Search NASA⌕ Search

SEARCH · Search NASA

Results for “mathematical regularization”

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 163 records · Page 9

Optimal Control of Heterogeneous Systems with Endogenous Domain of Heterogeneity

The paper deals with optimal control of heterogeneous systems, that is, families of controlled ODEs parameterized by a parameter running over a domain called domain of heterogeneity. The main novelty in the paper is that the domain of heterogeneity is endogenous: it may depend on the control and on the state of the system. This extension is crucial for several economic applications and turns out to rise interesting mathematical problems. A necessary optimality condition is derived, where one of the adjoint variables satisfies a differential inclusion (instead of equation) and the maximization of the Hamiltonian takes the form of 'min-max'. As a consequence, a Pontryagin-type maximum principle is obtained under certain regularity conditions for the optimal control. A formula for the derivative of the objective function with respect to the control from L{sub {infinity}} is presented together with a sufficient condition for its existence. A stylized economic example is investigated analytically and numerically.

97 MATHEMATICS AND COMPUTING↗

Dealing with Measurement Uncertainties as Nuisance Parameters in Bayesian Model Calibration

In the presence of model discrepancy, the calibration of physics-based models for physical parameter inference is a challenging problem. Lack of identifiability between calibration parameters and model discrepancy requires additional identifiability constraints to be placed on the model discrepancy to obtain unique physical parameter estimates. If these assumptions are violated, the inference for the calibration parameters can be systematically biased. In many applications, such as in dynamic material property experiments, many of the calibration inputs refer to measurement uncertainties. Here, we develop a metric for identifying overfitting of these measurement uncertainties, propose a prior capable of reducing this overfitting, and show how this leads to a diagnostic tool for validation of physical parameter inference. The approach is demonstrated for a benchmark example and applied for a material property application to perform inference on the equation of state parameters of tantalum.

97 MATHEMATICS AND COMPUTING↗

Boosting the Performance of Plug-and-Play Priors via Denoiser Scaling

Plug-and-play priors (PnP) is an image reconstruction framework that utilizes an image denoiser as an imaging prior. Unlike traditional regularized inversion, PnP does not require the prior to be expressible in the form of a regularization function. This flexibility enables PnP algorithms to exploit the most effective image denoisers, leading to their state-of-the-art performance in various imaging tasks. However, many powerful denoisers, such as the ones based on convolutional neural networks (CNNs), do not have tunable parameters that would allow controlling their influence within PnP. To address this issue, in this paper, we introduce a scaling parameter that adjusts the magnitude of the denoiser input and output. We theoretical justify the denoiser scaling from the perspectives of proximal optimization, statistical estimation, and consensus equilibrium. Finally, we provide numerical experiments demonstrating the ability of denoiser scaling to systematically improve the performance of PnP for denoising CNN priors that do not have explicitly tunable parameters.

Boosting↗

Multilevel-in-width training for deep neural network regression

A common challenge in regression is that for many problems, the degrees of freedom required for a high-quality solution also allows for overfitting. Regularization is a class of strategies that seek to restrict the range of possible solutions so as to discourage overfitting while still enabling good solutions, and different regularization strategies impose different types of restrictions. In this paper, we present a multilevel regularization strategy that constructs and trains a hierarchy of neural networks, each of which has layers that are wider versions of the previous network's layers. Here, we draw intuition and techniques from the field of Algebraic Multigrid (AMG), traditionally used for solving linear and nonlinear systems of equations, and specifically adapt the Full Approximation Scheme (FAS) for nonlinear systems of equations to the problem of deep learning. Training through V-cycles then encourage the neural networks to build a hierarchical understanding of the problem. We refer to this approach as multilevel-in-width to distinguish from prior multilevel works which hierarchically alter the depth of neural networks. The resulting approach is a highly flexible framework that can be applied to a variety of layer types, which we demonstrate with both fully connected and convolutional layers. We experimentally show with PDE regression problems that our multilevel training approach is an effective regularizer, improving the generalize performance of the neural networks studied.

