Filters








278,668 Hits in 5.2 sec

Parallelizing Gaussian Process Calculations inR

Christopher J. Paciorek, Benjamin Lipshitz, Wei Zhuo, Prabhat, Cari G. Kaufman, Rollin C. Thomas
2015 Journal of Statistical Software  
Gaussian process regression in an R package called bigGP that relies on C and MPI.  ...  We consider parallel computation for Gaussian process calculations to overcome computational and memory constraints on the size of datasets that can be analyzed.  ...  ) and calls the C code from the R function or uses a parallel framework in R.  ... 
doi:10.18637/jss.v063.i10 fatcat:dtkf36zvxbcypg43di2ngcddye

PetRBF — A parallel O(N) algorithm for radial basis function interpolation with Gaussians

Rio Yokota, L.A. Barba, Matthew G. Knepley
2010 Computer Methods in Applied Mechanics and Engineering  
The parallel code is freely available in the open-source model.  ...  We have developed a parallel algorithm for radial basis function (RBF) interpolation that exhibits O(N) complexity,requires O(N) storage, and scales excellently up to a thousand processes.  ...  The same calculation as the one shown Figure 7 was performed in parallel. The results for 1, 2, 4, and 8 processes are shown in Figure 8 .  ... 
doi:10.1016/j.cma.2010.02.008 fatcat:teeh25zqa5bc7f3boe34vhjpze

GP3: A Sampling-based Analysis Framework for Gaussian Processes [article]

Armin Lederer, Markus Kessler, Sandra Hirche
2020 arXiv   pre-print
In order to overcome this issue, we propose a novel framework called GP3, general purpose computation on graphics processing units for Gaussian processes, which allows to solve many of the existing problems  ...  Gaussian process regression is a prominent example among those methods, which attracts growing attention due to its strong Bayesian foundations.  ...  Multi-resolution Analysis of Gaussian Processes Gaussian processes exhibit a strongly nonlinear mean function in general.  ... 
arXiv:2006.07871v1 fatcat:gpwszmm4mfcjvamurf7f2iq7ym

A GPU-based Affine and Scale Invariant Feature Transform Algorithm

Guofeng Tong
2013 Journal of Information and Computational Science  
Finally, the experiment shown that multicore parallel ASIFT algorithm greatly improves computing speed in the same precision as mononuclear serial ASIFT algorithm.  ...  By analyzing the algorithm's principle of ASIFT, this paper proposed multicore parallel ASIFT algorithm and achieved ASIFT algorithm's parallelization by CUDA architecture.  ...  stream in multiple processing units.  ... 
doi:10.12733/jics20102203 fatcat:l4uqqtb7pzeclie7qui3xr527e

Hardware implementation of radial-basis neural networks with Gaussian activation functions on FPGA

Volodymyr Shymkovych, Sergii Telenyk, Petro Kravets
2021 Neural computing & applications (Print)  
The implementation of the Gaussian activation functions of the hidden layer of the RBF network occupies 106 LUTs, and the speed of the Gaussian activation functions is 29.33 ns.  ...  Hardware implementation of RBF neural networks with such speed allows them to be used in real-time control systems for high-speed objects.  ...  In VHDL, each parallel process is represented by a separate operator process(in), when the input value in changes, the calculation begins in the block process(in).  ... 
doi:10.1007/s00521-021-05706-3 fatcat:hx7st7mfxrcc7aoqc7vjtr26wa

Expected Improvements for the Asynchronous Parallel Global Optimization of Expensive Functions: Potentials and Challenges [chapter]

Janis Janusevskis, Rodolphe Le Riche, David Ginsbourger, Ramunas Girdziusas
2012 Lecture Notes in Computer Science  
But Gaussian processes can also generate parallel optimization strategies. We focus here on a new, parameter free, parallel expected improvement criterion for asynchronous optimization.  ...  Sequential sampling strategies based on Gaussian processes are now widely used for the optimization of problems involving costly simulations.  ...  In the Bayesian Global Optimization settings considered here, the unknown f is represented a priori by a Gaussian Process (Y (x)) x∈R d , and is being approximated relying on the conditional distribution  ... 
doi:10.1007/978-3-642-34413-8_37 fatcat:3wuigyoxcrgrxg3gu6r7bu47qi

Parallel likelihood calculation for phylogenetic comparative models: The SPLITT C++ library

Venelin Mitov, Tanja Stadler, Tamara Münkemüller
2019 Methods in Ecology and Evolution  
We conclude that parallel pruning effectively accelerates the likelihood calculation and, thus, the statistical inference of Gaussian phylogenetic models.  ...  We implement several parallel traversal algorithms in the form of a generic C++ library for Serial and Parallel LIneage Traversal of Trees (SPLITT).  ...  Krzysztof Bartoszek for valuable insights on the Ornstein-Uhlenbeck process.  ... 
doi:10.1111/2041-210x.13136 fatcat:35trxpqjxjb6tdlompla73p7hu

