\documentclass{article} \usepackage{amsmath,amssymb,amsfonts} \usepackage{amsthm} \setlength{\oddsidemargin}{0.25 in} \setlength{\evensidemargin}{-0.25 in} \setlength{\topmargin}{-0.6 in} \setlength{\textwidth}{6.5 in} \setlength{\textheight}{8.5 in} \setlength{\headsep}{0.75 in} \setlength{\parindent}{0 in} \setlength{\parskip}{0.1 in} \newcounter{lecnum} % % The following macro is used to generate the header. % \newcommand{\lecture}[4]{ \pagestyle{myheadings} \thispagestyle{plain} \newpage \setcounter{lecnum}{#1} \setcounter{page}{1} \noindent \begin{center} \framebox{ \vbox{\vspace{2mm} \hbox to 6.28in { {{\bf 18.317~Combinatorics, Probability, and Computations on Groups} \hfill #2} } \vspace{4mm} \hbox to 6.28in { {\Large \hfill Lecture #1 \hfill} } \vspace{2mm} \hbox to 6.28in { {\it Lecturer: #3 \hfill Scribe: #4} } \vspace{2mm}}} \end{center} \markboth{Lecture #1: #2}{Lecture #1: #2} \vspace*{4mm} } \input{epsf} %usage: \fig{LABEL}{FIGURE-HEIGHT}{CAPTION}{FILENAME} \newcommand{\fig}[4]{ \begin{figure} \setlength{\epsfysize}{#2} \centerline{\epsfbox{#4}} \caption{#3} \label{#1} \end{figure} } \newtheorem{theorem}{Theorem} \newtheorem{lemma}[theorem]{Lemma} \newtheorem{proposition}[theorem]{Proposition} \newtheorem{claim}[theorem]{Claim} \newtheorem{corollary}[theorem]{Corollary} \theoremstyle{definition} \newtheorem{definition}[theorem]{Definition} \theoremstyle{remark} \newtheorem{example}{Example} \renewenvironment{proof}{{\bf Proof:}}{\hfill\rule{2mm}{2mm}} \newenvironment{mproof}[1]{{\bf #1}}{\hfill\rule{2mm}{2mm}} \DeclareMathOperator{\Aut}{Aut} \def\E#1{{\rm E}[#1]} \def\i{\hspace*{5mm}} \newcounter{lineno} \setcounter{lineno}{0} \def\n{\addtocounter{lineno}{1} \hbox to 7mm{\hfill \arabic{lineno}:}\hspace{2mm}} \def\resetn{\setcounter{lineno}{0}} \def\tqbfk{\tqbf_k^\exists} \def\ep{\epsilon} \def\erd{Erd\H{o}s} \begin{document} \lecture{31}{30 October 2001}{Igor Pak}{D. Jacob Wildstrom} \section*{Bias (continued)} \begin{theorem}[P. Hall] For a simple group $H$ and $G=H^m$, it follows that $\langle g_1,\ldots,g_k\rangle=G$ if and only if $\langle h_j^{(1)},h_j^{(2)},\ldots,h_j^{(k)}\rangle=H$ for all $j$ from $1$ to $m$. \end{theorem} We shall henceforth work with $H=A_n$, $m=\frac{n!}8$, $\varkappa=o(n)$, $G=H^m$, and $Q=Q_k$, which to refresh our memory, is simply the probability distribution of $g_1$ in $\bar{g}=(g_1,g_2,\ldots,g_k)\in\Gamma_k(G)$. \begin{theorem} There is a subset $B$ of $G$ such that, as $n\rightarrow\infty$, $\frac{|B|}{|G|}\rightarrow 1$ but $Q(B)\rightarrow 0$. \end{theorem} That is to say that there is a huge subset (approaching the full set) of $G$ which is hardly ever generated by a $k$-tuple of generators. We claim that, roughly, if $k\geq k$, then the values of $h_j=(h_j^{(1)},h_j^{(2)},\ldots,h_j^{(k)})$ are independent. Then we have the lemma: \begin{lemma} $|\Gamma_k(o)|=|\Gamma_k(H)|^m|(1-O(\frac1{n!