{ "types" : { "Bookmark" : { "pluralLabel" : "Bookmarks" }, "Publication" : { "pluralLabel" : "Publications" }, "GoldStandardPublication" : { "pluralLabel" : "GoldStandardPublications" }, "GoldStandardBookmark" : { "pluralLabel" : "GoldStandardBookmarks" }, "Tag" : { "pluralLabel" : "Tags" }, "User" : { "pluralLabel" : "Users" }, "Group" : { "pluralLabel" : "Groups" }, "Sphere" : { "pluralLabel" : "Spheres" } }, "properties" : { "count" : { "valueType" : "number" }, "date" : { "valueType" : "date" }, "changeDate" : { "valueType" : "date" }, "url" : { "valueType" : "url" }, "id" : { "valueType" : "url" }, "tags" : { "valueType" : "item" }, "user" : { "valueType" : "item" } }, "items" : [ { "type" : "Publication", "id" : "https://puma.ub.uni-stuttgart.de/bibtex/280d2faf3c7ef622c9ff634d9282d69c6/ipvs-sgs", "tags" : [ "myown","from:brunnme","grids,","SG++","sparse","Data-mining,","GPU," ], "intraHash" : "80d2faf3c7ef622c9ff634d9282d69c6", "interHash" : "1de8bc2524ca633988bae1b71b70c831", "label" : "Data-mining on adaptively refined sparse grids with higher order basis functions on GPUs", "user" : "ipvs-sgs", "description" : "", "date" : "2020-07-29 15:07:21", "changeDate" : "2020-07-29 13:07:21", "count" : 2, "pub-type": "misc", "year": "2016", "url": "http://www2.informatik.uni-stuttgart.de/cgi-bin/NCSTRL/NCSTRL_view.pl?id=BCLR-2479&engl=1", "author": [ "Malte Brunn" ], "authors": [ {"first" : "Malte", "last" : "Brunn"} ], "number": "2479","pages": "87","abstract": "Mankind produces enormous amounts of data through simulations, observations,\r\n and experiments. With improving technology such as more precise measurements\r\n and faster supercomputers this data grows rapidly and has to be stored,\r\n computed, and evaluated. Thus, efficient algorithms are needed to process all\r\n this data in order to extract useful information. Most of this data is\r\n high-dimensional and it depends on a variety of parameters. Methods based on\r\n regular grids fail to handle it reasonably since the complexity of the problem\r\n exponentially depends on the number of dimensions. However, methods for sparse\r\n grids reduce this complexity while maintaining the accuracy. In this work a\r\n parallel approach for regression and classification tasks on adaptively refined\r\n sparse grids with higher order basis functions is presented. The developed\r\n algorithms are tested and analyzed on graphic accelerator cards. The results of\r\n a previous work about a parallel approach for the hierarchisation on regular\r\n sparse grids are reused to develop highly parallel algorithms for the\r\n hierarchisation, multi-evaluation, and the transposed evaluation on adaptively\r\n refined sparse grids. Furthermore, several optimizations for these sparse grid\r\n operations are explained. The hierarchisation as well as the multi-evaluation\r\n make use of the hierarchical structure of the grid to reduce the problemâ\u20AC™s\r\n complexity, and the effects of sorting the evaluation points along a\r\n space-filling curve on the runtime of the operations are analyzed. The\r\n space-filling curve methods not only improve the cache usage and thread\r\n coherence of the parallel algorithms, but they also reduce the computational\r\n effort of algorithms based on the streaming approach. The transposed evaluation\r\n achieves speed-ups by factors from 2 to up to 11 compared to a parallel\r\n implementation without space-filling curve optimizations. The effects on the\r\n thread coherence and the cache usage improve the runtime of the hierarchical\r\n evaluation by a factor of up to 10 for adaptively refined grids. Additionally,\r\n a benchmark analysis of the developed methods with respect to the maximal\r\n polynomial degree shows that the degree for most of the basis functions is\r\n strongly limited by the grid level. Thus, the runtime and the complexity do not\r\n behave in a linear fashion with respect to the maximal polynomial degree as\r\n expected.", "language" : "English", "cr-category" : "F.2.1 Numerical Algorithms and Problems,\r\n G.0 Mathematics of Computing General,\r\n G.1.2 Numerical Analysis Approximation", "bibtexKey": "BCLR-2479" } , { "type" : "Publication", "id" : "https://puma.ub.uni-stuttgart.de/bibtex/280d2faf3c7ef622c9ff634d9282d69c6/brunnme", "tags" : [ "Data-mining,","GPU,","SG++","grids,","myown","sparse" ], "intraHash" : "80d2faf3c7ef622c9ff634d9282d69c6", "interHash" : "1de8bc2524ca633988bae1b71b70c831", "label" : "Data-mining on adaptively refined sparse grids with higher order basis functions on GPUs", "user" : "brunnme", "description" : "", "date" : "2020-07-27 15:58:26", "changeDate" : "2020-07-29 13:07:21", "count" : 2, "pub-type": "misc", "year": "2016", "url": "http://www2.informatik.uni-stuttgart.de/cgi-bin/NCSTRL/NCSTRL_view.pl?id=BCLR-2479&engl=1", "author": [ "Malte Brunn" ], "authors": [ {"first" : "Malte", "last" : "Brunn"} ], "number": "2479","pages": "87","abstract": "Mankind produces enormous amounts of data through simulations, observations,\r\n and experiments. With improving technology such as more precise measurements\r\n and faster supercomputers this data grows rapidly and has to be stored,\r\n computed, and evaluated. Thus, efficient algorithms are needed to process all\r\n this data in order to extract useful information. Most of this data is\r\n