97 MATHEMATICS AND COMPUTING↗

An inexact regularized Newton framework with a worst-case iteration complexity of $ {\mathscr O}(\varepsilon^{-3/2}) $ for nonconvex optimization

Abstract An algorithm for solving smooth nonconvex optimization problems is proposed that, in the worst-case, takes ${\mathscr O}(\varepsilon ^{-3/2})$ iterations to drive the norm of the gradient of the objective function below a prescribed positive real number $\varepsilon $ and can take ${\mathscr O}(\varepsilon ^{-3})$ iterations to drive the leftmost eigenvalue of the Hessian of the objective above $-\varepsilon $. The proposed algorithm is a general framework that covers a wide range of techniques including quadratically and cubically regularized Newton methods, such as the Adaptive Regularization using Cubics (arc) method and the recently proposed Trust-Region Algorithm with Contractions and Expansions (trace). The generality of our method is achieved through the introduction of generic conditions that each trial step is required to satisfy, which in particular allows for inexact regularized Newton steps to be used. These conditions center around a new subproblem that can be approximately solved to obtain trial steps that satisfy the conditions. A new instance of the framework, distinct from arc and trace, is described that may be viewed as a hybrid between quadratically and cubically regularized Newton methods. Numerical results demonstrate that our hybrid algorithm outperforms a cubically regularized Newton method.

Mathematics↗

Alternatives to Contour Visualizations for Power Systems Data

Electrical grids are geographical and topological structures whose voltage states are challenging to represent accurately and efficiently for visual analysis. The current common practice is to use colored contour maps, yet these can misrepresent the data. We examine the suitability of four alternative visualization methods for depicting voltage data in a geographically dense distribution system-Voronoi polygons, H3 tessellations, S2 tessellations, and a network-weighted contour map. We find that Voronoi tessellations and network-weighted contour maps more accurately represent the statistical distribution of the data than regular contour maps.

MATHEMATICS AND COMPUTING↗

Space-Split Algorithm for Sensitivity Analysis of Discrete Chaotic Systems With Multidimensional Unstable Manifolds

Accurate approximations of the change of a system's output and its statistics with respect to the input are highly desired in computational dynamics. Ruelle's linear response theory provides breakthrough mathematical machinery for computing the linear response of chaotic dynamical systems. In this paper, we propose an algorithm for sensitivity analysis of discrete chaos with an arbitrary number of positive Lyapunov exponents. We combine the concept of perturbation space-splitting, which regularizes Ruelle's original expression, together with measure-based parameterization of the expanding subspace. We use these tools to rigorously derive trajectory-following recursive relations that converge exponentially fast, and construct a memory-efficient Monte Carlo scheme for derivatives of the output statistics. Thanks to the regularization and lack of simplifying assumptions on the system's behavior, our method is immune to the common problems of other popular methods such as the exploding tangent solutions and unphysical shadowing directions. Here, we provide a ready-to-use algorithm, analyze its complexity, and demonstrate several numerical examples of sensitivity computation using physically-inspired low-dimensional systems.

97 MATHEMATICS AND COMPUTING↗

Enumeration of Regular Maps on Surfaces of a Given Genus

We enumerate the labelled and unlabelled d-regular maps on two-dimensional oriented surfaces of arbitrary genus g. The case of d-regular maps with a single face is considered separately in more detail.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Regularity of a Boundary Point for the p(x)-Laplacian

We study the behavior of solutions to the Dirichlet problem for the p(x)-Laplacian with a continuous boundary function. We prove the existence of a weak solution under the assumption that p is separated from 1 and ∞. We present a necessary and sufficient Wiener type condition for regularity of a boundary point provided that the exponent p has the logarithmic modulus of continuity at this point.

