1: %auto-ignore
2: \newcommand{\HH}{{\bf H}}
3: \newcommand{\Case}{\paragraph}
4: \paragraph{Abstract.}
5: \footnote{This research was completed when this author held a position at the
6: Institute for System Analysis, Moscow.}
7: Let $\Sigma_0,\dots,\Sigma_6$ be the graphs from Suzuki chain
8: \cite{Suz}. We classify connected locally $\Sigma_i$ graphs
9: $\Theta_{i+1}$ for $i=3,4,5$. If $i=3,4$ then $\Theta_{i+1}$ is
10: isomorphic to $\Sigma_{i+1}$, whereas $\Theta_6$ is isomorphic
11: either to $\Sigma_6$ or to its 3-fold antipodal cover
12: $3\Sigma_6$.
13: %
14: \section{Introduction and results}
15: \label{suz:sect1}
16: Recently there is an extensive study of the geometries of the
17: classical groups as point-line systems with fixed local structure.
18: This approach has led to many beautiful results. We refer the
19: reader to a recent paper of Cohen and Shult~\cite{CoSh} for a brief
20: survey about that.
21:
22: This paper is an attempt to look
23: at the Suzuki chain sporadic simple groups from the aforementioned
24: point of view.
25:
26: It is necessary to mention a lot of activity in this direction
27: under some additional assumptions. The first idea is to assume
28: some group action on the geometry (usually, it is a diagram
29: geometry). Such an approach lead to study of some presentations
30: for sporadic groups, and sometimes to proving the faithfulness of
31: these presentations.
32: For instance, see Ivanov~\cite{Ivnv:J4}, Shpectorov~\cite{ssh:Co2},
33: Soicher~\cite{Soi:Suz}.
34: Another direction is via assuming some global property of the
35: geometry, e.g. assuming that the related graph is strongly
36: regular or distance regular, see the book by
37: Brouwer~et~al.~\cite{BCN}. This book contains many results and many
38: references to results of this kind. See also
39: Cameron~et~al.~\cite{CHP}. It is worth to mention that here we do not
40: assume either any group action or global property.
41:
42: Note that graphs with constant neighbourhood often arise as collinearity
43: graphs of diagram geometries, see Buekenhout~\cite{Bue:geo:spo}.
44: Neumaier~\cite{Neu} observed that the graphs from Suzuki chain produce
45: nice flag-transitive diagram geometries for the related groups.
46:
47: Throughout the paper we consider undirected graphs without loops
48: and multiple edges.
49: Given a graph $\Gamma$, let us denote the set of vertices by
50: $\V=\V\Gamma$, the set of edges by $\E=\E\Gamma$. Let $X\subseteq
51: \V\Gamma$.
52: We denote by $\<X\>=\<X\>_{\Gamma}$ the subgraph $\Xi$
53: of $\Gamma$ {\em induced} by $X\/$
54: (i.e. $\V\Xi=X,\/\E\Xi=\{ (u,v)\in \E\Gamma|u,v\in X\}$).
55: Given two graphs $\Gamma$ and $\Delta$, the graph $\Gamma\cup\Delta$
56: (resp. the graph $\Gamma\cap\Delta$) is
57: the graph with the vertex set $\V\Gamma\cup \V\Delta$ (resp.
58: $\V\Gamma\cap \V\Delta$) and the edge set $\E\Gamma\cup \E\Delta$
59: (resp. $\E\Gamma\cap \E\Delta$). Given $v\in \V\Gamma$, we denote
60: $\Gamma_i(v)=\<\{x\in \V\Gamma|x$ at distance $i$ from $v\}\>$,
61: and $\Gamma_1(v)=\Gamma(v)$.
62: Furthermore, $\Gamma(X)=\bigcap_{x\in X} \Gamma(x)$.
63: To simplify the notation we use
64: $\Gamma(v_1,\dots,v_k)$ instead of $\Gamma(\{v_1,\dots,v_k\})$ and
65: $u\in\Gamma_i(\dots)$ instead of $u\in \V\Gamma_i(\dots)$.
66: As usual, $v=v(\Gamma)=|\V\Gamma|$, $k=k(\Gamma)=v(\Gamma(x))$,
67: $\lambda=\lambda(x,y)=\lambda(\Gamma)=v(\Gamma(x,y))$, where
68: $x\in\V\Gamma$, $y\in\Gamma(x)$. Let $y\in\Gamma_2(x)$. We denote
69: $\mu=\mu(x,y)=\mu(\Gamma)=v(\Gamma(x,y))$. Of course, we use
70: $k$, $\lambda$, $\mu$ if it makes sense, i.e. if those numbers
71: are independent on the
72: particular choice of the corresponding vertices.
73: If $\Delta$ is a (proper) subgraph of
74: $\Gamma$ we denote this fact as $\Delta\subseteq\Gamma$ (resp.
75: $\Delta\subset\Gamma$).
76:
77: We denote the complete $n$-vertex graph by $K_n$, the complete
78: multipartite graph with the $m$ parts of equal size $n$
79: by $K_{n\times m}$, the circuit of length $n$ by
80: $C_n$, the empty graph by $\emptyset$.
81:
82: $\Aut(\Gamma)$ denotes the automorphism group of $\Gamma$.
83: Our group-theoretic notation is as in \cite{Atl}.
84:
85: A {\em strongly regular} graph $\Gamma$ is a connected regular
86: graph of valency $k$, such that
87: $\lambda=\lambda(\Gamma)=\lambda(u,v)$ and
88: $\mu=\mu(\Gamma)=\mu(u,w)>0$ are independent on the choice of
89: $u\in \V\Gamma$,
90: $v\in\Gamma(u)$ and $w\in V\Gamma\setminus(\{u\}\cup\Gamma(u))$.
91: Note that $\mu>0$ implies the diameter of $\Gamma$ is less than
92: or equal to two.
93:
94: Let $\Gamma$, $\Delta$ be two graphs. We say that $\Gamma$ is
95: {\em locally} $\Delta$ if $\Gamma(v)\cong\Delta$ for any $v\in \V(\Gamma)$.
96:
97: Let $\Gamma,\ \overline\Gamma$ be two graphs. We say that $\Gamma$ is a
98: {\em cover} of $\overline\Gamma$ if there exists a mapping $\varphi$
99: from $\V\Gamma$ to $\V\overline\Gamma$ which maps edges to edges and for
100: any $v \in \V\Gamma$ the restriction of $\varphi$ to $\Gamma(v)$ is an
101: isomorphism onto $\overline\Gamma(\varphi(v))$. Note that, since
102: the latter assumption in the definition of cover is usually
103: omitted, our definition of cover is a bit nonobvious.
