arXiv subjects

Stefan M. Wild

Publications and source records attributed to Stefan M. Wild.

53 records · Page 3Linked to original sources

arXiv2020

Tuning Multigrid Methods with Robust Optimization

Local Fourier analysis is a useful tool for predicting and analyzing the performance of many efficient algorithms for the solution of discretized PDEs, such as multigrid and domain decomposition methods. The crucial aspect of local Fourier analysis is that it can be used to minimize an estimate of the spectral radius of a stationary iteration, or the condition number of a preconditioned system, in terms of a symbol representation of the algorithm. In practice, this is a "minimax" problem, minimizing with respect to solver parameters the appropriate measure of work, which involves maximizing over the Fourier frequency. Often, several algorithmic parameters may be determined by local Fourier analysis in order to obtain efficient algorithms. Analytical solutions to minimax problems are rarely possible beyond simple problems; the status quo in local Fourier analysis involves grid sampling, which is prohibitively expensive in high dimensions. In this paper, we propose and explore optimization algorithms to solve these problems efficiently. Several examples, with known and unknown analytical solutions, are presented to show the effectiveness of these approaches.

math.OC↗

arXiv2020

A Method for Convex Black-Box Integer Global Optimization

We study the problem of minimizing a convex function on a nonempty, finite subset of the integer lattice when the function cannot be evaluated at noninteger points. We propose a new underestimator that does not require access to (sub)gradients of the objective but, rather, uses secant linear functions that interpolate the objective function at previously evaluated points. These linear mappings are shown to underestimate the objective in disconnected portions of the domain. Therefore, the union of these conditional cuts provides a nonconvex underestimator of the objective. We propose an algorithm that alternates between updating the underestimator and evaluating the objective function. We prove that the algorithm converges to a global minimum of the objective function on the feasible set. We present two approaches for representing the underestimator and compare their computational effectiveness. We also compare implementations of our algorithm with existing methods for minimizing functions on a subset of the integer lattice. We discuss the difficulty of this problem class and provide insights into why a computational proof of optimality is challenging even for moderate problem sizes.

math.OC↗

arXiv2019

Adaptive Sampling Quasi-Newton Methods for Derivative-Free Stochastic Optimization

We consider stochastic zero-order optimization problems, which arise in settings from simulation optimization to reinforcement learning. We propose an adaptive sampling quasi-Newton method where we estimate the gradients of a stochastic function using finite differences within a common random number framework. We employ modified versions of a norm test and an inner product quasi-Newton test to control the sample sizes used in the stochastic approximations. We provide preliminary numerical experiments to illustrate potential performance benefits of the proposed method.

math.OC↗

arXiv2019

Derivative-free optimization methods

In many optimization problems arising from scientific, engineering and artificial intelligence applications, objective and constraint functions are available only as the output of a black-box or simulation oracle that does not provide derivative information. Such settings necessitate the use of methods for derivative-free, or zeroth-order, optimization. We provide a review and perspectives on developments in these methods, with an emphasis on highlighting recent developments and on unifying treatment of such problems in the non-linear optimization and machine learning literature. We categorize methods based on assumed properties of the black-box functions, as well as features of the methods. We first overview the primary setting of deterministic methods applied to unconstrained, non-convex optimization problems where the objective function is defined by a deterministic black-box oracle. We then discuss developments in randomized methods, methods that assume some additional structure about the objective (including convexity, separability and general non-smooth compositions), methods for problems where the output of the black-box oracle is stochastic, and methods for handling different types of constraints.

math.OC↗

arXiv2019

Simultaneous Sensing Error Recovery and Tomographic Inversion Using an Optimization-based Approach

Tomography can be used to reveal internal properties of a 3D object using any penetrating wave. Advanced tomographic imaging techniques, however, are vulnerable to both systematic and random errors associated with the experimental conditions, which are often beyond the capabilities of the state-of-the-art reconstruction techniques such as regularizations. Because they can lead to reduced spatial resolution and even misinterpretation of the underlying sample structures, these errors present a fundamental obstacle to full realization of the capabilities of next-generation physical imaging. In this work, we develop efficient and explicit recovery schemes of the most common experimental error: movement of the center of rotation during the experiment. We formulate new physical models to capture the experimental setup, and we devise new mathematical optimization formulations for reliable inversion of complex samples. We demonstrate and validate the efficacy of our approach on synthetic data under known perturbations of the center of rotation.

