Search NASA⌕ Search

SEARCH · Search NASA

Results for “gradient methods”

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 685 records · Page 38

Oscillator strengths, first-order properties, and nuclear gradients for local ADC(2)

We describe theory and implementation of oscillator strengths, orbital-relaxed first-order properties, and nuclear gradients for the local algebraic diagrammatic construction scheme through second order. The formalism is derived via time-dependent linear response theory based on a second-order unitary coupled cluster model. The implementation presented here is a modification of our previously developed algorithms for Laplace transform based local time-dependent coupled cluster linear response (CC2LR); the local approximations thus are state specific and adaptive. The symmetry of the Jacobian leads to considerable simplifications relative to the local CC2LR method; as a result, a gradient evaluation is about four times less expensive. Test calculations show that in geometry optimizations, usually very similar geometries are obtained as with the local CC2LR method (provided that a second-order method is applicable). As an exemplary application, we performed geometry optimizations on the low-lying singlet states of chlorophyllide a.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Dataset from: "Adjoint methods for quasisymmetry of vacuum fields on a surface"

Adjoint methods can speed up stellarator optimisation by providing gradient information more efficiently compared to finite-difference evaluations. Adjoint methods are herein applied to vacuum magnetic fields, with objective functions targeting quasisymmetry and a rotational transform value on a surface. To measure quasisymmetry, a novel way of evaluating approximate flux coordinates on a single flux surface without the assumption of a neighbourhood of flux surfaces is proposed. The shape gradients obtained from the adjoint formalism are evaluated numerically and verified against finite-difference evaluations.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Completeness and coverage of open-access freshwater fish distribution data in the United States

Aim: Primary biodiversity data constitute observations of particular species at given points in space and time. Open-access databases provide unprecedented access to these data, but their utility depends on how complete species inventories are at a given surveying resolution and how uniformly distributed surveys are in space and time. Our aim was to characterize the completeness of freshwater fish data across spatial scales and among habitat types (lotic stream reaches and lentic waterbodies), and assess coverage bias in space and time. Location: Freshwater habitats of the contiguous United States. Methods: We compiled fish occurrence records from the Global Biodiversity Information Facility (GBIF), the Multi-state Aquatic Resources Information Server (MARIS), and three federally-administered surveys (EMAP, NAQWA, NRS). Next, we evaluated survey completeness across seven spatial resolutions representing nested stream catchments, and separately for lotic and lentic habitats. Lastly, we tested for bias in survey coverage along environmental and temporal gradients for each habitat type. Results: A total of 39,855,157 records representing 1925 species were available from 1800 to 2014. The percent of catchments surveyed increased as spatial resolution decreased while the density of occurrence records within surveyed catchments decreased. Species richness also increased with catchment size consistent with putative species–area relationships. Occurrence records were distributed among 72,114 lotic reaches and 3,325 lentic waterbodies throughout the contiguous US. Seven percent (N=5,055) of surveyed lotic reaches and 23% (N=771) of surveyed lentic waterbodies were well-surveyed based on the most conservative thresholds for two completeness indices. Most well-surveyed lotic reaches (95%) and undammed lentic waterbodies (97%) were completed since 1980, and most were surveyed during only one 20-year time interval (83% and 56% for lotic and lentic waterbodies, respectively). Coverage analyses indicated that contemporary and future climates corresponding to regions along the west coast and gulf coast are the most under-represented for both lotic and lentic habitats. Small lentic waterbodies (surface area < 0.1 km2) are also under-represented. Main Conclusions: Contemporary surveys of lotic reaches and, to a lesser extent, lentic waterbodies are plentiful and generally representative of climatic and landscape characteristics. Data compilation efforts should focus on adding historical data to public repositories. Clustering of well-surveyed waterbodies in MARIS-participating states highlights the value of state-agency data and demonstrates the need for dissemination of these data by more state agencies.

