\documentstyle[11pt,psfig]{article}

\setlength{\parskip}{.05in}

\setlength{\textwidth}{6.20in}

\setlength{\textheight}{8.25in}

\setlength{\evensidemargin}{-.2in}

\setlength{\oddsidemargin}{-.2in}

\setlength{\topmargin}{-.25in}

 

\def\squarebox#1{\hbox to #1{\hfill\vbox to #1{\vfill}}}

 

\newcommand{\finish}{\centerline{\rule{14cm}{1.5mm}}}

\newcommand{\qed}{\hspace*{\fill}

\vbox{\hrule\hbox{\vrule\squarebox{.667em}\vrule}\hrule}\smallskip}

 

\newenvironment{proof}{\noindent {\bf Proof:} \hspace{.677em}}%

                      {\qed}

\newenvironment{proofr}[1]{\noindent {\bf Proof of

{#1}:}\hspace{.677em}}%

                      {\qed}

 

 

 

\vspace{-.5in}

\title{  Introduction to Probability\\

       }

\author{

        {\bf Nader H.~Bshouty }\\

 {\small Department of Computer Science}\\

 {\small Technion 32000}\\

 {\small Israel}\\

 {\small {\tt e-mail:~bshouty@cs.technion.ac.il}} \\

 }

 

\newcommand{\singlespacing}%

{\renewcommand{\baselinestretch}{1.0}\small\normalsize}

 

\newcommand{\comment}[1]{}

 

%Macro to draw pretty boxes around things

\newlength{\boxwidth}

\setlength{\boxwidth}{\textwidth}

\addtolength{\boxwidth}{-27 pt}

\newcommand{\boxer}[1]{\begin{center}

                        \fbox{\fbox{

                                \begin{minipage}{\boxwidth}

                                \vspace*{.2in}

% Copyright (C) 1988, all rights reserved.

 

 

\maketitle

 

 

 

\label{style}

 

                                #1

                                \vspace*{.2in}

                                \end{minipage}}}

                        \end{center}}

 

\def\cjfigures{norm}

\def\cjfigure#1#2#3{ % psfig command , caption , label

        {

        \begin{figure}[htb]

        \centerline{\psfig{#1}}

        \caption{#2}

        \label{#3}

        \end{figure}

        }

}

\def\dir{/home/profs/bshouty/SUBMITTED/23-MonotoneTheory/OTHER/}

 

 

