Refine
Year of publication
Document Type
- Article (2185)
- ZIB-Report (1758)
- In Proceedings (1325)
- Master's Thesis (284)
- Book chapter (166)
- Doctoral Thesis (164)
- Other (99)
- Bachelor's Thesis (69)
- Book (66)
- In Collection (66)
Keywords
- integer programming (36)
- mixed integer programming (34)
- KOBV (33)
- optimal control (32)
- Annual Report (28)
- Kooperativer Bibliotheksverbund Berlin-Brandenburg (24)
- Mixed Integer Programming (21)
- Bibliotheksverbund (20)
- Integer Programming (19)
- mixed-integer programming (19)
Institute
- Numerical Mathematics (1507)
- Mathematical Optimization (1271)
- Visual and Data-centric Computing (1141)
- Visual Data Analysis (1017)
- ZIB Allgemein (951)
- Computational Nano Optics (499)
- Distributed Algorithms and Supercomputing (489)
- Modeling and Simulation of Complex Processes (409)
- Visual Data Analysis in Science and Engineering (378)
- Therapy Planning (315)
Branching rules revisited
(2004)
Mixed integer programs are commonly solved with linear programming based branch-and-bound algorithms. The success of the algorithm strongly depends on the strategy used to select the variable to branch on. We present a new generalization called {\sl reliability branching} of today's state-of-the-art {\sl strong branching} and {\sl pseudocost branching} strategies for linear programming based branch-and-bound algorithms. After reviewing commonly used branching strategies and performing extensive computational studies we compare different parameter settings and show the superiority of our proposed newstrategy.
MIPLIB 2003
(2005)
This paper reports on the fourth version of the Mixed Integer Programming Library. Since ({\sc miplib}) is to provide a concise set of challenging problems, it became necessary to purge instances that became too easy. We present an overview of the 27 new problems and statistical data for all 60 instances.
Branching Rules Revisited
(2005)
MIPLIB 2003
(2006)
Starting with the description of the Traveling Salesmen Problem formulation as given by van Vyve and Wolsey in the article Approximate extended formulations'', we investigate the effects of small variations onto the performance of contemporary mixed integer programming solvers. We will show that even minor changes in the formulation of the model can result in performance difference of more than a factor of 1000. As the results show it is not obvious which changes will result in performance improvements and which not.
Given a general mixed integer program (MIP), we automatically detect block structures in the constraint matrix together with the coupling by capacity constraints arising from multi-commodity-flow formulations. We identify the underlying graph and generate cutting planes based on cuts in the detected network. Our implementation adds a separator to the branch-and-cut libraries of SCIP and CPLEX. We make use of the complemented mixed integer rounding framework (cMIR) but provide a special purpose aggregation heuristic that exploits the network structure. Our separation scheme speeds-up the computation for a large set of MIPs coming from network design problems by a factor of two on average.
An error controlled finite elemente method (FEM) for solving stationary Schrödinger equations in three space dimensions is proposed. The method is based on an adaptive space discretization into tetrahedra and local polynomial basis functions of order $p=1$--$5$ defined on these tetrahedra. According to a local error estimator the triangulation is automatically adapted to the solution. Numerical results for standard problems appearing in vibrational motion and molecular structure calculations are presented and discussed. Relative precisions better than 1e-8 are obtained. For equilateral H$_3^{++}$ the adaptive FEM turns out to be superior to global basis set expansions in the literature. Our precise FEM results exclude in a definite manner the stability or metastability of equilateral H$_3^{++}$ in its groundstate.
Using the perturbational-variational Rayleigh-Ritz matrix formalism, the 1/Z-expansion for the ground state of the isoelectronic $H_2$ sequence in the range of the internuclear distance $0.2\le R \le 9.0$ is calculated. Also lower bounds of the radius of convergence, based on Kato's theory of linear operators, are given. The numerical results of the 1/Z-expansion can be compared with the exact results and do not converge in the whole R-range. This behavior is in qualitative agreement with the lower bounds for the radius of convergence and enlights some still open properties of 1/Z- expansions for this sequence in the literature. {\bf PACS:} 31.15 + q; 31.20 Di; 31.20 Tz.
