Search NASA⌕ Search

SEARCH · Search NASA

Results for “adjoint method”

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 109 records · Page 6

Comparison of hybrid methods for global variance reduction in shielding calculations

For Monte Carlo shielding problems that calculate a mesh tally over the entire problem, the statistical uncertainties computed for each voxel can vary widely. This can lead to unacceptably long run times in order to reduce the uncertainties in all areas of the problem to a reasonably low level. Hybrid methods - using estimates from deterministic calculations to create importance maps for variance reduction in Monte Carlo calculations - have been successfully used to optimize the calculation of specific tallies. For the global problem, several methods have been proposed to create importance maps that distribute Monte Carlo particles in such a way as to achieve a more uniform distribution of relative uncertainty across the problem. The goal is to compute a mesh tally with nearly the same relative uncertainties in the low flux/dose areas as in the high flux/dose areas. Methods based on only forward deterministic estimates and methods using both forward and adjoint deterministic methods have been implemented in the SCALE/MAVRIC package and have been compared against each other by computing global mesh tallies on several representative shielding problems. Methods using both forward and adjoint estimates provide better performance for computing more uniform relative uncertainties across a global mesh tally. (authors)

97 MATHEMATICS AND COMPUTING↗

Adjoint-based deviational Monte Carlo methods for phonon transport calculations

In the field of linear transport, adjoint formulations exploit linearity to derive powerful reciprocity relations between a variety of quantities of interest. We develop an adjoint formulation of the linearized Boltzmann transport equation for phonon transport. We use this formulation for accelerating deviational Monte Carlo simulations of complex, multiscale problems. Benefits include significant computational savings via direct variance reduction, or by enabling formulations which allow more efficient use of computational resources, such as formulations which provide high resolution in a particular phase-space dimension (e.g., spectral). We show that the proposed adjoint-based methods are particularly well suited to problems involving a wide range of length scales (e.g., nanometers to hundreds of microns) and lead to computational methods that can calculate quantities of interest with a cost that is independent of the system characteristic length scale, thus removing the traditional stiffness of kinetic descriptions. Applications to problems of current interest, such as simulation of transient thermoreflectance experiments or spectrally resolved calculation of the effective thermal conductivity of nanostructured materials, are presented and discussed in detail.

42 ENGINEERING↗

Semi-analytical sensitivity analysis for nonlinear transient problems

Efficient analytical sensitivity computations are essential elements of gradient-based optimization schemes; however, they can be difficult to implement. This implementation issue is often resolved by adopting the semi-analytical method which exhibits the efficiency of the analytical methods and the ease of implementation of the finite difference method. However, care must be taken as semi-analytical sensitivities may exhibit errors due to truncation and round-off. Additional errors are introduced if the convergence tolerance of the primal analysis is not sufficiently small. This paper gives a general overview and some new developments of the analytical and semi-analytical sensitivity analyses for nonlinear steady-state, transient, and dynamic problems. We discuss the restrictive assumptions, accuracy, and consistency of these methods. Both adjoint and direct differentiation methods are studied. Numerical examples are provided.

42 ENGINEERING↗

Comparison Of Hybrid Methods For Global Variance Reduction In Shielding Calculations

For Monte Carlo shielding problems that calculate a mesh tally over the entire problem, the relative uncertainties computed for each voxel can vary widely. This can lead to unacceptably long run times in order to reduce the uncertainties in all areas of the problem to a reasonably low level. Hybrid methods using estimates from deterministic calculations to create importance maps for variance reduction in Monte Carlo calculations have been successfully used to optimize the calculation of specific tallies. For the global problem, several methods have been proposed that create importance maps that distribute Monte Carlo particles in such a way as to achieve a more uniform distribution of relative uncertainty across the problem. The goal is to compute a mesh tally with nearly the same relative uncertainties in the low flux/dose areas as in the high flux/dose areas. Methods based on only forward deterministic estimates and methods using both forward and adjoint deterministic methods have been implemented the SCALE/MAVRIC package and have been compared against each other by computing global mesh tallies on several representative shielding problems. Methods using both forward and adjoint estimates provide better performance for computing more uniform relative uncertainties across a global mesh tally.