97 MATHEMATICS AND COMPUTING↗

A comprehensive and fair comparison of two neural operators (with practical extensions) based on $\mathrm{FAIR}$ data

Neural operators can learn nonlinear mappings between function spaces and offer a new simulation paradigm for real-time prediction of complex dynamics for realistic diverse applications as well as for system identification in science and engineering. Herein, we investigate the performance of two neural operators, which have shown promising results so far, and we develop new practical extensions that will make them more accurate and robust and importantly more suitable for industrial-complexity applications. The first neural operator, DeepONet, was published in 2019 (Lu et al., 2019), and its original architecture was based on the universal approximation theorem of Chen & Chen (1995). The second one, named Fourier Neural Operator or FNO, was published in 2020, and it is based on parameterizing the integral kernel in the Fourier space. DeepONet is represented by a summation of products of neural networks (NNs), corresponding to the branch NN for the input function and the trunk NN for the output function; both NNs are general architectures, e.g., the branch NN can be replaced with a CNN or a ResNet. According to Kovachki et al. (2021), FNO in its continuous form can be viewed conceptually as a DeepONet with a specific architecture of the branch NN and a trunk NN represented by a trigonometric basis. In order to compare FNO with DeepONet computationally for realistic setups, we develop several extensions of FNO that can deal with complex geometric domains as well as mappings where the input and output function spaces are of different dimensions. We also develop an extended DeepONet with special features that provide inductive bias and accelerate training, and we present a faster implementation of DeepONet with cost comparable to the computational cost of FNO, which is based on the Fast Fourier Transform. Here we consider 16 different benchmarks to demonstrate the relative performance of the two neural operators, including instability wave analysis in hypersonic boundary layers, prediction of the vorticity field of a flapping airfoil, porous media simulations in complex-geometry domains, etc. We follow the guiding principles of FAIR (Findability, Accessibility, Interoperability, and Reusability) for scientific data management and stewardship. The performance of DeepONet and FNO is comparable for relatively simple settings, but for complex geometries the performance of FNO deteriorates greatly. We also compare theoretically the two neural operators and obtain similar error estimates for DeepONet and FNO under the same regularity assumptions.

42 ENGINEERING↗

Boundary values of the Schwarzian derivative of a regular function

Regular functions f in the half-plane Imz>0 admitting an asymptotic expansion f(z)=c{sub 1}z+c{sub 2}z{sup 2}+c{sub 3}z{sup 3}+{gamma}(z)z{sup 3}, where c{sub 1}>0, Imc{sub 2}=0 and the angular limit <lim{sub z{yields}0{gamma}}(z)=0, are considered. For various conditions on the function f inequalities for the real part of the Schwarzian derivative S{sub f}(0)=6(c{sub 3}/c{sub 1}-c{sub 2}{sup 2}/c{sub 1}{sup 2}) are established. These inequalities complement and refine some known versions of Schwarz's lemma. The results obtained are close to the well-known Burns-Krantz rigidity theorem on regular self-maps and its generalizations due to Tauraso, Vlacci and Shoikhet. Bibliography: 16 titles.

97 MATHEMATICS AND COMPUTING↗

Chaotic dynamics in a periodically driven spin-1 condensate

We use periodically modulated magnetic fields to drive spin-1 Bose-Einstein condensates (BECs) and study the corresponding spin-mixing dynamics. Due to the time-dependent driving, this system permits chaotic dynamics depending on the drive parameters, which could not occur in previous studies. From the investigation of the Poincare sections, we find there exist complex trajectories in the phase space, leading to very complicated structures of the phase space with mixed regular and chaotic regions. By calculating the quasienergy levels of the corresponding Floquet operators, the signatures of quantum chaos are also found in this system. The level spacing distribution is very close to the Poisson distribution or Wigner distribution when the corresponding classical dynamics is regular or chaotic.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

