Patience

lechat · Post by **lechat** » Fri Sep 05, 2008 5:18 pm

Hi adum!

I liked you Patience Challenge!
I could really sit back and relax for... For how long exactly?

My estimate is 8 years give or take

...This would really be THE CHALLENGE!!!

Post by **adum** » Fri Sep 05, 2008 5:26 pm

the_impaler · Post by **the_impaler** » Sun Sep 07, 2008 1:09 am

There are only 2 options left then :
- use the future Java version (the one that used in "Portal to the Future")
- make gerbils to spin the wheel faster

lechat · Post by **lechat** » Sun Sep 07, 2008 7:34 pm

Oh! Great!

Thank you for the hint about Portal to the Future!

I solved it!

michuber · Post by **michuber** » Mon Nov 24, 2008 7:29 am

Have i missed something?

I removed the time consuming line from the for loop and found many l's in seconds that would pass the while. But none of them is the correct answer.

Where am i wrong?

gfoot · Post by **gfoot** » Mon Nov 24, 2008 9:36 am

Me too. Do we need to find the particular one that would come out of the seeded RNG first?

MichaBln · Post by **MichaBln** » Mon Nov 24, 2008 12:16 pm

Hi,

I solved that challenge by optimizing and thinking about what really is relevant for getting the right answer ... still it took a few mins till it printed out the right number ... so if it took just secondes and you didn't do more than taking away that slowing-down-lines inside the loop ... i guess something else is wrong.

michuber · Post by **michuber** » Mon Nov 24, 2008 2:20 pm

I found the last 5 digits xxxxx and tested xxxxx+i*100000. For i=111 the last 9 digits match (took 5 sec.). So this should be a correct answer... The first number from the RNG was i=7670024504486xxxxx, incorrect too...

MichaBln · Post by **MichaBln** » Mon Nov 24, 2008 2:36 pm

Hi,

I don't really get why you test the numbers the way you do ... all i can tell you, that the correct number occurs much later in the sequence than i=111, MUCH later. I can't tell you how much later, that would spoil to much.

I don't understand what you are doing (but i guess alot of approaches might work here ... so your approach might be correct).

This is really about optimization, so keep the approach from the original and just speed it up.
Ok, here is one major hint: BigIntegers are very slow ... and you can solve this without them.

If anyone thinks that spoils to much just tell me and I'll edit that part out.

snibril · Post by **snibril** » Mon Nov 24, 2008 9:50 pm

If you speed up this than you get the answer after:
snibril@hera:~/hacker$ time java Patience
XXXXXXXXXXXXXXXX

real 0m0.546s
user 0m0.452s
sys 0m0.044s

homeas · Post by **homeas** » Sat Jan 31, 2009 8:41 pm

whoever uses J# from Microsoft (I do it for laziness with VS 2005 ...)

-- oh dear, he/she will get a completely different solution.

there is obviously no common RNG ...

gfoot · Post by **gfoot** » Mon Feb 02, 2009 3:43 pm

The Java Random class is explicitly defined to use a certain algorithm, in order to assure repeatability of the generated sequences regardless of implementation or architecture. I'm not a Java expert, but I'd guess that if J# veers from the standard implementation then it's broken - maybe it provides a compatibility option though?

That said, maybe J# is documented as being based on a totally different implementation of the standard library. I dunno. I guess J# just isn't Java, in the same way that "managed" C++ is not C++.

megabreit · Post by **megabreit** » Mon Feb 02, 2009 7:05 pm

Erm... I don't get it... but from my point of view the solution is completely independent
from the RNG. The RNG defines just the starting point of each calculation.
One should find the same solution just by trying all possibilities with brute force...
although this would take "a bit" longer to computer.

nto · Post by **nto** » Thu Dec 03, 2009 9:18 am

megabreit wrote:Erm... I don't get it... but from my point of view the solution is completely independent
from the RNG. The RNG defines just the starting point of each calculation.
One should find the same solution just by trying all possibilities with brute force...
although this would take "a bit" longer to computer.

This would only be the case if there is exactly one number that satisfies the criteria.

KindeX · Post by **KindeX** » Wed Sep 15, 2010 1:26 pm

I think approximate every 1 000 000 000th number fits the criteria.
So the question is if the checker accepts only one predefined correct answer, or checks every every answer with some algorithm.
I've found many numbers that fits criteria, but site does not accept it.