\newcommand{\unset}[2]{#1_{#2 \leftarrow *}}

\newcommand{\proj}[3]{#1 | (#2 \!\leftarrow\! #3)}

\newcommand{\unary}[2]{#1_{#2}}

\newcommand{\fA}{\unary{f}{A}}

\newcommand{\fB}{\unary{f}{B}}

\newcommand{\fC}{\unary{f}{C}}

\newcommand{\fD}{\unary{f}{D}}

\newcommand{\fAx}{\unary{f}{A}(x)}

\newcommand{\fBx}{\unary{f}{B}(x)}

\newcommand{\error}{\mbox{ERROR}}

\newcommand{\twobytwo}[4]{\left( \! \begin{array}{cc} #1 & #2 \\ #3 & #4

\end{array} \! \right)}

\newcommand{\twobyone}[2]{\left(\!\begin{array}{c} #1 \\ #2

\end{array}\!\right)}

\newcommand{\onebytwo}[2]{( \! \begin{array}{cc} #1 & #2 \end{array} \! )}

\newcommand{\assign}[3]{#1_{#2 \leftarrow #3}}

\newcommand{\assignpair}[5]{#1_{#2 \leftarrow #3,#4 \leftarrow #5}}

\newcommand{\op}{\triangleright}

\newcommand{\M}{{\cal M}}

\newcommand{\D}{{\cal D}}

\newcommand{\F}{{\cal F}}

\newcommand{\T}{{\cal T}}

\newcommand{\bB}{\mbox{\bf B}}

\newcommand{\bE}{\mbox{\bf E}}

\newcommand{\bT}{\mbox{\bf T}}

\newcommand{\bY}{\mbox{\bf Y}}

\newcommand{\bN}{\mbox{\bf N}}

\newcommand{\bL}{\mbox{\bf L}}

\newcommand{\bM}{\mbox{\bf M}}

\newcommand{\bH}{\mbox{\bf H}}

\newcommand{\bO}{\mbox{\bf O}}

\newcommand{\OR}{{\vee}}

\newcommand{\bor}{\bigvee}

\newcommand{\AND}{{\wedge}}

\newcommand{\band}{\bigwedge}

\newcommand{\imply}{\Rightarrow}

\newcommand{\equ}{\Leftrightarrow}

\newcommand{\eqdef}{\stackrel{\Delta}{=}}

\newtheorem{lemmasection}{Lemma}[section]

\newtheorem{lemma}{Lemma}[subsection]

\newtheorem{Rule}{Rule}

\newtheorem{theoremsection}{Theorem}[section]

\newtheorem{theorem}{Theorem}[subsection]

\newtheorem{corolarysection}{Corolary}[section]

\newtheorem{corolary}{Corolary}[subsection]

\newtheorem{definitionsection}{Definition}[section]

\newtheorem{definition}{Definition}[subsection]

\newtheorem{exercise}{Exercise}[subsection]

\newtheorem{example}{Example}

 

 

\begin{document}

 

%\tableofcontents

\date{}

\bibliographystyle{alpha}

\maketitle

\vspace{-.2in}

\thispagestyle{empty}

\pagenumbering{arabic}

\setlength{\unitlength}{.6mm}

%\begin{abstract}

%Here we give few inequalities we usually use in COLT.

%\end{abstract}

\medskip

 

\section{Combinatorics}

 

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

 

 

\subsection{Simple Rules In Combinatorics}

 

The {\bf rule of sum} says that the number of ways to choose

an element from one of two disjoints sets is the sum of

the cardinalities of the sets.

That is, if $A$ and $B$ are two finite sets with no members

in common, then $|A\cup B|=|A|+|B|$.

 

The {\bf rule of product} says that the number of ways to choose

an ordered pair is the number of ways to choose the first

element from $A$ and the second element from $B$.

That is, if $A$ and $B$ are two finite sets, then

$|A\times B|=|A|\cdot |B|$.

 

 

\subsection{Simple Counting}

\noindent

{\bf Strings}

 

A {\bf string} over a set $S$ is a sequence of elements of $S$.

For example, there are 8 binary strings of length 3:

000,001,010,011,100,101,110,111.

A $k$-string over a set $S$ can be viewed as an element

of the Cartesian product $S^k$ of $k$-tuples; thus, there are

$|S|^k$ strings of length $k$.

 

 

\noindent

{\bf Permutations}

 

A {\bf permutation} of a finite set $S$ is an ordered sequence

of all the elements of $S$, with each element appearing

exactly once. For example, if $S=\{a,b,c\}$, there are 6

permutations of $S$:

abc,acb,bac,bca,cab,cba.

There are $n!$ permutations of a set of $n$ elements.

A {\bf $k$-permutation} of $S$ is an ordered sequence of $k$

of $S$, with no element appearing  more than once in the

sequence.

The number of $k$-permutations of an $n$-set is:

$$n(n-1)(n-2)\cdots (n-k+1)=\frac {n!}{(n-k)!}$$

 

 

\noindent

{\bf Combinations}

 

A {\bf $k$-combination} of an $n$-set $S$ is simply a $k$-subset

of $S$. There are six 2-combinations of the 4-set $\{a,b,c,d\}$:

$ab,ac,ad,bc,bd,cd$.

The number of $k$-combinations of an $n$-set.

$$\frac{n!}{k!(n-k)!}$$

 

 

 

\subsection{Equalities}

 

\noindent

{\bf Binomial coefficients}

 

We use the notation ${n\choose k}$ (read "$n$ choose $k$")

to denote the number of $k$-combinations of an $n$-set.

