COCO (COmparing Continuous Optimisers) is a platform for systematic and sound comparisons of real-parameter global optimisers. COCO provides benchmark function testbeds, experimentation templates which are easy to parallelize, and tools for processing and visualizing data generated by one or several optimizers. The COCO platform has been used for the Black-Box-Optimization-Benchmarking (BBOB) workshops that took place during the GECCO conference in 2009, 2010, 2012, 2013, and in 2015-2018 (and that will take place again in 2019). It was also used at the IEEE Congress on Evolutionary Computation (CEC'2015) in Sendai, Japan.
The COCO experiment source code has been rewritten in the years 2014-2015 and the current production code is available on our COCO github page. The old code is still available at the downloads page and shall be used for experiments on the noisy test suite until this test suite will be available in the new code as well.
To subscribe to (or unsubscribe from) the bbob discussion mailing list follow this link http://lists.lri.fr/cgi-bin/mailman/listinfo/bbob-discuss .
To receive announcements related to the BBOB workshops simply send an email to BBOB team bbob_at_lri.fr with the title “register to BBOB announcement list”
Internal wiki: https://gforge.inria.fr/plugins/mediawiki/wiki/coco
The figures show selected results from BBOB 2009. The empirical runtime distribution is shown for six subgroups of the BBOB functions. Click on the respective figures for more details. (At the top of this page the aggregated results over all functions are shown). More plots (for all supported COCO test suites) can be found in the http://coco.gforge.inria.fr/ppdata-archive/ppdata-archive.
The current 24 noiseless test functions are
1 Separable Functions | |
---|---|
f1 | Sphere Function |
f2 | Ellipsoidal Function |
f3 | Rastrigin Function |
f4 | Büche-Rastrigin Function |
f5 | Linear Slope |
2 Functions with low or moderate conditioning | |
f6 | Attractive Sector Function |
f7 | Step Ellipsoidal Function |
f8 | Rosenbrock Function, original |
f9 | Rosenbrock Function, rotated |
3 Functions with high conditioning and unimodal | |
f10 | Ellipsoidal Function |
f11 | Discus Function |
f12 | Bent Cigar Function |
f13 | Sharp Ridge Function |
f14 | Different Powers Function |
4 Multi-modal functions with adequate global structure | |
f15 | Rastrigin Function |
f16 | Weierstrass Function |
f17 | Schaffers F7 Function |
f18 | Schaffers F7 Functions, moderately ill-conditioned |
f19 | Composite Griewank-Rosenbrock Function F8F2 |
5 Multi-modal functions with weak global structure | |
f20 | Schwefel Function |
f21 | Gallagher's Gaussian 101-me Peaks Function |
f22 | Gallagher's Gaussian 21-hi Peaks Function |
f23 | Katsuura Function |
f24 | Lunacek bi-Rastrigin Function |
Only f1 and f5 are purely quadratic or linear respectively.
See also N. Hansen et al (2010): Comparing Results of 31 Algorithms from the Black-Box Optimization Benchmarking BBOB-2009. Workshop Proceedings of the GECCO Genetic and Evolutionary Computation Conference 2010, ACM. 1)
The current 30 noisy test functions are