Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Monte Carlo Sampling”

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 379 records · Page 21

An Investigation of the Performance of the Unified Monte Carlo Method of Neutron Cross Section Data Evaluation

The Unified Monte Carlo method (UMC) has been suggested to avoid certain limitations and approximations inherent to the well-known Generalized Least Squares (GLS) method of nuclear data evaluation. This contribution reports on an investigation of the performance of the UMC method in comparison with the GLS method. This is accomplished by applying both methods to simple examples with few input values that were selected to explore various features of the evaluation process that impact upon the quality of an evaluation. Among the issues explored are: i) convergence of UMC results with the number of Monte Carlo histories and the ranges of sampled values; ii) a comparison of Monte Carlo sampling using the Metropolis scheme and a brute force approach; iii) the effects of large data discrepancies; iv) the effects of large data uncertainties; v) the effects of strong or weak model or experimental data correlations; and vi) the impact of ratio data and integral data. Comparisons are also made of the evaluated results for these examples when the input values are first transformed to comparable logarithmic values prior to performing the evaluation. Some general conclusions that are applicable to more realistic evaluation exercises are offered.

99 GENERAL AND MISCELLANEOUS↗

Multifidelity Monte Carlo Estimation of Variance and Sensitivity Indices

Variance-based sensitivity analysis provides a quantitative measure of how uncertainty in a model input contributes to uncertainty in the model output. Such sensitivity analyses arise in a wide variety of applications and are typically computed using Monte Carlo estimation, but the many samples required for Monte Carlo to be sufficiently accurate can make these analyses intractable when the model is expensive. This paper presents a multifidelity approach for estimating sensitivity indices that leverages cheaper low-fidelity models to reduce the cost of sensitivity analysis while retaining accuracy guarantees via recourse to the original, expensive model. This paper develops new multifidelity estimators for variance and for the Sobol' main and total effect sensitivity indices. We discuss strategies for dividing limited computational resources among models and specify a recommended strategy. Results are presented for the Ishigami function and a convection-diffusion-reaction model that demonstrate up to $10\times$ speedups for fixed convergence levels. Finally, for the problems tested, the multifidelity approach allows inputs to be definitively ranked in importance when Monte Carlo alone fails to do so.

97 MATHEMATICS AND COMPUTING↗

Emulating the impact of additional proton-proton interactions in the ATLAS simulation by pre-sampling sets of inelastic Monte Carlo events

The accurate simulation of additional interactions at the ATLAS experiment for the analysis of proton-proton collisions delivered by the Large Hadron Collider presents a significant challenge to the computing resources. During the LHC Run (2015-2018) there were up to 70 inelastic interactions per bunch crossing, which need to be accounted for in Monte Carlo (MC) production. In this document, a new method to account for these additional interactions in the simulation chain is described. Instead of sampling the inelastic interactions and adding their energy deposits to a hard-scatter interaction one-by-one, the inelastic interactions are presampled, independent of the hard scatter, and stored as combined events. Consequently, for each hard-scatter interaction only one such presampled event needs to be added as part of the simulation chain. For the Run 2 simulation chain, with an average of 35 interactions per bunch crossing, this new method provides a substantial reduction in MC production CPU needs of around 20%, while reproducing the properties of the reconstructed quantities relevant for physics analyses with good accuracy.

97 MATHEMATICS AND COMPUTING↗

Algorithms in Diffraction Profile Analysis

This chapter introduces several approaches to crystallographic structure refinement from diffraction profile data including the Rietveld method, global optimization methods, and Bayesian modeling. First, the Rietveld method and related nonlinear least squares approaches to analyzing diffraction profiles are briefly discussed. We then review the most common algorithms within global optimization as alternative methods of structure solution: grid search, simulated annealing, and genetic algorithms. As the Rietveld method is the standard tool for analyzing diffraction profile data, we catalogue its challenges and limitations. A Bayesian approach to diffraction profile analysis is presented as an alternative, along with several sampling algorithms including Gibbs sampling, variants of Metropolis sampling, Hamiltonian Monte Carlo, and approximate Bayesian computing. These sampling algorithms are applied to the fitting of a neutron diffraction profile from a National Institute of Standards and Technology silicon standard reference material, and their results are compared.