We have $${n\choose k}=\frac {n!}{k!(n-k)}.$$

This formula is symmetric in $k$ and $n-k$:

$${n\choose k}={n\choose n-k}.$$

These numbers are also known as {\bf binomial coefficients}, due to

their appearance in the {\bf binomial expansion}:

$$(x+y)^n=\sum_{k=0}^n {n\choose k}x^k y^{n-k}$$

A special case: $$2^n=\sum_{k=0}^n {n\choose k}.$$

We also have

$${n\choose k}=\frac{n}{k} {n-1\choose k-1}$$

$${n\choose k}=\frac{n}{n-k} {n-1\choose k}$$

$${n\choose k}={n-1\choose k}+{n-1\choose k-1}$$

 

\subsection{Inequalities}

\noindent

{\bf Binomial bounds}

 

For $1\le k\le n$, we have the lower bound:

$${n\choose k}\ge \left(\frac {n}{k}\right)^k.$$

The upper bound:

$${n\choose k}\le \left(\frac {en}{k}\right)^k.$$

For all $0\le k\le n$ (see \ref{nklnnkknmknmk}),

$${n\choose k}\le \frac{n^n}{k^k(n-k)^{n-k}},$$

where for convenience we assume that $0^0=1$.

For $k=\lambda n$, where

$0\le \lambda \le 1$, this bound can be rewritten as (see R\ref{nclnnHl}):

$${n\choose \lambda n}\le 2^{nH(\lambda)},$$

where $H(\lambda)=-\lambda lg \lambda-(1-\lambda)lg(1-\lambda)$ is the

{\bf entropy function} and where,

for convenience, we assume that $0\lg 0=0$,

so that $H(0)=H(1)=0$.

For $0\le \lambda\le \frac{1}{2}$ we have (see R\ref{slnnciltnHl})

$$\sum_{0}^{\lambda n}{n\choose i}\le 2^{nH(\lambda)}.$$

 

For any $n\ge 0,j\ge 0,k\ge 0$, and $j+k\le n$,

$${n\choose j+k}\le {n\choose j}{n-j\choose k}.$$

The Stirling's approximation is

$$\log n!=\left(n+\frac{1}{2}\right)\log n-n-\frac{1}{2}\log(2\pi)+o(1),$$

$${2n\choose n}=\frac{2^{2n}}{\sqrt{\pi n}}

(1+O(1/n)).$$

 

 

\section{Probability}

 

\subsection{Introduction}

We define probability in terms of a {\bf sample space}

$S$, which is a set whose

elements are called {\bf elementary events}.

For flipping two distinguishable coins, we can view the sample space

$S={HH,HT,TH,TT}$.

An {\bf event} is a subset of the sample space $S$.

The event $S$ is called the

{\bf certain event}, and the event $\O$ is called

the {\bf null event}.

We say that two events $A$ and

$B$ are {\bf mutually} exclusive if $A\cap B=\O$.

 

 

\noindent

{\bf Axioms of probability}

 

A {\bf probability distribution} $\Pr \{\}$ on a sample

space $S$ is a mapping from

events of $S$ to real numbers such that the

following probability axioms are

satisfied:

\begin{enumerate}

\item $\Pr [A]\ge 0$ for any event $A$.

\item $\Pr [S]=1$.

\item $\Pr [A\cup B]=\Pr [A]+\Pr [B]$ for

any two mutually exclusive events $A$ and $B$.

\end{enumerate}

We call $\Pr [A]$ the {\bf probability} of the event $A$.

 

Results:

$$\Pr [\O ]=0,$$ if $A\subseteq B$, then $\Pr [A]\le \Pr [B]$.

For $\overline{A}$ (the {\bf complement} of $A$),

$$\Pr [\overline{A}]=1-\Pr [A].$$ For

any two events $A$ and $B$,

\begin{eqnarray*}

\Pr [A\cup B]&=&\Pr [A]+\Pr [B]-\Pr [A\cap B]\\

&\le& \Pr [A]+\Pr [B]\\

\end{eqnarray*}

 

 

\noindent

{\bf Discrete probability distributions}

 

A probability distribution is {\bf discrete}

if it is defined over a finite or

or countably infinite sample space.Then for any event $A$,$$\Pr

[A]=\sum_{s\in A} \Pr [s].$$

If $S$ is finite and every elementary event $s\in S$ has probability

$\Pr[s]=1/|S|$

then we have the {\bf uniform probability distribution} on $S$.

