The US Department of energy has invested US $13.7 million in scientific "data reduction" research

Heart of machine 2021-09-14 22:47:06

edit / Kaixia

 Doe investment 1370 Million dollars for Science 「 Data reduction 」 Research

9 month 2 Japan , The U.S. Department of energy (DOE) Declared as 9 Investment in research projects 1370 Thousands of dollars , These projects will promote the development of computer science and Applied Mathematics . These projects are led by five universities in eight states and five U.S. Department of energy national laboratories , Move will be resolved 、 Storage and processing science Observatory 、 The challenge of massive data sets generated by experimental facilities and supercomputers , Speed up the pace of scientific discovery .

 Doe investment 1370 Million dollars for Science 「 Data reduction 」 Research

With the upgrading and expansion of scientific user facilities , Their ability to generate large amounts of scientific data began to surpass scientists' streaming 、 The ability to archive and analyze this data . There is an urgent need to develop new mathematics and computer science and Technology , Reduce these data sets by removing trivial or duplicate data , While retaining important scientific information that may lead to discovery .

Although data reduction (data reduction) The need for technology is obvious , But scientists using these technologies must believe , They won't lose important scientific information , This is a key challenge . The research supported by the program must not only address the efficiency and effectiveness of data reduction technology , Its credibility must also be addressed .

「 National Science user facilities, including the science office of the U.S. Department of energy , Data are being produced that may lead to exciting and important scientific discoveries , But the scale of these data is bringing new challenges ,」, Senior deputy director of scientific computing research, office of science, U.S. Department of energy Barb Helland say .「 These findings can only be found when the data are controllable , And scientists trust the technology used to do this .」

The items selected in the announcement cover a wide range of topics , These themes promise important innovations in data reduction technologies , Including the use of advanced machine learning 、 Large scale statistical computing and new hardware accelerator technology . Examples of projects include :

  • Compress Stream data Methods : Researchers at Oak Ridge National Laboratory will develop a technology , By using a specific structure of scientific instruments or computer models , And integrate advanced machine learning technology , Compress data directly from the instrument or computer model , It also allows scientists to control certain characteristics of the data .
  • Methods of intelligently selecting and adjusting compression technology : Researchers at Texas State University will develop technology to search for potential data compression The vast space of Technology , And according to the user's fidelity 、 Select the best method according to the requirements of speed and memory use .
  • Compression method of related data sets : Researchers at the University of California, San Diego, will develop scalable technologies to compress multiple related data streams , for example , By using the relationship between data sets , Data streams from multiple sensors observing the same physical system .
  • Programming method of customizing hardware accelerator for streaming compression : Researchers at Fermi National Accelerator Laboratory will develop techniques for coding advanced compression and filtering , Including based on machine learning The technology of the method , As a custom hardware accelerator for various experimental settings , From particle physics experiments to electron microscopy .

Reference Content :

本文为[Heart of machine]所创,转载请带上原文链接,感谢