Guided Random Testing for Numerical Error Measurement

This is a project of investigating search based floating-point error estimation. We are currently working on efficient and effective search methods for error estimation. Meanwhile, we are also working on applying this technique on other fields like numerical program analysis and auto-tuning.

Releases

Publications

  • Efficient Search for Inputs Causing High Floating-point Errors (PPoPP'14)
    Wei-Fan Chiang, Ganesh Gopalakrishnan, Zvonimir Rakamaric, and Alexey Solovyev
    [paper][online content][bib][slides]

Project Members

  • Professors: Ganesh Gopalakrishnan and Zvonimir Rakamaric.
  • Postdoc: Alexey Solovyev
  • PhD student: Wei-Fan Chiang