The subject of this study is a multilevel Finite Element Method based on an error estimator and step by step grid refinement as an universal tool for solving time--independent Schrödinger--eigenvalue problems. Numerical results for standard problems appearing in vibrational motion and molecular electronic structure calculations are given and discussed.
This paper presents the new program package MACRON for the simulation of macromolecular kinetics including standard chemical reactions. Such problems lead to countable (possibly) infinite systems of ordinary differential equations (CODE's), which are numerically treated by the so-called discrete Galerkin method here. By a chemical compiler the required analytical preprocessing is performed, such that the complete reaction system, standard kinetics as well as macromolecular reactions, can be entered in the chemical formalism. Typical macromolecular reaction steps are chain addition, termination, chain transfer and degradation (cracking). In order to ensure efficiency and reliability, high sophisticated numerical routines are built within the package. MACRON can be used without a detailed knowledge of the used numerical methods. As an illustration the application of MACRON to some realistic problems is presented.
The description of chain length distributions in macromolecular reaction kinetics leads to so-called countable systems of differential equations. In particular, when the appearing reaction rate coefficients depend on the chain length of the reacting macromolecules itself, an efficient numerical treatment of these systems is very difficult. Then even the evaluation of the right-hand side of the system can become prohibitively expensive with respect to computing time. In this paper we show how the discrete Galerkin method can be applied to such problems. The existing algorithm CODEX is improved by use of a multiplicative error correction scheme for time discretization and a new type of numerical preprocessing by means of a Gauss summation. Both ideas are exemplary for a wide class of approximation types and are described very briefly here. The new numerical techniques are tested on an example from soot formation, where the coagulation of molecules is modeled in terms of reaction coefficients depending on the surface of the particles and their collision frequency.
We employ the adaptive resolution approach AdResS, in its recently developed Grand Canonicallike version (GC-AdResS) [Wang et al. Phys.Rev.X 3, 011018 (2013)], to calculate the excess chemical potential, $μ^{ex}$, of various liquids and mixtures. We compare our results with those obtained from full atomistic simulations using the technique of thermodynamic integration and show a satisfactory agreement. In GC-AdResS the procedure to calculate $μ^{ex}$ corresponds to the process of standard initial equilibration of the system; this implies that, independently of the specific aim of the study, $μ^{ex}$, for each molecular species, is automatically calculated every time a GC-AdResS simulation is performed.
A great amount of material properties is strongly influenced by dislocations, the carriers of plastic deformation. It is therefore paramount to have appropriate tools to quantify dislocation substructures with regard to their features, e.g., dislocation density, Burgers vectors or line direction. While the transmission electron microscope (TEM) has been the most widely-used equipment implemented to investigate dislocations, it usually is limited to the two-dimensional (2D) observation of three-dimensional (3D) structures. We reconstruct, visualize and quantify 3D dislocation substructure models from only two TEM images (stereo pairs) and assess the results. The reconstruction is based on the manual interactive tracing of filiform objects on both images of the stereo pair. The reconstruction and quantification method are demonstrated on dark field (DF) scanning (S)TEM micrographs of dislocation substructures imaged under diffraction contrast conditions. For this purpose, thick regions (>300 nm) of TEM foils are analyzed, which are extracted from a Ni-base superalloy single crystal after high temperature creep deformation. It is shown how the method allows 3D quantification from stereo pairs in a wide range of tilt conditions, achieving line length and orientation uncertainties of 3% and 7°, respectively. Parameters that affect the quality of such reconstructions are discussed.
