Talk:Last letter-first letter: Difference between revisions

m
(→‎Number of names: Rules are rules)
m (→‎Number of names: mathemagic)
Line 2:
This is equivalent to asking for the longest path in a directed graph, which is an NP-complete problem. That doesn't rule it out as a task, but running it on a graph with 646 nodes might take too long. Also, since it seems to be rare to use external files to store the data (or at least I haven't seen many tasks that call for that) having all the names in each example would get really cluttered. [[User:MagiMaster|MagiMaster]] 20:52, 5 June 2011 (UTC)
: Re input data; I can always host input data if that makes it convenient to work on appropriate data. --[[User:Short Circuit|Michael Mol]] 23:12, 5 June 2011 (UTC)
:: That's not really the problem. A quick breadth first search implementation gave me 646 pokemon names, 1793 pokemon name pairs, 518546 pokemon name triples, 14745709 pokemon name quadruples, and I had to reboot my machine to recover from my attempt to represent all pokemon name quintuples. Inspecting this sequence suggests a search space on the order of 28^<sup>28</sup> or about 3e403&times;10<sup>40</sup> sequences to investigate. That exceeds the storage capacity of my laptop. (Not to mention, exceeds my patience). Note that finding a long sequence is not very hard, it's finding the '''longest''' that makes this task hard. --[[User:Rdm|Rdm]] 20:08, 6 June 2011 (UTC)
::: If the part about using all the pokemon names was changed, it'd be an interesting task. 20 or so names/words/nodes should demonstrate the process well enough without being too slow. [[User:MagiMaster|MagiMaster]] 21:37, 6 June 2011 (UTC)
 
Anonymous user