deck.txt

Is it anyone that has a deck.txt file that is possible to solve? I would like to test my program.

I found one my self:
QC
9D
1D
3D
10C
2D
5H
5C
QS
JS
7H
KS
JH
QD
KD
JC
KC
4C
9C
7C
3C
2C
8C
4D
JD
1C
8D
10D
6C
7D
6D
3S
6H
5S
8H
5D
1H
10S
2H
3H
9S
1S
8S
4H
2S
6S
KH
7S
4S
10H
QH
9H

My code can't manage this one, and yet it does do a logical move at every turn and never misses anything.

This brings up an interesting point: klondike isn't all that simple a game, and there are many ways to do things. Coding what "feels" right at first gave me a score of 210 in 65 turns. I rearranged the logic a few times and found that doing completely legal moves could also yeild 195 points in 61 turns or 140 points in 61 turns. All these were with the same deck, specified in the previous post.

I'm not 100% sure my score is being calculated correctly, but I am sure my program is only making legal moves and isn't "missing" any moves. What I end up with at the end are three entirely different pictures.

My question is, how many decks will be used to test our programs? Just because 2 or 3 end up giving worse results doesn't mean they will in the long run.

I surely will be making my program take multiple paths to see which one turns out best. But if my code runs 4 times slower and gets on average 10 or 15 points more, is that a worthwhile tradeoff? Or should I focus on speed instead? What's a good speed to shoot for, 100 ms?

A program that scores more points will always outrank a program that scores less. Regardless of speed.

That said, yes I will run multiple decks. Most like 4 decks. Each deck will be run through multiple times. Averages will be taken.

The key to this whole challenge is making your program choose the best legal moves in the best order ;)

> "program take multiple paths to see which one turns out best. "

I think it would break the rule:
> There is no cheating allowed. Your program can only take into account the cards to are "face up" at any given time.

IMHO, if you are playing several times with the same deck, you are cheating, because you are seeing hidden cards twice.

On the other hand the algorithm here is the most
important thing in this contest, and I feel that it is not right. It is like the hidden auction where every $0.01 counts.
Maybe there could be more that one winner? ;-)

BTW: that textarea field in the 'Post reply' HTML form should have more columns.

If anyone is interested:
I have uploaded PHP code
for generating random list
of cards (deck.txt file).
It is available in the Miscelanous code area:

http://codewalkers.com/seecode/101.html

However, as you can effectivly go through the deck 3 times, you should be able to remember the order that the cards (are already in). So you may wish to use a basic set of rules to go through the deck the first time, then pause while you use your new knowledge to work out a better stratagy, then use the next two passes of the deck to get a higher score.

Z

Quote:
You will have 3 passes through the stock

it is stock not whole deck.

Since you go through the stock 3 cards at a time, is it cheating to remember the in-between cards?

is the above posted deck-file really able to be solved? my program is not finished yet and it doesnt solve the deck-file. i only get 130 points :-(
:-)

I don't see anything in the rules against that. If you were really playing, you could remember those cards. In fact, professional players (yes there is such a thing for this game) probably do remember those cards....

The above deck.txt file is not solvable within the rules. You need to pass it 5 times to solve it (atleast I did.)

Below is 2 outputs from my code:
1) The maximum number of passes was 10 and the game was solved in 5.

The Game has been solved!
Total score: xxx
Total cards moved: xxx
Number of stock passes: 5 of 10 possible.
Creation time: x.xxxxxxx seconds.

2) The maximum number of passes was set to 3 and the game was not solved.

The game could not be solved!
Total score: xxx
Total cards moved: xxx
Number of stock passes: 3 of 3 possible.
Creation time: x.xxxxxxx seconds.

I've removed some info from the outputs.

"A program that scores more points will always outrank a program that scores less. Regardless of speed."

Does that mean if I create a recursive program,
which will always come with a best solution, but
will take ages to finish will always be the best?

"A program that scores more points will always outrank a program that scores less. Regardless of speed."

I think that statement is pretty clear.....

But to answer you, yes, if your program can gain more points than any other program, you will be the winner. If two programs score the same amount of points, then it will come down to speed.