Singh, Susheela↗

Influence of Thermophysical Property Variability on Thermal Predictions of Ti-6Al-4V

The additive manufacturing (AM) industry has experienced rapid growth in recent decades as industrial interest in the process has grown. Because of this, there are many groups interested in simulations of the AM process. However, the influence of thermophysical property variability on the melt pool geometry during processing of various AM alloys is unclear. The goal of my work at NASA Langley Research Center (LaRC) was to characterize this influence on Ti-6Al-4V during AM processing alongside developing a tool to enable equivalent studies on other AM materials. To facilitate this process, a database tool was developed to contain and organize thermophysical data previously reported by primary sources. The specific thermophysical properties of interest for this study were density, specific heat capacity, and conductivity. Thermal diffusivity was calculated using the other three properties. This data was then fit with a Gaussian distribution and then randomly sampled from using Monte Carlo random value sampling. Using the Rosenthal equation, this sample data was used to simulate the temperature field during additive manufacturing and extract the melt pool geometry. Specifically, the melt pool’s length, width, and depth. This process was repeated an arbitrary number of times, with the default being 1000. Once this information was obtained, histograms were made showing the distributions of the sizes of the melt pool’s length, width, and depth. Representative statistical metrics of the distributions were calculated (mean, standard deviation, and coefficient of variance). This work found that at the simulated processing values, the melt pool’s width and depth had a 5.7% variation while the length had a 3.2% variation. Additionally, a tool was constructed to graph a thermal color map representation of the melt pool for specific, arbitrary values of density, specific heat, and conductivity. This tool allowed for more efficient plotting of individual queries of the thermophysical properties. The results that were observed in the research were that values reported in literature vary and this variance can have a significant impact on the simulation. Predictions of the melt pool’s dimensions show that length has a standard deviation of approximately 6 μm, width has a standard deviation of approximately 5 μm, and depth can vary by approximately 3 μm. Considering the mean sizes of length, width, and depth are 185 μm, 88 μm, and 44 μm respectively, such a deviation is significant. This shows that depth, for example, could be more than 10% larger or smaller than expected. This demonstrates that variability in the reported thermophysical properties are not negligible and should be expected to have an influence on the results of the laser powder bed fusion additive manufacturing process. When simulating this process in the future, measures should be taken to account for this discrepancy and the uncertainty involved

Justin Martin↗

Replica-exchange Wang Landau sampling: pushing the limits of Monte Carlo simulations in materials sciences

We describe the study of thermodynamics of materials using replica-exchange Wang Landau (REWL) sampling, a generic framework for massively parallel implementations of the Wang Landau Monte Carlo method. To evaluate the performance and scalability of the method, we investigate the magnetic phase transition in body-centered cubic (bcc) iron using the classical Heisenberg model parameterized with first principles calculations. We demonstrate that our framework leads to a significant speedup without compromising the accuracy and precision and facilitates the study of much larger systems than is possible with its serial counterpart.

Perera, Meewanage Dilina N↗

Characterization of the Crab Pulsar's Timing Noise

We present a power spectral analysis of the Crab pulsar's timing noise, mainly using radio measurements from Jodrell Bank taken over the period 1982-1989, an interval bounded by sparse data sampling and a large glitch. The power spectral analysis is complicated by nonuniform data sampling and the presence of a steep red power spectrum that can distort power spectra measurement by causing severe power 'leakage'. We develop a simple windowing method for computing red noise power spectra of uniformly sampled data sets and test it on Monte Carlo generated sample realizations of red power-law noise. We generalize time-domain methods of generating power-law red noise with even integer spectral indices to the case of noninteger spectral indices. The Jodrell Bank pulse phase residuals are dense and smooth enough that an interpolation onto a uniform time series is possible. A windowed power spectrum is computed revealing a periodic or nearly periodic component with a period of 568 +/- 10 days and a l/f(exp 3) power-law noise component in pulse phase with a noise strength S(sub infinity)=(1.24 +/- 0.067) x 10(exp 16) cycles(exp 2)/sec(exp 2) over the analysis frequency range f=0.003- 0.1 cycles/day. This result deviates from past analyses which characterized the pulse phase timing residuals as either l/f(sub 4) power-law noise or a quasiperiodic process. The analysis was checked using the Deeter polynomial method of power spectrum estimation that was developed for the case of nonuniform sampling, but has lower spectral resolution. The timing noise is consistent with a torque noise spectrum rising with analysis frequency as f implying blue torque noise, a result not predicted by current models of pulsar timing noise. If the periodic or nearly periodic component is due to a binary companion, we find a mass function f(M) = (6.8 +/- 2.4) x 10(exp -16) solar mass and a companion mass, M(sub c) is greater than or equal to 3.2 solar mass assuming a Crab pulsar mass of 1.4 solar mass.