A great amount of material properties is strongly influenced by dislocations, the carriers of plastic deformation. It is therefore paramount to have appropriate tools to quantify dislocation substructures with regard to their features, e.g., dislocation density, Burgers vectors or line direction. While the transmission electron microscope (TEM) has been the most widely-used equipment implemented to investigate dislocations, it usually is limited to the two-dimensional (2D) observation of three-dimensional (3D) structures. We reconstruct, visualize and quantify 3D dislocation substructure models from only two TEM images (stereo-pairs) and assess the results. The reconstruction is based on the manual interactive tracing of filiform objects on both images of the stereo-pair. The reconstruction and quantification method are demonstrated on dark field (DF) scanning (S)TEM micrographs of dislocation substructures imaged under diffraction contrast conditions. For this purpose, thick regions (> 300 nm) of TEM foils are analyzed, which are extracted from a Ni-base superalloy single crystal after high temperature creep deformation. It is shown how the method allows 3D quantification from stereo-pairs in a wide range of tilt conditions, achieving line length and orientation uncertainties of 3 % and 7°, respectively. Parameters that affect the quality of such reconstructions are discussed.
Enhanced Photon Upconversion Using Erbium-Doped Nanoparticles Interacting with Silicon Metasurfaces
(2021)
The operation of a railway network as large as Deutsche Bahn's Intercity Express (ICE) hinges on a number of factors, such as the availability of personnel and the assignment of physical vehicles to a timetable schedule, a problem known as the rolling stock rotation problem (RSRP). In this paper, we consider the problem of creating an alternative timetable in the case that there is a long-term disruption, such as a strike, and the effects that this alternative timetable has on the resulting vehicle rotation plan. We define a priority measure via the Analytic Hierarchy Process (AHP) to determine the importance of each trip in the timetable and therefore which trips to cancel or retain. We then compare our results with those of a limited timetable manually designed by Deutsche Bahn (DB). We find that while our timetable results in a more expensive rotation plan, its flexibility lends itself to a number of simple improvements. Furthermore, our priority measure has the potential to be integrated into the rolling stock rotation optimization process, in particular, the Rotation Optimizer for Railways (ROTOR) software, via the cost function. Ultimately, our method provides the foundation for an automated way of creating a new timetable quickly, and potentially in conjunction with a new rotation plan, in the case of a limited scenario.
Inferring Proteolytic Processes from Mass Spectrometry Time Series Data Using Degradation Graphs
(2012)
We propose a mathematical optimization model and its solution for joint chance constrained DC Optimal Power Flow. In this application, it is particularly important that there is a high probability of transmission limits being satisfied, even in the case of uncertain or fluctuating feed-in from renewable energy sources. In critical network situations where the network risks overload, renewable energy feed-in has to be curtailed by the transmission system operator (TSO). The TSO can reduce the feed-in in discrete steps at each network node. The proposed optimization model minimizes curtailment while ensuring that there is a high probability of transmission limits being maintained. The latter is modeled via (joint) chance constraints that are computationally challenging. Thus, we propose a solution approach based on the robust safe approximation of these constraints. Hereby, probabilistic constraints are replaced by robust constraints with suitably defined uncertainty sets constructed from historical data. The ability to discretely control the power feed-in then leads to a robust optimization problem with decision-dependent uncertainties, i.e. the uncertainty sets depend on decision variables. We propose an equivalent mixed-integer linear reformulation for box uncertainties with the exact linearization of bilinear terms. Finally, we present numerical results for different test cases from the Nesta archive, as well as for a real network. We consider the discrete curtailment of solar feed-in, for which we use real-world weather and network data. The experimental tests demonstrate the effectiveness of this method and run times are very fast. Moreover, on average the calculated robust solutions only lead to a small increase in curtailment, when compared to nominal solutions.