}))$. \end{lemma} \begin{proof} The number of automorphisms of $\Gamma_k(H)$ is, as we discussed earlier, $\alpha_k(G)$, which must exceed $\frac{H^k}{2|\Aut(H)|}=\frac12\frac{(\frac n2)^k}{n!}=N$. Now, $|\Gamma_k(G)|$ is equal to the product of $|\Gamma_k(H)|^m$ and the probability that each generated $k$-tuple is in a distinct orbit. This we can easily calculate to be $(1-\frac1N)(1-\frac2N)\cdots(1-\frac mN)$, which exceeds $(1-\frac mN)^m$ and thus $(1-\frac{m^2}N)$. Using the equation $m=\frac{n!}8$ and $N\geq\frac{(n!)^3}{32}$, the above factor can be easily shown to exceed $1-\frac1{2n!}$ \end{proof} Let $A_n$ be generated by $(h_1^1,h_1^2,\ldots,h_1^k)$. We know that with probability $\approx\frac1n$ (specifically, $\frac1n\pm\frac1{n^3}$), $h_1^1$ moves the first element. What would the specific probability tell us about $g_1$? We start by looking at $\phi_k(A_n)$, which would be 1 minus the probability of ``bad events''. What sort of ``bad events'' might we have in mind? They can be characterized by $h_1,\ldots,h_k\in M$ for some maximal subgroup $M$ of $H$. There are really only 3 types of maximal subgroups in $H$: those with one fixed point, those with a pair of elements forming an orbit, and those with two fixed points. The probability of generating any of these is easily calculated: $$\phi_k(A_n)=1-\frac1{n^k}n-\frac1{n(n-1)^k}\binom n2+\frac1{2(n(n-1))^k}\binom n2=1-\frac 1{n^{k-1}}+O(\frac1{n^{2(k-1)}})$$ Let $\mathcal A$ be the event $(h_1,\ldots,h_k)\in\Gamma_k(H)$, and $\mathcal B$ be the event that $h_1=1$. By the above, $\Pr(\mathcal A)=1-\frac1{n^k-1}+O(\frac1{n^{2(k-1)}})$ and $\Pr(\mathcal B)=\frac1n$. So what is $\Pr(\mathcal B|\mathcal A)$? Well, it is equal to $\frac{\Pr(\mathcal A|\mathcal B)}{\Pr(A)}\Pr(B)$, and we may interpret $\Pr(\mathcal A|\mathcal B)$ as such; either $h_2,\ldots,h_k$ fix the first element or they fix an element not equal to the first. Calculating the probabilities, it follows that $$\Pr(\mathcal A|\mathcal B)=1-\frac2{n^{k-1}}+O(\frac1{n^{2(k-1)}})$$ and thus $$\Pr(\mathcal B|\mathcal A)=\frac1n\left(\frac{1-\frac2{n^{k-1}}+O(\frac1{n^{2(k-1)}})}{1-\frac1{n^{k-1}}+O(\frac1{n^{2(k-1)}})}\right)=\frac1n\left(1-\frac1{n^{k-1}}+O(\frac1{n^{2(k-1)}})\right)$$ so $P=\frac1n-\frac1{nk}+O(\frac1{n^{2k-1}})$. So, if we were to plot the number of generating sets giving us $h_1(1)=1$ for $g$ uniform and $g_1\in\Gamma_k(G)$, we wil have peaks at $\frac1n$ and $\frac1n-\frac1{n^k}$ respectively, and we may return to our original result by choosing $B\subset G$ such that $\{g=(h_1,\ldots,h_m)\}$ in which the number of generating sets in which $h_i(1)=1$ exceeds $m(\frac1n-\frac1{2n^2})$. Using the Chernoff bounds, we find that $|B|\approx |G|(1-\frac1{n!})\rightarrow 1$, and that $Q_k(B)\approx\frac1{n!}\rightarrow0$. \end{document}