Stefan Martin Wild

Stefan Martin Wild

Auto-generated profile, based on publicly available information

Stefan Martin Wild

Applied Mathematics and Computational Research Division, Lawrence Berkeley National Laboratory, Berkeley, USA

Research Focus & Metrics

Academic Career Length

23Years (2004-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
13 August 2010 in Physical Review C
Verified
Journal Article
Closed access
MKM. Kortelainen
M. Kortelainen
TLT. Lesinski
T. Lesinski
JMJ. Moré
J. Moré
WNW. Nazarewicz
W. Nazarewicz
JSJ. Sarich
J. Sarich
SWS. Wild
S. Wild
We carry out state-of-the-art optimization of a nuclear energy density of Skyrme type in the framework of the Hartree-Fock-Bogoliubov theory. The particle-hole and particle-particle channels are optimized simultaneously, and the experimental data set includes both spherical and deformed nuclei. The new model-based, derivative-free optimization algorithm used in this work has been found to be significantly better than standard optimization methods in terms of reliability, speed, accuracy, and precision. The resulting parameter set unedf0 results in good agreement with experimental masses, radii, and deformations and seems to be free of finite-size instabilities. An estimate of the reliability of the obtained parameterization is given, based on standard statistical methods. We discuss new physics insights offered by the advanced covariance analysis.

1 January 2009 in SIAM Journal on Optimization
Verified
Journal Article
Closed access
JMJorge J. Moré
Jorge J. Moré
SWStefan M. Wild
Stefan M. Wild
We propose data profiles as a tool for analyzing the performance of derivative-free optimization solvers when there are constraints on the computational budget. We use performance and data profiles, together with a convergence test that measures the decrease in function value, to analyze the performance of three solvers on sets of smooth, noisy, and piecewise-smooth problems. Our results provide estimates for the performance difference between these solvers, and show that on these problems, the model-based solver tested performs better than the two direct search solvers tested.

8 February 2012 in Physical Review C
Verified
Journal Article
Open access
PDF file
MKM. Kortelainen
M. Kortelainen
JMJ. McDonnell
J. McDonnell
WNW. Nazarewicz
W. Nazarewicz
PRP.-G. Reinhard
P.-G. Reinhard
JSJ. Sarich
J. Sarich
SWS. M. Wild
S. M. Wild
A new Skyrme-like energy density suitable for studies of strongly elongated nuclei was determined in the framework of the Hartree-Fock-Bogoliubov theory using the recently developed model-based, derivative-free optimization algorithm pounders. A sensitivity analysis at the optimal solution has revealed the importance of states at large deformations in driving the parameterization of the functional. The good agreement with experimental data on masses and separation energies, achieved with the previous parameterization unedf0, is largely preserved. In addition, the new energy density unedf1 gives a much improved description of the fission barriers in 240Pu and neighboring nuclei.

7 May 2013 in Physical Review Letters
Verified
Journal Article
Open access
PDF file
AEA. Ekström
A. Ekström
GBG. Baardsen
G. Baardsen
CFChristian Forssén
Christian Forssén
GHG. Hagen
G. Hagen
MHM. Hjorth-Jensen
M. Hjorth-Jensen
SWS. M. Wild
S. M. Wild
We optimize the nucleon-nucleon interaction from chiral effective field theory at next-to-next-to-leading order (NNLO). The resulting new chiral force NNLOopt yields χ2≈1 per degree of freedom for laboratory energies below approximately 125 MeV. In the A=3, 4 nucleon systems, the contributions of three-nucleon forces are smaller than for previous parametrizations of chiral interactions. We use NNLOopt to study properties of key nuclei and neutron matter, and we demonstrate that many aspects of nuclear structure can be understood in terms of this nucleon-nucleon interaction, without explicitly invoking three-nucleon forces.

1 May 2019 in Acta Numerica
Verified
Journal Article
Closed access
PDF file
JLJeffrey Larson
Jeffrey Larson
MMMatt Menickelly
Matt Menickelly
SWStefan M. Wild
Stefan M. Wild
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.

10 February 2019
Verified
Report
Closed access
PDF file
NBNathan Baker
Nathan Baker
FAFrank Alexander
Frank Alexander
TBTimo Bremer
Timo Bremer
AHAric Hagberg
Aric Hagberg
YKYannis Kevrekidis
Yannis Kevrekidis
SLSteven Lee
Steven Lee
The U.S. Department of Energy's Office of Scientific and Technical Information