown constant term states, which is shown to be unbiased. This conversion results in linear state and observation models, hence the standard Kalman filter can be applied. The actual target state is computed by application of back scaling with the estimate of unknown scale factor. We evaluate the approaches in desktop simulations.

Target tracking↗

COSMIC EMULATION: FAST PREDICTIONS FOR THE GALAXY POWER SPECTRUM

The halo occupation distribution (HOD) approach has proven to be an effective method for modeling galaxy clustering and bias. In this approach, galaxies of a given type are probabilistically assigned to individual halos in N-body simulations. In this paper, we present a fast emulator for predicting the fully nonlinear galaxy–galaxy auto and galaxy–dark matter cross power spectrum and correlation function over a range of freely specifiable HOD modeling parameters. The emulator is constructed using results from 100 HOD models run on a large ΛCDM N-body simulation, with Gaussian Process interpolation applied to a PCA-based representation of the galaxy power spectrum. The total error is currently ∼1% in the auto correlations and ∼2% in the cross correlations from z = 1 to z = 0, over the considered parameter range. We use the emulator to investigate the accuracy of various analytic prescriptions for the galaxy power spectrum, parametric dependencies in the HOD model, and the behavior of galaxy bias as a function of HOD parameters. Additionally, we obtain fully nonlinear predictions for tangential shear correlations induced by galaxy–galaxy lensing from our galaxy–dark matter cross power spectrum emulator. All emulation products are publicly available at http://www.hep.anl.gov/cosmology/CosmicEmu/emu.html.

79 ASTRONOMY AND ASTROPHYSICS↗

A Gaussian process based surrogate approach for the optimization of cylindrical targets

Simulating direct-drive inertial confinement experiments presents significant computational challenges, both due to the complexity of the codes required for such simulations and the substantial computational expense associated with target design studies. Machine learning models, and in particular, surrogate models, offer a solution by replacing simulation results with a simplified approximation. In this study, we apply surrogate modeling and optimization techniques that are well established in the existing literature to one-dimensional simulation data of a new cylindrical target design containing deuterium–tritium fuel. These models predict yields without the need for expensive simulations. We find that Bayesian optimization with Gaussian process surrogates enhances sampling efficiency in low-dimensional design spaces but becomes less efficient as dimensionality increases. Nonetheless, optimization routines within two-dimensional and five-dimensional design spaces can identify designs that maximize yield, while also aligning with established physical intuition. Optimization routines, which ignore constraints on hydrodynamic instability growth, are shown to lead to unstable designs in 2D, resulting in yield loss. However, routines that utilize 1D simulations and impose constraints on the in-flight aspect ratio converge on novel cylindrical target designs that are stable against hydrodynamic instability growth in 2D and achieve high yield.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Uncertainty quantification for nuclear density functional theory and information content of new measurements

Statistical tools of uncertainty quantification can be used to assess the information content of measured observables with respect to present-day theoretical models, to estimate model errors and thereby improve predictive capability, to extrapolate beyond the regions reached by experiment, and to provide meaningful input to applications and planned measurements. To showcase new opportunities offered by such tools, we make a rigorous analysis of theoretical statistical uncertainties in nuclear density functional theory using Bayesian inference methods. By considering the recent mass measurements from the Canadian Penning Trap at Argonne National Laboratory, we demonstrate how the Bayesian analysis and a direct least-squares optimization, combined with high-performance computing, can be used to assess the information content of the new data with respect to a model based on the Skyrme energy density functional approach. Employing the posterior probability distribution computed with a Gaussian process emulator, we apply the Bayesian framework to propagate theoretical statistical uncertainties in predictions of nuclear masses, two-neutron dripline, and fission barriers. Overall, we find that the new mass measurements do not impose a constraint that is strong enough to lead to significant changes in the model parameters. In addition, the example discussed in this study sets the stage for quantifying and maximizing the impact of new measurements with respect to current modeling and guiding future experimental efforts, thus enhancing the experiment-theory cycle in the scientific method.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

A mathematical study of a random process proposed as an atmospheric turbulence model

A random process is formed by the product of a local Gaussian process and a random amplitude process, and the sum of that product with an independent mean value process. The mathematical properties of the resulting process are developed, including the first and second order properties and the characteristic function of general order. An approximate method for the analysis of the response of linear dynamic systems to the process is developed. The transition properties of the process are also examined.

Sidwell, K.↗

Hybrid Parameter Search and Dynamic Model Selection for Mixed-Variable Bayesian Optimization

Herein this article presents a new type of hybrid model for Bayesian optimization (BO) adept at managing mixed variables, encompassing both quantitative (continuous and integer) and qualitative (categorical) types. Our proposed new hybrid models (named hybridM) merge the Monte Carlo Tree Search structure (MCTS) for categorical variables with Gaussian Processes (GP) for continuous ones. hybridM leverages the upper confidence bound tree search (UCTS) for MCTS strategy, showcasing the tree architecture’s integration into Bayesian optimization. Our innovations, including dynamic online kernel selection in the surrogate modeling phase and a unique UCTS search strategy, position our hybrid models as an advancement in mixed-variable surrogate models. Numerical experiments underscore the superiority of hybrid models, highlighting their potential in Bayesian optimization.

97 MATHEMATICS AND COMPUTING↗

Multitask methods for predicting molecular properties from heterogeneous data

Data generation remains a bottleneck in training surrogate models to predict molecular properties. We demonstrate that multitask Gaussian process regression overcomes this limitation by leveraging both expensive and cheap data sources. In particular, we consider training sets constructed from coupled-cluster (CC) and density functional theory (DFT) data. We report that multitask surrogates can predict at CC-level accuracy with a reduction in data generation cost by over an order of magnitude. Of note, our approach allows the training set to include DFT data generated by a heterogeneous mix of exchange–correlation functionals without imposing any artificial hierarchy on functional accuracy. More generally, the multitask framework can accommodate a wider range of training set structures—including the full disparity between the different levels of fidelity—than existing kernel approaches based on Δ-learning although we show that the accuracy of the two approaches can be similar. Consequently, multitask regression can be a tool for reducing data generation costs even further by opportunistically exploiting existing data sources.

Chemistry↗