Towards the Automorphism Conjecture I: Combinatorial Control

Let G be a permutation group on the set X. The degree of G is the size |X| of the underlying set X. For every x\in X, we define G\cdot x:=\{ \sigma (x):\sigma \in G\} and call it the (G-)orbit of x. We call G transitive iff, for one (and hence for all) x\in X, we have G\cdot x=X.

Let G be a permutation group on the set X. A subset B\subseteq X is called a (G-)block iff, for all \sigma \in G, we have

\sigma [B]=B or \sigma [B]\cap B=\emptyset. A block is called nontrivial iff it is not a singleton and not equal to X. A transitive permutation group G is called primitive iff it has no nontrivial G-blocks.

Let n\in {\mathbb{N}} \setminus \{ 1\}. For n \in \{ 5,6,7,8,9,10,11,12,13,15,16,23,24\}, we set d_n to be equal to the value in the last column of Table ?. For n\in \{ 2,3,4\}, we set d_2 :={1\over 2} = {\lg (2!)\over 2}, d_3 :=0.8617\geq {\lg (3!)\over 3} and d_4 :=1.1463\geq {\lg (4!)\over 4}. We set I :=\{ 2,3,4;5,6,7,8,9,10,11,12,13,15,16,23,24\} and, for n\not\in I, we set d_n :=2^n.

Let G be a permutation group on the set X and let C\subseteq X. We define G\cdot C:=\{ \sigma [C]:\sigma \in G\}.

Let G be a permutation group on the set X and let B\subseteq X be a block. For \sigma \in G, and Q\subseteq G\cdot B, we define the action \sigma \! \Rsh ^{Q} on Q by, for all C\in Q, setting \sigma \! \Rsh ^{Q} (C):=\sigma [C]. Let A be a block that contains B and let G' be a subgroup of G. We define A[G\cdot B]:=\{ B'\in G\cdot B: B'\subseteq A\} and the action of G' on A[G\cdot B] as G'\! \Rsh ^{G\cdot B} _A := \{ \sigma \! \Rsh ^{G\cdot B} |_{A[G\cdot B]} : \sigma \in G', \sigma [A]\subseteq A\}.(Unless the author is mistaken, the G\! \Rsh ^{G\cdot B} _A are sections of G. However, the definition of sections might include other structures and, to the author’s knowledge, does not explicitly refer to the block structures we define here.) When A is equal to the underlying set X, we will omit the subscript.

Let G be a permutation group on the set X, let B\subseteq X be a block and let S\subsetneq B be a block. Then (S,B) is called a primitive pair for G iff G\! \Rsh ^{G\cdot S} _B is primitive. We call {|B|\over |S|} the width of the primitive pair.

Let G be a permutation group on the set X. Two blocks B,B'\subseteq X are called conjugate blocks iff there is a \sigma \in G such that B'=\sigma [B]. For any primitive pair (S,B) and any \sigma \in G, we define \sigma (S,B):=(\sigma [S],\sigma [B]). Two primitive pairs (S,B) and (S',B') are called conjugate primitive pairs iff there is a \sigma \in G such that \sigma (S,B)=(S',B'). Let {\cal S} be a set of blocks, let {\cal P} be a set of primitive pairs, and let G'\subseteq G. Then we define G' \cdot {\cal S} := \{ \sigma [S]:S\in {\cal S}, \sigma \in G'\} and G' \cdot {\cal P} := \{ \sigma (S,B):(S,B)\in {\cal P}, \sigma \in G'\}.

Let G be a transitive permutation group on the set X. A sequence \{ x \} =B_0 \subsetneq B_1 \subsetneq \cdots \subsetneq B_{m} =X is called a primitive nesting for G iff, for all j\in \{ 1, \ldots , m\}, we have that (B_{j-1} , B_j ) is a primitive pair.