Scott, D. M.↗

Bayesian Inference in Probabilistic Risk Assessment -- The Current State of the Art

Markov chain Monte Carlo approaches to sampling directly from the joint posterior distribution of aleatory model parameters have led to tremendous advances in Bayesian inference capability in a wide variety of fields, including probabilistic risk analysis. The advent of freely available software coupled with inexpensive computing power has catalyzed this advance. This paper examines where the risk assessment community is with respect to implementing modern computational-based Bayesian approaches to inference. Through a series of examples in different topical areas, it introduces salient concepts and illustrates the practical application of Bayesian inference via Markov chain Monte Carlo sampling to a variety of important problems.

99 GENERAL AND MISCELLANEOUS↗

Forward and inverse uncertainty quantification using multilevel Monte Carlo algorithms for an elliptic non-local equation

Our paper considers uncertainty quantification for an elliptic nonlocal equation. In particular, it is assumed that the parameters which define the kernel in the nonlocal operator are uncertain and a priori distributed according to a probability measure. It is shown that the induced probability measure on some quantities of interest arising from functionals of the solution to the equation with random inputs is well-defined,s as is the posterior distribution on parameters given observations. As the elliptic nonlocal equation cannot be solved approximate posteriors are constructed. The multilevel Monte Carlo (MLMC) and multilevel sequential Monte Carlo (MLSMC) sampling algorithms are used for a priori and a posteriori estimation, respectively, of quantities of interest. Furthermore, these algorithms reduce the amount of work to estimate posterior expectations, for a given level of error, relative to Monte Carlo and i.i.d. sampling from the posterior at a given level of approximation of the solution of the elliptic nonlocal equation.

97 MATHEMATICS AND COMPUTING↗

Multilevel Monte Carlo Estimation of Unbiased Expectation via Sample Reuse and the Low Variance Estimation of Asymptotic Rates

A new variant of the multilevel Monte Carlo estimator [5, 3, 9, 12] is presented for the estimation of expectation statistics that utilizes sample reuse in specified levels, explicitly removes approximation error bias associated with numerically computed output quantities of interest that have an asymptotic limit behavior, and permits a low variance estimate of the asymptotic rate of convergence to that limit. In addition, it is shown that this new multilevel Monte Carlo variant can yield a computational cost savings. A review of Monte Carlo and multilevel Monte Carlo estimators is presented that includes analysis of expected value, expected mean squared error, and the calculation of optimized multilevel sample size parameters. The multilevel Monte Carlo estimator produces estimates of expectation for numerically approximated output quantities of interest that are biased by approximation error. When the quantity of interest can be modeled as the asymptotic limit of numerically approximated output quantities of interest, it is theoretically possible to remove this approximation error bias in the multilevel Monte Carlo estimator. In actual implementations, however, this procedure is unreliable due to statistical variability and inaccuracy in estimating the needed asymptotic limit. Analysis and numerical experiment show that the proposed variant of the multilevel Monte Carlo method greatly reduces (in some cases eliminates) the statistical variability in this limit estimation.

Barth, Timothy↗

A Sensitivity Study of Radiative Fluxes at the Top of Atmosphere to Cloud-Microphysics and Aerosol Parameters in the Community Atmosphere Model CAM5

In this study, we investigated the sensitivity of net radiative fluxes (FNET) at the top of atmosphere (TOA) to 16 selected uncertain parameters mainly related to the cloud microphysics and aerosol schemes in the Community Atmosphere Model version 5 (CAM5). We adopted a quasi-Monte Carlo (QMC) sampling approach to effectively explore the high dimensional parameter space. The output response variables (e.g., FNET) were simulated using CAM5 for each parameter set, and then evaluated using generalized linear model analysis. In response to the perturbations of these 16 parameters, the CAM5-simulated global annual mean FNET ranges from -9.8 to 3.5 W m-2 compared to the CAM5-simulated FNET of 1.9 W m-2 with the default parameter values. Variance-based sensitivity analysis was conducted to show the relative contributions of individual parameter perturbation to the global FNET variance. The results indicate that the changes in the global mean FNET are dominated by those of cloud forcing (CF) within the parameter ranges being investigated. The size threshold parameter related to auto-conversion of cloud ice to snow is confirmed as one of the most influential parameters for FNET in the CAM5 simulation. The strong heterogeneous geographic distribution of FNET variation shows parameters have a clear localized effect over regions where they are acting. However, some parameters also have non-local impacts on FNET variance. Although external factors, such as perturbations of anthropogenic and natural emissions, largely affect FNET variations at the regional scale, their impact is weaker than that of model internal parameters in terms of simulating global mean FNET in this study. The interactions among the 16 selected parameters contribute a relatively small portion of the total FNET variations over most regions of the globe. This study helps us better understand the CAM5 model behavior associated with parameter uncertainties, which will aid the next step of reducing model uncertainty via calibration of uncertain model parameters with the largest sensitivity.

