Seminar: Towards Automated Performance Analysis of Python Programs with VyPR
|Starts:||14:00 23 Oct 2019|
|Ends:||15:00 23 Oct 2019|
|What is it:||Seminar|
|Organiser:||Department of Computer Science|
|Who is it for:||University staff, Adults|
Join us for the next Computer Science Mercury talk with speaker Joshua Dawes
Analysing the performance of programs is typically a manual effort, involving profiling certain quantities via manual instrumentation of code, and performing offline analysis. While this approach works well and is used heavily in industry, it limits us to analysing program runs with respect to only very simple constraints and can result in loss of information about which parts of code generated certain observations.
During this talk, a Runtime Verification-inspired approach is described for monitoring programs at runtime with respect to complex constraints, while using automatic instrumentation to minimise intrusion into the program source code. Further, given more complex constraints, the problem of explanation of failure to satisfy such constraints is addressed, forming a contribution to an underdeveloped area of Runtime Verification.
The talk is concluded by a report of applications of the publicly available VyPR framework (http://cern.ch/vypr) at the CMS Experiment at CERN.
Role: Computer Science PG student
Organisation: University of Manchester & CERN
Travel and Contact Information
Kilburn Lecture Theatre 1.5