nasa caltech_jpl
GTOC X
10th Global Trajectory Optimisation Competition

May 21, 2019, 8 p.m. UTC

June 12, 2019, 8 p.m. UTC

Timeline

The competition is over.

Score using any number of fast ships?

avatar

DietmarW

Did anyone else try to determine the maximal score with all constraints of GTOCX intact beside the number of fast ships? We plan to continue to work on this interesting problem after the competition and this could be an indicator if we are "on the right track". We get about J=4000 without the fast ships <= 2 constraint. With real scores now > 2000 we have doubts that this is the real limit. Our idea is to use the algorithm we have as a tool to evaluate initial settler distributions with a legal number of ships starting from Sol to determine where to settle next in the initial phase.

avatar

IngoA

Hi Dietmar,
nice to read you. Are you back from vacation?

And questions for clarification:
* You believe to have an upper bound for the variant
with arbitrarily many fast ships, right?

* What would be your upper bound, if you have ony "slow" settler
ships, but arbitrarily many?

* And what would be the score if you settle all stars and have
B-value = 1 ?

Cheers, Ingo.

avatar

DietmarW

Finally back from vacation. The score settling all 100000 stars is about J = 7. I don't claim to have an upper bound but just a result my algorithms achieves. If a much higher score is achievable it needs improvement if it becomes part of a strategy to compute a real solution. This is the reason I asked the question. With slow settler ships my result is much lower, about J = 500, but I didn't try to optimise this.

I get the feeling NUDT followed really carefully what I wrote about game theory, submitting as late as possible is a winning strategy. Or they learned their lesson last time when JPL caught them at the finish line. Tsinghuas strategy to submit so early a competitor can react is a bit harder to follow. There was almost no bonus left at that time to compensate for the disadvantage. Would be interesting to here from Tsinghua after the competition why they did it. Ok, NUDTs actual 2300 are hard to beat anyway, but this was not known at that time. At least not to us. May be at NUDT there is a whistleblower?

What is interesting is that the NUDTs dv ratio is > 2.0 now. This is something I also observed with my unlimited fast ship experiment, a dv ratio even > 2.5.

avatar

DietmarW

Ok, 2615.15076 now with a "bad" dv ratio? Now I am confused. But nevertheless, congratulations to this exceptional result. Following what happened today I would predict: Its probably not be the final one. You don't need any game theory at all to win with such scores.

avatar

DietmarW

Was ESA at any GTOC competition in the past ever beaten by a factor > 2 ? Don't think so. Now its time to strike back if possible.

avatar

IngoA

Esa-ACT marching forward. If they will get rank 2 or better in the end
I will spend them a six-pack of Tsingtao beer (my favorite in China).

Ingo.

avatar

DietmarW

Only two minutes left and only NUDT is still very active. 3101, thats really impressive! This answers the initial question here: 4000 without fast ship limit is probably way too low.

Please log in to leave a comment.