[Colloq] Thesis Proposal - From the Approximate to the Exact: Solving Floating-Point Formulas via Proxy Theories - Jaideep Ramachandran - Wednesday, June 8, 2:00pm WVH 366

[Smith, Greg]

Title:  From the Approximate to the Exact: Solving Floating-Point Formulas via Proxy Theories
Speaker: Jaideep Ramachandran
Date: Wednesday, June 8
Time: 2:00 PM
Location: 366WVH

Abstract:
Most programs that perform numerical computations use floating-point arithmetic to represent and process real-valued quantities. While efficient, floating-point arithmetic exhibits properties very distinct from what we are used to with real numbers. This can make it unintuitive and often renders programs that use it unpredictable, requiring dedicated analysis tools. Our thesis is that numerical reductions that bridge the gap between approximate-but-efficient and exact-but-costly analysis can be used to build efficient solvers for floating-point formulas. We propose a framework for building such solvers that can be used to reason formally about numeric programs. Our framework offers an algorithmic architecture for bridging the above-mentioned gap via the notion of proxy theories, and for effective and precise mappings from solutions found in a proxy theory to a satisfying floating-point assignment to the given formula. We investigate the usefulness of our framework using two substantially distinct proxy theory instances, and propose to apply it for generating test cases for numeric programs.


Thank you,

Greg Smith
Project Manager, College of Computer and Information Science
Northeastern University
202 West Village H
Boston, MA  02115
Phone: 617.373.3237
Fax: 617.373.5121
Facebook<https://www.facebook.com/ccisatnu>|Instagram<https://instagram.com/ccisatnu/>|LinkedIn<https://www.linkedin.com/groups/Northeastern-University-College-Computer-Information-1943637?gid=1943637&mostPopular=&trk=tyah&trkInfo=idx%3A1-1-1%2CtarId%3A1426606862845%2Ctas%3ANortheastern+University+College+of+Com>|Twitter<https://twitter.com/CCISatNU>