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 145 records · Page 8

Optimal checkpointing for adjoint multistage time-stepping schemes

Here, we consider checkpointing strategies that minimize the number of recomputations needed when performing discrete adjoint computations using multistage time-stepping schemes that require computing several substeps within one complete time step. Specifically, we propose two algorithms that can generate optimal checkpoint-ing schedules under weak assumptions. The first is an extension of the seminal Revolve algorithm adapted to multistage schemes. The second algorithm, named CAMS, is developed based on dynamic programming, and it requires the least number of recomputations when compared with other algorithms. The CAMS algorithm is made publicly available in a library with bindings to C and Python. Numerical results show that the proposed algorithms can deliver up to two times the speedup compared with that of classical Revolve. Moreover, we discuss the utilization of the CAMS library in mature scientific computing libraries and demonstrate the ease of using it in an adjoint workflow. The proposed algorithms have been adopted by the PETSc TSAdjoint library. Their performance has been demonstrated with a large-scale PDE-constrained optimization problem on a leadership-class supercomputer. This work is a significant extension of the authors' conference paper.

97 MATHEMATICS AND COMPUTING↗

Nuclear data uncertainty propagation and modeling uncertainty impact evaluation in neutronics core simulation

Uncertainty analysis is a critical requirement in reactor simulation as it is used to quantify the reliability of best-estimate calculation. A comprehensive uncertainty analysis should characterize all sources of uncertainties in a computationally-feasible and scientifically-defendable manner. Here we employ a well-established reduced order modeling (ROM) based uncertainty quantification methodology to propagate uncertainties throughout neutronic calculations. ROM relies on recent advances in randomized data mining techniques applied to large data streams. In our proposed implementation, the nuclear data uncertainties are first propagated from multi-group level through lattice physics calculation to generate few-group parameter uncertainties, described using a vector of mean values and a covariance matrix. Employing an ROM-based compression of the covariance matrix, the few-group uncertainties are then propagated through downstream core simulation in a computationally efficient manner. This straightforward approach, albeit efficient as compared to brute force forward and/or adjoint-based methods, often employs a number of assumptions that have been unquestioned in the literature of neutronic uncertainty analysis. This manuscript argues that these assumptions could introduce another source of uncertainty referred to as modeling uncertainties, whose magnitude needs to be quantified in tandem with nuclear data uncertainties. Thus, our primary goal is to explore the interactions between these two uncertainty sources in order to assess whether modeling uncertainties have an impact on parameter uncertainties. To explore this endeavor, the impact of a number of modeling assumptions on core attributes uncertainties is quantified. The study employs a CANDU reactor model, with Serpent and NEWT as lattice physics solvers and NESTLE-C as core simulator. The modeling assumptions investigated include those related with the uncertainty propagation method employed, e.g., deterministic vs. stochastic, the few-group energy structure employed to represent the cross-sections, the resonance treatment in lattice physics calculation, the reference values for the cross-section, and the number of samples employed to render ROM compression. Results indicate that some of the modeling assumptions could have a non-negligible impact on the core responses propagated uncertainties, highlighting the need for a more comprehensive approach to combine parameter and modeling uncertainties.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Adjoint-Based Inversion of Geodetic Data for Sources of Deformation and Strain

An adjoint-based formulation leads to a particularly efficient approach for inverting geodetic measurements for the source of the deformation. Specifically, the quantities necessary to iteratively improve the fit to the observations can be computed with just three forward calculations, one to obtain the current residuals, another to solve the adjoint problem, and a third to compute the step length. An inversion algorithm utilizing the adjoint-based gradient is applied to a set of Interferometric Synthetic Aperture Radar (InSAR) data gathered between 2016 and 2018 over the Tulare Basin in California's Central Valley. Because the measured deformation is due to groundwater withdrawal, a penalty function is included in the inversion to avoid placing aquifer volume change in locations that are far from any documented wells. The solution of the inverse problem provides estimates of aquifer compaction that provide a match to the observed range changes while honoring the well data. The solution indicates an average aquifer volume loss of 2.17 km 3 /year over the two year period from January 2016 to January 2018, encompassing one drought year (2016) and one wet year (2017). Finally, this magnitude of lost volume is compatible with the 3.1 km 3 /year decrease in water volume for the entire Central Valley, estimated from GRACE satellite gravity data.