60 APPLIED LIFE SCIENCES↗

Technical Note: Dose gradients and prescription isodose in orthovoltage stereotactic radiosurgery

Purpose: The purpose of this work is to examine the trade-off between prescription isodose and dose gradients in orthovoltage stereotactic radiosurgery. Methods: Point energy deposition kernels (EDKs) describing photon and electron transport were calculated using Monte Carlo methods. EDKs were generated from 10 to 250 keV, in 10 keV increments. The EDKs were converted to pencil beam kernels and used to calculate dose profiles through isocenter from a 4π isotropic delivery from all angles of circularly collimated beams. Monoenergetic beams and an orthovoltage polyenergetic spectrum were analyzed. The dose gradient index (DGI) is the ratio of the 50% prescription isodose volume to the 100% prescription isodose volume and represents a metric by which dose gradients in stereotactic radiosurgery (SRS) may be evaluated. Results: Using the 4π dose profiles calculated using pencil beam kernels, the relationship between DGI and prescription isodose was examined for circular cones ranging from 4 to 18 mm in diameter and monoenergetic photon beams with energies ranging from 20 to 250 keV. Values were found to exist for prescription isodose that optimize DGI. Conclusions: The relationship between DGI and prescription isodose was found to be dependent on both field size and energy. Examining this trade-off is an important consideration for designing optimal SRS systems.

60 APPLIED LIFE SCIENCES↗

Derivative-free stochastic optimization via adaptive sampling strategies

In this paper, we present a novel derivative-free framework for solving unconstrained stochastic optimization problems. Many problems in fields ranging from simulation optimization to reinforcement learning to quantum computing involve settings where only stochastic function values are obtained via a zeroth-order oracle, which has no available gradient information and necessitates the usage of derivative-free optimization methodologies. Our approach includes estimating gradients using stochastic function evaluations and integrating adaptive sampling techniques to control the accuracy in these stochastic approximations. Our framework encapsulates several gradient estimation techniques, including standard finite-difference, Gaussian smoothing, sphere smoothing, randomized coordinate finite-difference, and randomized subspace finite-difference methods. We provide theoretical convergence guarantees for our framework and analyze the worst-case iteration and sample complexities associated with each gradient estimation method. Finally, we demonstrate the empirical performance of the methods on logistic regression and nonlinear least squares problems.

Adaptive sampling↗

Adjoint methods for quasi-symmetry of vacuum fields on a surface

Adjoint methods can speed up stellarator optimisation by providing gradient information more efficiently compared with finite-difference evaluations. Adjoint methods are herein applied to vacuum magnetic fields, with objective functions targeting quasi-symmetry and a rotational transform value on a surface. Here to measure quasi-symmetry, a novel way of evaluating approximate flux coordinates on a single flux surface without the assumption of a neighbourhood of flux surfaces is proposed. The shape gradients obtained from the adjoint formalism are evaluated numerically and verified against finite-difference evaluations.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Novel Relativistic Electronic Structure Theories for Actinide-Containing Compounds

Actinides of importance to basic energy sciences contain electrons moving at speed comparable to the speed of light. Reliable computational simulation of these electrons and hence actinide chemistry requires accurate description of relativistic effects. The present project advances computational actinide chemistry with development of new methodologies, algorithms, and computer programs in relativistic quantum chemistry, as well as applications to actinide chemistry and spectroscopy. A new “electrons-only” exact two-component approach has been developed to provide efficient treatments of relativistic effects, while maintaining chemical accuracy. New computational algorithms developed here extend the applicability of relativistic electron-correlation methods to larger molecules. The method-development work in this project also features the first implementation of analytic gradient technique for relativistic electron-correlation methods, which provides significantly enhanced ability to compute properties for molecules containing actinides. The applicability and usefulness of these new methods and computer programs have been demonstrated in calculations of actinide-containing molecules to facilitate understanding of actinide chemistry and spectroscopy.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Static Subspace Approximation for Random Phase Approximation Correlation Energies: Applications to Materials for Catalysis and Electrochemistry