FW-CADIS↗

Adjoint DSMC for nonlinear Boltzmann equation constrained optimization

Applications for kinetic equations such as optimal design and inverse problems often involve finding unknown parameters through gradient-based optimization algorithms. Based on the adjoint-state method, we derive two different frameworks for approximating the gradient of an objective functional constrained by the nonlinear Boltzmann equation. While the forward problem can be solved by the DSMC method, it is difficult to efficiently solve the high-dimensional continuous adjoint equation obtained by the “optimize-then-discretize” approach. This challenge motivates us to propose an adjoint DSMC method following the “discretize-then-optimize” approach for Boltzmann-constrained optimization. We also analyze the properties of the two frameworks and their connections. Here, several numerical examples are presented to demonstrate their accuracy and efficiency.

97 MATHEMATICS AND COMPUTING↗

Comparison of the Monte Carlo Adjoint-Weighted and Differential Operator Perturbation Methods

Two perturbation theory methodologies are implemented for k-eigenvalue calculations in the continuous-energy Monte Carlo code, MCNP6. A comparison of the accuracy of these techniques, the differential operator and adjoint-weighted methods, is performed numerically and analytically. Typically, the adjoint-weighted method shows better performance over a larger range; however, there are exceptions.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Adjoint Klein-Nishina Sampling Methods: Efficiency, Speed and Applications

Three new rejection sampling methods for generating samples from the adjoint Klein-Nishina cross section are discussed: the two-branch rejection sampling procedure, the three-branch linear rejection sampling procedure and the three-branch inverse rejection sampling procedure. These methods have all been implemented in the Framework for REsearch in Nuclear ScIence and Engineering (FRENSIE). The efficiency and sample generation rate of each of these methods are evaluated to characterize the methods and to make recommendations regarding their use. The use of these methods in realistic transport simulations is also evaluated by incorporating a scattering function into the sampling process. Furthermore, the results of an infinite medium problem are presented to verify that the sampling procedure can be used in an adjoint Monte Carlo simulation to generate results that are in agreement with an equivalent forward simulation.

97 MATHEMATICS AND COMPUTING↗

Multiscale modeling high-order methods and data-driven modeling

Projection-based reduced-order models (ROMs) comprise a promising set of data-driven approaches for accelerating the simulation of high-fidelity numerical simulations. Standard projection-based ROM approaches, however, suffer from several drawbacks when applied to the complex nonlinear dynamical systems commonly encountered in science and engineering. These limitations include a lack of stability, accuracy, and sharp a posteriori error estimators. This work addresses these limitations by leveraging multiscale modeling, least-squares principles, and machine learning to develop novel reduced-order modeling approaches, along with data-driven a posteriori error estimators, for dynamical systems. Theoretical and numerical results demonstrate that the two ROM approaches developed in this work - namely the windowed least-squares method and the Adjoint Petrov - Galerkin method - yield substantial improvements over state-of-the-art approaches. Additionally, numerical results demonstrate the capability of the a posteriori error models developed in this work.

97 MATHEMATICS AND COMPUTING↗

Diagnosing the sensitivity of grounding-line flux to changes in sub-ice-shelf melting

Abstract. Using a numerical ice flow model, we study changes in ice shelf buttressing and grounding-line flux due to localized ice thickness perturbations, a proxy for localized changes in sub-ice-shelf melting. From our experiments, applied to idealized (MISMIP+) and realistic (Larsen C) ice shelf domains, we identify a correlation between a locally derived buttressing number on the ice shelf, based on the first principal stress, and changes in the integrated grounding-line flux. The origin of this correlation, however, remains elusive from the perspective of a theoretical or physically based understanding. This and the fact that the correlation is generally much poorer when applied to realistic ice shelf domains motivate us to seek an alternative approach for predicting changes in grounding-line flux. We therefore propose an adjoint-based method for calculating the sensitivity of the integrated grounding-line flux to local changes in ice shelf geometry. We show that the adjoint-based sensitivity is identical to that deduced from pointwise, diagnostic model perturbation experiments. Based on its much wider applicability and the significant computational savings, we propose that the adjoint-based method is ideally suited for assessing grounding-line flux sensitivity to changes in sub-ice-shelf melting.

