Berlin 6 Open Access Conference: Sergey Fomel

Reproducible Research

Sergey Fomel

The University of Texas at Austin

Outline• Personal experience

• Why do reproducible research?

• How to do reproducible research?

* Reproducible computational experiments

Personal Experience• Jon Claerbout and RR at Stanford

•Madagascar open-source software

• CiSE special issue (Jan-Feb 2009)– David Donoho et al– Randall LeVeque – Roger Ping and Sandrah Eckel– Victoria Stodden

Jon Claerbout’s Story

• 1987: Sunview experience– Interactive programs are slavery

• 1992: LaTeX + cake– Rebuilding books by a single command

Reproducible Research at SEP• Stanford Exploration Project

– Founded in 1973– 2 Ph.D. students per year

• Reproducible research– From CD-ROMs to WWW– From cake to GNU make– 2001 CiSE paper

• The principal beneficiary is the author

The Madagascar Project• Multidimensional data analysis• Started in 2006• Open community• Open source (GPL)• Three levels

– Building blocks in C– Recipes in Python/SCons– Papers in LaTeX + SCons

http://ahay.org

Personal Experience• Jon Claerbout and RR at Stanford

•Madagascar open-source software

• CiSE special issue (Jan-Feb 2009)– David Donoho et al– Randall LeVeque – Roger Ping and Sandrah Eckel– Victoria Stodden

CiSE Reproducible Research• David Donoho, Arian Maleki, Inam Rahman,

Morteza Shahram, Victoria Stodden

• 15 years of reproducible research in computational harmonic analysis– MATLAB– WaveLab: 690 citations– “Striving for reproducibility imposes a

discipline that leads to better work.”

CiSE Reproducible Research• Randall J. LeVeque

• Python tools for reproducible research on hyperbolic problems– Fortran + Python– Clawpack: 7,000 registered users– “Scientific and mathematical journals

are filled with pretty pictures of computational experiments that the reader has no hope of repeating.”

CiSE Reproducible Research• Roger D. Peng and Sandrah P. Eckel

• Distributed reproducible research using cached computations– R language– Cacher package– “We propose that a modular research

approach lends itself more naturally to reproducible results.”

CiSE Reproducible Research• Victoria Stodden

• The legal framework for reproducible research in the sciences– Licensing and copyright– ORL (Open Research License)– “We need a license designed with the

needs of computational researchers in mind.”

Why Reproducible? Science is the systematic enterprise of

gathering knowledge about the universe and organizing and condensing that knowledge into testable laws and theories. The success and credibility of science are anchored in the willingness of scientists to independent testing and replication by other scientists. This requires the complete and

open exchange of data, procedures and materials.

Open-Source Software

“Abandoning the habit of secrecy in favor of process transparency and peer review was the crucial step by which alchemy became chemistry. In the same way, it is beginning to appear that open-source development may signal

the long-awaited maturation of

software development as a

discipline.” Eric S. Raymond

How to Do Reproducible Computational Experiments?

• Code attached to published results

• Continuous maintenance

• Previous results used for testing– Test-driven development

• Lessons from open-source – Intellectual property– Community http://ahay.org