Modeling complex materials using high-fidelity, ab initio methods at low cost is a fundamental goal for quantum chemical software packages. The GW approximation and random phase approximation (RPA) provide a unified description of both electronic structure and total energies using the same physics in a many-body perturbative approach that can be more accurate than generalized-gradient density functional theory (DFT) methods. However, GW/RPA implementations have historically been limited to either specific materials classes or application toward small chemical systems. Here, the static subspace approximation allows for reduced cost full-frequency GW/RPA calculations and has previously been benchmarked thoroughly for GW calculations. Here, we describe our approach to including partial occupations of electronic orbitals in full-frequency GW and RPA calculations for the study of electrocatalysts. We benchmarked RPA total energy calculations using the subspace approximation across a diverse test suite of materials for a variety of computational parameters. The benchmarking quantifies the impact of different extrapolation procedures for representing the static polarizability at infinite screened cutoff, and shows that using screened cutoffs above 20-25 Ryd result in diminishing accuracy returns for predicting RPA total energies. Additionally, for moderately sized electrocatalytic models, 2-3 times fewer computational resources are used to compute RPA total energies by representing the static polarizability with 20-30% of the static subspace basis, with an error of approximately 0.01 eV or better in RPA adsorption energy calculations. Finally, we show that for these electrochemical models RPA can shift DFT adsorption energy shifts by up to 0.5 eV and that GW can frequently shift DFT eigenvalues of surface and adsorbate states by approximately 0.5-1 eV.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Radiation of Gas Layer over Hot Surface

A method is presented for evaluation the radiation flux produced by a gas layer near a heated surface, where the gas temperature depends on a distance from the surface. This method refers to small temperature gradients and operates with an effective radiation temperature for each frequency, as well as with the width of the gas absorption band. These parameters are determined by the absorption spectrum of atoms or gas molecules, and also by the shape of the spectral line for the radiative transition between certain states of atomic particles of a gas. The possibilities of this method are demonstrated by examples of emission of photons from the solar photosphere, as well as emission of CO{sub 2} molecules in the atmospheres of the Earth and Venus.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

A comparative study of calibration techniques for finite strain elastoplasticity: Numerically-exact sensitivities for FEMU and VFM

Accurate identification of material parameters is crucial for predictive modeling in computational mechanics. Here, the two primary approaches in the experimental mechanics community for calibration from full-field digital image correlation data are known as finite element model updating (FEMU) and the virtual fields method (VFM). In VFM, the objective function is a squared mismatch between internal and external virtual work or power. In FEMU, the objective function quantifies the weighted mismatch between model predictions and corresponding experimentally measured quantities of interest. It is minimized by iteratively updating the parameters of an FE model. While FEMU is seen as more flexible, VFM is commonly used instead of FEMU due to its considerably greater computational expense. However, comparisons between the two methods usually involve approximations of gradients or sensitivities with finite difference schemes, thereby making direct assessments difficult. Hence, in this study, we compare VFM and FEMU in the context of numerically-exact sensitivities obtained through local sensitivity analyses and the application of automatic differentiation software. To this end, we conduct a series of test cases to assess both methods under practical challenges using a finite strain elastoplasticity model.

Automatic differentiation↗

Design Criteria for High-Gradient Radio-Frequency Linacs

This article will review methods used at the SLAC National Accelerator Laboratory and other world accelerator laboratories to design high-gradient normal conducting accelerating structures. A quest for compact radio-frequency linacs fueled decades of studies toward a higher accelerating gradient. A major phenomena limiting the increase of the gradient is vacuum radio-frequency breakdown; therefore, this paper will address the breakdown physics and discuss approaches that reduce the breakdown probability. This discussion will cover both the electrical design and fabrication technology of the accelerating structures to achieve practical operating accelerating gradients in excess of 100 MV/m. Most of the data described here were obtained during the development of 11 GHz linacs for electron–positron linear colliders, so extrapolation of the results to other frequencies should be performed cautiously.

