Login | Register

RoboZZle Forums

Forums / Puzzles / 4 bits

Thread for puzzle 4 bits. Hints are OK, but please no full solutions.

 

Log in to comment.

Getting this one to 25 was very simple once you use the correct algorithm. No hack needed, the solution is general.

edsaw, 106 months ago.

I think I've found those now; unfortunately they've lost the nice structural symmetry that I was trying to hint at.

keba, 108 months ago.

Sorry for my flooding this thread. I applied new method to 1086 and 1087 and got lengths 19 and 25. By the way 25 doesn't seem to be the shortest possible:)

Smylic, 108 months ago.

I found strength to try 1091 once again (it kept me on the alert). And I found out solution with length 12, but it was easily optimizable. Now I have length 11:)

Smylic, 108 months ago.

Thanks for hint. I've easily found optimization for 1131, but still don't know how to apply it to 1091 or other puzzle. Some day I'll think more about it... Or I'll implement good brute-forcer which would help me:) (for a long time I have many ideas about it, but I'm too lazy to implement)

Smylic, 108 months ago.

The technique is also used by the length 13 solution in 1131 (in that one, F1/F2 handle only the turning, the binary logic is all in F3/F4).

keba, 108 months ago.

You removed my question where is difference. Now I see that I miss some optimizations in every binary puzzle. I returned to 1091, optimized from 18 to 14, but don't see any optimization permitting to reach 12. It seems I need to invent some other technique of solving binary puzzles if I want to get shortest solutions...

I see boasting about beating record is rather bad idea, 'cause some players try to beat it again. And my solution looses status of the shortest one. But this adds additional interest:)

Smylic, 108 months ago.

It continues to amaze me how much these binary puzzles can be optimised. Revisiting this one with the optimisation edsaw found in 1091 gives impressive results.

keba, 108 months ago.

@markbyers: My compliments for your ability to solve 1-bit puzzle using 2*1-2 = 0 functions:)

Smylic, 108 months ago.

I used several optimizations in puzzle 1086, but still have one excess command. In this puzzle I had slight problem to fit solution, but my first successful attempt tied shortness and trivial optimization allowed to beat the record. Still don't know why this two puzzles are so different for me...

Smylic, 108 months ago.

well, more accurate would be to say that i use 1 function for each bit plus f1 plus return, only that as the last bit is simple i dont need a function for it, so, instead of n+2 functions for the general case i need n+1 (n-1 from bits, f1 and return)

Gercho, 118 months ago.

One function for each bit? You guys are clearly doing it a different way from me. I must be missing something. Hmm...

markbyers, 118 months ago.

I did the exact same thing I had already thought I was trying to do, but this time it somehow worked, so for whatever reason the old way must have had a bug I was not seeing, and this time I didn't introduce the bug. I didn't have any revelation of something I was not seeing after all.

beaza, 118 months ago.

well, i adapted easyly my function from binary to this puzzle, i did just one more optimization that i think i could have used on binary but didnt realized back then though i m not sure, my f1 are practically identical for both (only a little larger for this one obviously)
and i use one function for each bit + 1 for return in both puzzles, in binary used one for optimization to be able to fit the others in the space available since i needed 1 large return function.

Gercho, 118 months ago.

easier when you solved the other one...

gercho's comment after solving the puzzle, 118 months ago.

I used the same general method. In both cases the majority of the difficulty was in getting the solution to fit into the provided space. In the first case the difficulty was due to the unusual funciton lengths. My first solution required four functions. It took me a while to see how I could make use of the functions of length 2 and 3.

In the second problem my extrapolated solution required six functions but even in designer mode I was running out of space. I think in general for my solution I would need 2*n-2 functions to solve n-bits, assuming the functions are of unlimited length.

I found a way to get it to fit in five functions, and I think it would work in general but it requires a lot of commands in F1 (more than 10). Most of the time I spent on this puzzle was on finding ways to move commands from F1 to other functions.

markbyers, 118 months ago.

There are dirty dirty hacks in both my solutions, given the length of 37 I think I know what you have missed.

You found this one harder, even after solving the previous puzzle? Funny, I found it quite easy to extrapolate my solution. Did you use the same general method for the two puzzles?

eBusiness, 118 months ago.

Nice puzzle. I found this one even more difficult than your first binary counting puzzle.

I think I am missing some trick in both these puzzles. Your shortest solution is shorter than mine in both cases. This time you were slightly more generous with the function lengths. Because of this, I was somehow able to cram my solution into the provided space. :)

markbyers, 118 months ago.

Previous page (1 / 1) Next page

©2009 Igor Ostrovsky and other contributors. Terms of use