Changes in knee shape and geometry resulting from total knee arthroplasty can affect patients in numerous important ways: pain, function, stability, range of motion, and kinematics. Quantitative data concerning these changes have not been previously available, to our knowledge, yet are essential to understand individual experiences of total knee arthroplasty and thereby improve outcomes for all patients. The limiting factor has been the challenge of accurately measuring these changes. Our study objective was to develop a conceptual framework and analysis method to investigate changes in knee shape and geometry, and prospectively apply it to a sample total knee arthroplasty population. Using clinically available computed tomography and radiography imaging systems, the three-dimensional knee shape and geometry of nine patients (eight varus and one valgus) were compared before and after total knee arthroplasty. All patients had largely good outcomes after their total knee arthroplasty. Knee shape changed both visually and numerically. On average, the distal condyles were slightly higher medially and lower laterally (range: +4.5 mm to −4.4 mm), the posterior condyles extended farther out medially but not laterally (range: +1.8 to −6.4 mm), patellofemoral distance increased throughout flexion by 1.8–3.5 mm, and patellar thickness alone increased by 2.9 mm (range: 0.7–5.2 mm). External femoral rotation differed preop and postop. Joint line distance, taking cartilage into account, changed by +0.7 to −1.5 mm on average throughout flexion. Important differences in shape and geometry were seen between pre-total knee arthroplasty and post-total knee arthroplasty knees. While this is qualitatively known, this is the first study to report it quantitatively, an important precursor to identifying the reasons for the poor outcome of some patients. Using the developed protocol and visualization techniques to compare patients with good versus poor clinical outcomes could lead to changes in implant design, implant selection, component positioning, and surgical technique. Recommendations based on this sample population are provided. Intraoperative and postoperative feedback could ultimately improve patient satisfaction.
Mit dem Voranschreiten der Technologie erhalten die öffentlichen Verkehrsmittel eine größere Bedeutung. Die Beförderung mehrerer Personen eröffnet der Gesellschaft viele Möglichkeiten, unter Anderem den Vorteil der Zeitersparnis. Die Dauer des Verkehrswegs mit öffentlichen Verkehrsmitteln ist häufig geringer, als die mit individuellen Verkehrsmitteln. Jedes öffentliche Transportmittel ist mit einem Fahrplan versehen. Dieser bietet Passagieren, die öffentliche Verkehrsmittel öfter nutzen, eine Strukturierung und Planung ihrer Zeit. Dabei lassen sich Taktfahrpläne aufgrund ihres periodischen Verhaltens leicht einprägen. Dieses periodische Verhalten ist durch mathematische Modellierungen darstellbar. Das persönliche Nutzverhalten vieler Bürger im Personenverkehr ist auf die öffentlichen Verkehrsmittel beschränkt. Diese beinhalten im Gegensatz zum individuellen Verkehrsmittel eine Wartezeit. Dabei stellt sich die Frage, ob man anhand mathematischer Modelle diese Wartezeit minimieren kann. Eine bekannte mathematische Modellierung dieses Problems ist das Periodic Event Scheduling Problem (PESP). Die optimale Planung eines periodischen Taktfahrplanes steht im Vordergrund.
Während ich dieses Problem betrachtet habe, wurde ich auf das Rechnen mit linearen Gleichungssystemen modulo T aufmerksam. Bei periodischen Taktfahrplänen wird ein einheitliches zeitliches Muster, welches sich nach T Minuten wiederholt, betrachtet. Das dabei zu betrachtende Lösungsproblem eröffnet ein Teilgebiet der Mathematik, welches bislang nicht im Vordergrund stand: Das Lösen linearer Gleichungen modulo T, wobei T für die Zeit in Minuten steht und somit 60 ist. Da 60 keine Primzahl ist, kann – wie im Laufe der Arbeit präsentiert – das lineare Gleichungssystem nicht mehr über einen Körper gelöst werden. Lineare Gleichungssysteme werden nun über Nicht-Körpern betrachtet. Die Literatur weist sowohl im deutschsprachigem als auch im englischsprachigen Raum wenig Umfang bezüglich linearer Gleichungssysteme über Nicht-Körper auf. Der Bestand an Fachliteratur bezüglich den Themen lineare diophantische Gleichungssysteme, Hermite-
Normalform und Smith-Normalform ist zurzeit gering, dennoch erreichbar, beispielsweise in [1], welches in dieser Bachelorarbeit genutzt wurde. Insbesondere wurde ich bei der Suche nach geeigneter Literatur zu linearen Gleichungssystemen über Restklassenringe, die keinen Körper bilden, nicht fündig. Dabei recherchierte ich sowohl in den Universitätsbibliotheken als auch in webbasierenden Suchmaschinen.