aerosol↗

Hierarchical Gaussian Random Field Sampling for Multilevel Markov Chain Monte Carlo: Coupling Stochastic Partial Differential Equation and the Karhunen–Loève Decomposition

This work introduces structure preserving hierarchical decompositions for sampling Gaussian random fields (GRFs) within the context of multilevel Bayesian inference in high-dimensional space. Existing scalable hierarchical sampling methods, such as those based on stochastic partial differential equations (SPDEs), often reduce the dimensionality of the sample space at the cost of accuracy of inference. Other approaches, such that those based on Karhunen-Loève (KL) expansions, offer sample space dimensionality reduction but sacrifice GRF representation accuracy and ergodicity of the Markov chain Monte Carlo (MCMC) sampler and are computationally expensive for high-dimensional problems. The proposed method integrates the dimensionality reduction capabilities of KL expansions with the scalability of SPDE-based sampling, thereby providing a robust, unified framework for high-dimensional uncertainty quantification (UQ) that is scalable and accurate, preserves ergodicity, and offers dimensionality reduction of the sample space. The hierarchy in our multilevel algorithm is derived from the geometric multigrid hierarchy. By constructing a hierarchical decomposition that maintains the covariance structure across the levels in the hierarchy, the approach enables efficient coarse-to-fine sampling while ensuring that all samples are drawn from the desired distribution. The effectiveness of the proposed method is demonstrated on a benchmark subsurface flow problem, demonstrating its effectiveness in improving computational efficiency and statistical accuracy. Furthermore, our proposed technique is more efficient and accurate and displays better convergence properties than existing methods for high-dimensional Bayesian inference problems.

Gaussian random fields↗

Neural refinement of sample weights

Monte Carlo simulations are an essential tool in particle physics data analysis. Events are typically generated alongside weights that redistribute the cross section of the simulated process across the phase space. These weights can be negative, and several post hoc methods have been developed to eliminate or mitigate the negative values. All of these methods share the common strategy of approximating the average weight as a function of phase space. We introduce an alternative approach, which, instead of reweighting to the average, refines the initial weights with a scaling transformation, utilizing a phase space-dependent factor. Since this new refinement method does not need to model the full weight distribution, it can be more accurate. High-dimensional and unbinned phase space is processed using neural networks for the refinement method. In addition to the refinement method, we introduce a new resampling protocol, which can be used in conjunction with any weight transformation to not only preserve the average weight but also the statistical uncertainties of the initial distribution. Using both realistic and synthetic examples, we show that the new neural refinement method is able to match or exceed the accuracy of similar weight transformations and that the new resampling protocol is simpler in implementation than previous methods while exhibiting equivalent statistical properties.

Artificial neural networks↗

Aerospace Applications of Weibull and Monte Carlo Simulation with Importance Sampling

Recent developments in reliability modeling and computer technology have made it practical to use the Weibull time to failure distribution to model the system reliability of complex fault-tolerant computer-based systems. These system models are becoming increasingly popular in space systems applications as a result of mounting data that support the decreasing Weibull failure distribution and the expectation of increased system reliability. This presentation introduces the new reliability modeling developments and demonstrates their application to a novel space system application. The application is a proposed guidance, navigation, and control (GN&C) system for use in a long duration manned spacecraft for a possible Mars mission. Comparisons to the constant failure rate model are presented and the ramifications of doing so are discussed.

Bavuso, Salvatore J.↗

Monte Carlo Event Generation with Continuous Normalizing Flows

We apply continuous normalizing flows trained with the flow matching method to the problem of phase-space sampling in Monte Carlo event generation for high-energy collider physics. Focusing on lepton-pair and top-quark pair production with multiple jets, the two computationally most expensive processes at the Large Hadron Collider, we train helicity-conditioned continuous normalizing flows to remap the random numbers used in matrix element evaluation. Compared to standard methods, we achieve unweighting efficiency improvements by factors of up to 184 and 25 for the two processes at their respective highest jet number, at the cost of an increased evaluation time. When combining the advantages of continuous normalizing flows with the fast evaluation times of coupling-layer-based flows, using the RegFlow approach, we find parton-level unweighted event generation walltime gains of about a factor of 10 at the highest jet numbers. These substantial gains highlight the promise of samplers based on machine learning for next-generation collider experiments.