104:
105: Suppose we have a chain of graphs $\Sigma_1,\dots,\Sigma_n$, such that
106: $\Sigma_i$ is locally $\Sigma_{i-1},\ i=2,\dots,n$. Then for any complete
107: $k$-vertex subgraph $\Upsilon$ of $\Sigma_m$, $1<k<m\leq n$
108: we have $\Sigma_m(\V\Upsilon)\cong\Sigma_{m-k}$.
109:
110: {\em Suzuki chain} \cite{Suz} consists of the following graphs
111: $\Sigma_1,\dots,\Sigma_6$.
112:
113: $\Sigma_1$ is the empty four-vertex graph.
114:
115: $\Sigma_2$ is the incidence graph of the (unique) $2-(7,4,2)$ design.
116:
117: $\Sigma_3,\dots,\Sigma_6$ are strongly regular with the parameters
118: $(v,k,\lambda,\mu)$ equal to
119: $$(36, 14, 4, 6),\ (100, 36, 14, 12),\ (416, 100, 36, 20),\ (1782,
120: 416, 100, 96)$$ respectively.
121: The automorphism groups of $\Sigma_i$ are $S_4$, ${\it PGL}_2 (7)$, $G_2 (2)$,
122: $J_2\!:\!2$, $\Aut(G_2(4))$, $\Aut(\Suz)$ respectively.
123:
124: Note that $\Sigma_i,\ i=2,\dots ,6$ is locally $\Sigma_{i-1}$.
125: Besides $\Sigma_6$, there exists another connected locally $\Sigma_5$ graph
126: $3\Sigma_6$ -- the 3-fold antipodal cover of $\Sigma_6$.
127: $3\Sigma_6$ is
128: a distance transitive graph (see \cite{BCN}) with the intersection
129: array $$\{ 416,315,64,1;1,32,315,416\}$$. It has been
130: constructed by Soicher~\cite{Soi:Suz} (see also \cite{Soi:drg}).
131:
132: \begin{thm}
133: Let $\Theta_{i+1}$ be a connected locally $\Sigma_i$
134: graph, $i=3,4,5$. Then $\Theta_4\cong\Sigma_4$, $\Theta_5\cong\Sigma_5$,
135: $\Theta_6\cong\Sigma_6$ or $3\Sigma_6$.
136: \label{suz:MTH}
137: \end{thm}
138:
139: Note that under the additional assumption that $\Aut(\Theta_{i+1})$ is
140: transitive on the ordered $k$-cliques of $\Theta_{i+1}$ for any $k\leq i$,
141: this statement is proved in \cite{Soi:Suz} using a coset enumeration.
142:
143: \medskip
144: {\sl Remark.} Results from \cite{Soi:Suz} and our Theorem~\ref{suz:MTH}
145: immediately lead to
146: a computer-free proof of the faithfulness of the well-known presentations for
147: groups from Suzuki chain, given in \cite{Atl}. Namely if $G_n$ is given by
148: the presentation
149: $$\bigl\langle_{g_n}
150: \!\!\!\!\!\circ \!{\phantom{00000}\over \phantom{0}}
151: {\vphantom{\bigl\langle}}_{g_{n-1}}\!\!\!\!\!\!\!\!\!\!\!\!\!\circ \ \cdots\
152: {\vphantom{\bigl\langle}}_{g_3}
153: \!\!\!\!\!\!\!\circ \!\!{\phantom{00000}\over \phantom{0}}
154: {\vphantom{\bigl\langle}}_{g_2}
155: \!\!\!\!\!\!\!\circ \!\!{{\phantom{00} 8 \phantom{00}}\over \phantom{0}}
156: {\vphantom{\bigl\langle}}_{g_1}
157: \!\!\!\!\!\!\!\circ \!\!{\phantom{00000}\over \phantom{0}}
158: {\vphantom{\bigl\langle}}_{g_0}
159: \!\!\!\!\!\!\!\circ \!\!{\phantom{00000}\over \phantom{0}}
160: {\vphantom{\bigl\langle}}_a
161: \!\!\!\!\!\circ\ |\
162: (g_1 g_2)^4=a,\ (g_0 g_1 g_2 g_3)^8=1 \bigr\rangle, $$
163: then $G_4 \cong\Aut(J_2),\ G_5 \cong\Aut(G_2(4)),\ G_6 \cong 3\Aut(\Suz)$.
164:
165: \section{Proof of Theorem~\protect\ref{suz:MTH}}
166: \label{suz:sect2}
167: Trying to simplify the reading of Sect.~\ref{suz:sect2},
168: we give a sketch of proof here.
169:
170: Let $\Gamma=\Theta_i$. Pick $u\in \V\Gamma$ and look at
171: $\Gamma(u,v)$ for each $v\in\Gamma_2(u)$.
172: First, we establish the characterization of $\Gamma(u,v)$ in terms
173: of their local structure (Lemma~\ref{suz:L1}~(1)). Then we classify the
174: connected components of the possible candidates to $\Gamma(u,v)$
175: (Proposition~\ref{suz:P3}). Next, we list all the possible groupings of
176: $\Gamma(u,v)$ from the connected components (Lemma~\ref{suz:L5}).
177: This list
178: and Lemma~\ref{suz:L1}~(2) allows us to determine the edges between
179: $\Gamma(u)$ and $\Gamma_2(u)$. Using Proposition~\ref{suz:P2} we express the
180: adjacency in $\Gamma_2(u)$ in terms of intersections of
181: $\Gamma(u,v)$-subgraphs (Sect.~\ref{suz:sect23}
182: ($i=4,5$), Sect.~\ref{suz:sect24} ($i=6$), also
183: Lemmas~\ref{suz:L6} and~\ref{suz:L7}
184: are used). In cases $i=4,5$ we are done by Lemma~\ref{suz:L4},
185: which states that $\Gamma(u,v)$ are "hyperplanes" of $\Gamma(v)$.
186: So holds in a subcase (Case $c(\Omega)=3$) of the case $i=6$, as well.
187: In the remaining subcase (Case $c(\Omega)=1$, Case $c(\Omega)=2$
188: immediately leads to a contradiction) of the case $i=6$ we show
189: that $\Gamma$ is a (unique) covering of $\Sigma_6$, hence
190: $\Gamma\cong 3\Sigma_6$.
191:
192: \subsection{Suzuki's construction }
193: \label{suz:sect21}
194:
195: We include the necessary information about the construction of
196: $\Sigma_i,\/$ given by Suzuki~\cite{Suz}, $i=3,\dots,6$ (see also \cite{Atl}).
