Group items tagged

is this really as revolutionary as praised? optimisation guys please ... full paper here: http://arxiv.org/pdf/1508.04874v1.pdf

They use a 'separation oracle' meaning that the paper is theoretical.

Combination of the radio occultation method plus change in the orbital plan in order to observe new target (the volcanic outgassing from Io satellite) by passing though the magnetosphere of Jupiter. It poses a challenge in the sensitivity of the electronics. Maybe it is accepted if the risk is low. Idea for the next GTOC challenge maybe? Optimising the cost on the electronics vs the frequency of passages?

1) it has to actively work and produce a result that's what you intended 2) the design must be robust. 3) it should be efficient 4) it should minimize unintended consequences. "But we have to establish a formal, mathematically precise mechanism to measure complexity and adaptability . . . [where] adaptability means the system elements have sufficient margin, and can serve multiple purposes." "We need to break the paradigm of long cycles from design to product" some interesting questions....

what is the outcome of the CDF study ? I think actually that optimisation is not at all the key point. As it is stressed in this news, it is robustness (points 2 and 4). This is something we should think about ...

SYSTEM OF SYSTEMS, SYSTEM OF SYSTEMS!!! :-D

The usefulness of having a robust trajectory :) ... They have to wait 6 more years for another date with Venus ...

"optimised trajectory" of course not, robust definitely! It was the subject of my paper presented at the AAS (the one in San Diego) "Designing robust interplanetary trajectories subject to one temporary engine failure". The problem here is that they do not have enough fuel for a correction maneuver that would allow to come back to Venus earlier, and break for a VOI. A robust scenario could have alloted the best amount of fuel and time to be able to recover from almost all possible unplanned events. In the paper, I introduce some confidence regions such that I get the robust control for p% chance of mission success in case m% chance of problem with the propulsion system.

You should run your method on this scenario and see if you could get a trajectory with a shorter come back time using the same spacecraft.... would be a big selling point for a new trajectory design approach

a follow-up question: Can quantum computer be efficient for global optimisation ?

Via slashdot. Is that new? Must be of interest for global optimisation.

related conspiracy: politicians know that P=NP but they do not let us know the proof

did you put money on this? I would love to see it :)

A real scientific study...

For Dario, as they quote him heavily...

Yep, I was told already a few months ago from Storn that this is indeed the best adaptive DE ever made. It is already in the road map for pagmo 1.2. According to pagmo it humiliates CMAES !!!

And in case you didn't notice, the first author is an undergraduate student.

For PyGMO 2.0 ...

any idea how their brain manages this ?

Called the reflex arc: http://en.wikipedia.org/wiki/Reflex_arc

On this note, there is the stunningly optimised flight path of the Tiger beetle http://www.sciencedaily.com/releases/2014/04/140415133815.htm

Dario - perhaps worth giving a look to be up-to-date... There's even an article "Improving Classical and Decentralized Differential Evolution with New Mutation Operator and Population Topologies". They quote our CEC paper, but not the ParCo.

Isn't it a bit a paper-killing quote?

:) It's in the context of a review of the work that's been done about DE with island model in general, they don't evaluate. Pity they didn't refer to the ParCo article on topologies, as it was a bit more extensive and more focused on the method (as they do in the article) rather than on the problem (as was our CEC paper, if I recall well).

of any interest to us?

This is an "old" stuff. It is used, as far as I remember, in IPOPT, the Interior Point OPTimiser, that I use quite often (e.g. gtoc4) ...

for the ACT architects .... Can we do the same for the Moon Village? We brainstorm on some mathematical simplified objectives for growing the settlement (taking inputs from the modular growth, resources, terrain suitability etc ....), we define some simple rules for growth and we optimize. ..... easy peasy (i am serious)

i agree, with most of the parameters that would actually be really cool. but doesn't it get very messy once economy plays a large factor?

We can start studying the ideal case, or add also some economical constraints on the settlement layout ...