Dolgashev, Valery A. (ORCID:0000000166715402)↗

A robust spectral element implementation of the $k - τ$ RANS model in Nek5000/NekRS

The $k - ω$ Reynolds Averaged Navier Stokes (RANS) model is one of the industry standard approaches for modeling of turbulent flows. It performs better than the $k - ϵ$ model for low Reynolds number flows and is also more suitable for boundary layers with adverse pressure gradients. Major drawback of the model, however, is that the asymptotic value of $ω$ at the walls is singular, necessitating the use of a contrived “sufficiently” large value for $ω$ as the boundary condition for its transport equation. Here, this invariably leads to the solution being sensitive to near wall grid spacing. While an acceptable solution for low order (finite volume) methods, the excessive near wall gradients lead to persistent numerical stability issues in high order codes. To alleviate the problem, specifically in the context of the high order spectral element code Nek5000, a regularized $k - ω$ approach was formulated in our prior work (Tomboulides et al., 2018). The formulation, however, relies on the use of wall distance and its gradients for modeling the closure terms and can pose problems for simulations in complex geometries. This work presents a novel implementation of the $k - τ$ RANS model in Nek5000, where $τ = 1/ω$, eliminating the need for regularization, owing to the asymptotically bounded behavior of the source terms in the $τ$ transport equation, and also eliminating dependence on wall distance. Robustness and stability of the $k - τ$ model is ensured through implicit treatment of the source terms and their careful numerical implementation and demonstrated through several cases aimed at verification and validation. Studies include both canonical and engineering relevant problems, viz., turbulent channel flow, pipe flow, backward facing step, flow over NACA 0012 airfoil and flow in a T-junction. Results from the $k - τ$ model are shown to be consistent with regularized $k - ω$ model and also with the $k - ω$ SST model in OpenFOAM (for select studies). Comparison with experimental data is also shown, where available, to bolster validation efforts for the $k - τ$ model implementation through prediction of key turbulent quantities of interest.

Nek5000↗

Efficient approach to obtain free energy gradient using QM/MM MD simulation

The efficient computational approach denoted as charge and atom dipole response kernel (CDRK) model to consider polarization effects of the quantum mechanical (QM) region is described using the charge response and the atom dipole response kernels for free energy gradient (FEG) calculations in the quantum mechanical/molecular mechanical (QM/MM) method. CDRK model can reasonably reproduce energies and also energy gradients of QM and MM atoms obtained by expensive QM/MM calculations in a drastically reduced computational time. This model is applied on the acylation reaction in hydrated trypsin-BPTI complex to optimize the reaction path on the free energy surface by means of FEG and the nudged elastic band (NEB) method.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Tool employing homotopy-based approaches in finding the controlling unstable equilibrium point in the electric power grid

The analysis tool employs a computer-implemented algorithm that uses homotopy-based approaches to map the solution from the exit point to the controlling unstable equilibrium point (UEP). The computational time is reduced significantly by using an approximate exit point rather than computing an accurate exit point as it is required in finding the controlling UEPs using traditional transient stability direct methods. In addition, this method eliminates the necessity of computing the minimum gradient point (MGP) which is a key element in using Newton methods. These properties provide an advantage to homotopy-based approaches over traditional iterative methods in terms of both speed of computation and reliability of finding solutions.

97 MATHEMATICS AND COMPUTING↗

Gradient-Based Optimization of Wind Farms with Different Turbine Heights: Preprint

Turbine wakes reduce power production in a wind farm. Current wind farms are generally built with turbines that are all the same height, but if wind farms included turbines with different tower heights, the cost of energy (COE) may be reduced. We used gradient-based optimization to demonstrate a method to optimize wind farms with varied hub heights. Our study includes a modified version of the FLORIS wake model that accommodates three-dimensional wakes integrated with a tower structural model. Our purpose was to design a process to minimize the COE of a wind farm through layout optimization and varying turbine hub heights. Results indicate that when a farm is optimized for layout and height with two separate height groups, COE can be lowered by as much as 5%-9%, compared to a similar layout and height optimization where all the towers are the same. The COE has the best improvement in farms with high turbine density and a low wind shear exponent.

17 WIND ENERGY↗

DPM: A deep learning PDE augmentation method with application to large-eddy simulation

A framework is introduced that leverages known physics to reduce overfitting in machine learning for scientific applications. The partial differential equation (PDE) that expresses the physics is augmented with a neural network that uses available data to learn a description of the corresponding unknown or unrepresented physics. Training within this combined system corrects for missing, unknown, or erroneously represented physics, including discretization errors associated with the PDE's numerical solution. For optimization of the network within the PDE, an adjoint PDE is solved to provide high-dimensional gradients, and a stochastic adjoint method (SAM) further accelerates training. Additionally, the approach is demonstrated for large-eddy simulation (LES) of turbulence. High-fidelity direct numerical simulations (DNS) of decaying isotropic turbulence provide the training data used to learn sub-filter-scale closures for the filtered Navier–Stokes equations. Out-of-sample comparisons show that the deep learning PDE method outperforms widely-used models, even for filter sizes so large that they become qualitatively incorrect. It also significantly outperforms the same neural network when a priori trained based on simple data mismatch, not accounting for the full PDE. Measures of discretization errors, which are well-known to be consequential in LES, point to the importance of the unified training formulation's design, which without modification corrects for them. For comparable accuracy, simulation runtime is significantly reduced. A relaxation of the typical discrete enforcement of the divergence-free constraint in the solver is also successful, instead allowing the DPM to approximately enforce incompressibility physics. Since the training loss function is not restricted to correspond directly to the closure to be learned, training can incorporate diverse data, including experimental data.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Pseudo-Z symmetric space-times

In this paper, we investigate Pseudo-Z symmetric space-time manifolds. First, we deal with elementary properties showing that the associated form A{sub k} is closed: in the case the Ricci tensor results to be Weyl compatible. This notion was recently introduced by one of the present authors. The consequences of the Weyl compatibility on the magnetic part of the Weyl tensor are pointed out. This determines the Petrov types of such space times. Finally, we investigate some interesting properties of (PZS){sub 4} space-time; in particular, we take into consideration perfect fluid and scalar field space-time, and interesting properties are pointed out, including the Petrov classification. In the case of scalar field space-time, it is shown that the scalar field satisfies a generalized eikonal equation. Further, it is shown that the integral curves of the gradient field are geodesics. A classical method to find a general integral is presented.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Scalable Simulation of Pressure Gradient-Driven Transport of Rarefied Gases in Complex Permeable Media Using Lattice Boltzmann Method

Accurate representations of slip and transitional flow regimes present a challenge in the simulation of rarefied gas flow in confined systems with complex geometries. In these regimes, continuum-based formulations may not capture the physics correctly. This work considers a regularized multi-relaxation time lattice Boltzmann (LB) method with mixed Maxwellian diffusive and halfway bounce-back wall boundary treatments to capture flow at high Kn. The simulation results are validated against atomistic simulation results from the literature. We examine the convergence behavior of LB for confined systems as a function of inlet and outlet treatments, complexity of the geometry, and magnitude of pressure gradient and show that convergence is sensitive to all three. The inlet and outlet boundary treatments considered in this work include periodic, pressure, and a generalized periodic boundary condition. Compared to periodic and pressure treatments, simulations of complex domains using a generalized boundary treatment conserve mass but require more iterations to converge. Convergence behavior in complex domains improves at higher magnitudes of pressure gradient across the computational domain, and lowering the porosity deteriorates the convergence behavior for complex domains.

42 ENGINEERING↗