Aufgrund dem geringen Bestand an Fachliteratur in diesem Kontext, war ich gezwungen, an vielen Stellen eigene logische Verknüpfungen zu konzipieren und zu beweisen. Dies brachte viele Schwierigkeiten mit sich, die mit bestmöglichem Verständnis bearbeitet wurden.
Abseits der Zugänglichkeit der Literatur, finde ich es sehr überraschend, dass sich viele Professoren der Mathematik mit diesem Themenbereich nicht beschäftigten. Insbesondere gingen von den Dozenten, die ich um Literaturempfehlung bat, kein Werk aus. Damit
wurde das Thema "Lineare Gleichungssysteme Modulo T" einerseits eine große Herausforderung, andererseits eine große Motivation, da ich mit dieser Bachelorarbeit vielen Interessenten der Mathematik als Sekundärliteratur dienen kann.
Surgical tool detection is attracting increasing attention from the medical image analysis community. The goal generally is not to precisely locate tools in images, but rather to indicate which tools are being used by the surgeon at each instant. The main motivation for annotating tool usage is to design efficient solutions for surgical workflow analysis, with potential applications in report generation, surgical training and even real-time decision support. Most existing tool annotation algorithms focus on laparoscopic surgeries. However, with 19 million interventions per year, the most common surgical procedure in the world is cataract surgery. The CATARACTS challenge was organized in 2017 to evaluate tool annotation algorithms in the specific context of cataract surgery. It relies on more than nine hours of videos, from 50 cataract surgeries, in which the presence of 21 surgical tools was manually annotated by two experts. With 14 participating teams, this challenge can be considered a success. As might be expected, the submitted solutions are based on deep learning. This paper thoroughly evaluates these solutions: in particular, the quality of their annotations are compared to that of human interpretations. Next, lessons learnt from the differential analysis of these solutions are discussed. We expect that they will guide the design of efficient surgery monitoring tools in the near future.
Im Oktober 2012 hatte die Deutsche Forschungsgemeinschaft (DFG) ein Förderprogramm zur Neuausrichtung überregionaler Informationsservices ausgeschrieben, um eine umfassende Reorganisation bestehender Infrastrukturen anzustoßen, die mit den Empfehlungen des Wissenschaftsrates zur Zukunft der bibliothekarischen Verbundsysteme in Deutschland gefordert wurde. Im Themenfeld „Bibliotheksdateninfrastruktur und Lokale Systeme“ der DFG-Ausschreibung wurde das vom Hessischen Bibliotheksinformationssystem (HeBIS), vom Bibliotheksverbund Bayern (BVB) und vom Kooperativen Bibliotheksverbund Berlin-Brandenburg (KOBV) beantragte Vorhaben „Cloudbasierte Infrastruktur für Bibliotheksdaten“ (CIB) bewilligt. Das Projekt zielt auf die Überführung bibliothekarischer Workflows und Dienste in cloudbasierte Arbeitsumgebungen und die sukzessive Ablösung traditioneller Verbund- und Lokalsysteme durch internationale Systemplattformen. Zu den Arbeitspaketen des Vorhabens gehört u. a. die Einbindung von Norm- und Fremddatenangeboten und von weiteren Services in diese Plattformen.
