-- nonconvergence of a series. with application in statistics
- From: elodie.gillain@xxxxxxxxx
- Date: Mon, 21 Jan 2008 15:39:44 -0800 (PST)
I am trying to show that
when Xi=+/- 3^(i-1) with probability 1/2 each, the mean Xbar goes
to +infty with probability 1/2 and goes to -infty with probability
1/2.
I have a hint telling me to evaluate the smallest value that (X1 + · ·
· + Xn) /n
can take on when Xn = 3^(n-1).
I suppose that the smallest value of Xbar_n when Xn = 3^(n-1)
must be reached when
X_{n-1} = - 3^(n-2)
X_{n-2} = + 3^(n-3)
X_{n-3} = - 3^(n-4)
....
X_{1} = + 3^(0)
In which case
(X1 + · · · + Xn) /n
=[3^(0) - 3^(1) + 3^(2)+......+3^(n-1)]/n
=[(1- (-3)^{n+1} )/(1+3)]/n
=[(1- (-3)^n)]/4n
Now (-3)^n/4n goes to -infty or +infty depending on n because
4n=o((-3)^n)
My argument is really not rigorous and is probably inaccurate, and I
still need
to show that -infty or +infty are reached with probability 1/2, and
1/2.
Can somebody help me improve my argument?
I appreciate your help/pointers.
.
- Follow-Ups:
- Re: -- nonconvergence of a series. with application in statistics
- From: Robert Israel
- Re: -- nonconvergence of a series. with application in statistics
- Prev by Date: Re: -- Draft unofficial <sci.math> FAQ. Comments please.
- Next by Date: Re: -- Draft unofficial <sci.math> FAQ. Comments please.
- Previous by thread: the prophet
- Next by thread: Re: -- nonconvergence of a series. with application in statistics
- Index(es):
Relevant Pages
|
Loading