1: %% AMS-LaTeX
2: %% *********
3: %% project: Paper 3
4: %% version: GAMMA
5: %% file: part 01 (Introduction)
6: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
7:
8: %!TEX root = paper3-2006.tex
9:
10: \section{Introduction}
11: \label{section:introduction}
12:
13: In 1951, Oystein Ore (see \cite{Ore:1951:src}) conjectured that
14: all elements in every non-abelian finite simple group
15: are commutators.
16: In terms of \emph{commutator width}, the question is
17: whether the commutator width of every
18: non-abelian finite simple group is~$1$.
19: This question still remains open.
20: However, using the Classification of Finite Simple Groups,
21: it was shown by John Wilson that there exists a (not found explicitly)
22: common upper bound on the commutator widths of all finite simple groups
23: (see~\cite{Wilson:1996:fogt}).
24: %^added on 2007-02-02 (4 lines)
25:
26: In 1977, Martin Isaacs (see \cite{Isaacs:1977:ccs})
27: noted that no simple group, finite or infinite,
28: was known to have commutator width greater than~$1$.
29: In 1999, Valerij Bardakov posed the following question
30: (see Problem 14.13 in \cite{kn14:1999:knupgt-eng}):
31: \begin{quote}
32: Does there exist a (finitely presented) simple group of infinite
33: commutator width?
34: \end{quote}
35:
36: Simple groups of infinite commutator width,
37: realised as groups of certain surface diffeomorphisms,
38: appeared in
39: \cite{BargeGhys:1992:cEM-fr,GambaudoGhys:2004:cds} in 1992.
40: The infinity of the commutator width is established there by
41: constructing \emph{nontrivial homogeneous quasi-morphisms}.
42: However, the question of existence of \emph{finitely generated\/}
43: simple groups of commutator width greater than $1$ seems to have been
44: open until now.
45:
46: In this paper it is shown that presentations
47: (by generators and defining relations)
48: of finitely generated infinite simple groups
49: of infinite commutator width, as well as of large finite commutator width,
50: can be constructed using methods of small-cancellation theory.
51: This approach is rather flexible and can yield groups with various
52: additional properties.%
53: %% Minor correction 2008-07-03:
54: \footnote{In particular, groups can be constructed so that to admit
55: no nontrivial homogeneous quasi-morphisms
56: (it suffices to make the \emph{stable commutator length\/} of
57: each element of a group equal $0$).
58: This is not actually done in the paper.}
59:
60: \begin{definition}
61: The \emph{commutator\/} of two group elements $x$ and $y$,
62: denoted $[x,y]$, is $xyx^{-1}y^{-1}$.
63: The \emph{commutator length\/} of an element $g$ of the derived
64: subgroup of a group $G$,
65: denoted $\cl_G(g)$, is the minimal $n$ such that there exist elements
66: $x_1$, \dots, $x_n$, $y_1$, \dots, $y_n$ in $G$ such that
67: $g=[x_1,y_1]\dots[x_n,y_n]$.
68: The commutator length of the identity element is $0$.
69: The \emph{commutator width\/} of a group $G$, denoted $\cw(G)$,
70: is the maximum of the commutator lengths of the elements of its
71: derived subgroup $[G,G]$.
72: \end{definition}
73:
74: %% Definition corrected 2008-07-03:
75: \begin{definition}
76: The \emph{conjugate\/} of a group element $g$ by a group element $h$,
77: denoted $g^h$, is $hgh^{-1}$.
78: A nontrivial group $G$ is called \emph{$n$-boundedly simple\/} if
79: %and only if
80: for every two nontrivial elements $g,h\in G$,
81: the element $h$ is the product of $n$ or fewer
82: conjugates of $g^{\pm1}$, \ie,
83: $$
84: (\exists m\le n)\,
85: (\exists \sigma_1,\dots,\sigma_m\in\{\pm1\})\,
86: (\exists x_1,\dots,x_m\in G)\,
87: (g=(h^{\sigma_1})^{x_1}\dots(h^{\sigma_m})^{x_m}).
88: $$
89: A group $G$ is called \emph{boundedly simple\/} if
90: %and only if
91: it is $n$-boundedly simple for some natural~$n$.
92: \end{definition}
93:
94: %\begin{definition}
95: %The \emph{conjugate\/} of a group element $g$ by a group element $h$,
96: %denoted $g^h$, is $hgh^{-1}$.
97: %A group $G$ is called \emph{$n$-boundedly simple\/} if
98: %%and only if
99: %for every two nontrivial elements $g,h\in G$,
100: %the element $h$ is the product of $n$ or fewer
101: %conjugates of $g^{\pm1}$, \ie,
102: %$$
103: %(\exists m\le n)\,
104: %(\exists \sigma_1,\dots,\sigma_m\in\{\pm1\})\,
105: %(\exists x_1,\dots,x_m\in G)\,
106: %(g=(h^{\sigma_1})^{x_1}\dots(h^{\sigma_m})^{x_m}).
107: %$$
108: %%\begin{multline*}
109: %%(\exists m\le n)\,
110: %%(\exists \sigma_1,\dots,\sigma_m\in\{\pm1\})\,
111: %%(\exists x_1,\dots,x_m\in G)\\
112: %%(g=(h^{\sigma_1})^{x_1}\dots(h^{\sigma_m})^{x_m}).
