a5

[Lonnie Princehouse]

People have been asking questions of the form, "Do I have to do XYZ for assignment 5?"

So I'll try to clarify our expectations for this assignment...

Your README should explain your naturalist algorithm at a high level, and the reasoning behind your design decisions. You don't need to get a positive score, but your algorithm should be able to reliably collect all of the animals for any map we throw at you, given enough time.  (The RandomWalkNaturalist example doesn't satisfy this, because there is a small probability that it will never end; it could go back and forth between the same two nodes forever).  

There's no strict requirement to use any particular algorithm from class, but if you invent your own ad-hoc algorithm, you should be able to explain why it's a good choice. Bogo-algorithms that don't really work are not likely to get full credit. It is possible to get a positive score by using only algorithms we've covered.

And since this is the final assignment, we expect your implementation to be copacetic =)  Write your best Java.

[Tim]

With minimal coding I am able to get the program to collect all the
animals. However, the algorithm isn't the most efficient because it
returns a score of -110890, and a time of .16 seconds. So my question
is, is the program graded just on the fact that TheNaturalist can
successfully collect all the animals, or is it graded for some level
of efficiency also?

[Daniel Wyleczuk-Stern]

Can you tell us what the "good" scores usually are? If getting
positive is good, how positive is really good? Essentially what I'm
asking is what are historically the best scores?

On Nov 24, 2:00 am, Lonnie Princehouse <lon...@cs.cornell.edu> wrote:

[Peter Tseng]

Well, historically we don't really know anything because we haven't done
this game before, but on a map with 41 animals, theoretically your score
could be 5100 + exploration - moves.

If it helps give any idea, my implementation gets more than 5100 on the
default map (random seed 0), and I think it's possible to do even better
than mine because I haven't tried as hard as I could to optimize my path
finding. There are other maps where this score is much lower: on random
seed 1234567, I only manage a score of around 3000.

Peter

[asong]

Quick question:

For the fastest clock time part of the competition, is that the time
it takes for the program to run with --headless
or is the least number of moves?

[Lonnie Princehouse]

Clock time with --headless.  It's not necessarily the least number of moves; it's quite plausible that the program with least number of moves might be using an expensive algorithm that takes a lot of clock time to think about each move.

--
You received this message because you are subscribed to the Google Groups "Cornell CS 2110" group.
To post to this group, send email to cornell...@googlegroups.com.
To unsubscribe from this group, send email to cornell-cs211...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/cornell-cs2110?hl=en.