A {\bf fair coin} is

one which the probability of obtaining a head is the same as

the probability of obtaining a tail, that is, 1/2.

 

 

\subsection{Conditional probability and Independence}

 

Conditional probability formalizes the notion of having prior partial

knowledge of the

outcome of an experiment. The conditional probability of an event $A$

given that another

event $B$ occurs is defined to be:

$$\Pr [A|B]=\frac{\Pr[A\cap B]}{\Pr [B]}$$

Whenever $\Pr [B]\ne 0$,

(we read ``$\Pr [A|B]$'' as "the probability of

$A$ given $B$.")

Two events are {\bf independent} if,

$\Pr [A\cap B]=\Pr[A] \Pr [B]$, which

is equivalent, if

$\Pr [B]\ne 0$, to the condition $\Pr [A|B]= \Pr [A]$.

A collection $A_1,A_2,...,A_n$ of events

is said to be {\bf pairwise independent}

if $\Pr [A_i\cap A_j]=\Pr [A_i]\Pr [A_j]$.

For all $1\le i<j\le n$.

We say that they are ({\bf mutually}) {\bf independent} if

every $k$-subset $A_{i_1},A_{i_2},...,A_{i_k}$

of collection, where $2\le k\le n$ and $1\le

i_1<i_2<\cdots <i_k\le n$, satisfies:

$$\Pr[A_{i_1} \cap A_{i_2} \cap \cdots \cap A_{i_k} ]=\Pr [A_{i_1} ]

\Pr [A_{i_2} ]\cdots \Pr [A_{i_k} ]$$

 

\noindent

{\bf Bayes's Theorem}

 

$$\Pr [A|B]=\frac{\Pr[A] \Pr[B|A]}{\Pr [B]}.$$

From Bayes's theorem: $$\Pr[A|B]=\frac {\Pr [A] \Pr [B|A]} {\Pr

[A] \Pr [B|A]+\Pr [\overline{A}] \Pr [B|\overline{A}]}$$

 

\noindent

{\bf Boole's inequality:} For any finite or countably infinite sequence of events

$A_1 ,A_2 ,...,$

$$\Pr [A_1 \cup A_2 \cup \cdots] \le \Pr [A_1]+\Pr [A_2]+ \cdots.$$

For collection of events $A_1 ,A_2 ,...,A_n$,

\begin{eqnarray*}

\Pr [A_1 \cap A_2 \cap \cdots\cap A_n]&=&\Pr [A_1]\cdot \Pr

[A_2|A_1]\cdot \Pr [A_3|A_1\cap A_2 ]\cdots\\

&&\Pr [A_n|A_1 \cap A_2 \cap \cdots \cap A_{n-1}]\\

\end{eqnarray*}

 

 

\subsection {Discrete Random Variables and Expectation}

 

A ({\bf discrete}) {\bf random variable} $X$ is a function from a finite or

countably infinite sample space $S$ to the real numbers.

For random variable $X$ and a real number $x$, we define the event $X=x$ to be

$[s\in S:X(s)=x]$; thus, $\Pr [X=x]=\sum_{[s\in S:X(s)=x]} \Pr [s]$.

The function: $f(x)=\Pr [X=x]$, is the {\bf probability density function} of

the random variable $X$.

 

It is common for several random variables to be defined on the same sample

space . If $X$ and $Y$ are random variables, the function:

$f(x,y)= \Pr [X=x$ and $Y=y]$

Is the {\bf joint probability density function} of $X$ and $Y$. For a fixed value

$y$, $\Pr [Y=y]=\sum_x \Pr [X=x$ and $Y=y]$.