113: %%\end{multline*}
114: %A group $G$ is called \emph{boundedly simple\/} if
115: %%and only if
116: %it is $n$-boundedly simple for some natural~$n$.
117: %\end{definition}
118:
119: Every boundedly simple group is simple, but the converse
120: is not generally true
121: (\eg, for an infinite alternating group).
122:
123: \begin{remark}
124: %For each natural $n$, the class of $n$-boundedly simple groups is
125: %definable by a formula of the restricted predicate calculus.
126: %The class of all simple groups is not.
127: A group is boundedly simple if and only if
128: each of its ultrapowers is simple.
129: If a group is $n$-boundedly simple, then all its ultrapowers are
130: $n$-boundedly simple.
131: \end{remark}
132:
133: \begin{theorem}
134: \label{theorem:bsglcw}
135: For every natural\/ $n$\textup,
136: there exists a torsion-free\/ $2$-generated simple group\/ $G$
137: with a rank-\/$2$ free subgroup\/ $H$ such that\/\textup:
138: \begin{enumerate}
139: \item
140: for every\/ $g\in G$ and every\/ $x\in G\setminus\{1\}$\textup,
141: there exist\/ $y_1$\textup{, \dots,} $y_{2n+2}$ in\/ $G$ such that\/
142: $g=x^{y_1}\dots x^{y_{2n+2}}$\textup; and
143: \item
144: for every\/ $h\in H\setminus\{1\}$ and for every\/ $m\ge2n$\textup,
145: $\cl_G(h^m)>n$
146: \end{enumerate}
147: \textup(in particular\textup, $G$ is\/ $(2n+2)$-boundedly simple\textup,
148: and\/ $n+1\le\cw(G)\le2n+2$\textup{).}
149: Moreover\textup, there exists such a group\/ $G$ with decidable
150: word and conjugacy problems\textup.
151: \end{theorem}
152:
153: Note that Theorem~\ref{theorem:bsglcw} improves the result of
154: Theorem~2 in~\cite{Muranov:2005:dsmcbgbsg}.
155:
156: \begin{theorem}
157: \label{theorem:sgicw}
158: There exists a torsion-free\/ $2$-generated simple group\/ $G$
159: with a rank-\/$2$ free subgroup\/ $H$ such that
160: for every\/ $h\in H\setminus\{1\}$\textup,
161: $$
162: \lim_{n\to+\infty}\cl_G(h^n)=+\infty
163: $$
164: \textup(in particular\textup, $G$ has
165: infinite commutator width\/\textup{).}
166: Moreover\textup, there exists such a group\/ $G$ with decidable
167: word and conjugacy problems\textup.
168: \end{theorem}
169:
170: The theorems are proved by providing examples of groups
171: which satisfy the required properties.
172: These groups are presented by generators and defining relations
173: in Section~\ref{section:group_presentations}.
174: (The constructed presentations are recursive,
175: as follows from the proof of Proposition~\ref{proposition:Gn_G_swcp}.)
176:
177: The properties of simplicity or bounded simplicity for the constructed
178: groups follow directly from the imposed relations, but the
179: existence of free non-cyclic subgroups and estimates on the commutator
180: lengths of their elements are obtained through a nontrivial
181: analysis of van Kampen diagrams on spheres with handles.
182:
183: To show that the commutator length of a given element $g$ of
184: a constructed group is greater than $n$, it is proved that
185: if $\Delta$ is a van Kampen diagram on a sphere with handles and a hole
186: such that some group word representing the element $g$ ``reads''
187: on the boundary of $\Delta$,
188: then the number of handles is greater than $n$.
189: This is done by assuming that the number of handles
190: is not greater than $n$, which gives a lower bound on the
191: Euler characteristic, and coming to a contradiction.
192:
193: The contradiction is obtained as follows.
194: The hole in the diagram is covered with an extra face
195: so as to make the diagram closed.
196: Some arcs of the diagram are selected and distributed
197: among the faces.
198: This is done in such a manner that the sum of the lengths
199: of the arcs associated to each face is small,
200: significantly less than half of the perimeter of that face,
201: but almost all edges of the diagram lie on selected arcs.
202: %^corrected on 2007-01-31, before was: <the following 1 line>
203: %but almost all {the} edges of the diagram lie on selected arcs.
204: This eventually leads to a contradiction with the fact that
205: in a closed diagram the number of edges is half the sum
206: of the perimeters of faces.
207:
208: Certainly, the least obvious part of the proof is
209: distributing ``almost all'' edges among the faces,
210: while associating ``few'' edges to each face.
211: For this purpose the group presentations are constructed
212: with small-cancellation-type conditions.
213: These conditions allow one to choose a system of selected arcs
214: in a specific way.
215: The selected arcs together cover almost all edges of the diagram,
216: and are short relative to the perimeters of incident faces.
217: Using the bound on the Euler characteristic of the diagram
218: (determined by the number of handles),
219: and one combinatorial lemma by Philip Hall,
220: the selected arcs can be distributed among the faces so as
221: to have ``very few'' relatively ``short'' and
222: ``a few'' relatively ``very short''
223: arcs associated to each face.
224:
225: The approach used in this paper is similar to that of
226: \cite{Muranov:2005:dsmcbgbsg}.
227: There are improvements and generalizations which allow one
228: to obtain better estimates and
229: to deal with diagrams on arbitrary surfaces.
230: \par
231:
232: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
233: