5
$\begingroup$

Let's suppose there are $n$ real numbers $a_0 < ... < a_n$ uniformly selected from interval [0, 1). If one knows $k$ numbers on consecutive positions $a_i < ... < a_{i+k-1}$ how good is $(k - 1) / (a_{i+k-1} - a_i)$ an estimator for $n$? What other estimators are possible/better?

NOTE: $n >> k$.

  • 0
    What if there are no consecutive numbers?2011-01-17
  • 0
    @Trevor: Edited for clarity. Does it answer your question?2011-01-17
  • 0
    How about the error margin for 95% confidence?2011-01-17
  • 1
    if we increase $n$, how does $k$ increase?2011-01-17
  • 0
    n & k are not correlated.2011-01-17

3 Answers 3