Let G be a permutation group on the set X and let (S,B) be a primitive pair for G. Then the primitive pair (S,B) is called a factorial pair for G iff G\! \Rsh ^{G\cdot S} _B contains the alternating group A_{|B[G\cdot S]|} (B[G\cdot S]) acting on the set of blocks B[G\cdot S].

Let U be an ordered set without nontrivial order-autonomous antichains. Let G^* be a subgroup of {\rm Aut} (U). The set of all G^*-orbits will be denoted {\cal D}. We will call (U,{\cal D} ,G^* ) a standard configuration.

For every G^*-orbit D\in {\cal D}, we define \Lambda ^* (D):= \left\{ \Phi |_D: \Phi \in G^* \right\}, where the superscript shall be our indication that we work with the subgroup G^*.

Let (U,{\cal D} , G^* ) be a standard configuration. For every D\in {\cal D}, we can choose a primitive nesting \{ x^D \} =B_0 ^D \subsetneq B_1 ^D \subsetneq \cdots \subsetneq B_{m_D} ^D =D for \Lambda ^* (D). Upon making such a choice for every D\in {\cal D}, we call the set {\cal B} :=\left\{ B_j ^D : D\in {\cal D} , j\in \{ 0,\ldots , m_D \} \right\} a primitive nesting collection for (U,{\cal D},G^* ). We let {\cal F}\subseteq (G^* \cdot {\cal B} )^2 be the set of all factorial pairs (S,B)\in (G^* \cdot {\cal B} )^2 of width w\geq 5.

We call (U,{\cal D} , G^* ; {\cal B}), as well as, when factorial pairs are specifically addressed, (U,{\cal D} , G^* ; {\cal B} ,{\cal F} ), a nested standard configuration.

Let (U,{\cal D} , G^* ; {\cal B} ) be a nested standard configuration. For every C\in G^* \cdot {\cal B}, we define D_C \in {\cal D} to be the unique orbit that contains C. For every C\in {\cal B}, we define the nesting depth of C to be the unique index j such that C=B_j ^{D_C}. Moreover, when the nesting depth of C\in {\cal B} is j<m_{D_C}, we define C^\dag to be the unique block in {\cal B} such that \left( C,C^\dag \right) is a primitive pair, that is, \left( C,C^\dag \right) =\left( B_j ^{D_C} ,B_{j+1} ^{D_C} \right). For every conjugate \widetilde{C} \in \Lambda ^* (D_C ) \cdot C of C, we define the successor \widetilde{C} ^\dag \in \Lambda ^* (D_C ) \cdot C^\dag to be the unique conjugate of C^\dag such that \left( \widetilde{C} , \widetilde{C} ^\dag \right) is a primitive pair.

Let (U,{\cal D} , G^* ) be a standard configuration, let G' be a subgroup of G^*, and let {\cal K} be a set of blocks for G^*. We define

In cases in which {\cal K} can be easily enumerated as {\cal K} =\{ B_1 , \ldots , B_n \}, we will write G' _{\searrow B_1 , \ldots , B_n } instead of G' _{\searrow \{ B_1 , \ldots , B_n \} }.

Let (U,{\cal D} , G^* ; {\cal B}) be a nested standard configuration, let {\cal K} \subseteq {\cal B} and let B\in {\cal B}\setminus {\cal K}. We define

Let (U,{\cal D} , G^* ; {\cal B}) be a nested standard configuration. A vector \overrightarrow{{\cal K}} =(B_1 , \ldots , B_k )\in {\cal B} ^k is a reverse depth enumeration iff the following hold.

1. For all distinct i, j\in \{ 1, \ldots , k\}, we have B_i \not= B_j.

2. For all j\in \{ 1, \ldots , k\}, the set {\cal K} _j :=\{ B_1 , \ldots , B_j \} is upper block closed, that is, for all K\in {\cal K}_j and B\in {\cal B}, we have that K\subseteq B implies B\in {\cal K}_j.