We define two random variables $X$ and $Y$ to be {\bf independent} if for all $x$

and $y$, the events $X=x$ and $Y=y$ are independent or, equivalently, if for

all $x$ and $y$, we have $\Pr [X=x$ and $Y=y]=\Pr[X=x]\Pr[Y=y]$.

 

 

\noindent

{\bf Expected value of a random variable}

 

The {\bf expected value} (or, synonymously, {\bf expectation} or {\bf mean}) of a discrete

random variable $X$ is $$E [X]=\sum_x x\Pr [X=x].$$

Sometimes the expectation of $X$ is denoted by $\mu_x$ or , when the

random is apparent from context, simply by $\mu$.

The expectations of the sum of two random variables is the sum of their

expectations, that is, $$E [X+Y]=E [X]+[Y].$$

If $X$ is any random variable, any function $g(x)$ defines a new random

variable $g(X)$. If the expectation of $g(X)$ is defined, then

$$E [g(X)]=\sum_x g(x) \Pr [X=x].$$

Letting $g(x)=ax$, we have for any constant $a$, $$E [aX]=aE[X].$$

$$E [aX+Y]=aE [X]+E [Y].$$

When two random variables $X$ and $Y$ are independent and each has a defined

expectation, $E [XY]=E [X] E [Y].$

In general, when $n$ random variable $X_1,X_2,...,X_n$ are mutually

independent, $E [X_1 X_2 \cdots X_n]=E [X_1]E [X_2]\cdots E[X_n].$

When a random variable $X$ takes on values from the natural numbers

$N=\{0,1,2,...\}$, there is a nice formula for its expectation:

$$E[X]=\sum_{i=1}^\infty \Pr [X\ge i].$$

 

 

\noindent

{\bf Variance and standard deviation}

 

The {\bf variance} of a random variable $X$ with mean $E [X]$ is:

\begin{eqnarray*}

Var [X]&=&E [(X-E [X])^2]\\

&=&E [X^2]-E^2 [X]\\

\end{eqnarray*}

We have

$$E [X^2]=Var [X]+E^2 [X]$$

$$Var [aX]=a^2 Var [X]$$

When $X$ and $Y$ are independent random variables, $Var [X+Y]=Var [X]+Var[Y].$

In general, if $n$ random variables $X_1,X_2,...,X_n$ are pairwise independent,

then $$Var [\sum_{i=1}^n X_i]=\sum_{i=1}^n Var [X_i].$$

The {\bf standard deviation} of a random variable $X$ is the positive square root

of the variance of $X$.

 

 

\noindent

{\bf Markov's inequality:} (see Rule \ref{Markov})

 

For a nonnegative random variable $X$

$$\Pr [X\ge t]\le \frac{E [X]}{t}$$

for all $t>0$ and

$$\Pr[X\ge \lambda E[X]]\le \frac{1}{\lambda}.$$

See more results in \cite{G} Appendix A.

 

\noindent

{\bf Chebychev's inequality:}

 

$$\Pr[|X-E[X]|\ge k]\le \frac{Var[X]}{k^2}.$$

 

\noindent

{\bf Bienayme-Chebyschev}

Let $X_1,\ldots,X_m$ be pairwise independent random

variables such that $\bE[X_i]=\mu$ and $Var[X_i]=\sigma^2$. Then

$$\Pr\left[\left|\frac{\sum_{i=1}^m X_i}{m}-\mu\right|

\ge \lambda \right]

\le \frac{\sigma^2}{\lambda^2 m}.$$

See the proof in \cite{G} Appendix A.

 

\subsection{Geometric and Binomial distributions}

 

A coin flip is an instance of a {\bf Bernoulli trail}, which is defined as an

experiment with only two possible outcomes: {\bf success}, which occurs with

probability $p$, and {\bf failure}, which occurs with probability $q=1-p$.

The trails are mutually independent ok. each has the same probability

$p$ for success.

 

 

\noindent

{\bf The geometric distribution}

 

Suppose we have a sequence of Bernoulli trails, each with a probability

$p$ of success and a probability $q=1-p$ of failure.

