Berlin, Germany

Zuse Institute Berlin
Berlin, Germany

The Zuse Institute Berlin is a research institute for applied mathematics and computer science on the campus of Freie Universität Berlin in Dahlem.The ZIB was founded by law as a statutory establishment and as a non-university research institute of the State of Berlin in 1984. In close interdisciplinary cooperation with the Berlin universities and scientific institutions Zuse Institute implements research and development in the field of information technology with a particular focus on application-oriented algorithmic Mathematics and practical Computer science. ZIB also provides high-performance computer capacity as an accompanying service as part of the Network of high performance computers in Northern Germany ).Konrad Zuse, born in Berlin in 1910, is the namesake of the ZIB. Wikipedia.

Time filter
Source Type

Mukhopadhyay A.,Zuse Institute Berlin
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2017

This paper proposes a fully automatic supervised segmentation technique for segmenting the great vessel and blood pool of pediatric cardiac MRIs of children with Congenital Heart Defects (CHD). CHD affects the overall anatomy of heart, rendering model-based segmentation framework infeasible, unless a large dataset of annotated images is available. However, the cardiac anatomy still retains distinct appearance patterns, which has been exploited in this work. In particular, Total Variation (TV) is introduced for solving the 3D disparity and noise removal problem. This results in homogeneous appearances within anatomical structures which is exploited further in a Random Forest framework. Context-aware appearance models are learnt using Random Forest (RF) for appearance-based prediction of great vessel and blood pool of an unseen subject during testing. We have obtained promising results on the HVSMR16 training dataset in a leave-one-out cross-validation. © Springer International Publishing AG 2017.

Zhao J.,University of Stuttgart | Frank B.,University of Stuttgart | Burger S.,Zuse Institute Berlin | Giessen H.,University of Stuttgart
ACS Nano | Year: 2011

We introduce angle-controlled colloidal nanolithography as a fast and low-cost fabrication technique for large-area periodic plasmonic oligomers with complex shapes and tunable geometry parameters. We investigate the optical properties and find highly modulated plasmon modes in oligomers with triangular building blocks. Fundamental modes, higher-order modes, as well as Fano resonances due to coupling between bright and dark modes within the same complex structure are present, depending on polarization and structure geometry. Our process is well-suited for mass fabrication of novel large-area plasmonic sensing devices and nanoantennas. © 2011 American Chemical Society.

Weiser M.,Zuse Institute Berlin
BIT Numerical Mathematics | Year: 2014

Spectral deferred correction methods for solving stiff ODEs are known to converge reasonably fast towards the collocation limit solution on equidistant grids, but show a less favourable contraction on non-equidistant grids such as Radau-IIa points. We interprete SDC methods as fixed point iterations for the collocation system and propose new DIRK-type sweeps for stiff problems based on purely linear algebraic considerations. Good convergence is recovered also on non-equidistant grids. The properties of different variants are explored on a couple of numerical examples. © 2014 Springer Science+Business Media Dordrecht

We propose and analyze an interior point path-following method in function space for state constrained optimal control. Our emphasis is on proving convergence in function space and on constructing a practical path-following algorithm. In particular, the introduction of a pointwise damping step leads to a very efficient method, as verified by numerical experiments. © 2012 Springer-Verlag Berlin Heidelberg and Mathematical Optimization Society.

Weber M.,Zuse Institute Berlin
Studies in Classification, Data Analysis, and Knowledge Organization | Year: 2013

In this chapter, PCCAC is described as a special spectral clustering algorithm which is applicable for molecular simulation data. From a mathematical point of view, only PCCAC is able to correctly identify the physical timescales of molecular motion. In order to decrease the statistical error of this timescales analysis, an adaptive clustering algorithm is necessary. © Springer-Verlag Berlin Heidelberg 2013.

