Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userTerabyte IDE RAID-5 Disk Arrays
97
0
0.0
(
0
)
Authors
D. A. Sanders
Ask ChatGPT about the research
No Arabic abstract
High energy physics experiments are currently recording large amounts of data and in a few years will be recording prodigious quantities of data. New methods must be developed to handle this data and make analysis at universities possible. We examine some techniques that exploit recent developments in commodity hardware. We report on tests of redundant arrays of integrated drive electronics (IDE) disk drives for use in offline high energy physics data analysis. IDE redundant array of inexpensive disks (RAID) prices now are less than the cost per terabyte of million-dollar tape robots! The arrays can be scaled to sizes affordable to institutions without robots and used when fast random access at low cost is important.
rate research
Read More
We discuss the traditional criterion for discovery in Particle Physics of requiring a significance corresponding to at least 5 sigma; and whether a more nuanced approach might be better.
New heterogeneous computing paradigms on dedicated hardware with increased parallelization, such as Field Programmable Gate Arrays (FPGAs), offer exciting solutions with large potential gains. The growing applications of machine learning algorithms in particle physics for simulation, reconstruction, and analysis are naturally deployed on such platforms. We demonstrate that the acceleration of machine learning inference as a web service represents a heterogeneous computing solution for particle physics experiments that potentially requires minimal modification to the current computing model. As examples, we retrain the ResNet-50 convolutional neural network to demonstrate state-of-the-art performance for top quark jet tagging at the LHC and apply a ResNet-50 model with transfer learning for neutrino event classification. Using Project Brainwave by Microsoft to accelerate the ResNet-50 image classification model, we achieve average inference times of 60 (10) milliseconds with our experimental physics software framework using Brainwave as a cloud (edge or on-premises) service, representing an improvement by a factor of approximately 30 (175) in model inference latency over traditional CPU inference in current experimental hardware. A single FPGA service accessed by many CPUs achieves a throughput of 600--700 inferences per second using an image batch of one, comparable to large batch-size GPU throughput and significantly better than small batch-size GPU throughput. Deployed as an edge or cloud service for the particle physics computing model, coprocessor accelerators can have a higher duty cycle and are potentially much more cost-effective.
A spectral fitter based on the graphics processor unit (GPU) has been developed for Borexino solar neutrino analysis. It is able to shorten the fitting time to a superior level compared to the CPU fitting procedure. In Borexino solar neutrino spectral analysis, fitting usually requires around one hour to converge since it includes time-consuming convolutions in order to account for the detector response and pile-up effects. Moreover, the convergence time increases to more than two days when including extra computations for the discrimination of $^{11}$C and external $gamma$s. In sharp contrast, with the GPU-based fitter it takes less than 10 seconds and less than four minutes, respectively. This fitter is developed utilizing the GooFit project with customized likelihoods, pdfs and infrastructures supporting certain analysis methods. In this proceeding the design of the package, developed features and the comparison with the original CPU fitter are presented.
We present an open-source Mathematica importer for CERN ROOT files. Taking advantage of Mathematicas import/export plug-in mechanism, the importer offers a simple, unified interface that cleanly wraps around its MathLink-based core that links the ROOT libraries with Mathematica. Among other tests for accuracy and efficiency, the importer has also been tested on a large (~5 Gbyte) file structure, D3PD, used by the ATLAS experiment for offline analysis without problems. In addition to describing the installation and usage of the importer, we discuss how the importer may be further improved and customized. A link to the package can be found at: http://library.wolfram.com/infocenter/Articles/7793/ and a related presentation is at: http://cd-docdb.fnal.gov/cgi-bin/DisplayMeeting?conferenceid=522
The many ways in which machine and deep learning are transforming the analysis and simulation of data in particle physics are reviewed. The main methods based on boosted decision trees and various types of neural networks are introduced, and cutting-edge applications in the experimental and theoretical/phenomenological domains are highlighted. After describing the challenges in the application of these novel analysis techniques, the review concludes by discussing the interactions between physics and machine learning as a two-way street enriching both disciplines and helping to meet the present and future challenges of data-intensive science at the energy and intensity frontiers.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