197:
198: Let $\Delta=\Sigma_{i-1}$, $\Gamma=\Sigma_i$ and $H=\Aut(\Delta)$
199: for $i=3,\dots,6$.
200: In fact, $\Gamma$ can be defined in terms of $\Delta$ and $H$.
201:
202: Let $\infty$ be an extra symbol. Let $S$ be the
203: conjugacy class of $2$-subgroups $Z=\<z\>$ of $H$, where $z$ is a
204: 2A-involution of $H$, $i=3,4,5$. In the remaining case $i=6$ $S$ is the
205: conjugacy class of $2^2$-subgroups $Z=\<x,y\>$ of $H$,
206: such that $[H:N_H(Z)]=1365$ and $x,\/y$ are 2A-involutions of $H$.
207: Then $\V\Gamma=\{\infty\} \cup \V\Delta \cup S.\/$
208:
209: $\E\Gamma$ is exactly as follows. The vertex $\infty$ is adjacent to each
210: vertex
211: from $\V\Delta$, two vertices from $\V\Delta$ are adjacent if they are
212: adjacent
213: as vertices of $\Delta$, a vertex $x\in S\/$ is adjacent to a vertex
214: $v\in \V\Delta$ if a nontrivial element of $x$, considering as a subgroup
215: of $H$, fixes $v$, two vertices $x,y\in S\/$ are adjacent if $x,\/y$,
216: considering as subgroups of $H$, do not commute, but there exists $z\in S\/$
217: commuting with both of them.
218:
219: \subsection{Neighbourhood of two vertices at distance two}
220: \label{suz:sect22}
221: We start with a simple general fact. Let $\Delta$ be a connected
222: graph satisfying the following property \medskip
223:
224: \begin{itemize}
225: \item[$(*)$] For any $u\in \V\Delta$ and
226: $v\in \V\Delta\setminus (\V\Delta(u)\cup\{ u \})$\\
227: 1)~the subgraph $\Delta(u,v)$ is isomorphic to some $M_{\Delta}$,
228: whose isomorphism type is independent on the particular choice of $u,\/ v
229: $.\\
230: 2)~for $w\in \V\Delta\setminus (\V\Delta(u)\cup\{ u \}),\/w\not=v,\/$
231: $\Delta(u,v)\not=\Delta(u,w)$.
232: \end{itemize}
233:
234: \begin{lem}
235: Let $\Gamma$ be locally $\Delta$ graph, where $\Delta$
236: satisfies $(*)$. For any $u\in \V\Gamma$ and $v\in\Gamma_2(u)$
237: the following holds
238: \begin{itemize}
239: \item[$(1)$] $\Gamma(u,v)$ is locally $M_{\Delta}$,
240: \item[$(2)$] for $w\in\Gamma_2(u),\ w\not=v\/$, we have
241: $\Gamma(u,v)\not=\Gamma(u,w)$.
242: \end{itemize}
243: \label{suz:L1}
244: \end{lem}
245:
246: \begin{pf}
247: By 1) of (*), we have $\Gamma(u,v,y)\cong M_{\Delta}$ for
248: any $y\in \Gamma(u,v)$. The first claim is proved.
249: By way of contradiction, let $w\in\Gamma_2(u),\/w\not=v\/$,\break
250: $\Gamma(u,v)=\Gamma(u,w)$. Pick $y\in\Gamma(u,v)$.
251: Looking at the common neighbourhood of $u,\/v$ and $u,\/w$
252: in $\Gamma(y)\cong\Delta$, we obtain a contradiction. \end{pf}
253:
254: Note that $\Delta=\Sigma_i$ ($i=3,4,5$) satisfies (*) above.
255: Indeed, the stabilizer of $u\in \V\Delta$ in $\Aut(\Delta)$ acts
256: transitively and primitively on
257: $\Delta_2(u)$. The transitivity implies 1) of (*), while the
258: primitivity
259: 2) of (*). Thus Lemma~\ref{suz:L1} holds for locally $\Sigma_i$ graphs
260: $\Gamma$ for any $i=3,4,5$. In particular, for any
261: $u\in\V\Gamma$ and $v\in\Gamma_2(u)$ the subgraph $\Gamma(u,v)$
262: is a locally $M_{\Sigma_i}$ subgraph of $\Gamma(u)\cong\Sigma_i$
263: (i=3,4,5). We turn to the classification of the
264: locally $M_{\Sigma_i}$ subgraphs of $\Sigma_i$ ($i=3,4,5$).
265:
266: First, we need a technical statement. Note that in fact we will prove
267: slightly more than we will really need. For a graph $\Gamma$ and $g\in
268: \Aut(\Gamma)$ we denote the
269: subgraph induced by the vertices fixed by $g$ as $Fix(g)=Fix_\Gamma(g)$.
270:
271: \begin{pr}\label{suz:P2}
272: Let $g$ be a 2A-involution in $\Aut(\Sigma_i)$, $i=3,4,5$. Then
273: the number of 2A-involutions $h$ corresponding to particular isomorphism
274: type of the subgraph $Fix(g)\cap Fix(h)$ is presented in
275: Table~\ref{suz:tab1}.
276: \end{pr}
277: \begin{table}
278: \begin{center}
279: \begin{tabular}{|c|c|c|c|c|} \hline
280: $i$ &\multicolumn{4}{|c|}{$Fix(g)\cap Fix(h)$}\\ \hline
281: 3 &6 &24 &32 &--\\ \cline{2-5}
282: &$K_{4\times 2}$ &$K_{2\times 2}$ &$K_3$ &--\\ \hline
283: 4 &10 &80 &160 &64 \\ \cline{2-5}
284: &$K_{4\times 3}$ &$K_{2\times 3}$ &$K_4$ &$\emptyset$ \\ \hline
285: 5 &20 &320 &1024 &--\\ \cline{2-5}
286: &$K_{4\times 4}$ &$K_{2\times 4}$ &$K_5$ &--\\ \hline
287: \end{tabular}
288: \end{center}
289: \caption{ Number of 2A-involutions $h\in Aut(\Sigma_i)$
290: vs. the isomorphism type of $Fix(g)\cap Fix(h)$. }
291: \label{suz:tab1}
292: \end{table}
293: \begin{pf}
294: We follow the notation of Sect.~\ref{suz:sect21}
295: Denote $\Gamma=\Sigma_i$, $\Omega=Fix(g)$, $G=\Aut(\Gamma)$.
