TY - JOUR
T1 - Reproducible research in computational harmonic analysis
AU - Donoho, David L.
AU - Maleki, Arian
AU - Shahram, Morteza
AU - Rahman, Inam Ur
AU - Stodden, Victoria
N1 - Funding Information:
The US National Science Foundation partially supported our work through its statistics and probability program (David Donoho and Iain Johnstone, co-PIs; DMS 92-09130, DMS 95-05150, DMS 00-772661, and DMS 05-05303), its optimization program (Stephen Boyd, PI), its focused research group (Multi-scale Geometric Analysis, FRG DMS-0140698, David Donoho, PI), an information technology research project, a knowledge diffusion initiative project (Amos Ron, PI; KDI Towards Ideal Data Representation; and ITR Multiresolution Analysis of The Global Internet), and its signal-processing program. In addition, the US Air Force Office of Scientific Research partially supported our work through a multi-university research initiative project (David Castanon, PI); as did the Office of Naval Research (Comotion: Claire Tomlin, PI). Finally, DARPA partially supported our work through three projects: Efficient Mathematical Algorithms for Signal Processing, 1998–2000; Efficient Multiscale Methods for Automatic Target Recognition, 2001– 2002; and GeoStar: Efficient Multiscale Representation of Geopotential Data, 2005–2007. Thanks to the program officers at these agencies for guidance and interest; we only mention a few: Mary Ellen Bock, Doug Cochran, John Cozzens, Dennis Healy, Reza Malek-Madani, Wen Masters, Jong-Shi Pang, Carey Schwartz, Jon Sjogren, Gabor Szekely, Grace Yang, and Yazhen Wang.
PY - 2009/1
Y1 - 2009/1
N2 - Wavelab, based on the Matlab quantitative computing environment, is the first effort to provide a toolbox specifically designed to reproduce results in a series of computational science papers. The toolboxes in Wavelab framework include Atomizer for sparse representation of signals, Beamlab for multiscale geometric analysis, and Sparselab for multiscale analysis of manifold valued data. Various researchers have compared their own novel algorithms with the algorithms first published in the Sparselab package, which allows them to reproduce papers in Sparselab. The Symmlab package, which includes core algorithms and generates synthetic data, disseminates new problems and illustrates new data types. The arguments raised by researchers against reproducibility include reproducibility undermining the creation of intellectual capita; and reproducibility destroying time-honored motivations for collaboration.
AB - Wavelab, based on the Matlab quantitative computing environment, is the first effort to provide a toolbox specifically designed to reproduce results in a series of computational science papers. The toolboxes in Wavelab framework include Atomizer for sparse representation of signals, Beamlab for multiscale geometric analysis, and Sparselab for multiscale analysis of manifold valued data. Various researchers have compared their own novel algorithms with the algorithms first published in the Sparselab package, which allows them to reproduce papers in Sparselab. The Symmlab package, which includes core algorithms and generates synthetic data, disseminates new problems and illustrates new data types. The arguments raised by researchers against reproducibility include reproducibility undermining the creation of intellectual capita; and reproducibility destroying time-honored motivations for collaboration.
UR - http://www.scopus.com/inward/record.url?scp=58149133580&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=58149133580&partnerID=8YFLogxK
U2 - 10.1109/MCSE.2009.15
DO - 10.1109/MCSE.2009.15
M3 - Article
AN - SCOPUS:58149133580
SN - 1521-9615
VL - 11
SP - 8
EP - 18
JO - Computing in Science and Engineering
JF - Computing in Science and Engineering
IS - 1
M1 - 4720218
ER -