An important aspect of computaional science: reproducibility

There are often two basic requires for the computational investigations:

1.      Validation, largely based on the experimental data or theories; 2. Reproducibility, can be contested by other investigators.

And to be published,

Computational studies usually need to provide significant in-depth understanding of the scientific problems. This can usually be achieved because the conditions can be well controlled in computational work.

Some highlights:

A critical barrier to reproducibility in many cases is that the computer code is no longer available.

Perhaps the biggest barrier to reproducible research is the lack of a deeply ingrained culture that simply requires reproducibility for all scientific claims.

The field of science will not change overnight, but simply bringing the notion of reproducibility to the forefront and making it routine will make a difference. Ultimately, developing a culture of reproducibility in which it currently does not exist will require time and sustained effort from the scientific community.

If open source code becomes true when a contribution is published, I think, the competition will be only in creative ideas and understanding, not the ways to solve problems. It seems that the contribution of the developers of novel approaches is minimized.

How about the experimental methods? How can they become accessible to others to reproduce the results? Seems a long way to run!

Feed: Science: Current Issue
Posted on: Friday, 2 December 2011 4:15 AM
Author: Roger D. Peng
Subject: [Special Issue Perspective] Reproducible Research in Computational Science

Author: Roger D. Peng

View article...