Brokener Keys

[snibril]

My program only works for numbers with a difference smaller than 300. Has someone a solution without this limitation?

[tog]

I think you could enlarge the difference by just multiplying another 9 . But I wonder if there is a solution that isn't limited to some range of numbers.

[efe]

My program is limited by the size of the memory.
It works only for numbers in the range [2..16384].

Due to the cycle limit I had to narrow it down even more to [2..1024].

[m!nus]

numbers are 1-256 so no need for bigger limits.
did you do it with lots of spaces aswell?

[efe]

No, there are no spaces in my program.
It's only 24 instructions long and it can be made even shorter.

[snibril]

I don't like my solution:

1<0<1<-25*0^*3*+2*c {Spaces till pos. 600}p! {min. 600Spaces}0<p

[m!nus]

efe: i'd love to see how your method works, because mine, with its 2KB of spaces isn't very nice.

[efe]

OK, here is my solution:

Code: Select all

...

It's pretty neat, isn't it?

BTW: To print the minimum without :,? and / instructions I only need 18 instructions:

Code: Select all

...

[snibril]

I had a similar idea, but I didn't find a function that always give me a zero except for one specific value. I didn't think to use the memory for this.

[the_impaler]

OK, here is my solution:

Code: Select all

...

It's pretty neat, isn't it?

BTW: To print the minimum without :,? and / instructions I only need 18 instructions:

Code: Select all

...

Nice.
I have similar solution but the loop is in other direction and so it's 60 instructions long.

Code: Select all

.....

I had to take special precautions for neighboring numbers and for equal numbers
It works for numbers starting with 1 and with difference no more 1242 to fit into cycle limit.

[Karian]

I more or less had the same solution as snibril, the same concept anyway. I can find a nice evolution is my solutions for this line of puzzles. They are getting more and more specific to the input that is used in the tests.

For the first puzzle, I had one that even worked with negative numbers, only limitation was that the difference squared is still an integer. I haven't tested it with large numbers myself, but I think it was still making the cycle limit.

For the second one it only worked for positive numbers, but the limitation about the difference dropped. The code was actually a lot nicer then the first challenge, and I don't see any problems with the cycle limit for any size of numbers.

And now for the last one, it is limited to the small range that the test cases are made of. When the range extends, it will become pretty hard to get it working within the cycle limit.

[theStack]

Nice.
I have similar solution but the loop is in other direction and so it's 60 instructions long.

Loop? It is possible to do loops without condition testing? Hm good to know... thanks for the hint
Well all I can imagine is infinity loops but somewhen you have to stop.

I had an idea about implementing the challenge's solution quite nice and short but I just thought loops are not possible anyway so I forgot that idea and did it with the dirty filling spaces solution (mine is >7k, I considered that the differences between the numbers could be quite large)

It's a pity that the codes are not there anymore but I guess they would reveal too much for the "Brokenest keys" challenge

[therethinker]

My code, I guess, was like efes. It **** until it finds a number.

Code: Select all

...

EDIT: After reading the Brokenest Keys thread, apparently this technique gives it away

[jonik555]

This I invented for challenge before, and it worked perfectly also in this

Code: Select all

20<>11<>32> 2<0^<8*g1+2>56+c        2<<1-<p!

[arthur]

my solution only works for positive integers
and require the small number smaller than the maximum cycles allowed

Code: Select all

0<1<+60<61<00>01>>>00^<g1+45*c-p