Let the random variable $X$ be the number of trails needed to obtain a success.

$$\Pr [X=k]=q^{k-1} p$$

A probability distribution satisfying this is said to be a {\bf geometric distribution}.

$$E [X]=1/p$$

$$Var [X]=q/p^2$$

 

 

\noindent

{\bf The binomial distribution}

 

Define the random variable $X$ to be the number of success in $n$ trails.

$$\Pr [X=x]={n\choose k}p^k q^{n-k}$$

A probability distribution satisfying this is said to be a {\bf binomial  distribution} .

Define: $$b(k;n,p)={n\choose k}p^k (1-p)^{n-k}$$

We have: $E [X]=np$ and $Var [X]=npq$.

The binomial distribution $b(k;n,p)$ increases as $k$ runs from 0 to $n$

until it reaches the mean $np$, and then it decreases.

 

Let $n\ge0$, let $0<p<1$, let $q=1-p$, and let $0\le k\le n$.

Then, $$b(k;n,p)\le \left(\frac{np}{k}\right)^k \left(\frac{nq}{n-k}\right)^{n-k}.$$

 

For $0\le k\le n$, $$b(k;n,1/2)\le 2^{nH(k/n)-n}.$$

 

 

\noindent

{\bf The tails of the binomial distribution}

 

Let $X_1,X_2,\ldots,X_n$ be independent random variables such that

$X_i\in \{0,1\}$ and $E[X_i]=p$.

Then this is equivalent to

a sequence of $n$ Bernoulli trails , where success occurs

with probability $p$. Let $X$ be the random variable denoting the

total number of successes. Then for $0\le k\le n$, the probability

of at least $k$ success is:

$$ \Pr \left[\sum_{i=1}^n X_i\ge k\right]=\sum_{i=k}^n b(i;n,p)

\le {n\choose k}p^k. $$

 

For $np<k<n$, the probability of more than $k$ successes is:

$$\Pr \left[\sum_{i=1}^nX_i>k\right]=\sum_{i=k+1}^n b(i;n,p)

<\frac {(n-k)p}{k-np}b(k;n,p),$$

and

$$

\Pr \left[\frac{\sum_{i=1}^nX_i}{n}-p\ge \lambda\right]

\le\left(\frac{pe}{\lambda}\right)^{\lambda n}.

$$

 

$$ \Pr \left[\sum_{i=1}^nX_i\le k\right]=\sum_{i=1}^k b(i;n,p)

\le{n\choose k}(1-p)^{n-k}.  $$

 

For $0<k<np$, the probability of fewer than $k$ success is:

$$\Pr \left[\sum_{i=1}^nX_i<k\right]=\sum_{i=0}^{k-1} b(i;n,p)

<\frac {kq}{np-k} b(k;n,p).$$

 

 

 

{\bf Chernoff}

Let $X_1,\ldots,X_m$ be independent random

variables such that $X_i\in \{0,1\}$ and $\bE[X_i]=p$ Then

$$\Pr\left[\frac{\sum_{i=1}^n X_i}{n}-p

\ge \lambda p \right]

\le e^{-\lambda^2np/3}.$$

$$\Pr\left[p-\frac{\sum_{i=1}^n X_i}{n}

\ge \lambda p\right] \le e^{-\lambda^2np/2}.$$

See another bound in \cite{G} Appendix A.

 

{\bf Hoeffding}

Let $X_1,\ldots,X_m$ be independent random

variables such that $X_i\in [a,b]$ and $\bE[X_i]=p$.

Then

$$\Pr\left[\left|\frac{\sum_{i=1}^n X_i}{n}-p\right|\ge \lambda \right]

\le 2e^{-2\lambda^2n/(b-a)^2}.$$

 

 

 

{\bf Other Results.}

Consider independent random variables $X_1,\ldots,X_n$

(a sequence of $n$ Bernoulli trails)

where $E[X_i]=p_i$ (where in the $i$th trail, for

$i=1,2,...,n$, success occurs with probability $p_i$ and failure occurs with

probability $q_i=1-p_i$).

