However, we can see that on some of the boards, the winning entry did significantly better. The boards that it won on were those with the large valued 80’s randomly around the board.
Surprising to me, the 1000 node solver did noticably better on a couple of the tall skinny boards like this:
That leaves me wondering if someone could have done a quick check to see if the board had 30 colors or so, and then tried the simpler solver. It looks like this would have made some significant gains… A good contest, congratulations to all that participated.
To leave a comment, please click here to sign in to your MathWorks Account or create a new one.