Timeline for Recursive sequence of binomial random variables
Current License: CC BY-SA 3.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 8, 2015 at 23:40 | answer | added | Matthew Junge | timeline score: 4 | |
| Jul 8, 2015 at 15:34 | comment | added | mathjunge | This is a toy version of a more difficult problem. It's coming up in the context of the Frog Model. Fix an increasing sequence $a_k$. Add $k$ balls uniformly randomly to $a_k$ bins. Letting $e_k$ be the number of non-empty bins, we then add $k+e_k$ balls to $a_{k+1}$ (new/fresh) bins and continue the algorithm. | |
| Jul 8, 2015 at 15:31 | comment | added | mathjunge | My mistake. I'm interested in a left tail bound. Azuma's or Chernoff's inequalities seem like natural tools, but I'm, so far, unsuccessful appyling them. | |
| Jul 8, 2015 at 15:29 | history | edited | mathjunge | CC BY-SA 3.0 |
I want a "left" tail bound
|
| Jul 8, 2015 at 12:41 | comment | added | Douglas Zare | The right tail bounds seem much easier than the left tail bounds. For example, Markov's inequality says $P[X_k > (1+p+\epsilon)^k] \le (\frac{1+p}{1+p+\epsilon})^k$. | |
| Jul 8, 2015 at 12:14 | comment | added | Brendan McKay | Your "perhaps" is a left tail bound, but you ask for a right tail bound? | |
| Jul 8, 2015 at 6:34 | comment | added | Johan Wästlund | Seems to me like a nice question, what is the context? | |
| Jul 8, 2015 at 1:17 | history | asked | mathjunge | CC BY-SA 3.0 |