58 GEOSCIENCES↗

Global adjoint tomography—model GLAD-M25

SUMMARY Building on global adjoint tomography model GLAD-M15, we present transversely isotropic global model GLAD-M25, which is the result of 10 quasi-Newton tomographic iterations with an earthquake database consisting of 1480 events in the magnitude range 5.5 ≤ Mw ≤ 7.2, an almost sixfold increase over the first-generation model. We calculated fully 3-D synthetic seismograms with a shortest period of 17 s based on a GPU-accelerated spectral-element wave propagation solver which accommodates effects due to 3-D anelastic crust and mantle structure, topography and bathymetry, the ocean load, ellipticity, rotation and self-gravitation. We used an adjoint-state method to calculate Fréchet derivatives in 3-D anelastic Earth models facilitated by a parsimonious storage algorithm. The simulations were performed on the Cray XK7 ‘Titan’ and the IBM Power 9 ‘Summit’ at the Oak Ridge Leadership Computing Facility. We quantitatively evaluated GLAD-M25 by assessing misfit reductions and traveltime anomaly histograms in 12 measurement categories. We performed similar assessments for a held-out data set consisting of 360 earthquakes, with results comparable to the actual inversion. We highlight the new model for a variety of plumes and subduction zones.

58 GEOSCIENCES↗

Global adjoint tomography: first-generation model

We present the first-generation global tomographic model constructed based on adjoint tomography, an iterative full-waveform inversion technique. Synthetic seismograms were calculated using GPU-accelerated spectral-element simulations of global seismic wave propagation, accommodating effects due to 3-D anelastic crust & mantle structure, topography & bathymetry, the ocean load, ellipticity, rotation, and self-gravitation. Fréchet derivatives were calculated in 3-D anelastic models based on an adjoint-state method. The simulations were performed on the Cray XK7 named ‘Titan’, a computer with 18 688 GPU accelerators housed at Oak Ridge National Laboratory. The transversely isotropic global model is the result of 15 tomographic iterations, which systematically reduced differences between observed and simulated three-component seismograms. Our starting model combined 3-D mantle model S362ANI with 3-D crustal model Crust2.0. We simultaneously inverted for structure in the crust and mantle, thereby eliminating the need for widely used ‘crustal corrections’. We used data from 253 earthquakes in the magnitude range 5.8 ≤ M w ≤ 7.0. We started inversions by combining ~30 s body-wave data with ~60 s surface-wave data. The shortest period of the surface waves was gradually decreased, and in the last three iterations we combined ~17 s body waves with ~45 s surface waves. We started using 180 min long seismograms after the 12th iteration and assimilated minor- and major-arc body and surface waves. The 15th iteration model features enhancements of well-known slabs, an enhanced image of the Samoa/Tahiti plume, as well as various other plumes and hotspots, such as Caroline, Galapagos, Yellowstone and Erebus. Furthermore, we see clear improvements in slab resolution along the Hellenic and Japan Arcs, as well as subduction along the East of Scotia Plate, which does not exist in the starting model. Point-spread function tests demonstrate that we are approaching the resolution of continental-scale studies in some areas, for example, underneath Yellowstone. Here, this is a consequence of our multiscale smoothing strategy in which we define our smoothing operator as a function of the approximate Hessian kernel, thereby smoothing gradients less wherever we have good ray coverage, such as underneath North America.

58 GEOSCIENCES↗

Fréchet kernels based on a fractional viscoacoustic wave equation

Incorporating the seismic attenuation into the waveform inversion framework could not only improve the accuracy of the velocity model but also provide an additional Q model. Recently, we proposed a viscoacoustic wave equation assisted by the fractional Laplacian operators to accurately model the wave propagation in heterogeneous attenuating media with computational efficiency. The explicit presence of Q as a coefficient in this equation suggests the potential to conveniently develop its full waveform inversion scheme. In this study, we utilize the adjoint-state method to formulate the computation of the Fréchet kernels with respect to both velocity and attenuation based on this wave equation. These kernels will play a fundamental role in the viscoacoustic multiparameter waveform inversion.

Xing, Guangchi↗

Ecosystem biogeochemistry model parameterization: Do more flux data result in a better model in predicting carbon flux?

