For the current REF see the REF 2021 website REF 2021 logo

Output details

11 - Computer Science and Informatics

University College London

Return to search Previous output Next output
Output 0 of 0 in the submission
Output title

Automatic detection of floating-point exceptions.

Type
E - Conference contribution
Name of conference/published proceedings
POPL
Volume number
-
Issue number
-
First page of article
549
ISSN of proceedings
-
Year of publication
2013
Number of additional authors
3
Additional information

<07> In 1996, the Ariane 5 rocket self-destructed at a loss of $500m. Cyber-physics systems, like the Ariane 5 or Google's driverless car, are ubiquitous and safety critical. To interact with the physical world, they approximate real numbers. Exceptions occur when these approximations fail. Indeed, the Ariane 5 disaster occurred when a large value overflowed a storage location. This work presents the first tool that uses symbolic execution to find inputs that trigger such exceptions, so programmers can fix them. It discovered over 2,000 exceptions in the mature and widely used GNU scientific library. (18%)

Interdisciplinary
-
Cross-referral requested
-
Research group
None
Citation count
0
Proposed double-weighted
No
Double-weighted statement
-
Reserve for a double-weighted output
No
Non-English
No
English abstract
-