שאלה 2 סעיף 3

[Gleb Kodinets]

I think you mean "at most" and not "at least". Probability to fail is
"at most" 1/2. Else for any minimal p (for example 0 or 1/N) the
probability of z1<Theta*N*(1-epsilon) will be much greater then 1/2.
Am I right?

[Edo Liberty]

Yes, the probability of failure should be at most one half.

Also, you are right that this failure probability can be much much smaller that 1/2.

This is what you are asked to show later in the exercise.

Edo 

---

From: Gleb Kodinets <gle...@gmail.com>
To: Algorithms In Data Mining-tau <algorithms-in-...@googlegroups.com>
Sent: Sunday, November 27, 2011 1:42 AM
Subject: [algorithms-in-data-mining-tau] שאלה 2 סעיף 3

[shlomih]

This is a bit confusing... what's written in the exrecise as "at least
0.5" in question 3 is correct, right? it' shouldn't be altered to "at
most 0.5".
what's mentioned there is the complete event to the failure... not the
failure itself

On 27 נובמבר, 10:42, Edo Liberty <edo.libe...@ymail.com> wrote:
> Yes, the probability of failure should be at most one half.
> Also, you are right that this failure probability can be much much smaller that 1/2.
> This is what you are asked to show later in the exercise.
> Edo
>
>
>
> >________________________________

> > From: Gleb Kodinets <gleb...@gmail.com>

> >To: Algorithms In Data Mining-tau <algorithms-in-...@googlegroups.com>
> >Sent: Sunday, November 27, 2011 1:42 AM
> >Subject: [algorithms-in-data-mining-tau] שאלה 2 סעיף 3
>
> >I think you mean "at most" and not "at least". Probability to fail is
> >"at most" 1/2. Else for any minimal p (for example 0 or 1/N) the
> >probability of z1<Theta*N*(1-epsilon) will be much greater then 1/2.

> >Am I right?-הסתר טקסט מצוטט-
>
> -הראה טקסט מצוטט-