Search NASA⌕ Search

SEARCH · Search NASA

Results for “algorithmic differentiation”

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 559 records · Page 31

Decomposition of Feynman integrals by multivariate intersection numbers

We present a detailed description of the recent idea for a direct decomposition of Feynman integrals onto a basis of master integrals by projections, as well as a direct derivation of the differential equations satisfied by the master integrals, employing multivariate intersection numbers. We discuss a recursive algorithm for the computation of multivariate intersection numbers, and provide three different approaches for a direct decomposition of Feynman integrals, which we dub the straight decomposition, the bottom-up decomposition, and the top-down decomposition. These algorithms exploit the unitarity structure of Feynman integrals by computing intersection numbers supported on cuts, in various orders, thus showing the synthesis of the intersection-theory concepts with unitarity-based methods and integrand decomposition. We perform explicit computations to exemplify all of these approaches applied to Feynman integrals, paving a way towards potential applications to generic multi-loop integrals.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

On Two-Scale Modelling of Heat and Mass Transfer

Modelling of macroscopic behaviour of materials, consisting of several layers or components, whose microscopic (at least stochastic) analysis is available, as well as (more general) simulation of non-local phenomena, complicated coupled processes, etc., requires both deeper understanding of physical principles and development of mathematical theories and software algorithms. Starting from the (relatively simple) example of phase transformation in substitutional alloys, this paper sketches the general formulation of a nonlinear system of partial differential equations of evolution for the heat and mass transfer (useful in mechanical and civil engineering, etc.), corresponding to conservation principles of thermodynamics, both at the micro- and at the macroscopic level, and suggests an algorithm for scale-bridging, based on the robust finite element techniques. Some existence and convergence questions, namely those based on the construction of sequences of Rothe and on the mathematical theory of two-scale convergence, are discussed together with references to useful generalizations, required by new technologies.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

KIVA-Update June 2012

Development of a fractional step, a Predictor-Corrector Split (PCS), or what is often known as a projection method combining hp-adaptive system in a Finite Element Method (FEM) for combustion modeling has been achieved. This model will advance the accuracy and range of applicability of the KIVA combustion model and software used typically for internal combustion engine modeling. This abstract describes a PCS hp-adaptive FEM model for turbulent reactive flow spanning all velocity regimes and fluids that is being developed for the new KIVA combustion algorithm, particularly for internal combustion engines. The method and general solver is applicable to Newtonian and non- Newtonian fluids and also for incompressible solids, porous media, solidification modeling, and fluid structure interaction problems. The fuel injection and injector modeling could easily benefit from the capability of solving the fluid structure interaction problem in an injector, helping to understand cycle to cycle variation and cavitation. This is just one example where the new algorithm differs from the old, in addition to handling Conjugate Heat Transfer (CHT), although there a numerous features that makes the new system more robust and accurate. In these ways, the PCS hp-adaptive algorithm does not compete with commercial software packages, those often used in conjunction with the currently distributed KIVA codes for engine combustion modeling. In addition, choosing a local ALE method on immersed moving parts represented by overset grid that is 2nd order spatially accurate, allows for easy grid generation from CAD to fluid grid while also provide for robustness in handling any possible moving parts configuration without any code modifications. The combined methods employed produce a minimal amount of computational effort as compared to fully resolved grids at the same accuracy. We demonstrate the solver on benchmark problems for the all flow regimes as follows: (1) 2-D backward-facing step using h-adaption, (2) 2-D driven cavity, (3) 2-D natural convection in a differentially heat cavity with h-adaptation, (4) NACA 0012 airfoil in 2-D, (5) supersonic flows over compression ramps, (6) 2-D natural convection in a differentially heat cavity with hp-adaptation, (7) 3-D natural convection in a differentially heat sphere with hp-adaptation. In addition, we show the new moving parts algorithm for working for a 2-D piston; the immersed moving parts method also for valves and pistons, vanes, etc... The movement is performed using an overset grid method and is 2nd order accurate in space, and never produces a tangle grid, that is, robust system at any resolution and any parts configuration. We also show CHT for the currently distributed KIVA-4mpi software and some fairly automatic grid generation using Sandia's Cubit unstructured grid generator. A new electronic web-based manual for KIVA-4 has been developed as well.

02 PETROLEUM↗

FDEM v.1.0

Algorithm FOEM numerically solves a system of six, coupled, first-order, linear, inhomogeneous partial differential equations (called the "EH-system") with an explicit, time-domain, finite-difference (FD) The six dependent variables three electric field vector components and three magnetic field FOEM vector components) are stored on staggered spatial and temporal grids. Discrete FO operators approximating the partial derivatives possess 2nd-order accuracy in time and Nth-order accuracy in space. The 3D spatial grid is rectangular, and grid intervals are uniform in each coordinate direction. FD formulae for updating the wavefield variables are appropriately non-dimensionalized and normalized so that single-precision arithmetic calculations yield adequate accuracy. This software will be used to generate electromagnetic responses for typical geological/geophysical models of interest in the petroleum, mining, geotechnical, geothermal, hydrological, environmental assessment, and underground storage industries. Additionally, US government agencies (including the military) involved in monitoring of explosions (chemical or nuclear) or remote surveillance of underground facilities may be interested in using algorithm FOEM. Finally, this software may be of use to organizations involved in fundamental research and applications development in the area of electromagnetic wave propagation (e.g. national laboratories, academic departments. industry).

Source record↗

A Positive-Definite, WENO-Limited, High-Order Finite Volume Solver for 2-D Transport on the Cubed Sphere Using an ADER Time Discretization

Modern computer architectures reward added computation if it reduces algorithmic dependence, reduces data movement, increases accuracy/robustness, and improves memory accesses. The driving motive for this study is to develop a numerical algorithm that respects these constraints while improving accuracy and robustness. This study introduces the ADER-DT (Arbitrary DERivatives in time and space-differential transform) time discretization to positive-definite, weighted essentially nonoscillatory (WENO)-limited, finite volume transport on the cubed sphere in lieu of semidiscrete integrators. The cost of the ADER-DT algorithm is significantly improved from previous implementations without affecting accuracy. A new function-based WENO implementation is also detailed for use with the ADER-DT time discretization. While ADER-DT costs about 1.5 times more than a fourth-order, five-stage strong stability preserving Runge-Kutta (SSPRK4) method, it is far more computationally dense (which is advantageous on accelerators such as graphics processing units), and it has a larger effective maximum stable time step. ADER-DT errors converge more quickly with grid refinement than SSPRK4, giving 6.5 times less error in the L ∞ norm than SSPRK4 at the highest refinement level for smooth data. For nonsmooth data, ADER-DT resolves C 0 discontinuities more sharply. For a complex flow field, ADER exhibits less phase error than SSPRK4. In conclusion, improving both accuracy and robustness as well as better respecting modern computational efficiency requirements, we believe the method presented herein is competitive for efficiently transporting tracers over the sphere for applications targeting modern computing architectures.

97 MATHEMATICS AND COMPUTING↗

KADATH: A spectral solver for theoretical physics

KADATH is a library that implements spectral methods in a very modular manner. It is designed to solve a wide class of problems that arise in the context of theoretical physics. Several types of coordinates are implemented and additional geometries can be easily encoded. Partial differential equations of various types are discretized by means of spectral methods. The resulting system is solved using a Newton-Raphson iteration. Doing so, KADATH is able to deal with strongly non-linear situations. The algorithms are validated by applying the library to four different problems of contemporary physics, in the fields of gauge field theory and general relativity.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Efficient ptychographic phase retrieval via a matrix-free Levenberg-Marquardt algorithm

The phase retrieval problem, where one aims to recover a complex-valued image from far-field intensity measurements, is a classic problem encountered in a range of imaging applications. Modern phase retrieval approaches usually rely on gradient descent methods in a nonlinear minimization framework. Calculating closed-form gradients for use in these methods is tedious work, and formulating second order derivatives is even more laborious. Additionally, second order techniques often require the storage and inversion of large matrices of partial derivatives, with memory requirements that can be prohibitive for data-rich imaging modalities. We use a reverse-mode automatic differentiation (AD) framework to implement an efficient matrix-free version of the Levenberg-Marquardt (LM) algorithm, a longstanding method that finds popular use in nonlinear least-square minimization problems but which has seen little use in phase retrieval. Furthermore, we extend the basic LM algorithm so that it can be applied for more general constrained optimization problems (including phase retrieval problems) beyond just the least-square applications. Since we use AD, we only need to specify the physics-based forward model for a specific imaging application; the first and second-order derivative terms are calculated automatically through matrix-vector products, without explicitly forming the large Jacobian or Gauss-Newton matrices typically required for the LM method. We demonstrate that this algorithm can be used to solve both the unconstrained ptychographic object retrieval problem and the constrained “blind” ptychographic object and probe retrieval problems, under the popular Gaussian noise model as well as the Poisson noise model. We compare this algorithm to state-of-the-art first order ptychographic reconstruction methods to demonstrate empirically that this method outperforms best-in-class first-order methods: it provides excellent convergence guarantees with (in many cases) a superlinear rate of convergence, all with a computational cost comparable to, or lower than, the tested first-order algorithms.

36 MATERIALS SCIENCE↗

Measurement of inclusive jet and dijet cross-sections in proton-proton collisions at $$ \sqrt{s}=13 $$ TeV with the ATLAS detector

Inclusive jet and dijet cross-sections are measured in proton-proton collisions at a centre-of-mass energy of 13 TeV. The measurement uses a dataset with an integrated luminosity of 3.2 fb -1 recorded in 2015 with the ATLAS detector at the Large Hadron Collider. Jets are identified using the anti- k t algorithm with a radius parameter value of R = 0 . 4. The inclusive jet cross-sections are measured double-differentially as a function of the jet transverse momentum, covering the range from 100 GeV to 3.5 TeV, and the absolute jet rapidity up to | y | = 3. The double-differential dijet production cross-sections are presented as a function of the dijet mass, covering the range from 300 GeV to 9 TeV, and the half absolute rapidity separation between the two leading jets within | y | < 3, y * , up to y * = 3. Next-to-leading-order, and next-to-next-to-leading-order for the inclusive jet measurement, perturbative QCD calculations corrected for non-perturbative and electroweak effects are compared to the measured cross-sections.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

A collaborative detection approach for internal steam leakage of tyre vulcanization workshop with artificial immune algorithm

In manufacturing industry, internal leakage of steam trap usually results in great steam waste. In particular, internal steam leakage in tyre vulcanization workshop has significant influences on its production safety and energy efficiency. In practice, internal steam leakage problem often is ignored and it tends to be difficult to detect this leakage due to lack of comprehensive flow meter or method. This paper presents a collaborative detection method for this problem in tyre vulcanization workshop with artificial immune algorithm. In the method, internal leakage of steam trap is defined as nonself antigens, and steam pressure differentials between steam pipe and steam rooms (or bladders) are extracted as epitopes of antigens. Furthermore, periodic energy efficiency and steam pressure of vulcanizer as the danger signals are simultaneously detected. Energy efficiency is represented by damaged cells which will be identified firstly for locating the leaking steam straps through antibodies. Furthermore, the self-adaptive danger thresholds for energy efficiency are evaluated through a steam consumption model and the Levenberg–Marquardt back propagation algorithm. An immune-based clustering algorithm aiNet is then adopted to generate antibodies (detectors) for detection on the steam pressure of vulcanizer. Finally, a case study is implemented to validate this method, which shows that collaborative detection method allows locating the specific leaking steam trap and is a feasible tool to reduce steam waste and ensure the safety of steam supply.

42 ENGINEERING↗

Improving ADAM through an implicit-explicit (IMEX) time-stepping approach

The ADAM optimizer, often used in machine learning for neural network training, corresponds to an underlying ordinary differential equation (ODE) in the limit of very small learning rates. Here, this work shows that the classical ADAM algorithm is a first-order implicit-explicit (IMEX) Euler discretization of the underlying ODE. Employing the time discretization point of view, we propose new extensions of the ADAM scheme obtained by using higher-order IMEX methods to solve the ODE. Based on this approach, we derive a new optimization algorithm for neural network training that performs better than classical ADAM on several regression and classification problems.

97 MATHEMATICS AND COMPUTING↗

A method for smoothing multiple yield functions

A method is presented that produces a single C 2 differentiable and convex yield function from a plastic model that contains multiple yield surfaces that are individually C 2 differentiable and convex. The method is suitable for implementation in numerical computer codes. Here, the error incurred through the smoothing procedure is quantified. The method is illustrated through a number of examples. Two simple finite-element models are solved to compare this method with existing approaches.

42 ENGINEERING↗

Population parametrization of costly black box models using iterations between SAEM algorithm and kriging

In this article we focus on parametrization of black box models from repeated measurements among several individuals (population parametrization). We introduce a variant of the SAEM algorithm, called KSAEM algorithm, which couples the standard SAEM algorithm with the dynamic construction of an approximate metamodel. The costly evaluation of the genuine black box is replaced by a kriging step, using a basis of precomputed values, basis which is enlarged during SAEM algorithm to improve the accuracy of the metamodel in regions of interest.

97 MATHEMATICS AND COMPUTING↗

On convergent finite difference schemes for variational–PDE-based image processing

We study an adaptive anisotropic Huber functional-based image restoration scheme. Using a combination of L2–L1 regularization functions, an adaptive Huber functional-based energy minimization model provides denoising with edge preservation in noisy digital images. We study a convergent finite difference scheme based on continuous piecewise linear functions and use a variable splitting scheme, namely the Split Bregman (In: Goldstein and Osher, SIAM J Imaging Sci 2(2):323–343, 2009) algorithm, to obtain the discrete minimizer. Experimental results are given in image denoising and comparison with additive operator splitting, dual fixed point, and projected gradient schemes illustrates that the best convergence rates are obtained for our algorithm.

97 MATHEMATICS AND COMPUTING↗

Direct pore-level modeling of incompressible fluid flow in porous media

We present a dynamic particle-based model for direct pore-level modeling of incompressible viscous fluid flow in disordered porous media. The model is capable of simulating flow directly in three-dimensional high-resolution micro-CT images of rock samples. It is based on moving particle semi-implicit (MPS) method. We modify this technique in order to improve its stability for flow in porous media problems. Using the micro-CT image of a rock sample, the entire medium, i.e., solid and fluid, is discretized into particles. The incompressible Navier-Stokes equations are then solved for each particle using the MPS summations. The model handles highly irregular fluid-solid boundaries effectively. An algorithm to split and merge fluid particles is also introduced. To handle the computational load, we present a parallel version of the model that runs on distributed memory computer clusters. The accuracy of the model is validated against the analytical, numerical, and experimental data available in the literature. The validated model is then used to simulate both unsteady- and steady-state flow of an incompressible fluid directly in a representative elementary volume (REV) size micro-CT image of a naturally-occurring sandstone with 3.398 {mu}m resolution. We analyze the quality and consistency of the predicted flow behavior and calculate absolute permeability using the steady-state flow rate.

97 MATHEMATICS AND COMPUTING↗

Semiclassical Analysis of Constrained Quantum Systems

Exact procedures that follow Dirac's constraint quantization of gauge theories are usually technically involved and often difficult to implement in practice. We overview an 'effective' scheme for obtaining the leading order semiclassical corrections to the dynamics of constrained quantum systems developed elsewhere. Motivated by the geometrical view of quantum mechanics, our method mimics the classical Dirac-Bergmann algorithm and avoids direct reference to a particular representation of the physical Hilbert space. We illustrate the procedure through the example of a relativistic particle in Minkowski spacetime.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Aerodynamic optimization of supersonic compressor cascade using differential evolution on GPU

Differential Evolution (DE) is a powerful stochastic optimization method. Compared to gradient-based algorithms, DE is able to avoid local minima but requires at the same time more function evaluations. In turbomachinery applications, function evaluations are performed with time-consuming CFD simulation, which results in a long, non affordable, design cycle. Modern High Performance Computing systems, especially Graphic Processing Units (GPUs), are able to alleviate this inconvenience by accelerating the design evaluation itself. In this work we present a validated CFD Solver running on GPUs, able to accelerate the design evaluation and thus the entire design process. An achieved speedup of 20x to 30x enabled the DE algorithm to run on a high-end computer instead of a costly large cluster. The GPU-enhanced DE was used to optimize the aerodynamics of a supersonic compressor cascade, achieving an aerodynamic loss minimization of 20%.

97 MATHEMATICS AND COMPUTING↗

Highly parallelisable simulations of time-dependent viscoplastic fluid flow with structured adaptive mesh refinement

We present the extension of an efficient and highly parallelisable framework for incompressible fluid flow simulations to viscoplastic fluids. The system is governed by incompressible conservation of mass, the Cauchy momentum equation, and a generalised Newtonian constitutive law. In order to simulate a wide range of viscoplastic fluids, we employ the Herschel-Bulkley model for yield-stress fluids with nonlinear stress-strain dependency above the yield limit. We utilise Papanastasiou regularisation in our algorithm to deal with the singularity in apparent viscosity. The resulting system of partial differential equations is solved using the IAMR (Incompressible Adaptive Mesh Refinement) code, which uses second-order Godunov methodology for the advective terms and semi-implicit diffusion in the context of an approximate projection method to solve adaptively refined meshes. By augmenting the IAMR code with the ability to simulate regularised Herschel-Bulkley fluids, we obtain efficient numerical software for time-dependent viscoplastic flow in three dimensions, which can be used to investigate systems not considered previously due to computational expense. We validate results from simulations using this new capability against previously published data for Bingham plastics and power-law fluids in the two-dimensional lid-driven cavity. In doing so, we expand the range of Bingham and Reynolds numbers which have been considered in the benchmark tests. Moreover, extensions to time-dependent flow of Herschel-Bulkley fluids and three spatial dimensions offer new insights into the flow of viscoplastic fluids in this test case, and we provide missing benchmark results for these extensions.

42 ENGINEERING↗

The nonexistence of a paddlewheel effect in superionic conductors

Since the 1980s, the paddlewheel effect has been suggested as a mechanism to boost lithium-ion diffusion in inorganic materials via the rotation of rotor-like anion groups. However, it remains unclear whether the paddlewheel effect, defined as large-angle anion group rotations assisting Li hopping, indeed exists; furthermore, the physical mechanism by which the anion-group dynamics affect lithium-ion diffusion has not yet been established. In this work, we differentiate various types of rotational motions of anion groups and develop quaternion-based algorithms to detect, quantify, and relate them to lithium-ion motion in ab initio molecular dynamics simulations. Our analysis demonstrates that, in fact, the paddlewheel effect, where an anion group makes a large angle rotation to assist a lithium-ion hop, does not exist and thus is not responsible for the fast lithium-ion diffusion in superionic conductors, as historically claimed. Instead, we find that materials with topologically isolated anion groups can enhance lithium-ion diffusivity via a more classic nondynamic soft-cradle mechanism, where the anion groups tilt to provide optimal coordination to a lithium ion throughout the hopping process to lower the migration barrier. This anion-group disorder is static in nature, rather than dynamic and can explain most of the experimental observations. Our work substantiates the nonexistence of the long-debated paddlewheel effect and clarifies any correlation that may exist between anion-group rotations and fast ionic diffusion in inorganic materials.

36 MATERIALS SCIENCE↗