Respiratory viral infections (RVIs) are common reasons for healthcare consultations. The inpatient management of RVIs consumes significant resources. From 2009 to 2014, we assessed the costs of RVI management in 4776 hospitalized children aged 0–18 years participating in a quality improvement program, where all ILI patients underwent virologic testing at the National Reference Centre followed by detailed recording of their clinical course. The direct (medical or non-medical) and indirect costs of inpatient management outside the ICU (‘non-ICU’) versus management requiring ICU care (‘ICU’) added up to EUR 2767.14 (non-ICU) vs. EUR 29,941.71 (ICU) for influenza, EUR 2713.14 (non-ICU) vs. EUR 16,951.06 (ICU) for RSV infections, and EUR 2767.33 (non-ICU) vs. EUR 14,394.02 (ICU) for human rhinovirus (hRV) infections, respectively. Non-ICU inpatient costs were similar for all eight RVIs studied: influenza, RSV, hRV, adenovirus (hAdV), metapneumovirus (hMPV), parainfluenza virus (hPIV), bocavirus (hBoV), and seasonal coronavirus (hCoV) infections. ICU costs for influenza, however, exceeded all other RVIs. At the time of the study, influenza was the only RVI with antiviral treatment options available for children, but only 9.8% of influenza patients (non-ICU) and 1.5% of ICU patients with influenza received antivirals; only 2.9% were vaccinated. Future studies should investigate the economic impact of treatment and prevention of influenza, COVID-19, and RSV post vaccine introduction.
In the highly competitive area of telecommunications, cost, quality, and network management are among the most important aspects to be considered when designing a network. We study the problem of dimensioning a telecommunication network that is still operating in case of a failure of a network component. Given a demand between each pair of nodes of a telecommunication network and a finite set of possible capacities for each edge of the network, we consider the problem of deciding what capacity to install on each edge of the network in order to minimize the building cost of the network and to satisfy the demand between each pair of nodes, even if a network component fails. The routing of the demands must satisfy the following additional restrictions: (a) there is a maximum number of nodes allowed in each path between any pair of nodes (path length restriction), and (b) there is a maximum percentage of the demand between each pair of nodes that can be routed through any network component (diversification restriction). Moreover, the chosen capacities must be such that, for every single node or single edge failure, a certain percentage of the demand between any pair of nodes is reroutable (i.e. it ``survives'' the particular failure). We formulate the problem as a mixed integer linear programming problem and present a cutting plane algorithm as well as several heuristics for its solution. Furthermore, we discuss several ways to implement survivability into a telecommunication network.
Given a communication demand between each pair of nodes of a network we consider the problem of deciding what capacity to install on each edge of the network in order to minimize the building cost of the network and to satisfy the demand between each pair of nodes. The feasible capacities that can be leased from a network provider are of a particular kind in our case. There are a few so-called basic capacities having the property that every basic capacity is an integral multiple of every smaller basic capacity. An edge can be equipped with a capacity only if it is an integer combination of the basic capacities. We treat, in addition, several restrictions on the routings of the demands (length restriction, diversification) and failures of single nodes or single edges. We formulate the problem as a mixed integer linear programming problem and develop a cutting plane algorithm as well as several heuristics to solve it. We report on computational results for real world data.
Designing low-cost networks that survive certain failure situations is one of the prime tasks in the telecommunication industry. In this paper we survey the development of models for network survivability used in practice in the last ten years. We show how algorithms integrating polyhedral combinatorics, linear programming, and various heuristic ideas can help solve real-world network dimensioning instances to optimality or within reasonable quality guarantees in acceptable running times. The most general problem type we address is the following. Let a communication demand between each pair of nodes of a telecommunication network be given. We consider the problem of choosing, among a discrete set of possible capacities, which capacity to install on each of the possible edges of the network in order to (i) satisfy all demands, (ii) minimize the building cost of the network. \noindent In addition to determining the network topology and the edge capacities we have to provide, for each demand, a routing such that (iii) no path can carry more than a given percentage of the demand, (iv) no path in the routing exceeds a given length. \noindent We also have to make sure that (v) for every single node or edge failure, a certain percentage of the demand is reroutable. \noindent Moreover, for all failure situations feasible routings must be computed. The model described above has been developed in cooperation with a German mobile phone provider. We present a mixed-integer programming formulation of this model and computational results with data from practice.