Reliability of terrestrial ecosystem models highly depends on the quantity and quality of thedata that have been used to calibrate the models. Nowadays, in situ observations of carbon fluxes areabundant. However, the knowledge of how much data (data length) and which subset of the time seriesdata (data period) should be used to effectively calibrate the model is still lacking. This study uses theAmeriFlux carbon flux data to parameterize the Terrestrial Ecosystem Model (TEM) with an adjoint-baseddata assimilation technique for various ecosystem types. Parameterization experiments are thus conductedto explore the impact of both data length and data period on the uncertainty reduction of the posteriormodel parameters and the quantification of site and regional carbon dynamics. We find that: the modelis better constrained when it uses two-year data comparing to using one-year data. Further, two-year datais sufficient in calibrating TEM’s carbon dynamics, since using three-year data could only marginallyimprove the model performance at our study sites; the model is better constrained with the data thathave a higher‘‘climate variability’’than that having a lower one. The climate variability is used to measurethe overall possibility of the ecosystem to experience all climatic conditions including drought and extremeair temperatures and radiation; the U.S. regional simulations indicate that the effect of calibration datalength on carbon dynamics is amplified at regional and temporal scales, leading to large discrepanciesamong different parameterization experiments, especially in July and August. Our findings areconditioned on the specific model we used and the calibration sites we selected. The optimal calibrationdata length may not be suitable for other models. However, this study demonstrates that there may exist athreshold for calibration data length and simply using more data would not guarantee a better modelparameterization and prediction. More importantly, climate variability might be an effective indicator ofinformation within the data, which could help data selection for model parameterization. As a result, we believe ourfindings will benefit the ecosystem modeling community in using multiple-year data to improve modelpredictability.

59 BASIC BIOLOGICAL SCIENCES↗

Nonlinear Optimal Control of Electron Dynamics Within Hartree-Fock Theory

Consider the problem of determining the optimal applied electric field to drive a molecule from an initial state to a desired target state. For even moderately sized molecules, solving this problem directly using the exact equations of motion—the time-dependent Schrödinger equation (TDSE)—is numerically intractable. Here, we present a solution of this problem within time-dependent Hartree-Fock (TDHF) theory, a mean field approximation of the TDSE. Optimality is defined in terms of minimizing the total control effort while maximizing the overlap between desired and achieved target states. We frame this problem as an optimization problem constrained by the nonlinear TDHF equations; we solve it using trust region optimization with gradients computed via a custom-built adjoint state method. For three molecular systems, we show that with very small neural network parametrizations of the control, our method yields solutions that achieve desired targets within acceptable constraints and tolerances.

97 MATHEMATICS AND COMPUTING↗

Prompt Neutron Lifetime for the NBSR Reactor

In preparation for the proposed conversion of the National Institute of Standards and Technology (NIST) research reactor (NBSR) from high-enriched uranium (HEU) to low-enriched uranium (LEU) fuel, certain point kinetics parameters must be calculated. We report here values of the prompt neutron lifetime that have been calculated using three independent methods. All three sets of calculations demonstrate that the prompt neutron lifetime is shorter for the LEU fuel when compared to the HEU fuel and longer for the equilibrium end-of-cycle (EOC) condition when compared to the equilibrium startup (SU) condition for both the HEU and LEU fuels.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Exploring gauge-fixing conditions with gradient-based optimization

Lattice gauge fixing is required to compute gauge-variant quantities, for example those used in RI-MOM renormalization schemes or as objects of comparison for model calculations. Recently, gauge-variant quantities have also been found to be more amenable to signal-to-noise optimization using contour deformations. These applications motivate systematic parameterization and exploration of gauge-fixing schemes. This work introduces a differentiable parameterization of gauge fixing which is broad enough to cover Landau gauge, Coulomb gauge, and maximal tree gauges. The adjoint state method allows gradient-based optimization to select gauge-fixing schemes that minimize an arbitrary target loss function.