We set {\cal K} :=\{ B_1 , \ldots , B_k \}, and, for every B \in {\cal K}, we define \overleftarrow{\cal K} ( B ):= {\cal K}_{j-1 }, where j is the index of B in the reverse depth enumeration. For simpler notation, we will write \Lambda ^* \left( \overleftarrow{\cal K} \not\searrow B \right) instead of \Lambda ^* \left( \overleftarrow{\cal K} (B)\not\searrow B \right).

We call \left( U,{\cal D} , G^* ; {\cal B}, \overrightarrow{{\cal K}} \right) a partially enumerated nested standard configuration. In case {\cal K}={\cal B}, we will call \left( U,{\cal D} , G^* ; \overrightarrow{\cal B}\right), as well as \left( U,{\cal D} , G^* ; \overrightarrow{\cal B}, {\cal F}\right), an enumerated nested standard configuration.

Let \left( U,{\cal D} , G^* ; {\cal B}, \overrightarrow{{\cal K}} \right) be a partially enumerated nested standard configuration. We call the function \gamma _{\cal K} ^* :{\cal K} \to [1, \infty ) a control function iff the following hold.

1. \displaystyle{ \prod _{C\in {\cal K} } \left| \Lambda ^* \left( \overleftarrow{\cal K} \not\searrow C \right) \right| \leq \prod _{C\in {\cal K} } \gamma _{\cal K} ^* (C) .}

2. For all C\in {\cal K} of depth at least 1 such that \left( C, C^\dag \right) is of width w, but not factorial of width w\geq 5, we have \gamma _{\cal K} ^* (C) \leq \left| \Lambda ^* \left( \overleftarrow{\cal K} \not\searrow C \right) \right| \leq 2^{d_w |\Lambda ^* (D_C )\cdot C|}.

We also define {\cal K}_f to be the set of all S\in {\cal K} such that \left( S, S^\dag \right) is factorial of width w\geq 5, we define {\cal K}_0 to be the set of all S\in {\cal K} that have depth 0, and we define the set of (\gamma _{\cal K} ^*-)uncontrolled factors in {\cal K} to be

Let \left( U,{\cal D} , G^* ; {\cal B}, \overrightarrow{{\cal K}} \right) be a partially enumerated nested standard configuration, and let \gamma _{\cal K} ^* be a control function. For every D\in {\cal D}, we define {\cal K} ^D := {\cal K} \cap \left\{ B_0 ^D , B_1 ^D , \ldots , B_{m_D -1} ^D \right\}, {\cal K} ^{\gamma ,D} _u := {\cal K} _u ^\gamma \cap \left\{ B_0 ^D , B_1 ^D , \ldots , B_{m_D -1} ^D \right\},

Let (U,{\cal D} , G^* ) be a standard configuration, let D,E\in {\cal D}, let S be a block of \Lambda ^* (D), and let C be a block of \Lambda ^* (E).

1. We say that C is (nontrivially) woven with S and write C\curlyvee S iff there are elements c\in C and s\in S that are comparable, as well as elements c'\in C and s'\in S that are incomparable.

2. We say that S is made by C and write C\diamondsuit S iff C>S or C<S.

3. We say that C is incomparable to (every element in) S and write C\| S iff no element c\in C is comparable to any element s\in S.

In case S\diamondsuit C or S\| C, we will also say that C is (trivially) woven with S.

Let (U,{\cal D} , G^* ) be a standard configuration, let D,E\in {\cal D}, let S_1 ,S_2 be blocks of \Lambda ^* (D), and let C_1 , C_2 be blocks of \Lambda ^* (E). Then we say that (C_1 , S_1 ) and (C_2 ,S_2 ) are similarly woven and write (C_1 , S_1 )\bowtie (C_2 , S_2 ) iff there is a \Phi \in G^* such that \Phi [C_1 ]=C_2 and \Phi [S_1 ]=S_2.