54 ENVIRONMENTAL SCIENCES↗

Implementation of Material and Geometry Criticality Search in RMC Code

RMC is a continuous-energy Reactor Monte Carlo neutron and photon transport code being developed by Department of Engineering Physics at Tsinghua University, Beijing. For the use of simulation of reactor control or design of reactor with unknown material and geometry parameters, an accurate and efficient method of eigenvalue search is implemented in RMC code. The eigenvalue search focuses on two types of parameters. One is nuclide atom density, called material search, which includes the fuel density, enrichment or boron concentration. Perturbation theory of Monte Carlo Code, like differential operator method and adjoint-weighted perturbation method has been widely researched in sensitivity and uncertainty study. For material search, the perturbation algorithm needs to coupled with flexible material control block and optimized strategy. The other type of parameter is surface position, called geometry search, which includes fuel rod diameter, pitch and control rod position. MC21 iterated on surface position based on bisection method or regula-falsi method, and directly recalculated k-eff until it converged to target error interval. In this paper, the differential operator method, improved Newton-Raphson method is introduced. Based on these methods, the global material search is implemented. Then based on material search and finite method, global geometry search is implement. Finally, fuel enrichment, boron concentration and control rod position in normal PWR assembly are searched. The results show that RMC can search material and geometry parameters accurately and efficiently. In this paper, global eigenvalue search of material and geometry parameters have been implemented in RMC code. They are based on the differential operator method in Monte Carlo perturbation theory. The differential operator method samples δk/δρ{sub r}, δ{sup 2}k/δρ{sub r}{sup 2} along with k-eff estimator. Therefore it won't take up too much additional computational resource. For homogeneous material search, fission source effect has little effect of accuracy and can be ignored. The results of fuel enrichment and boron concentration searches in assembly show that improved Newton-Raphson method can iterate efficiently until k-eff is in the error interval. Based on material search and finite difference formula, the geometry parameter, like control rod axial position, can iterate until k-eff is in the error interval. However, the tiny perturbed region may not contain enough tracks to simulate. Besides, the large number density's relative change up to -1 may cause underestimation of Δk-eff, and then slope of k-x curve. This will cause non-convergence for Newton- Raphson method. (authors)

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Adjoint Error Estimation for Linear Advection

An a posteriori error formula is described when a statistical measurement of the solution to a hyperbolic conservation law in 1D is estimated by finite volume approximations. This is accomplished using adjoint error estimation. In contrast to previously studied methods, the adjoint problem is divorced from the finite volume method used to approximate the forward solution variables. An exact error formula and computable error estimate are derived based on an abstractly defined approximation of the adjoint solution. This framework allows the error to be computed to an arbitrary accuracy given a sufficiently well resolved approximation of the adjoint solution. The accuracy of the computable error estimate provably satisfies an a priori error bound for sufficiently smooth solutions of the forward and adjoint problems. The theory does not currently account for discontinuities. Computational examples are provided that show support of the theory for smooth solutions. The application to problems with discontinuities is also investigated computationally.