Bothmann, Enrico [CERN; Gottingen U.] (ORCID:00000↗

Direct Simulation Monte Carlo Studies of the Gas Sampling for the VATMOS-SR Mission Concept

VATMOS-SR (Venus ATMOSpheric - Sample Return) is a small spacecraft mission concept which would return a gas sample from the upper atmosphere of Venus to Earth for scientific analysis. This could be the first sample return mission for an extra-terrestrial atmosphere, and potentially the first sample return from an Earth-sized planet. The VATMOS-SR mission concept consists of a SmallSat atmospheric sampling probe (45 deg. sphere/cone geometry, <1 m diameter) that is designed to skim through the Venus upper atmosphere and acquire gas samples below the homopause altitude (around ~110 km altitude), where the different atmospheric gases are mixed. The velocity of the spacecraft where sampling would occur is expected to be between ~10.5 km/s and ~13.1 km/s, depending on the trajectory chosen. This presentation will discuss hypervelocity sampling in the upper atmosphere of Venus, with respect to the VATMOS-SR mission concept. VATMOS-SR would enable critical atmospheric measurements to form a full picture of how, why, and when Venus evolved to be so different from Earth and Mars. The abundances and isotopic compositions of volatile elements (such as N, C, S, O and the noble gases) in planetary atmospheres record volatile delivery during accretion, outgassing from planetary interiors, and atmospheric loss to space. Precise and accurate determinations of atmospheric volatile signatures are the key to understanding the origins and geodynamical evolution of Venus compared to the other terrestrial planets. Hypersonic sampling poses unique technical and scientific challenges. To ensure it is possible to relate the composition of the sampled gases to the free stream atmospheric composition, large-scale numerical simulations are employed to model the flow through the VATMOS-SR sampling system. In particular, an emphasis is placed on quantifying noble gas isotopic fractionation that occurs during the sample acquisition and transfer process in order to determine how measured isotopic ratios of noble gases in the sample compare to actual isotopic ratios in the Venusian atmosphere. The Direct Simulation Monte Carlo (DSMC) code SPARTA, an open source software package developed by Sandia National Laboratories, is used in this work. SPARTA, based on Bird’s DSMC method, is a molecular-level gas-kinetic technique. As SPARTA is able to model hypervelocity reacting flows in strong chemical and thermal non-equilibrium, this software package is well suited to determine relevant flow properties for the VATMOS-SR mission concept, and to numerically quantify the expected level of elemental and/or isotopic fractionation in the sample acquired by VATMOS-SR. This presentation will show results from 3D simulations correlating the noble gas isotopic fractionation in the gas acquired at hypervelocity speeds to its ambient atmosphere value. In particular, emphasis will be placed at Xenon isotopes of masses 128 and 130, as precise measurements of that ratio would yield a comparison to Earth’s atmosphere. Additionally, sensitivity studies that quantify the uncertainties due to the freestream parameters as well as the modeling parameters will be performed.

direct simulation Monte Carlo↗

TU-F-17A-05: Calculating Tumor Trajectory and Dose-Of-The-Day for Highly Mobile Tumors Using Cone-Beam CT Projections

Purpose: Cone-beam CT (CBCT) projection images provide anatomical data in real-time over several respiratory cycles, forming a comprehensive picture of tumor movement. We developed a method using these projections to determine the trajectory and dose of highly mobile tumors during each fraction of treatment. Methods: CBCT images of a respiration phantom were acquired, where the trajectory mimicked a lung tumor with high amplitude (2.4 cm) and hysteresis. A template-matching algorithm was used to identify the location of a steel BB in each projection. A Gaussian probability density function for tumor position was calculated which best fit the observed trajectory of the BB in the imager geometry. Two methods to improve the accuracy of tumor track reconstruction were investigated: first, using respiratory phase information to refine the trajectory estimation, and second, using the Monte Carlo method to sample the estimated Gaussian tumor position distribution. 15 clinically-drawn abdominal/lung CTV volumes were used to evaluate the accuracy of the proposed methods by comparing the known and calculated BB trajectories. Results: With all methods, the mean position of the BB was determined with accuracy better than 0.1 mm, and root-mean-square (RMS) trajectory errors were lower than 5% of marker amplitude. Use of respiratory phase information decreased RMS errors by 30%, and decreased the fraction of large errors (>3 mm) by half. Mean dose to the clinical volumes was calculated with an average error of 0.1% and average absolute error of 0.3%. Dosimetric parameters D90/D95 were determined within 0.5% of maximum dose. Monte-Carlo sampling increased RMS trajectory and dosimetric errors slightly, but prevented over-estimation of dose in trajectories with high noise. Conclusions: Tumor trajectory and dose-of-the-day were accurately calculated using CBCT projections. This technique provides a widely-available method to evaluate highly-mobile tumors, and could facilitate better strategies to mitigate or compensate for motion during SBRT.

60 APPLIED LIFE SCIENCES↗