Re: Proving Veritasiums riddle using Perl

[pfaut]

Very interesting problem and very interesting algorithm. Statistics are very often counter-intuitive.

I think setting up the boxes might be easier and faster with something like Array::Shuffle.

[cavac]

I think setting up the boxes might be easier and faster with something like Array::Shuffle.

Yes. And it would be way faster. My randomizer as probably the least efficient version possible.

The other variant would probably to just generate the array of numbers, then randomly select and splice.

I just wanted to do a qick proof of concept without looking up anything and hand-rolled algos (without modules) so i can easily play with every part to see what happens. I do Perl 40 hours a week at work¹, this one was a just-for-fun little home project without any code quality and/or maintainability requirements ;-)

---

¹ all to do with Point-of-Sales stuff and financial data. I nearly never get to use rand() based stuff, so i'm a bit rusty in that regard

[pfaut]

Cool program. It's much faster using a shuffle.

Your code:

% time perl prisoners-orig.pl
== SIMULATION 9900 of 10000 ==
Success:   3085
Fail:      6915
Fail rate: 69.15%
perl prisoners-orig.pl  13.09s user 0.01s system 99% cpu 13.101 total

Modified to use Array::Shuffle:

% time perl prisoners.pl     
== SIMULATION 9900 of 10000 ==
Success:   3175
Fail:      6825
Fail rate: 68.25%
perl prisoners.pl  0.71s user 0.00s system 99% cpu 0.716 total

Diff:

Read more... (1035 Bytes)