[Bbob-discuss] Request for BBOB algorithms.

Olivier Teytaud olivier.teytaud at gmail.com
Tue Feb 28 00:09:12 CET 2012


Dear Niko,

No, it's not that I am not entirely satisfied, or more precisely it's just
that I think there
is always room for improvement :-)

 In particular, when I downloaded raw data and made a few graphs, I got
very different results from the graphs I've seen in BBOB, so I felt I
should take time for downloading codes, reproducing experiments by varying
experimental conditions, to see stability of results.

 Please don't consider that comments
are aimed at claiming that the current status is not satisfactory, and see
this positively: it just means that I find the testbed interesting enough
for discussing it.

Thanks for pointing out that there are noisy functions with noise
non-decreasing to zero; this I was not aware of, and I find this
introduction great. I was in a early BBOB discussion a long time ago and
this was my main point then, so you see that my comments are not that bad
:-) . Maybe I was not aware of the introduction of these
non-zero-noise-at-optimum because the link
http://coco.gforge.inria.fr/lib/exe/fetch.php?id=bbob-2009-downloads&cache=cache&media=download2.0:bbobdocnoisyfunctions.pdf
on http://www.lri.fr/~hansen/publications.html is dead :-)


 Yes, I know that we can change the dimension very quickly in the code,
obviously.
The point is just that usually you don't present this kind of result. This
is precisely
why I'd like to run this, but if I write the interfacing myself, it will
not be very fair. This
is why I'd find that interesting, for testing the stability of algorithms,
to have the exact conditions in which the algorithms are run.

I agree that it's not E log(fitness), I was imprecise; but log(ERT) vs
log(running time) is somehow emphasizing long term results and very precise
results, somehow similarly to E log(fitness-fitness*) vs E(running time).

For the 32bits/64 bits: some years ago, one of the organizers told me that
some of the algorithms used by the organizers were run in 32bits, and some
in 64 bits. Maybe this has been corrected in later runs - I was far from
BBOB recently :-)

Best regards,
Olivier
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lri.fr/pipermail/bbob-discuss/attachments/20120228/c0eb7541/attachment-0001.html>


More information about the bbob-discuss mailing list