Talk:Average loop length: Difference between revisions
→Ill Specified
(Ill Specified) |
|||
Line 3:
: I've watched the lecture; the problem in question is in the last 10 minutes and is only mentioned in passing. IIRC, it is at about 1:32:00 (and I forgot to note it down). If you've got the time though, it's ''well'' worth watching the whole thing, and the mathematics involved is extremely simple (yet talks about some very interesting properties of sequences). –[[User:Dkf|Donal Fellows]] 10:54, 9 January 2013 (UTC)
==Ill Specified==
The task, as written (and I have not studied the link) is completely ambiguous. If we had a specific f, there would be no problem. But because we cannot assume the simplest case (that each number in 1..N maps to only one other number), we cannot know how many numbers (on average) a number maps to. And, this is critical information. For example, we could have a mapping where most numbers map to most other numbers, or we could have a mapping where most numbers do not map to most other numbers - and either of these could be built using a uniformly random process. That said, it does look like a simple change could make the task description meaningful. It's something like: fN is generated using a random process which generates N+1 numbers, randomly selecting each from the range 0..N --[[User:Rdm|Rdm]] 14:05, 25 January 2013 (UTC)
| |||