296: Pick $\infty\in \V\Omega$, denote $\Delta=\Gamma(\infty)$. It
297: follows from the Suzuki's construction given in Sect.~\ref{suz:sect21}
298: that $\Omega$ is locally $M_\Gamma$. It should be mentioned that
299: $\Omega_2(\infty)=\{x \in g^{\Aut(\Delta)}|(xg)^2=1\}$. Let us
300: denote by $\Xi_h$ the subgraph $\Omega\cap Fix(h)$ for $h\in g^G$.
301:
302: \Case{$i=3$} There exist two distinct subgraphs $\Xi$ of $\Omega$
303: isomorphic to $K_{4\times 2}$ and containing $\infty$. It is easy
304: to check that
305: $\Xi$ determines exactly two $h\in g^G$ such that
306: $\Xi_h=\Xi$. Hence there exist exactly
307: $2\cdot 2v(\Omega)/v(K_{4\times 2})=6$ involutions $h\in g^G$ such that
308: $\Xi_h\cong K_{4\times 2}$.
309:
310: It is well-known that the action of $G$ on $g^G$ coincides with the
311: action on the set of points of a generalized hexagon \HH\ of order
312: (2,2) and that
313: the corresponding subdegrees are $1,6,24,32$ according to the distance
314: in the collinearity graph of \HH. Hence if $h'\in g^G$ corresponds to
315: the subdegree 24 then there exists
316: $h\in g^G$ such that $\Xi_h\cong Fix(h)\cap Fix(h')\cong K_{4\times 2}$.
317: Hence $v(\Xi_{h'})\geq 4$. On the other hand it is easy to find
318: elements
319: $h',h''\in g^G$ with $\Xi_{h'}\cong K_{2\times 2}$ and $\Xi_{h''}\cong K_3$.
320: The proof for $i=3$ is complete.
321:
322: \Case{$i=4$} By induction, there exist 6 elements $h_j\in g^G$ such that
323: $\Xi_{h_j}(\infty)\cong K_{4\times 2}$. Pick $y=h_1$ and $u\in
324: \V\Xi_y(\infty)$. Since $v(\Xi_y(u))\geq 1+k(K_{4\times 2})$, we have
325: $\Xi_y(u)\cong K_{4\times 2}$. Hence a connected component $\Upsilon$
326: of $\Xi_y$ is locally $K_{4\times 2}$. Thus $\Upsilon\cong
327: K_{4\times3}$ (cf. \cite[Proposition~1.1.5]{BCN}).
328: It can be easily shown that for any $w \in \V\Omega$ the subgraph
329: $\Omega(w)\cap\Upsilon$ is nonempty. Hence $\Xi_y=\Upsilon$. Thus there exist
330: exactly $6v(\Omega)/v(K_{4\times 3})=10$ involutions $h\in g^G$ such that
331: $\Xi_h\cong K_{4\times 3}$.
332:
333: There exist 24 elements $h'_j\in g^G$ such that
334: $\Xi_{h'_j}(\infty)\cong K_{2\times 2}$. Pick $y=h'_1$ and $u\in
335: \V\Xi_y(\infty)$.
336: Since $\Xi_y(u)$ contains a 2-path, we have $\Xi_y(u)\cong K_{2\times 2}$ or
337: $K_{4\times 2}$. The latter case is impossible. Indeed, as we have already
338: proved, if the latter case holds, we have $\Xi_y\cong K_{3\times
339: 4}$. Thus $\Xi_y(\infty)\cong K_{2\times 4}$, contradiction.
340: Hence a connected component $\Upsilon$ of $\Xi_y$
341: is locally $K_{2\times 2}$. Thus $\Upsilon\cong K_{2\times 3}$.
342: Therefore a connected component of $\Xi_y$ distinct from
343: $\Upsilon$ (if such a component exists at all) is a
344: singleton $s\in \V\Xi_y$. Looking at the neighbourhood of $s$, we obtain a
345: contradiction, since $\Xi_y(s)\cong\emptyset$. Hence $\Xi_y=\Upsilon$.
346: Thus there exist exactly $24v(\Omega)/v(K_{2\times 3})=80$
347: elements $h'\in g^G$ such that $\Xi_{h'}\cong K_{2\times 3}$.
348:
349: There exist 32 elements $h''_j\in g^G$ such that
350: $\Xi_{h''_j}(\infty)\cong K_3$. Pick $y=h''_1$ and $u\in \V\Xi_y(\infty)$.
351: Similarly to the previous case, we obtain that a connected component
352: $\Upsilon$ of $\Xi_y$ is locally $K_3$. Thus $\Upsilon\cong K_4$.
353: Again, looking at $\Xi_y$ as at a subgraph of $\Omega$, we obtain
354: $\Xi_y=\Upsilon$.
355: Thus there exist exactly $32v(\Omega)/v(K_4)=160$ elements $h''\in
356: g^G$ such that $\Xi_{h''}\cong K_4$.
357:
358: Clearly for the other $h\in g^G$ we have $\Xi_h\cong\emptyset$.
359: The proof in the case $i=4$ is complete.
360:
361: \Case{$i=5$} The argument runs parallel to the argument for the case $i=4$.
362: \end{pf}
363:
364: \begin{pr}
365: Any connected locally $M_{\Sigma_i}$ subgraph
366: of $\Sigma_i$ coincides with the subgraph $Fix(g)$ for a
367: 2A-involution $g\in \Aut(\Sigma_i)$ $(i=3,4,5)$.
368: \label{suz:P3}
369: \end{pr}
370:
371: \begin{pf}
372: Denote $\Gamma=\Sigma_i$. Let $\Omega$ be a connected locally
373: $M_\Gamma$ subgraph
374: of $\Gamma$, $i=3,4,5$. We will frequently use the construction of $\Gamma$
375: given in Sect.~\ref{suz:sect21}, as well as the corresponding notation.
376: Without loss of generality $\infty\in \V\Omega$.
377:
378: \Case{$i=3$} Recall that $\Delta$ is the incidence graph of the
379: unique $2-(7,4,2)$ design. Since the latter design is complementary to
380: $\Pi=PG(2,2)\/$, the equivalent description of $\Delta$ (and $\Gamma$)
381: can be given in terms of $\Pi$. Namely, the vertices of $\Delta$ are
382: the points and the lines of $\Pi$, a point $p$ is adjacent to a line
383: $l$ if $p$ does not belong to $l$.
384:
385: For each 2A-involution $v$ of $H$ there exists a unique flag
386: (= incident point-line pair) $(p,l)$ such that $v$ leaves all points on
387: $l$ and all lines through $p$ fixed. Thus it is easy to check that
388: the subgraph $M_\Gamma=\Gamma(\infty,v)$ is isomorphic to
389: $K_1\cup K_1\cup C_4$.