Let $\mu =E\left[\sum_{i=1}^n X_i\right]$. Then for $r>\mu$,

$$\Pr \left [\sum_{i=1}^n X_i-\mu \ge r\right]\le \left(\frac{\mu e}{r}\right)^r$$

$$\Pr [X-\mu\ge r]\le e^{-r^2/2n}$$

 

\section{Examples}

\begin{example}

{\bf Random Halving}: A game where you randomly uniformly choose

an integer $1\le i_1\le n$ and then $1\le i_2\le i_1$ until you

get $1$. Show that the expected number of steps is

$1+\sum_{i=1}^{n-1} 1/i$.

\end{example}

{\bf Solution:} Let $X_n$ be the expected number of steps then

$X_1=1$ and

$$E[X_n]=1+E_{1\le i\le n}[E[X_i]].$$

Now

$$

\frac{n+1}{n} E[X_{n+1}]-E[X_n]=

\frac{1}{n}+\frac{1}{n}E[X_{n+1}]$$ and therefore

$E[X_{n+1}]=E[X_{n}]+(1/n)$.

 

 

{\bf Acknowledgement:} To Vivian Bshouty for helping me with this

survey.

 

\begin{thebibliography}{10}

 

\bibitem[AS]{AS}

N. Alon and J. H. Spencer.

\newblock The Probabilistic Method.

 

\bibitem[CLR]{CLR}

T. Cormen, C. E. Leiserson and R. L. Rivest.

Introduction to Algorithms.

 

\bibitem[G]{G}

O. Goldreich.

Modern Cryptography, Probabilistic Proofs and Pseudo-randomness.

 

\bibitem[MR]{MR}

R. Motwani and P. Raghavan.

Randomized Algorithms.

 

 

\end{thebibliography}

 

 

 

\newpage

\section{Proofs}

\begin{Rule}

{\bf Markov's inequality:}

$$\Pr [X\ge t]\le \frac{E [X]}{t}$$

for all $t>0$.

\label{Markov}

{\bf Proof.} We have

$$E[X]=\sum_x\Pr[X=x]\cdot x\ge \sum_{x\ge t}\Pr[X=x]x

\ge \Pr[X\ge t] t.\Box$$

 

\end{Rule}

\begin{Rule}\label{nklnnkknmknmk}

We have

$${n\choose k}\le \frac{n^n}{k^k(n-k)^{n-k}}.$$

\end{Rule}

{\bf Proof.} We have

$$n^n=(k+(n-k))^n\ge {n\choose k}k^k (n-k)^{n-k}.\Box$$

 

\begin{Rule}\label{nclnnHl}

For $0\le \lambda \le 1$ we have

$${n\choose \lambda n}\le 2^{nH(\lambda)},$$

where $H(\lambda)=-\lambda lg \lambda-(1-\lambda)lg(1-\lambda)$.

\end{Rule}

{\bf Proof.} Use Rule \ref{nklnnkknmknmk}.$\Box$

 

\begin{Rule}\label{slnnciltnHl}

For $0\le \lambda\le \frac{1}{2}$ we have

$$\sum_{0}^{\lambda n}{n\choose i}\le 2^{nH(\lambda)}.$$

\end{Rule}

{\bf Proof.} We have

\begin{eqnarray*}

1&=&(\lambda+(1-\lambda))^n\ge

\sum_{i=1}^{\lambda n}{n\choose i}\lambda^i(1-\lambda)^{n-i}

=\sum_{i=1}^{\lambda n}{n\choose i}(1-\lambda)^n\left(

\frac{\lambda}{1-\lambda}\right)^{\lambda i}\\

&\le& \sum_{i=1}^{\lambda n}{n\choose

i}(1-\lambda)^n\left(\frac{\lambda}{1-\lambda}\right)^{\lambda n}

=2^{-n H(\lambda)}\sum_{0}^{\lambda n}{n\choose i}.\Box

\end{eqnarray*}

\end{document}