Personal tools
You are here: Home Events Lab Lunch by Daniël Reijsbergen

Lab Lunch by Daniël Reijsbergen

— filed under:

Recomputability of scientific experiments

What
  • Lab Lunch
When Jan 27, 2015
from 01:00 PM to 02:00 PM
Add event to calendar vCal
iCal

A scientific experiment is said to be 'reproducible' if it can be repeated by a different group of researchers. An example from the social sciences would be a description of the procedure to select a sample population, the type of information gathered, and the statistical methodology used to analyse the dataset. In the 'computational' sciences, a large part of the experimental side of research is the execution by a computer of a sequence of command lines. In this case, the level of detail with which an experiment can be repeated is much higher. This stronger property is called 'recomputability'. Different views on recomputability exist: e.g., is sharing source code enough, or should the experiment be stored on a virtual machine that is shared in its entirety (including the operating system) with other researchers? In this talk I will give a brief overview of the points raised by the speakers at a recent workshop on recomputable research. The idea is more to raise discussion than to present original research. 

Document Actions