390:
391: It is easy to check that {\em any} (generated) subgraph of
392: $\Delta$ of the latter shape is $\Gamma(\infty,v)$ for some
393: $v\in \Gamma_2(\infty)$. Indeed, $\Delta$ has the remarkable property
394: that any 2-path in $\Delta$ lies in a unique $C_4$ (i.e. $\Delta$
395: is a {\em rectagraph} (see \cite{BCN})). Moreover,
396: $\Aut(\Delta)$ acts transitively on the $C_4$-subgraphs of
397: $\Delta$. Finally, it is
398: straightforward to check that there exist exactly two vertices of $\Delta$
399: which are nonadjacent to any vertex of a subgraph $\Xi$ of
400: $\Delta$ isomorphic to $C_4$.
401:
402: Thus, without loss of generality, we may assume that
403: $\Gamma(\infty,v)\subset \Omega$ for some $v\in \Gamma_2(\infty)$. Let
404: $\Xi$ be the $C_4$-subgraph of $\Omega(\infty)$, $u\in \V\Xi$.
405: Since $\Gamma(u)\cong\Delta$ is a rectagraph, a quick
406: look at $\Xi(u)$ gives us $v\in \V\Omega$. Next, $v$ is adjacent to each
407: vertex in $\Omega(\infty)$. Hence any $w\in \Gamma_2(\infty)\cap\Gamma(v)$
408: does not belong to $\V\Omega$.
409:
410: Let $x\in \Omega(\infty)\setminus \V\Xi$.
411: The unique $C_4$-subgraph of $\Omega(x)$ lies in
412: $\Gamma_2(\infty)$. On the other hand $\Gamma(x,v)\subset\Gamma_2(\infty)$.
413: Hence for $Q=\Gamma_2(\infty)\cap\Gamma_2(v)\cap\Gamma(x)$ we have
414: $v(Q)=4$. Thus $Q\subset \Omega$.
415:
416: Consider $v$ as a 2A-involution of $H$. It is easy to
417: see, using \cite{Atl}, that $v$ is a 2A-involution of $\Aut(\Gamma)$. Therefore
418: its fixed vertices are
419: $\{\infty\}\cup\{v\}\cup \V\Gamma(\infty,v)\cup \V Q\subseteq \V\Omega$. Since
420: the subgraph $Fix(v)$ induced by the vertices fixed by $v$ is connected and
421: locally
422: $M_\Gamma$, we have equality in the latter inclusion. This completes
423: the proof in the case $i=3$.
424:
425: \Case{$i=4$} By induction, any subgraph of $\Sigma_3$ isomorphic to
426: $\Gamma(\infty,v)$ for some $v\in \Gamma_2(\infty)$ is in fact $Fix(g)$ for
427: some 2A-involution $g\in H\/$. Hence, without loss of generality, we may
428: assume
429: that $\Gamma(\infty,v)\subset \Omega\/$.
430:
431: Let $u\in \Omega(\infty),\ v\in \Omega_2(\infty)\cap\Omega(u)\/$.
432: Remembering the structure of $\Omega(u)$, we have
433: $v(\Omega(\infty,u,v))\geq 4$. Hence, by Proposition~\ref{suz:P2}
434: we have
435: $v(\Omega(\infty,v))\geq 8$. So $v(\Omega_2(\infty))\leq 7$. Let $Q$ be
436: the set of {\em all} possible candidates to $\V\Omega_2(\infty)$. Since
437: $\lambda(\Omega)=6$, there are precisely $60$ edges of $\E\Omega$ going
438: from $\V\Omega(\infty)$ to $\V\Omega_2(\infty)$. On the other hand precisely
439: $60$ edges are going from $Q$ to $\V\Omega(\infty)$. Hence
440: $Q=\V\Omega_2(\infty)$.
441: In particular, a 2A-involution $w\in H$ belongs to $Q$. Next, $w$ is
442: a 2A-involution of $\Aut(\Gamma)$. Therefore $\Omega=Fix(w)$.
443:
444: \Case{$i=5$} The argument runs parallel to the argument in the
445: previous case.
446: \end{pf}
447:
448: We have already known the connected components of locally $M_\Gamma$
449: subgraphs $\Omega$. Let us study the relationship between $\Omega$
450: and the vertices outside $\V\Omega$.
451:
452: \begin{lem}
453: Let $\Omega$ be connected locally $M_\Gamma$,
454: $v\in \V\Gamma\setminus \V\Omega$, $\Gamma(v)\cap \Omega\not=\emptyset$,
455: where $\Gamma=\Sigma_i$ $(i=3,4,5)$.
456: Then $\Gamma(v)\cap\Omega\cong K_{2\times (i-1)}$.
457: If $i=3,4$ then for any $v\in \V\Gamma\setminus \V\Omega$ the subgraph
458: $\Gamma(v)\cap \Omega\not=\emptyset$.
459: In the remaining case $i=5$ for any $v\in \V\Gamma\setminus \V\Omega$
460: either the subgraph
461: $\Gamma(v)\cap \Omega$ is nonempty or $v$ lies in a uniquely determined
462: connected locally $M_\Gamma$ subgraph.
463: \label{suz:L4}
464: \end{lem}
465:
466: \begin{pf}
467: If $i=3,4$ then the centralizer of a 2A-involution in $\Aut(\Gamma)$
468: has two orbits on $\V\Gamma$ (look at the scalar product of permutation
469: characters, given in \cite{Atl}). Hence the isomorphism type of
470: $\Xi=\Xi_v=\Gamma(v)\cap \Omega$ is independent on the choice of
471: $v\in \V\Gamma\setminus \V\Omega$. Next,
472: $$v(\Xi)(v(\Gamma)-v(\Omega))=v(\Omega)(k(\Gamma)-k(\Omega)).$$
473: Thus $v(\Xi)=4$ or $6$ according to $i=3$ or $4$.
474:
475: If $i=3$ it is easy to
476: check that $k(\Xi)=2$. Indeed, pick a
477: vertex $u\in \V\Xi$. Look at $\Delta=\Gamma(u)$. Recall that
478: $\V\Omega(u)$ is the set of fixed points of a 2A-involution $g\in
479: \Aut(\Delta)$.
480: $C_{\Aut(\Delta)}(g)$ has one orbit on $\V\Delta\setminus \V\Omega(u)$.
481: Performing
482: the same calculation as above, we have $k(\Xi)=v(\Omega(u)\cap\Delta(v))=2$.
483: The proof in the case $i=3$ is complete.