97 MATHEMATICS AND COMPUTING↗

Advancing attenuation estimation through integration of the Hessian in multiparameter viscoacoustic full-waveform inversion

Accurate seismic attenuation models of subsurface structures not only enhance subsequent migration processes by improving fidelity, resolution, and facilitating amplitude-compliant angle gather generation but also provide valuable constraints on subsurface physical properties. Leveraging full-wavefield information, multiparameter viscoacoustic full-waveform inversion ( Q-FWI) simultaneously estimates seismic velocity and attenuation ( Q) models. However, a major challenge in Q-FWI is the contamination of crosstalk artifacts, where inaccuracies in the velocity model are mistakenly mapped to the inverted attenuation model. While incorporating the Hessian is expected to mitigate these artifacts, the explicit implementation is prohibitively expensive due to its formidable computational cost. In this study, we formulate and develop a Q-FWI algorithm via the Newton-conjugate gradient (CG) framework, where the search direction at each iteration is determined through an internal CG loop. In particular, the Hessian is integrated into each CG step in a matrix-free fashion using the second-order adjoint-state method. We find through synthetic experiments that our Newton-CG Q-FWI significantly mitigates crosstalk artifacts compared with the limited-memory Broyden-Fletcher-Goldfarb-Shanno method and the CG method, albeit with a notable computational cost. In the discussion of several key implementation details, we also determine the significance of the approximate Gauss-Newton Hessian, the second-order adjoint-state method, and the two-stage inversion strategy.

Geochemistry & Geophysics↗

A general approach to seismic inversion with automatic differentiation

Imaging Earth structure or seismic sources from seismic data involves minimizing a target misfit function, and is commonly solved through gradient-based optimization. The adjoint-state method has been developed to compute the gradient efficiently; however, its implementation can be time-consuming and difficult. We develop a general seismic inversion framework to calculate gradients using reverse-mode automatic differentiation. The central idea is that adjoint-state methods and reverse-mode automatic differentiation are mathematically equivalent. Here, the mapping between numerical PDE simulation and deep learning allows us to build a seismic inverse modeling library, ADSeismic, based on deep learning frameworks, which supports high performance reverse-mode automatic differentiation on CPUs and GPUs. We demonstrate the performance of ADSeismic on inverse problems related to velocity model estimation, rupture imaging, earthquake location, and source time function retrieval. ADSeismic has the potential to solve a wide variety of inverse modeling applications within a unified framework.

58 GEOSCIENCES↗

Interlocked optimization and fast gradient algorithm for a seismic inverse problem

Highlights: {yields} A 2D extension of the 1D nonlinear inversion of well-seismic data is given. {yields} Appropriate regularization yields a well-determined large scale inverse problem. {yields} An interlocked optimization loop acts as an efficient preconditioner. {yields} The adjoint state method is used to compute the misfit function gradient. {yields} Domain decomposition method yields an efficient parallel implementation. - Abstract: We give a nonlinear inverse method for seismic data recorded in a well from sources at several offsets from the borehole in a 2D acoustic framework. Given the velocity field, approximate values of the impedance are recovered. This is a 2D extension of the 1D inversion of vertical seismic profiles . The inverse problem generates a large scale undetermined ill-conditioned problem. Appropriate regularization terms render the problem well-determined. An interlocked optimization algorithm yields an efficient preconditioning. A gradient algorithm based on the adjoint state method and domain decomposition gives a fast parallel numerical method. For a realistic test case, convergence is attained in an acceptable time with 128 processors.

97 MATHEMATICS AND COMPUTING↗

A Full Multigrid Method for the Self-Adjoint Angular Flux Equation Based on Cellwise Block Jacobi Iteration

Previous work by Densmore et al. has demonstrated that multigrid methods for the self-adjoint angular flux (SAAF) equation with cellwise block Jacobi smoothing can be extremely effective at solving for the angular flux in one dimension. The extension of that work to two dimensions demonstrated that while the smoother works well, the use of a direct coarse-grid approximation can severely limit the rate of convergence for optically thin cells. In that paper, this convergence limit was evaded through the use of a partially Galerkin coarse-grid approximation. However, relying on a Galerkin coarse-grid approximation is not ideal as constructing the coarse-grid discretization in this approach is complicated. The motivation for this summary is to find a method that can work without the need for a Galerkin coarse-grid approximation. The Full Multigrid (FMG) method is well known as a method that can theoretically solve a problem to its discretization accuracy with optimal linear computational complexity, that is in just O(N) operations for N unknowns. Indeed, we develop an FMG method for the 2-D SAAF equation that yields discretization accuracy and does not require the use of a Galerkin coarse-grid approximation. Until recently, aside from proofs about the order of the error for FMG, little work had been done in quantifying when and how FMG can obtain discretization accuracy. However, recent work by Rodrigo et al. has introduced a Fourier analysis technique called worst-case right-hand side analysis used to calculate the FMG accuracy measure. The FMG accuracy measure is the upper bound on the constant of proportionality between the error of the FMG solution and the discretization error. In this summary, we develop and implement an FMG method for the 2-D Cartesian SAAF equations. We then perform worst-case right-hand side analysis and give numerical examples that demonstrate under what circumstances the method can solve the 2-D SAAF equations to discretization accuracy, for optically thin cells, without the need for a Galerkin coarse-grid approximation. (authors)

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Particle Sensitivity Analysis

We propose to develop a computational sensitivity analysis capability for Monte Carlo sampling-based particle simulation relevant to Aleph, Cheetah-MC, Empire, Emphasis, ITS, SPARTA, and LAMMPS codes. These software tools model plasmas, radiation transport, low-density fluids, and molecular motion. Our report demonstrates how adjoint optimization methods can be combined with Monte Carlo sampling-based adjoint particle simulation. Our goal is to develop a sensitivity analysis to drive robust design-based optimization for Monte Carlo sampling-based particle simulation - a currently unavailable capability.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Adjoint Fokker-Planck equation and runaway electron dynamics

The adjoint Fokker-Planck equation method is applied to study the runaway probability function and the expected slowing-down time for highly relativistic runaway electrons, including the loss of energy due to synchrotron radiation. In direct correspondence to Monte Carlo simulation methods, the runaway probability function has a smooth transition across the runaway separatrix, which can be attributed to effect of the pitch angle scattering term in the kinetic equation. However, for the same numerical accuracy, the adjoint method is more efficient than the Monte Carlo method. The expected slowing-down time gives a novel method to estimate the runaway current decay time in experiments. A new result from this work is that the decay rate of high energy electrons is very slow when E is close to the critical electric field. This effect contributes further to a hysteresis previously found in the runaway electron population.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Enhanced Data Efficiency Using Deep Neural Networks and Gaussian Processes for Aerodynamic Design Optimization

Adjoint-based optimization methods are attractive for aerodynamic shape design primarily due to their computational costs being independent of the dimensionality of the input space and their ability to generate high-fidelity gradients that can then be used in a gradient-based optimizer. This makes them very well suited for high-fidelity simulation based aerodynamic shape optimization of highly parametrized geometries such as aircraft wings. However, the development of adjoint-based solvers involve careful mathematical treatment and their implementation require detailed software development. Furthermore, they can become prohibitively expensive when multiple optimization problems are being solved, each requiring multiple restarts to circumvent local optima. In this work, we propose a machine learning enabled, surrogate-based framework that replaces the expensive adjoint solver, without compromising on predicting predictive accuracy. Specifically, we first train a deep neural network (DNN) from training data generated from evaluating the high-fidelity simulation model on a model-agnostic design of experiments on the geometry shape parameters. The optimum shape may then be computed by using a gradient-based optimizer coupled with the trained DNN. Subsequently, we also perform a gradient-free Bayesian optimization, where the trained DNN is used as the prior mean. We observe that the latter framework (DNN-BO) improves upon the DNN-only based optimization strategy for the same computational cost. Overall, this framework predicts the true optimum with very high accuracy, while requiring far fewer high-fidelity function calls compared to the adjoint-based method. Furthermore, we show that multiple optimization problems can be solved with the same machine learning model with high accuracy, to amortize the offline costs associated with constructing our models. Our methodology finds applications in the early stages of aerospace design. (C) 2021 Published by Elsevier Masson SAS.

Renganathan, S. Ashwin↗