Publications

Results 76–100 of 136

Search results

Jump to search filters

Optimality conditions for the numerical solution of optimization problems with PDE constraints :

Aguilo Valentin, Miguel A.; Ridzal, Denis R.

A theoretical framework for the numerical solution of partial di erential equation (PDE) constrained optimization problems is presented in this report. This theoretical framework embodies the fundamental infrastructure required to e ciently implement and solve this class of problems. Detail derivations of the optimality conditions required to accurately solve several parameter identi cation and optimal control problems are also provided in this report. This will allow the reader to further understand how the theoretical abstraction presented in this report translates to the application.

More Details
TYPE SAND Report YEAR 2014
DOIOSTI

Optimization-based conservative transport on the cubed-sphere grid

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Peterson, Kara J.; Bochev, Pavel B.; Ridzal, Denis R.

Transport algorithms are highly important for dynamical modeling of the atmosphere, where it is critical that scalar tracer species are conserved and satisfy physical bounds. We present an optimization-based algorithm for the conservative transport of scalar quantities (i.e. mass) on the cubed sphere grid, which preserves local solution bounds without the use of flux limiters. The optimization variables are the net mass updates to the cell, the objective is to minimize the discrepancy between these variables and suitable high-order cell mass update (the "target"), and the constraints are derived from the local solution bounds and the conservation of the total mass. The resulting robust and efficient algorithm for conservative and local bound-preserving transport on the sphere further demonstrates the flexibility and scope of the recently developed optimization-based modeling approach [1, 2]. © 2014 Springer-Verlag.

More Details
TYPE Conference YEAR 2014
DOIOSTIScopus

Edge remap for solids

Love, Edward L.; Robinson, Allen C.; Ridzal, Denis R.

We review the edge element formulation for describing the kinematics of hyperelastic solids. This approach is used to frame the problem of remapping the inverse deformation gradient for Arbitrary Lagrangian-Eulerian (ALE) simulations of solid dynamics. For hyperelastic materials, the stress state is completely determined by the deformation gradient, so remapping this quantity effectively updates the stress state of the material. A method, inspired by the constrained transport remap in electromagnetics, is reviewed, according to which the zero-curl constraint on the inverse deformation gradient is implicitly satisfied. Open issues related to the accuracy of this approach are identified. An optimization-based approach is implemented to enforce positivity of the determinant of the deformation gradient. The efficacy of this approach is illustrated with numerical examples.

More Details
TYPE SAND Report YEAR 2013
DOIOSTI

Designing experiments through compressed sensing

Young, Joseph G.; Ridzal, Denis R.

In the following paper, we discuss how to design an ensemble of experiments through the use of compressed sensing. Specifically, we show how to conduct a small number of physical experiments and then use compressed sensing to reconstruct a larger set of data. In order to accomplish this, we organize our results into four sections. We begin by extending the theory of compressed sensing to a finite product of Hilbert spaces. Then, we show how these results apply to experiment design. Next, we develop an efficient reconstruction algorithm that allows us to reconstruct experimental data projected onto a finite element basis. Finally, we verify our approach with two computational experiments.

More Details
TYPE SAND Report YEAR 2013
DOIOSTI
Results 76–100 of 136
Results 76–100 of 136