484:
485: By induction if $i=4$ then $\Xi$ is locally $C_4=K_{2\times 2}$. Such
486: a graph $\Xi$ is unique and isomorphic to $K_{2\times 3}$.
487:
488: Consider $\Sigma_5$ as the subgraph $\Delta=\Gamma(\infty)$
489: of $\Gamma=\Sigma_6$, as in Sect.~\ref{suz:sect21}. For a vertex $u\in
490: \Gamma_2(\infty)$ the group $H={\rm N}_{\Aut(\Delta)}(u)$ (recall that
491: $u$ is a $2^2$-subgroup of $\Aut(\Delta)$) has two orbits on $\Delta$.
492: By Sect.~\ref{suz:sect21} and Proposition~\ref{suz:P3}, $\Gamma(\infty,u)$
493: is the disjoint union of three copies of $\Omega$.
494: This enables us to calculate, for $v\in \V\Gamma(\infty)\setminus
495: \V\Gamma(u)$, the number $n$ of edges coming from $v$ to
496: $\Gamma(\infty,u)$. Namely $n=24$.
497: There exists $g\in H(v)$ of order 3 permuting cyclically the connected
498: components of $\Gamma(\infty,u)$.
499: Hence $v(\Xi)=n/3=8$. On the other hand $\Xi$ is locally $K_{2\times3}$.
500: Such a graph is isomorphic to $K_{2\times 4}$.
501:
502: Finally, the last statement of the lemma follows directly from the
503: aforementioned consideration of $\Sigma_5$ as a subgraph of $\Sigma_6$. \end{pf}
504:
505: Lemma~\ref{suz:L4} immediately implies
506:
507: \begin{lem}
508: Let $\Omega$ be locally $M_{\Sigma_i}$ $(i=3,4,5)$.
509: If $i=3,4$ then $\Omega$ is connected. If $i=5$ then there exists a unique
510: equivalence relation $\varphi$ with the class size three on the set of
511: connected locally $M_{\Sigma_i}$ subgraphs such that all connected
512: components of $\Omega$ lie in one class of $\varphi$. \qed
513: \label{suz:L5}
514: \end{lem}
515:
516: \subsection{The final part of proof for i=3 and 4}
517: \label{suz:sect23}
518: Let $\Delta=\Sigma_i$, $\Gamma$ be a connected locally $\Delta$ graph.
519: Let us identify $\Delta$ with $\Gamma(\infty)$ for $\infty\in \V\Gamma$.
520: Since $\Delta$ satisfies (*), by the classification of the locally $M_\Delta$
521: subgraphs of $\Delta$ given in Sect.~\ref{suz:sect22}, we have that
522: $\mu(\Gamma)$ equals to the number of the fixed points of some
523: 2A-involution $g\in H=\Aut(\Delta)$. Since
524: $$v(\Gamma_2(\infty))=k(\Gamma)\cdot (k(\Gamma)-k(\Delta)-1)/ \mu(\Gamma),$$
525: $v(\Gamma_2(\infty))=|g^H|$. Since $\Delta$ satisfies (*), by
526: Lemma~\ref{suz:L1} for a locally $M_\Delta$ subgraph
527: $\Omega$ of $\Delta$ there exists a unique $v=v_\Omega \in \Gamma_2(\infty)$
528: such that
529: $\Omega=\Gamma(\infty,v)$ and for two distinct locally $M_\Delta$ subgraphs
530: $\Omega,\;\Omega'$ we have $v_\Omega\not=v_{\Omega'}$. Thus we have
531: proved that we do not have any choice determining the edges from
532: $\Gamma(\infty)$ to $\Gamma_2(\infty)$.
533:
534: Let $v\in \Gamma_2(\infty)$, $u\in \Gamma(v)\setminus\Gamma(\infty)$. It
535: follows from Lemma~\ref{suz:L4} , applied to $\Gamma(v)\cong\Delta$, that
536: $\Gamma(\infty,v,u)\cong K_{2\times (i-1)}$
537: (in particular, $\Gamma$ has diameter two).
538: Next, $v(\Gamma_2(\infty)\cap\Gamma(v))=24$ if $i=3$ ($80$ if $i=4$). On
539: the other hand by Proposition~\ref{suz:P2}
540: if $i=3$ there exist exactly $24$ ($80$ if $i=4$) locally
541: $M_\Delta$ subgraphs of $\Delta$ intersecting $\Gamma(\infty,v)$ in a
542: subgraph of the shape $K_{2\times (i-1)}$. Thus the adjacency in
543: $\Gamma_2(\infty)$ is unique determined, as well.
544:
545: The proof of Theorem~\ref{suz:MTH} is complete for $i=3,4$.
546:
547: \subsection{The final part of proof for i=5}
548: \label{suz:sect24}
549: Let $\Delta=\Sigma_5$, $\Gamma$ be a connected locally $\Delta$ graph.
550: First we prove further lemmas about locally $M_\Delta$ subgraphs of $\Delta$.
551: Denote $G=\Aut(\Delta)$. Let $\Omega$ be a connected locally $M_\Delta$
552: subgraph of $\Delta$. Define the graph $\Lambda$ with
553: $\V\Lambda=\Omega^G,\ \E\Lambda=\{(\Omega',\Omega'')|\Omega',\Omega''\in
554: \Omega^G,\ \Omega'\cap\Omega''\cong K_{2\times 4}\}$.
555:
556: \begin{lem}
557: The graph $\Lambda$ is a connected graph of valency
558: $k(\Lambda)=320$, on which $G$ acts vertex- and edge-transitively.
559: For each $v\in\V\Lambda$ and $u\in\Lambda_2(v)$ we have $k(\Lambda(v,u))\leq
560: 32$.
561: \label{suz:L6}
562: \end{lem}
563:
564: \begin{pf}
565: The vertex-transitivity is clear. $k(\Lambda)=320$ follows immediately
566: from Proposition~\ref{suz:P2}.
567: Let $v\in\V\Lambda$. Consider the action of the stabilizer
568: $F$ of $\Omega$
569: in $G(v)$ on the set of 80 locally $M_{\Delta(\infty)}$ subgraphs
570: $\Xi_j(v)$ of $\Delta(v)$ such that $\Omega(v)\cap \Xi_j(v)\cong
571: K_{2\times 3}$ (cf. the case $i=4$ of Proposition~\ref{suz:P2}).