math.NA↗

arXiv2018

Comparing optimization strategies for force field parameterization

Classical molecular dynamics (MD) simulations enable modeling of materials and examination of microscopic details that are not accessible experimentally. The predictive capability of MD relies on the force field (FF) used to describe interatomic interactions. FF parameters are typically determined to reproduce selected material properties computed from density functional theory (DFT) and/or measured experimentally. A common practice in parameterizing FFs is to use least-squares local minimization algorithms. Genetic algorithms (GAs) have also been demonstrated as a viable global optimization approach, even for complex FFs. However, an understanding of the relative effectiveness and efficiency of different optimization techniques for the determination of FF parameters is still lacking. In this work, we evaluate various FF parameter optimization schemes, using as example a training data set calculated from DFT for different polymorphs of Ir$O_2$. The Morse functional form is chosen for the pairwise interactions and the optimization of the parameters against the training data is carried out using (1) multi-start local optimization algorithms: Simplex, Levenberg-Marquardt, and POUNDERS, (2) single-objective GA, and (3) multi-objective GA. Using random search as a baseline, we compare the algorithms in terms of reaching the lowest error, and number of function evaluations. We also compare the effectiveness of different approaches for FF parameterization using a test data set with known ground truth (i.e generated from a specific Morse FF). We find that the performance of optimization approaches differs when using the Test data vs. the DFT data. Overall, this study provides insight for selecting a suitable optimization method for FF parameterization, which in turn can enable more accurate prediction of material properties and chemical phenomena.

physics.comp-ph↗

arXiv2018

Robust Learning of Trimmed Estimators via Manifold Sampling

We adapt a manifold sampling algorithm for the nonsmooth, nonconvex formulations of learning that arise when imposing robustness to outliers present in the training data. We demonstrate the approach on objectives based on trimmed loss. Empirical results show that the method has favorable scaling properties. Although savings in time come at the expense of not certifying optimality, the algorithm consistently returns high-quality solutions on the trimmed linear regression and multiclass classification problems tested.

math.OC↗

arXiv2016

Doing Moore with Less -- Leapfrogging Moore's Law with Inexactness for Supercomputing

Energy and power consumption are major limitations to continued scaling of computing systems. Inexactness, where the quality of the solution can be traded for energy savings, has been proposed as an approach to overcoming those limitations. In the past, however, inexactness necessitated the need for highly customized or specialized hardware. The current evolution of commercial off-the-shelf(COTS) processors facilitates the use of lower-precision arithmetic in ways that reduce energy consumption. We study these new opportunities in this paper, using the example of an inexact Newton algorithm for solving nonlinear equations. Moreover, we have begun developing a set of techniques we call reinvestment that, paradoxically, use reduced precision to improve the quality of the computed result: They do so by reinvesting the energy saved by reduced precision.

cs.OH↗

arXiv2016

Origins and optimization of entanglement in plasmonically coupled quantum dots

A system of two or more quantum dots interacting with a dissipative plasmonic nanostructure is investigated in detail by using a cavity quantum electrodynamics approach with a model Hamiltonian. We focus on determining and understanding system configurations that generate multiple bipartite quantum entanglements between the occupation states of the quantum dots. These configurations include allowing for the quantum dots to be asymmetrically coupled to the plasmonic system. Analytical solution of a simplified limit for an arbitrary number of quantum dots and numerical simulations and optimization for the two- and three-dot cases are used to develop guidelines for maximizing the bipartite entanglements. For any number of quantum dots, we show that through simple starting states and parameter guidelines, one quantum dot can be made to share a strong amount of bipartite entanglement with all other quantum dots in the system, while entangling all other pairs to a lesser degree.

quant-ph↗

arXiv2016

Bayesian optimization under mixed constraints with a slack-variable augmented Lagrangian

An augmented Lagrangian (AL) can convert a constrained optimization problem into a sequence of simpler (e.g., unconstrained) problems, which are then usually solved with local solvers. Recently, surrogate-based Bayesian optimization (BO) sub-solvers have been successfully deployed in the AL framework for a more global search in the presence of inequality constraints; however, a drawback was that expected improvement (EI) evaluations relied on Monte Carlo. Here we introduce an alternative slack variable AL, and show that in this formulation the EI may be evaluated with library routines. The slack variables furthermore facilitate equality as well as inequality constraints, and mixtures thereof. We show how our new slack "ALBO" compares favorably to the original. Its superiority over conventional alternatives is reinforced on several mixed constraint examples.

stat.CO↗

arXiv2015

A Taxonomy of Constraints in Simulation-Based Optimization

The types of constraints encountered in black-box and simulation-based optimization problems differ significantly from those treated in nonlinear programming. We introduce a characterization of constraints to address this situation. We provide formal definitions for several constraint classes and present illustrative examples in the context of the resulting taxonomy. This taxonomy, denoted QRAK, is useful for modeling and problem formulation, as well as optimization software development and deployment. It can also be used as the basis for a dialog with practitioners in moving problems to increasingly solvable branches of optimization.

