12/8: Class Notes

What did people think of 7 letter word?
Daniel: It was helpful to have previous reports, previous code less helpful
Neha: It was helpful to know the approach of previous years. Knowing what not to do helped to figure out what to do.
EPK: Maybe it would be interesting to repeat multiple projects
Dan: We wasted a lot of time getting an architecture together when we could have been writing the code.
Elba: It was helpful that the first project was text based (and not geometric)
Flavio: It was nice that we could get some basic results quick

Mosquito?
EPK: I liked that it was visual
Daniel: It was interesting that nobody got a good discrete solution
Everyone thought that the competition was fun

Push?
Elba: Everyone kinda plateaued
Yang: It was an interesting game, that models what we see in real life
Neha: The piling was terrible

iSnork?
General consensus - many variables, very complex, quite a bit of room for further work

Most people voted to extend project 4 for next year

12/6 Class Notes

Tournament parameters:

Group 4 will create an extra player that does just danger avoidance, and then comes back to the boat.

D:
30, 40
R:
5, 8, 15
N:
1, 2, 7, 20, 30
Maps:
Lochness
Mines and treasure
G2 benchmark
Lochness, but it's happy, not dangerous
Piranha
Lochness, plus 3 happy creatures of the same species
G5 VERY happy
1 or 2 static happy creatures
Piranha plus a few good creatures
More than 26 species

1 map with very dense wildlife many different species, d=60

We'll narrow this down to 8-10 boards

How many different random seeds?
~100-1000, depending on time it all takes (will maximize # seeds to fit in 4 hr run)

We'll use a different configuration today...
D=40, R=8,N=7, Seed=1,291,659,224,861

G1: 1023
G2: 1591
G3: 1330
G4: 702
G5: 1217
G6: 836
G7: 1153

What did we learn today?
John: Cover the board!
Daniel: We are overly conservative when estimating the number of creatures instantiated
Lauren: We are overly optimistic with the same thing!
Flavio: Better to be home early than to be home late!


Weds: Final code due by noon. Course wrap-up.
Mon: Presentation and report. and peer review

12/1: Class Notes

Schedule update: We are swapping the class wrapup and presentation days. So presentations on the 13th. Course wrapup on the 8th. Final players due at noon on the 8th.

We ran a d30 piranha map:
G5: -887,000; 13 didn't make it home
G1: 35,000
G2: 35,000
Lauren: We stay in the boat until there's no danger in the radius.
Flavio: This is going to keep you locked in the boat!
G3: -21,000
Jesse: We got the max, then we didn't go back, so we lost it all
G4: -1,347,000
Zach: We don't properly avoid danger due to a bug
G6: -71,000; 7 DSQ
You should have a fuzzy condition test to see if you have the max

Mines:
G5: 0
G1: 5740
G2: -156
Whats a good meta strategy to realize that your current strategy needs to be changed?
David: Make milestones, and see if it's taking too long to ge there

Neha: Do path finding through static dangerous creatures

G3: 20740
G4: 1740
Zach: We should be exploring... but retraced our steps for some reason
G6: 1740

Lauren: It's important to have different strategies for static/moving danger.

Lauren: Visiting all of the creatures will be NP-Hard (traveling salesman)
DFed: Just use simple heuristics (see the closest, etc. only need to see 3 of each creature)

G5: 5005
G1: 4785
G2: 5321
G3: 6275
G4: Crash
G6: 5853