COCO data archives
Welcome to the website of the benchmarking data archives of the COCO (Comparing Continuous Optimizers) platform.
This website provides the official benchmarking data archives from numerical benchmarking experiments run on the COCO platform. For each test suite we provide a listing of the official data sets with additional information like authors, links to papers, source code etc. Data sets for the following test suites are available:
- bbob: 24 single-objective noiseless functions, 250+ data sets
- bbob-noisy: 24 single-objective noisy functions, 40+ data sets
- bbob-biobj: 55 bi-objective functions, 30+ data sets
- bbob-largescale: large-scale version of the 24 bbob functions (dimension up to 640), 10+ data sets
- bbob-mixint: mixed-integer versions of the 24 bbob functions, 8 data sets
- bbob-constrained: 54 constrained functions with varying number of (non-linear) constraints, 9 data sets
- sbox-cost: box-constrained version of 24 bbob functions, 2 data sets
These data can be directly used “by name” through the cocopp Python module.