572: This action is transitive, since
573: $\Xi_j(v)$ ($j=1,\dots,80$) constitute a suborbit of $\Aut(J_2)$ in its primitive
574: action of degree $315$. By Proposition~\ref{suz:P3}
575: each $\Xi_j(v)$ lifts to a unique
576: locally $M_\Delta$ subgraph $\Xi_j$ of $\Delta$. By Proposition~\ref{suz:P2}
577: we have $\Xi_j \cap \Omega \cong K_{2 \times 4}$. Hence $F$ acts transitively
578: on the $\Xi_j$ ($j=1,\dots,80$). Since the stabilizer of $\Omega$ in $G$
579: acts transitively on $\V\Omega$, the edge-transitivity is proved.
580:
581: The unique imprimitivity system of $G$ on $\V\Lambda$ is given by
582: the equivalence relation $\varphi$ defined in Lemma~\ref{suz:L5}
583: (cf. \cite{Atl}). Hence $\Lambda$ is connected.
584:
585: The last statement follows from the fact that $\Lambda$ appears as
586: the subgraph $\Gamma_2(\infty)$ of the graph $\Gamma\cong 3\Sigma_6$
587: and that $\mu(\Gamma)=32$.
588: \end{pf}
589:
590: Let us denote by $\overline\Omega$ the $\varphi$-equivalence class of
591: $\Omega$. Define the graph $\overline\Lambda$ by
592: $\V\overline\Lambda=\overline\Omega^G$ and
593: $\E\overline\Lambda=\{(\overline\Omega',\overline\Omega'')|
594: \overline\Omega',\overline\Omega''\in \overline\Omega^G,\
595: \Omega'\cap\Omega''\cong K_{2 \times 4}\}$.
596:
597: \begin{lem}
598: \label{suz:L7}
599: For each $(\overline\Omega',\overline\Omega'')$,
600: we have $\overline\Omega'\cap\overline\Omega''\cong
601: K_{2 \times 4}\cup K_{2 \times 4}\cup K_{2 \times 4}$.
602: The graph $\overline\Lambda$ is connected graph of valency
603: $k(\overline\Lambda)=320$.
604: \end{lem}
605:
606: \begin{pf}
607: Let $S$ be the set of $\Xi\in\Omega^G$ satisfying
608: $\Xi\cap\Upsilon\cong\ K_{2 \times 4}$ for some $\Upsilon$
609: $\varphi$-equivalent to $\Omega$. Since $\Omega$ does not contain a
610: subgraph $K_{2 \times 4}\cup K_{2 \times 4}$, there is a unique
611: such subgraph $\Upsilon$ for each $\Xi$ of $S$.
612: Hence the stabilizer $F$ of $\Omega$ in $G$ acts on $S$ transitively.
613: The subdegrees of $G$ on $\overline\Omega^G$ are $1,20,320,1024$.
614: Since $|S|=960,\ S=\bigcup_{\Xi\in S}\overline\Xi$. Hence
615: we have proved the first two claims of the lemma. The last one follows
616: from the primitivity of $G$ on $\overline\Omega^G$. \end{pf}
617:
618: Let us identify $\Delta$ with $\Gamma(\infty)$ for $\infty\in \V\Gamma$.
619: Choose a vertex $v\in \Gamma_2(\infty)$. Denote $\Omega=\Gamma(\infty,v)$.
620: According to the results of Sect.~\ref{suz:sect22},
621: we know the connected components of
622: $\Omega$ up to conjugacy. By Lemma~\ref{suz:L5}
623: we have several possibilities on the number of the
624: components $c(\Omega)$ of $\Omega$.
625:
626: \Case{$c(\Omega)=3$} We have $\Omega=\overline\Theta$
627: for a connected locally $M_\Delta$ subgraph $\Theta$ of $\Delta$. It
628: follows from Lemmas~\ref{suz:L4} and~\ref{suz:L5} that the subgraph
629: $\Xi(v)=\<\V\Gamma(v)\setminus\V\Omega\>$ lies in $\Gamma_2(\infty)$.
630: Clearly $v(\Xi(v))=v(\Delta)-v(\Omega)=320$. Let
631: $u \in \Xi(v)$. Then by Lemma~\ref{suz:L4} $\Gamma(\infty,v,u)
632: \cong K_{2 \times 4}\cup K_{2 \times 4}\cup K_{2 \times 4}$,
633: where the connected components of this subgraph lie in distinct
634: connected components of $\Omega$. Hence
635: $\Gamma(\infty,u)=\overline\Theta '$ for a connected locally $M_\Delta$
636: subgraph $\Theta'$ of $\Delta$. By Lemma~\ref{suz:L7}
637: a connected component $\Xi$ of
638: $\Gamma_2(\infty)$ containing $v$ is isomorphic to
639: $\overline\Lambda$. Counting in two ways the number of edges
640: from $\Gamma(\infty)$ to $\Gamma_2(\infty)$, we obtain that
641: $\Gamma_2(\infty)=\Xi$. Thus $\Gamma\cong\Sigma_6$ and the proof
642: in this case was completed.
643:
644: \Case{$c(\Omega)=1$} Consider $u \in \Gamma(v)\cap \Gamma_2(\infty)$.
645: By Lemma~\ref{suz:L4} $\Gamma(\infty,v,u) \cong K_{2 \times 4}$. Therefore
646: $\Gamma(\infty,u)$ is connected. Look at the connected component $\Upsilon$ of
647: $\Gamma_2(\infty)$ containing $v$. By Lemmas~\ref{suz:L6} and~\ref{suz:L4},
648: we have $k(\Upsilon)\geq 320$ and $\Upsilon$
649: contains $\Lambda$ as a (possibly non-generated) subgraph.
650: Calculating in two ways the number of edges coming from $\Gamma(\infty)$ to
651: $\Gamma_2(\infty)$, we obtain that $\Gamma_2(\infty)=\Upsilon$.
652:
653: Let us prove that $\Upsilon\cong\Lambda$, i.e. if $(u,v)\in\E\Upsilon$
654: then $\Gamma(\infty,u,v)\cong K_{2\times 4}$. Suppose contrary for
655: some $(u,v)\in\E\Upsilon$. We know that $\Gamma(v,u)\cong\Sigma_4$.
656: Let us determine to which layers of $\Gamma$ with respect to $\infty$ the
657: vertices of $\Gamma(v,u)$ do belong.
658: By Lemma~\ref{suz:L4} $\Gamma(\infty,u,v)=\emptyset$. Next, $u$ lies in a uniquely
659: determined locally $M_\Delta$ subgraph $\Omega'$ of $\Gamma(v)$.
660: Namely, $\Omega'\in\overline\Omega$ (cf. Lemma~\ref{suz:L5}).
661: Dually, $v$ lies in such a subgraph of $\Gamma(u)$.