Detmold, William↗

Invertible Neural Networks for Airfoil Design

We report the airfoil design problem, in which an engineer seeks a shape with desired performance characteristics, is fundamental to aerodynamics. Design workflows traditionally rely on iterative optimization methods using low-fidelity integral boundary-layer methods as higher-fidelity adjoint-based computational fluid dynamics methods are computationally expensive. Surrogate-based approaches can accelerate the design process but still rely on some iterative inverse design procedure. In this work, we leverage emerging invertible neural network (INN) tools to enable the rapid inverse design of airfoil shapes for wind turbines. INNs are specialized deep-learning models with well-defined inverse mappings. When trained appropriately, INN surrogate models are capable of forward prediction of aerodynamic and structural quantities for a given airfoil shape as well as inverse recovery of airfoil shapes with specified aerodynamic and structural characteristics. The INN approach offers a roughly 100 times speed-up compared to adjoint-based methods for inverse design. We demonstrate the INN tool for inverse design on three test cases of 100 airfoils each that satisfy the performance characteristics close to those of airfoils used in wind-turbine blades. All generated shapes satisfy the desired aerodynamic characteristics, demonstrating the success of the INN approach for inverse design of airfoils.

17 WIND ENERGY↗

Adjoint design sensitivity analysis of reduced atomic systems using generalized Langevin equation for lattice structures

An efficient adjoint design sensitivity analysis method is developed for reduced atomic systems. A reduced atomic system and the adjoint system are constructed in a locally confined region, utilizing generalized Langevin equation (GLE) for periodic lattice structures. Due to the translational symmetry of lattice structures, the size of time history kernel function that accounts for the boundary effects of the reduced atomic systems could be reduced to a single atom’s degrees of freedom. For the problems of highly nonlinear design variables, the finite difference method is impractical for its inefficiency and inaccuracy. However, the adjoint method is very efficient regardless of the number of design variables since one additional time integration is required for the adjoint GLE. Through numerical examples, the derived adjoint sensitivity turns out to be accurate and efficient through the comparison with finite difference sensitivity.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Consistent Adjoint Driven Importance Sampling using Space, Energy and Angle

For challenging radiation transport problems, hybrid methods combine the accuracy of Monte Carlo methods with the global information present in deterministic methods. One of the most successful hybrid methods is CADIS Consistent Adjoint Driven Importance Sampling. This method uses a deterministic adjoint solution to construct a biased source distribution and consistent weight windows to optimize a specific tally in a Monte Carlo calculation. The method has been implemented into transport codes using just the spatial and energy information from the deterministic adjoint and has been used in many applications to compute tallies with much higher figures-of-merit than analog calculations. CADIS also outperforms user-supplied importance values, which usually take long periods of user time to develop. This work extends CADIS to develop weight windows that are a function of the position, energy, and direction of the Monte Carlo particle. Two types of consistent source biasing are presented: one method that biases the source in space and energy while preserving the original directional distribution and one method that biases the source in space, energy, and direction. Seven simple example problems are presented which compare the use of the standard space/energy CADIS with the new space/energy/angle treatments.

Peplow, Douglas E.↗

Optimization of the Direct Discrete Method Using the Solution of the Adjoint Equation and its Application in the Multi-Group Neutron Diffusion Equation

Obtaining the set of algebraic equations that directly correspond to a physical phenomenon has been viable in the recent direct discrete method (DDM). Although this method may find its roots in physical and geometrical considerations, there are still some degrees of freedom that one may suspect optimize-able. Here we have used the information embedded in the corresponding adjoint equation to form a local functional, which in turn by its minimization, yield suitable dual mesh positioning.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Reverse-Mode Algorithmic Differentiation of an OpenMP-Parallel Compressible Flow Solver

y Reverse-mode algorithmic differentiation (AD) is an established method for obtaining adjoint derivatives of computer simulation applications. In computational fluid dynamics (CFD), adjoint derivatives of a cost function output such as drag or lift with respect to design parameters such as surface coordinates or geometry control points are a key ingredient for shape optimization, uncertainty quantification and flow control. The computational cost of CFD applications and their derivatives makes it essential to use high-performance computing hardware efficiently, including multi- and many-core architectures. Nevertheless, OpenMP is not supported in most AD tools, and previously shown methods achieve poor scalability of the derivative code. We present the AD of an OpenMP-parallelized finite volume compressible flow solver for unstructured meshes. Our approach enables us to reuse the parallelization of the original code in the computation of adjoint derivatives. The method works by identifying code segments that can be differentiated in reverse-mode without changing their memory access pattern. The OpenMP parallelization is integrated into the derivative code during the build process in a way that is robust to modifications of the original code and independent of the OpenMP support of the differentiation tool. We show the scalability of our adjoint CFD solver on test cases ranging from thousands to millions of finite volume mesh cells on CPUs with up to 16 threads as well as on an Intel XeonPhi card with 236 threads. We demonstrate that our approach is more practical to implement for production-sized CFD codes and produces more efficient adjoint derivative code than previously shown AD methods.

Huckelheim, Jan↗

Assessment of Multi-Step CADIS Parameter Sensitivity to Neutron Flux Spectrum in Shutdown Dose Rate Calculations

Shutdown dose rate (SDDR) analysis requires (a) a neutron transport calculation to estimate neutron flux fields, (b) an activation calculation to compute radionuclide inventories and associated photon sources, and (c) a photon transport calculation to estimate final dose rate. In some applications, accurate full-scale Monte Carlo (MC) SDDR simulations are needed for very large systems with massive amounts of shielding materials. For example, SDDR assessments are required throughout the biological shield (bio-shield) of the ITER experimental facility to evaluate the required waiting period after the shutdown of ITER and to identify the locations where human accessibility should be prohibited. The bio-shield is a large cylindrical concrete structure (30 m tall and 34 m in diameter) surrounding the very complex tokamak machine. Determining the effects on SDDR of important factors, such as the cross talk (interactions) between the different ports of ITER, is possible only through full-scale simulations that involve all the complex inner details of the ITER tokamak machine. These simulations are impractical because calculation of space- and energy-dependent neutron fluxes throughout the structural materials is needed to estimate the distribution of radioisotopes causing the SDDR. Determining a good importance function that can be used in biasing a neutron MC calculation for SDDR is complicated because it is difficult to explicitly express the SDDR response function, which depends on subsequent computational steps. The Multi-Step Consistent Adjoint Driven Importance Sampling (MS-CADIS) hybrid MC/deterministic method was developed to speed up the SDDR MC neutron transport calculation using an importance function that represents the neutron importance to the final SDDR. The MS-CADIS method uses the CADIS method, which has been Notice: This manuscript has been authored by UT-Battelle, LLC, under Contract No. DE-AC0500OR22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for the United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan). successfully used for more than a decade in radiation shielding applications to generate consistent source biasing and weight window (WW) variance reduction parameters to accelerate continuous-energy MC simulations. However, because MS-CADIS focuses on multi-step shielding calculations such as SDDR calculations, it develops an importance function for the initial radiation transport calculation (e.g., the neutron calculations in SDDR simulations) that represents the importance of particles to the final response of the overall simulation. To develop this importance function, the MS-CADIS method uses an adjoint calculation with a specific neutron adjoint source that represents the neutron's contribution to the final SDDR. Brute force calculation of the MS-CADIS neutron adjoint source requires a deterministic neutron transport calculation coupled with activation calculations in each element of the deterministic mesh. These calculations are needed to determine the MS-CADIS parameter for the neutron interactions that produce radioisotopes contributing to the SDDR. The MS-CADIS parameter is the microscopic cross section of the radioisotope-producing interaction, multiplied by the mass of this radioisotope existing at the end of the irradiation and decay scenario, and divided by the steady-state rate of the interaction. Because the activation calculations are performed for each radionuclide in each mesh element, implementation of the MS-CADIS method can be computationally expensive. However, these neutron transport and activation calculations are not needed if we assume that the radioisotopes are created due to single neutron interactions and their concentrations do not change as a result of burnup by neutron irradiation. The use of this approximation, which renders the MS-CADIS parameter independent of the neutron flux, seems reasonable in calculating the MS-CADIS importance function because this function is needed only to speed up the MC calculation, not to calculate the final SDDR. In this work, we investigate the validity of this approximation by investigating the behavior of the MS-CADIS parameter as a function of the neutron flux spectra in the Fusion Evaluated Nuclear Data Library (FENDL) Neutronics and Shielding Benchmark problem that contains representative isotopes and flux environments of fusion energy systems. (authors)

61 RADIATION PROTECTION AND DOSIMETRY↗