A Network Dimensioning Tool
(1996)
Designing low cost networks that survive certain failure situations belongs to one of the prime tasks in the telecommunications industry. In this paper we describe a mathematical model combining several aspects of survivability that are elsewhere treated in a hierarchical fashion. We present mathematical investigations of this integrated model, a cutting plane algorithm, as well as several heuristics for its solution. Moreover, we report computational results with real world data. The problem we address is the following. Suppose, between each pair of nodes in a region, a communication demand is given. We want to determine the topology of a telecommunication network connecting the given nodes and to dimension all potential physical links. For each link, the possible capacities are restricted to a given finite set. The capacities must be chosen such that the communication demands are satisfied, even if certain network components fail, and such that the network building costs are as small as possible. Moreover, for each pair of nodes and each failure situation, we want to determine the paths on which the demand between the nodes is routed.
We model a symmetric system of coupled oscillators as a graph with symmetry group $\gamma$. Each vertex of the graph represents an "oscillator" or a "cell" of reactants. The magnitude (concentration) of the reactants in the $ i $ th cell is represented by a vector $ x^i $. The edges represent the coupling of the cells. The cells are assumed to evolve by identical reaction-diffusion equation which depends on the sum of the reactants in the nearest neighbors. Thus the dynamics of the system is described by a nonlinear differential system \begin{flushleft} \[ \mbox {(*) \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ } \dot{x}^i = f (x^i,\sum_{j \in N_i} x^j), \mbox { \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ } \] \end{flushleft} where the sum ranges over the set $ N_i $ of neighbors of cell $ i $ . If $ f $ also has a symmetry (e.g., oddness), there are geometric conditions on the graph such that the nonlinear system $ (*) $ decouples globally into a product flow on certain sums of isotropy subspaces. Thus we may detect higher-dimensional tori of solutions of $ (*) $ which are not amenable to other types of analysis. We present a number of examples, such as bipartite graphs, complete graphs, the square, the octahedron, and a 6-dimensional cube.
Solving PDEs on unstructured grids is a cornerstone of engineering and scientific computing. Heterogeneous parallel platforms, including CPUs, GPUs, and FPGAs, enable energy-efficient and computationally demanding simulations.
In this article, we introduce the HPM C++-embedded DSL that bridges the abstraction gap between the mathematical formulation of mesh-based algorithms for PDE problems on the one hand and an increasing number of heterogeneous platforms with their different programming models on the other hand.
Thus, the HPM DSL aims at higher productivity in the code development process for multiple target platforms. We introduce the concepts as well as the basic structure of the HPM DSL, and demonstrate its usage with three examples. The mapping of the abstract algorithmic description onto parallel hardware, including distributed memory compute clusters, is presented.
A code generator and a matching back end allow the acceleration of HPM code with GPUs. Finally, the achievable performance and scalability are demonstrated for different example problems.
Solving partial differential equations on unstructured grids is a cornerstone of engineering and scientific computing. Nowadays, heterogeneous parallel platforms with CPUs, GPUs, and FPGAs enable energy-efficient and computationally demanding simulations. We developed the HighPerMeshes C++-embedded Domain-Specific Language (DSL) for bridging the abstraction gap between the mathematical and algorithmic formulation of mesh-based algorithms for PDE problems on the one hand and an increasing number of heterogeneous platforms with their different parallel programming and runtime models on the other hand. Thus, the HighPerMeshes DSL aims at higher productivity in the code development process for multiple target platforms. We introduce the concepts as well as the basic structure of the HighPer-Meshes DSL, and demonstrate its usage with three examples, a Poisson and monodomain problem, respectively, solved by the continuous finite element method, and the discontinuous Galerkin method for Maxwell’s equation. The mapping of the abstract algorithmic description onto parallel hardware, including distributed memory compute clusters is presented. Finally, the achievable performance and scalability are demonstrated for a typical example problem on a multi-core CPU cluster.
We study performance and scaling of the Berlin Quantum Chromodynamics Program (BQCD) on the SGI Altix 4700 at Leibniz Supercomputing Centre (LRZ). We employ different communication methods (MPI, MPI with two OpenMP threads per process, as well as the shmem library) and run the MPI version on the two types of nodes of that machine. For comparison with other machines we made performance measurements on an IBM p690 cluster and a Cray XT4.