What is the largest dimension of an unbounded continuous function tried in literature for testing optimization techniques? I mean benchmark functions such as Rosenbrock.
Is this a "big data" type question, in that, are you interested in the largest unbounded continuous problems ever *solved*? Or are you interested in functions for testing which have variable dimension (e.g., you can choose the dimension)?
This paper by Yu. Nesterov has problems up to more than a million variables, but I am sure that there are other papers with larger instances:
http://epubs.siam.org.proxy.lib.chalmers.se/doi/pdf/10.1137/100802001
The reason why I believe that must be the case is that I have written a paper on a constrained problem where the largest problem solved has more than 100 million variables.
Robert,
I meant functions for testing with variable dimensions. Thanks.
Michael,
I couldn't get the paper. Could you send me the title of the paper? Thanks.
If you mean my paper, it is awaiting approval after a revision, so I cannot distribute it.
Then could you send me the title of the paper by Yu. Nesterov? I can't have access to the contents of the link. Thanks.
Ok. :-)
EFFICIENCY OF COORDINATE DESCENT METHODS ON HUGE-SCALE OPTIMIZATION PROBLEMS
Hi Ehsan, in that I case I suggest the functions in
http://people.sc.fsu.edu/~jburkardt/m_src/test_opt/test_opt.html
that specify C
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Writing