Duy N.T.,Zuse Institute Berlin
Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention | Year: 2012

We propose a fully automatic method for tooth detection and classification in CT or cone-beam CT image data. First we compute an accurate segmentation of the maxilla bone. Based on this segmentation, our method computes a complete and optimal separation of the row of teeth into 16 subregions and classifies the resulting regions as existing or missing teeth. This serves as a prerequisite for further individual tooth segmentation. We show the robustness of our approach by providing extensive validation on 43 clinical head CT scans.

Pothkow K.,Zuse Institute Berlin | Hege H.-C.,Zuse Institute Berlin
Computer Graphics Forum | Year: 2013

An uncertain (scalar, vector, tensor) field is usually perceived as a discrete random field with a priori unknown probability distributions. To compute derived probabilities, e.g. for the occurrence of certain features, an appropriate probabilistic model has to be selected. The majority of previous approaches in uncertainty visualization were restricted to Gaussian fields. In this paper we extend these approaches to nonparametric models, which are much more flexible, as they can represent various types of distributions, including multimodal and skewed ones. We present three examples of nonparametric representations: (a) empirical distributions, (b) histograms and (c) kernel density estimates (KDE). While the first is a direct representation of the ensemble data, the latter two use reconstructed probability density functions of continuous random variables. For KDE we propose an approach to compute valid consistent marginal distributions and to efficiently capture correlations using a principal component transformation. Furthermore, we use automatic bandwidth selection, obtaining a model for probabilistic local feature extraction. The methods are demonstrated by computing probabilities of level crossings, critical points and vortex cores in simulated biofluid dynamics and climate data. © 2013 The Author(s) Computer Graphics Forum © 2013 The Eurographics Association and Blackwell Publishing Ltd.

Berthold T.,Zuse Institute Berlin
Operations Research Letters | Year: 2013

In modern MIP solvers, primal heuristics play a key role in finding high-quality solutions. However, classical performance measures reflect the impact of primal heuristics on the overall solving process badly. In this article, we introduce a new performance measure, the "primal integral", which depends on the quality of solutions and on the time when they are found. We compare five state-of-the-art MIP solvers w.r.t. the newly proposed measure, and show that heuristics improve their performance by up to 80%.© 2013 Elsevier B.V.

Berthold T.,Zuse Institute Berlin | Gleixner A.M.,Zuse Institute Berlin
Mathematical Programming | Year: 2014

We present Undercover, a primal heuristic for nonconvex mixed-integer nonlinear programs (MINLPs) that explores a mixed-integer linear subproblem (sub-MIP) of a given MINLP. We solve a vertex covering problem to identify a smallest set of variables to fix, a so-called cover, such that each constraint is linearized. Subsequently, these variables are fixed to values obtained from a reference point, e.g., an optimal solution of a linear relaxation. Each feasible solution of the sub-MIP corresponds to a feasible solution of the original problem. We apply domain propagation to try to avoid infeasibilities, and conflict analysis to learn additional constraints from infeasibilities that are nonetheless encountered. We present computational results on a test set of mixed-integer quadratically constrained programs (MIQCPs) and MINLPs. It turns out that the majority of these instances allows for small covers. Although general in nature, we show that the heuristic is most successful on MIQCPs. It nicely complements existing root-node heuristics in different state-of-the-Art solvers and helps to significantly improve the overall performance of the MINLP solver SCIP. © 2013 Springer-Verlag Berlin Heidelberg and Mathematical Optimization Society.

Gamrath G.,Zuse Institute Berlin
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2013

Strong branching is an important component of most variable selection rules in branch-and-bound based mixed-integer linear programming solvers. It predicts the dual bounds of potential child nodes by solving auxiliary LPs and thereby helps to keep the branch-and-bound tree small. In this paper, we describe how these dual bound predictions can be improved by including domain propagation into strong branching. Computational experiments on standard MIP instances indicate that this is beneficial in three aspects: It helps to reduce the average number of LP iterations per strong branching call, the number of branch-and-bound nodes, and the overall solving time. © Springer-Verlag 2013.

Loading Zuse Institute Berlin collaborators
Loading Zuse Institute Berlin collaborators