Parallel computing and the generation of basic plasma data

Vincent McKoy, Carl Winstead, Chuo-Han Lee
1998 Journal of Vacuum Science & Technology. A. Vacuum, Surfaces, and Films  
Comprehensive simulations of the processing plasmas used in semiconductor fabrication will depend on the availability of basic data for many microscopic processes that occur in the plasma and at the surface  ...  In this article, we report on the progress we have made in exploiting large-scale distributed-memory parallel computers, consisting of hundreds of interconnected microprocessors, to generate electron-collision  ...  If we expand the molecular orbitals in the Slater determinants i , j , and ⌽ m in Eqs. ͑2͒ and ͑3͒ in Cartesian Gaussian functions ͑r;R,l,m,n ͒ϭN ␣ ͑ xϪX ͒ l ͑ yϪY ͒ m ͑ zϪZ ͒ n ϫexp͑Ϫ␣͉rϪ 2 ͒, ͑4͒ where  ... 
doi:10.1116/1.580990 fatcat:3j2fyo2a5zfhdmpodxbamwh63u

Acceleration Techniques for Analysis of Microstrip Structures

R. Pomarnacki, A. Krukonis, V. Urbanavicius
2014 Elektronika ir Elektrotechnika  
In this paper we present three techniques for such computations acceleration: parallel algorithm implemented in computer cluster, sparse bound-matrix technique, and graphic processing unit in conjunction  ...  Accurate calculation of parameters of such structures with numerical techniques requires the solution of dense matrix equations involving thousands of unknowns.  ...  PARALLEL ALGORITHM AND COMPUTER CLUSTER Almost every calculation process, especially cyclic calculations, can be organized in parallel manner, when calculations are distributed among more when one computers  ... 
doi:10.5755/j01.eee.20.5.7109 fatcat:vp7bray6gfhcrn42jkx3gqga3q

Two-way partitioning of a recursive Gaussian filter in CUDA

Chang Won Lee, Jaepil Ko, Tae-Young Choe
2014 EURASIP Journal on Image and Video Processing  
In order to increase the parallelism of recursive Gaussian filters, we propose a two-way partitioned recursive Gaussian filter.  ...  This partition increases the parallelism because the filter is applied to the two blocks in parallel.  ...  Figure 6 6 Parallelism of the row-oriented step. In the first pass, B l , P c , and B r are computed in parallel.  ... 
doi:10.1186/1687-5281-2014-33 fatcat:4xkhfqfsonfafc6sfntpzxsqqa

Highly parallel steered mixture-of-experts rendering at pixel-level for image and light field data

Vasileios Avramelos, Ruben Verhack, Ignace Saenen, Glenn Van Wallendael, Bart Goossens, Peter Lambert
2018 Journal of Real-Time Image Processing  
SMoE has multiple applications in coding, scale-conversion, and general processing of image modalities.  ...  In this paper it is shown that on appropriate hardware, the OpenCL implementation can achieve 85fps and 22fps for respectively 1080p and 4K renderings of large models with more than 100.000 of Gaussian  ...  The assumption is that every block is processed in parallel.  ... 
doi:10.1007/s11554-018-0843-3 fatcat:pcubeilcizeu5ezx52wnv2udvi

Efficient rendering of regions of response in list-mode reconstruction for PET

Giancarlo Sportelli, Juan E. Ortuno, Andres Santos
2011 2011 IEEE Nuclear Science Symposium Conference Record  
In the current version it has been tested for reconstructing List-Mode (LM) data simulated with GATE on the rPET small animal scanner and used with a 2D Gaussian kernel specifically designed for the rPET  ...  EM parallelization has been achieved at per-event level on a 8 core dual CPU.  ...  Parallelization In order to parallelize the OSEM reconstruction algorithm we subdivide the LM dataset into several partitions, each processed by a different processing unit.  ... 
doi:10.1109/nssmic.2011.6153809 fatcat:k3544z76grghpf2t6mtzeo5ibm

On stability across a Gaussian product channel

Utsaw Kumar, Vijay Gupta, J. Nicholas Laneman
2011 IEEE Conference on Decision and Control and European Control Conference  
Gaussian product channel.  ...  The Gaussian product channel models a continuous-time waveform Gaussian channel, where the encoder transmits information to the receiver across multiple noisy paths.  ...  The controller calculates a control input U (k) and applies it to the process in (1).  ... 
doi:10.1109/cdc.2011.6160955 dblp:conf/cdc/KumarGL11 fatcat:4vedoxe73rhthh6x63lt6vmkym

Kubo–Greenwood electrical conductivity formulation and implementation for projector augmented wave datasets

L. Calderín, V.V. Karasiev, S.B. Trickey
2017 Computer Physics Communications  
It is MPI parallelized over k-points, bands, and plane waves, with an option to recover the plane wave processes for their use in band parallelization as well.  ...  New analytical results and a full implementation of the KG approach in an open-source Fortran 90 post-processing code for use with Quantum Espresso (J.Phys. Cond.  ...  wave parallelization for each band process.  ... 
doi:10.1016/j.cpc.2017.08.008 fatcat:3nm3bmvdnvcuhmpnyfhmnctnse

Numerical Simulation of the Formation of Hydrated Electron States

Alina Volokhova, Elena Zemlyanaya, Viktor Lakhno, Ilkizar Amirkhanov, Maxim Bashashin, Igor Puzynin, Taisiya Puzynina, Gh. Adam, J. Buša, M. Hnatič, D. Podgainy
2018 EPJ Web of Conferences  
calculation of the hydrated electron absorption band width.  ...  Effectiveness of parallel implementation is tested on the HybriLIT cluster.  ...  An MPI-based parallel algorithm is shown to provide the 8-11 times acceleration in comparison with the serial calculations.  ... 
doi:10.1051/epjconf/201817306013 fatcat:44jtmlw53nb55pugg5bskfh3su
« Previous Showing results 1 — 15 out of 278,668 results