662: Therefore the number of vertices of
663: $\Gamma(v,u)$ belonging to $\Gamma_3(\infty)$ cannot exceed $k(\Omega')=20$.
664: Thus, $\Gamma(v,u)$ has at least $80$ vertices at the distance $2$ from
665: $\infty$. Let $w$ be a vertex of this set. It is easy to check that
666: $\Gamma(\infty,v,w)\cong\Gamma(\infty,u,w)\cong K_{2\times 4}$.
667: One can consider $(v,w,u)$ as a 2-path in $\Lambda$. There are at least
668: $80$ such paths from $v$ to $u$, which contradicts the last part of
669: Lemma~\ref{suz:L6}. We are done.
670:
671: Let $w \in \Gamma_3(\infty)\cap\Gamma(v)$. Denote $\Theta=\Gamma(w)$.
672:
673: \begin{lem}
674: \label{suz:L8}
675: The subgraph $\Xi=\Gamma_2(\infty)\cap\Theta$ equals
676: $\Theta_2(t)$ for some $t\in \V\Theta$.
677: \end{lem}
678:
679: \begin{pf}
680: From the above we know that $w \in \V {\rm A}$ for a connected
681: locally $M_\Delta$ subgraph ${\rm A}\subset\Gamma_3(\infty)\cap\Gamma(v)$.
682: Thus $k(\Xi)=v(\Theta(v))-k({\rm A})=80$. By Proposition~\ref{suz:P3}
683: for any $u\in
684: \V\Xi$ the subgraph $\overline\Xi(u)=\Theta(u)\cap\Gamma_3(\infty)$ equals
685: $\Theta(u,t^u)$ for some $t^u \in \Theta_2(u)$. This faces us to the
686: problem of classification of the subgraphs of $\Theta$ satisfying the
687: local property above.
688:
689: Denote $t=t^v$. It suffices to show that
690: $\Pi^u=\overline\Xi(u)\cap\Theta(u,t)\setminus\overline\Xi(v)$
691: is nonempty for any $u\in\Xi(v)$. Indeed, since the pointwise
692: stabilizer of $v$ and $t$ in $\Aut(\Theta)$ acts transitively on
693: $\Theta(t)\setminus\overline\Xi(v)$, the sets $\Pi^u$ $(u\in\Xi(v))$
694: cover it, i.e. $\Theta(t)\cap\V\Xi=\emptyset$.
695:
696: Pick $u\in\Xi(v)$. Denote $\Pi=\Pi^u$. By Lemma~\ref{suz:L4},
697: $\Phi=\Theta(v,t,u)= \Theta(v)\cap\overline\Xi(u)$
698: is isomorphic to $K_{2\times 3}$.
699: Pick $(x,y)\in\E\Phi$. Consider $\Psi=\Theta(u,x,y)\cong\Sigma_2$.
700: The subgraph $\Phi(x,y)$ of $\Psi$ consists of two nonadjacent vertices
701: $a$ and $b$. We are interesting in ways of completion of $\{a,b\}$ to
702: the subgraph of fixed points of a 2A-involution of $\Aut(\Psi)$ (one
703: such way gives $\Theta(u,t,x,y)$). Since $avb$ is a 2-path in $\Psi$, the
704: involutions fixing both of $a$ and $b$ commute. Therefore if $t'$
705: gives us some $\Theta(u,t',x,y)$-subgraph we are interesting in,
706: $\Theta(u,t,x,y)\cap\Theta(u,t',x,y)$ contains $\{a,b\}$ properly.
707: Since $\Pi\subseteq\Theta(u,t,x,y)\cap\Theta(u,t',x,y)\setminus\{a,b\}$,
708: we are done. \end{pf}
709:
710: Next, let $w' \in \Theta(t)$. By Lemma~\ref{suz:L8}
711: $\Gamma(w')\cap\Gamma_2(\infty)=\Gamma(w')\cap\Gamma_2(t')$ for some
712: $t'\in\Gamma(w')$. Suppose $t\not=t'$. Since $t'\in \V\Theta$,
713: we then have $\Theta(w,t')\cap\Theta_2(t)\not=\emptyset$, a contradiction.
714: Thus $t=t'$.
715:
716: Thus $\Gamma(t)\subseteq\Gamma_3(\infty)$. Therefore $t\in\Gamma_4(\infty)$.
717: By standard calculations, we can verify that $\Gamma$ is distance-regular
718: with the same intersection array as $3\Sigma_6$.
719:
720: Note that we have shown that any nontrivial connected cover of $\Sigma_6$ is
721: distance-regular with the same intersection array as $3\Sigma_6$. Therefore
722: {\em any} (in our sense, see Sect.~\ref{suz:sect1})
723: nontrivial connected cover of $\Sigma_6$ is isomorphic to
724: $3\Sigma_6$. Hence in order to complete the proof, it suffices to show that
725: $\Gamma$ is a cover of $\Sigma_6$.
726:
727: Since $\Gamma_3(\infty)\cong\Sigma_5\cup\Sigma_5$, to be at the
728: distance 4 in $\Gamma$ is an equivalence relation on $\V\Gamma$.
729: Let $t\in\Gamma_4(\infty)$. Then for any $x\in\Gamma(\infty)$ there
730: exists a unique $y\in\Gamma(t)$ such that $x\in\Gamma_4(y)$. Hence
731: our relation is well-defined on $\E\Gamma$. Thus we may define the
732: quotient graph $\overline\Gamma$, whose vertices (resp. edges) are
733: equivalence classes of vertices (resp. edges) of $\Gamma$. Clearly,
734: for any $v\in \V\Gamma$ the
735: restriction of our quotient mapping to $\Gamma(v)$
736: is an isomorphism. Hence $\overline\Gamma$ is locally $\Sigma_5$.
737: Therefore $\overline\Gamma\cong\Sigma_6$.
738: The proof in this case was completed.
739:
740: \Case{$c(\Omega)=2$} It follows from the above that in this case
741: $v(\Gamma(\infty,u))=64$ for any $u \in \Gamma_2(\infty)$. Now the standard
742: calculation of the edges coming from $\Gamma(\infty)$ to $\Gamma_2(\infty)$
743: shows that $v(\Gamma_2(\infty))$ is not integer, a contradiction.
744:
745: The proof of Theorem~\ref{suz:MTH} is complete.
746:
747: \medskip
748: {\bf Acknowledgment.} Several hints about subgraphs of graphs under
749: consideration were found using the computer package COCO \cite{COCO}.
750: The author thanks the designer of COCO Igor~Faradzev for help in the
751: use of it. The author thanks Sergey~Shpectorov who read the first
752: version of the paper and suggested some improvements to it.
753: