hacker.org

I find the Mortal Coil Puzzle to be boring ... all you have to do is to partition the board into sections which have only 2 entries/exists and thus only 1-2 solutions and then combine them it seems?

Here is the code for C++ for a depth of 3 ... testing it with std::bitset. Using bitsets speeds up things dramatically. I'll now work on the weighting of the new heuristic and I'll also do some early-outs based on precomputed states which won't clear the outermost tiles. // addPiece const void addPi...

So still no hint from you guys? :P It seems as if representing the board as n=$depth-1 bitsets speeds up things dramatically ... Getting the required "flippower" of a board is a matter of calling the CPU's builtin "popcount" instruction (returning the number of bits set in the bi...

This was "inspired" by this post about ways to hash a board game's state ... http://stackoverflow.com/a/4301869 ... but using 2-3 sequential bits to represent a tile will cost a lot of computation (for individual access and manipulation) and so using a separate vector of bits (i.e. uint128...

Any hint on what you did after the 43-ish levels? You mentioned Zobrist hashing but that's not really speeding up things - does it? I just found out that the boost library now features very efficient "simulated" 128bit and 256bit int types and I'll try to represent the board as n=$depht-1 ...

Same goes for the OneOfUs Challenge. One of the most performant HP/HC solvers out there runs for a few minutes on the last level. Perl is at least 4-5 magnitudes slower ...

Well if you genuinely believe that the language doesn't make any difference then good luck. Try solving the Crossflip challenge with Perl on a standard computer ...

Who really cares? The site is dead and the ranking never made much sense.

The point is to improve your programming skills and your overall intelligence ... well at least the one the academic world cares about ...

After talking to 2-3 people I can confirm that they managed to solve levels beyond 44 using a totally different approach than iteration/recursion with pruning/backtracking/early outs. I expected something like that because even with a heuristic that prunes 2/3 of the entire search space lvl 44 alrea...

I am getting "your solution sucked" for every valid solution I submit lately ... not just for the Pusherboy game. Solution is correct and gets accepted (i.e. I progress to the next level) but I still get the error anyway ...

http://search.cpan.org/~ether/WWW-Mecha ... chanize.pm

Also I don't understand why people would help out with this problem. If a person lacks the very basic skill to automate simple http requests it's probably a good thing. We got enough ego driven flooders and script kiddies already.

IANA is no ISP and this site is about improving your skills and understanding in positive and productive ways and not about personal disputes with your email friends.

I even tried using a machine with 300gig of memory and 100cpus. Precalculating a lookup table for the smallest 5-6-7 pieces and using the inc-power based early out does not seem to be enough. At least not when you use every tile instead of using a tiles%2 or tiles%3 approach. I am currently busy wit...

Also forget Perl! I am also a Perl veteran but the problems here get so big that Perl just is the wrong choice in terms of speed, memory management, robustness, etc. ... you will just get frustrated. When I ported my Modulo solver from Perl to C++ I solved another 15 levels in like 10 seconds when P...

There are very performant algos out there and even binaries taking a specific input format to solve the HP/HC problem. Why re-invent the wheel when you already identified the kind of algorithm you need?