math.OC↗

arXiv2015

Modeling an Augmented Lagrangian for Blackbox Constrained Optimization

Constrained blackbox optimization is a difficult problem, with most approaches coming from the mathematical programming literature. The statistical literature is sparse, especially in addressing problems with nontrivial constraints. This situation is unfortunate because statistical methods have many attractive properties: global scope, handling noisy objectives, sensitivity analysis, and so forth. To narrow that gap, we propose a combination of response surface modeling, expected improvement, and the augmented Lagrangian numerical optimization framework. This hybrid approach allows the statistical model to think globally and the augmented Lagrangian to act locally. We focus on problems where the constraints are the primary bottleneck, requiring expensive simulation to evaluate and substantial modeling effort to map out. In that context, our hybridization presents a simple yet effective solution that allows existing objective-oriented statistical approaches, like those based on Gaussian process surrogates and expected improvement heuristics, to be applied to the constrained setting with minor modification. This work is motivated by a challenging, real-data benchmark problem from hydrology where, even with a simple linear objective function, learning a nontrivial valid region complicates the search for a global minimum.

stat.ME↗

arXiv2014

A Bayesian Approach for Parameter Estimation and Prediction using a Computationally Intensive Model

Bayesian methods have been very successful in quantifying uncertainty in physics-based problems in parameter estimation and prediction. In these cases, physical measurements y are modeled as the best fit of a physics-based model $η(θ)$ where $θ$ denotes the uncertain, best input setting. Hence the statistical model is of the form $y = η(θ) + ε$, where $ε$ accounts for measurement, and possibly other error sources. When non-linearity is present in $η(\cdot)$, the resulting posterior distribution for the unknown parameters in the Bayesian formulation is typically complex and non-standard, requiring computationally demanding computational approaches such as Markov chain Monte Carlo (MCMC) to produce multivariate draws from the posterior. While quite generally applicable, MCMC requires thousands, or even millions of evaluations of the physics model $η(\cdot)$. This is problematic if the model takes hours or days to evaluate. To overcome this computational bottleneck, we present an approach adapted from Bayesian model calibration. This approach combines output from an ensemble of computational model runs with physical measurements, within a statistical formulation, to carry out inference. A key component of this approach is a statistical response surface, or emulator, estimated from the ensemble of model runs. We demonstrate this approach with a case study in estimating parameters for a density functional theory (DFT) model, using experimental mass/binding energy measurements from a collection of atomic nuclei. We also demonstrate how this approach produces uncertainties in predictions for recent mass measurements obtained at Argonne National Laboratory (ANL).

physics.data-an↗

arXiv2014

Derivative-free optimization for parameter estimation in computational nuclear physics

We consider optimization problems that arise when estimating a set of unknown parameters from experimental data, particularly in the context of nuclear density functional theory. We examine the cost of not having derivatives of these functionals with respect to the parameters. We show that the POUNDERS code for local derivative-free optimization obtains consistent solutions on a variety of computationally expensive energy density functional calibration problems. We also provide a primer on the operation of the POUNDERS software in the Toolkit for Advanced Optimization.

physics.comp-ph↗

arXiv2014

Error Analysis in Nuclear Density Functional Theory

Nuclear density functional theory (DFT) is the only microscopic, global approach to the structure of atomic nuclei. It is used in numerous applications, from determining the limits of stability to gaining a deep understanding of the formation of elements in the universe or the mechanisms that power stars and reactors. The predictive power of the theory depends on the amount of physics embedded in the energy density functional as well as on efficient ways to determine a small number of free parameters and solve the DFT equations. In this article, we discuss the various sources of uncertainties and errors encountered in DFT and possible methods to quantify these uncertainties in a rigorous manner.

nucl-th↗

arXiv2013

Variable selection and sensitivity analysis using dynamic trees, with an application to computer code performance tuning

We investigate an application in the automatic tuning of computer codes, an area of research that has come to prominence alongside the recent rise of distributed scientific processing and heterogeneity in high-performance computing environments. Here, the response function is nonlinear and noisy and may not be smooth or stationary. Clearly needed are variable selection, decomposition of influence, and analysis of main and secondary effects for both real-valued and binary inputs and outputs. Our contribution is a novel set of tools for variable selection and sensitivity analysis based on the recently proposed dynamic tree model. We argue that this approach is uniquely well suited to the demands of our motivating example. In illustrations on benchmark data sets, we show that the new techniques are faster and offer richer feature sets than do similar approaches in the static tree and computer experiment literature. We apply the methods in code-tuning optimization, examination of a cold-cache effect, and detection of transformation errors.

stat.AP↗

arXiv2013

Computational Nuclear Quantum Many-Body Problem: The UNEDF Project

The UNEDF project was a large-scale collaborative effort that applied high-performance computing to the nuclear quantum many-body problem. UNEDF demonstrated that close associations among nuclear physicists, mathematicians, and computer scientists can lead to novel physics outcomes built on algorithmic innovations and computational developments. This review showcases a wide range of UNEDF science results to illustrate this interplay.

nucl-th↗