ON THE MODE OF DYNAMO ACTION IN A GLOBAL LARGE-EDDY SIMULATION OF SOLAR CONVECTION

In this paper, we examine the mode of dynamo action in the implicit large-eddy magnetohydrodynamical simulation of solar convection reported upon in Ghizaru et al. Motivated by the presence of a strong and well-defined large-scale axisymmetric magnetic component undergoing regular polarity reversals, we define the fluctuating component of the magnetic field as the difference between the total field and its zonal average. The subsequent analysis follows the physical logic and mathematical formulation of mean-field electrodynamics, whereby a turbulent electromotive force (EMF) is computed by the suitable averaging of cross-correlations between fluctuating flow and field components and expressed in terms of the mean field via a linear truncated tensorial expansion. We use singular value decomposition to perform a linear least-squares fit of the temporal variation of the EMF to that of the large-scale magnetic component, which yields the components of the full {alpha}-tensor. Its antisymmetric component, describing general turbulent pumping, is also extracted. The {alpha}-tensor so calculated reproduces a number of features already identified in local, Cartesian simulations of magnetohydrodynamical rotating convection, including an {alpha}{sub {phi}{phi}} component positive in the northern solar hemisphere, peaking at high latitudes, and reversing sign near the bottom of the convection zone; downward turbulent pumping throughout the convecting layer; and significant equatorward turbulent pumping at mid latitudes, and poleward at high latitudes in subsurface layers. We also find that the EMF contributes significantly to the regeneration of the large-scale toroidal magnetic component, which from the point of view of mean-field dynamo models would imply that the simulation operates as an {alpha}{sup 2}{Omega} dynamo. We find little significant evidence of {alpha}-quenching by the large-scale magnetic field. The amplitude of the magnetic cycle appears instead to be regulated primarily by a magnetically driven reduction of the differential rotation.

79 ASTRONOMY AND ASTROPHYSICS↗

A Relaxation Approach to Vector-Valued Allen–Cahn MPEC Problems

In this paper we consider a vector-valued Allen–Cahn MPEC problem. To derive optimality conditions we exploit a regularization–relaxation technique. The optimality system of the regularized–relaxed subproblems are investigated by applying the classical result of Zowe and Kurcyusz. Finally we show that the stationary points of the regularized–relaxed subproblems converge to weak stationary points of the limit problem.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

On the Equality of Values in the Monge and Kantorovich Problems

This paper is concerned with the study of conditions under which the Monge and Kantorovich problems with a continuous cost function on a product of two completely regular spaces and two given atomless Radon measures-projections on these spaces have equal values of the corresponding infima.

97 MATHEMATICS AND COMPUTING↗

The exact discontinuity of a partial wave along the left-hand cut and the exact N∕D method in non-relativistic scattering

We first deduce the analytical continuation in the complex planes of the initial and final three-momenta of the Lippmann–Schwinger equation in coupled or uncoupled partial-wave amplitudes. This result allows us to deduce a master equation whose solution is the exact discontinuity of the on-shell partial-wave amplitudes along the left-hand cut. This equation is always a linear non-singular integral equation whose solution is fixed exclusively by the knowledge of the potential, applicable to either regular or singular potentials. The capability of calculating exactly this discontinuity allows one to settle the exact N∕D method in two-body non-relativistic scattering for coupled and uncoupled waves. We exemplify this new advance in scattering theory by explicitly checking the agreement between the Lippmann–Schwinger equation with the corresponding solutions of the exact N∕D method for some examples that involve regular and singular potentials, either attractive or repulsive.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Whittaker-Hill equation and semifinite-gap Schroedinger operators

A periodic one-dimensional Schroedinger operator is called semifinite gap if every second gap in its spectrum is eventually closed. We construct explicit examples of semifinite-gap Schroedinger operators in trigonometric functions by applying Darboux transformations to the Whittaker-Hill equation. We give a criterion for the regularity of the corresponding potentials and investigate the spectral properties of the new operators.

97 MATHEMATICS AND COMPUTING↗