high-dimensional and it depends on a variety of parameters. Methods based on\r\n regular grids fail to handle it reasonably since the complexity of the problem\r\n exponentially depends on the number of dimensions. However, methods for sparse\r\n grids reduce this complexity while maintaining the accuracy. In this work a\r\n parallel approach for regression and classification tasks on adaptively refined\r\n sparse grids with higher order basis functions is presented. The developed\r\n algorithms are tested and analyzed on graphic accelerator cards. The results of\r\n a previous work about a parallel approach for the hierarchisation on regular\r\n sparse grids are reused to develop highly parallel algorithms for the\r\n hierarchisation, multi-evaluation, and the transposed evaluation on adaptively\r\n refined sparse grids. Furthermore, several optimizations for these sparse grid\r\n operations are explained. The hierarchisation as well as the multi-evaluation\r\n make use of the hierarchical structure of the grid to reduce the problemâ\u20AC™s\r\n complexity, and the effects of sorting the evaluation points along a\r\n space-filling curve on the runtime of the operations are analyzed. The\r\n space-filling curve methods not only improve the cache usage and thread\r\n coherence of the parallel algorithms, but they also reduce the computational\r\n effort of algorithms based on the streaming approach. The transposed evaluation\r\n achieves speed-ups by factors from 2 to up to 11 compared to a parallel\r\n implementation without space-filling curve optimizations. The effects on the\r\n thread coherence and the cache usage improve the runtime of the hierarchical\r\n evaluation by a factor of up to 10 for adaptively refined grids. Additionally,\r\n a benchmark analysis of the developed methods with respect to the maximal\r\n polynomial degree shows that the degree for most of the basis functions is\r\n strongly limited by the grid level. Thus, the runtime and the complexity do not\r\n behave in a linear fashion with respect to the maximal polynomial degree as\r\n expected.", "language" : "English", "cr-category" : "F.2.1 Numerical Algorithms and Problems,\r\n G.0 Mathematics of Computing General,\r\n G.1.2 Numerical Analysis Approximation", "bibtexKey": "BCLR-2479" } , { "type" : "Publication", "id" : "https://puma.ub.uni-stuttgart.de/bibtex/26809bfc73b7cafc5a109578f775d3a84/mgeiger", "tags" : [ "(computer","(mathematics),","Computational","GPU,","Ptex,","Rendering","Splines","Terrain,","Three-dimensional","Tools,","algorithms,","common","complete","computer","curve-based","curves,","data","data,","diffusion,","diffusion-based","displays,","extraction,","feature","feature-based","features,","field,","fields,","generation,","graphics),","graphics,","height","input","interactive","large-scale","layers,","manual","mapping,","modeling,","modelling","modelling,","multiple","offline","primitives,","prominent","real-time","rendering","rendering,","representation","representations,","scale","sparse","spline","structure,","structures,","surface","surface,","terrain","texture,","three-dimensional","toolset,","two-dimensional","vertical","visualisation,","volumetric","workflow," ], "intraHash" : "6809bfc73b7cafc5a109578f775d3a84", "interHash" : "eb89ebb1743ac1f23d0413247168fe06", "label" : "Feature-based volumetric terrain generation and decoration", "user" : "mgeiger", "description" : "", "date" : "2019-11-08 16:11:00", "changeDate" : "2019-11-08 15:14:53", "count" : 5, "pub-type": "article", "journal": "IEEE Transactions on Visualization and Computer Graphics", "year": "2019", "url": "", "author": [ "Michael Becher","Michael Krone","Guido Reina","Thomas Ertl" ], "authors": [ {"first" : "Michael", "last" : "Becher"}, {"first" : "Michael", "last" : "Krone"}, {"first" : "Guido", "last" : "Reina"}, {"first" : "Thomas", "last" : "Ertl"} ], "volume": "25","number": "2","pages": "1283--1296","abstract": "Two-dimensional height fields are the most common data structure used for storing and rendering of terrain in offline rendering and especially real-time computer graphics. By its very nature, a height field cannot store terrain structures with multiple vertical layers such as overhanging cliffs, caves, or arches. This restriction does not apply to volumetric data structures. However, the workflow of manual modelling and editing of volumetric terrain usually is tedious and very time-consuming. Therefore, we propose to use three-dimensional curve-based primitives to efficiently model prominent, large-scale terrain features. We present a technique for volumetric generation of a complete terrain surface from the sparse input data by means of diffusion-based algorithms. By combining an efficient, feature-based toolset with a volumetric terrain representation, the modelling workflow is accelerated and simplified while retaining the full artistic freedom of volumetric terrains. Feature Curves also contain material information that can be complemented with local details by using per-face texture mapping. All stages of our method are GPU-accelerated using compute shaders to ensure interactive editing of terrain. Please note that this paper is an extended version of our previously published work [1] .", "doi" : "10.1109/TVCG.2017.2762304", "bibtexKey": "becher_feature-based_2019" } ] }