MPI-INF Logo
Campus Event Calendar

Event Entry

New for: D1, D2, D3, D4, D5

What and Who

How to find a good program abstraction automatically?

Hongseok Yang
University of Oxford
SWS Distinguished Lecture Series
AG 1, AG 2, AG 3, AG 4, AG 5, SWS, RG1, MMCI  
Expert Audience
English

Date, Time and Location

Tuesday, 4 March 2014
10:30
60 Minutes
E1 5
002
Saarbrücken

Abstract


Recent years have seen the development of successful commercial
programming tools based on static analysis technologies, which
automatically verify intended properties of programs or find tricky bugs
that are difficult to detect by testing techniques. One of the key
reasons for this success is that these tools use clever strategies
for abstracting programs -- most details about a given program
are abstracted away by these strategies, unless they are predicted to be
crucial for proving a given property about the program or detecting a type
of program errors of interest. Developing such a strategy is, however,
nontrivial, and is currently done by a large amount of manual
engineering efforts in most tool projects. Finding a good abstraction
strategy automatically or even reducing these manual efforts involved in the
development of such a strategy is considered
one of the main open challenges in the area of program analysis.

In this talk, I will explain how I tried to address this challenge
with colleagues in the US and Korea in the past few years.
During this time, we worked on parametric program analyses, where
parameters for controlling the degree of program abstraction are
expressed explicitly in the specification of the analyses.
For those analyses, we developed algorithms for searching for a
desired parameter value with respect to a given program and a given property,
which use ideas from the neighboring areas of program analysis such as testing,
searching and optimisation. In my talk, I will describe the main ideas
behind these algorithms without going into technical details. I will focus on
intuitions about why and when these algorithms work. I will also
talk briefly about a few lessons that I learnt while working on this problem.

This talk is based on the joint work with Mayur Naik, Xin Zhang, Ravi Mangal,
Radu Grigore, Hakjoo Oh, Wonchan Lee, Kihong Heo, and Kwangkeun Yi.

Contact

Brigitta Hansen
0681 93039102
--email hidden

Video Broadcast

Yes
Kaiserslautern
G26
111
passcode not visible
logged in users only

Christian Klein, 10/13/2016 17:27
Brigitta Hansen, 06/13/2014 15:12 -- Created document.