Let (U,{\cal D} , G^* ) be a standard configuration, let D,E\in {\cal D}, let S be a block of \Lambda ^* (D), and let C be a block of \Lambda ^* (E). Let {\cal S}_1 , \ldots , {\cal S} _p be the equivalence classes of \bowtie on \Lambda ^* (D)\cdot S\times \Lambda ^* (E)\cdot C that contain nontrivially woven pairs C'\curlyvee S'. Then, for all j\in \{ 1, \ldots , p\} and all (C',S')\in {\cal S} _j, we will write C'\curlyvee _j S'. These relations technically are not symmetric, and they, as well as their number p, depend on the containing orbits, as well as the block systems involved. Because the direction, the containing orbits and the block systems will be clear from the context, similar to \diamondsuit and \|, they will not be indicated in the relation symbol \curlyvee _j or in the number p.

From now on, p will denote the number of \bowtie-equivalence classes that contain similarly nontrivially woven pairs C'\curlyvee S', and \curlyvee _1 , \ldots , \curlyvee _p will denote the corresponding weaving relations.

Let w,m\in {\mathbb{N}} and let S_1, \ldots , S_w be pairwise disjoint nonempty sets. We define (P_1 , \ldots , P_m ) to be a sorted partition of \{ S_1 , \ldots , S_w \} iff every P_j is a (not necessarily nonempty) subset of \{ S_1 , \ldots , S_w \} and the set \{ P_1 , \ldots , P_m \} \setminus \{ \emptyset \} is a partition of \{ S_1 , \ldots , S_w \}.

Let (U,{\cal D} , G^* ) be a standard configuration, let D,E\in {\cal D}, let C\subsetneq E be a \Lambda ^* (E)-block, let Z\subseteq \Lambda ^* (E)\cdot C, and let S be a block of \Lambda ^* (D). The canonical S-partition of Z is the sorted partition \left( S_\diamondsuit (Z), S_\| (Z), S_{\curlyvee _1 } (Z), \ldots , S_{\curlyvee _p } (Z) \right) of Z with the following sets.

1. The set S_\diamondsuit (Z):=\{ A\in Z: S\diamondsuit A\} of blocks in Z that are made by S.

2. The set S_\| (Z):=\{ A\in Z: S\| A\} of blocks in Z that are incomparable with S.

3. The sets S_{\curlyvee _j } (Z):=\{ A\in Z:S\curlyvee _j A\} of blocks in Z that are j-woven with S.

Let (U,{\cal D} , G^* ) be a standard configuration, let D,E\in {\cal D}, let C be a block of \Lambda ^* (E), let Z\subseteq \Lambda ^* (E)\cdot C, and let S be a block of \Lambda ^* (D). For any S'\in \Lambda ^* (D )\cdot S, we define the Z-home block of S', denoted H_{Z} (S'), to be the set of all S''\in \Lambda ^* (D )\cdot S such that, for all \bigstar \in \{ \diamondsuit , \| , \curlyvee _1 , \ldots , \curlyvee _p \}, we have S' _\bigstar (Z)=S'' _\bigstar (Z), that is, the canonical S'-partition of Z equals the canonical S''-partition of Z.

Let (U,{\cal D} , G^* ) be a standard configuration, let D,E\in {\cal D}, let (S,B) be a primitive pair for \Lambda ^* (D), and let C be a block of \Lambda ^* (E). Then we define C \blacktriangleright ^B S and say that C (together with B) controls S iff the C-canonical partition of B[\Lambda ^* (D)\cdot S] has more than one nonempty set.

Let (U,{\cal D} , G^* ) be a standard configuration, let D,E\in {\cal D}, let (S,B) be a primitive pair for \Lambda ^* (D), and let (C,T) be a primitive pair for \Lambda ^* (E). We say that (S,T) and (C,T) are mutually controlling, and we write (S,T)\matrix{\blacktriangleright \vspace*{-.1in} \cr \blacktriangleleft \cr } (C,T), iff C \blacktriangleright ^B S and S \blacktriangleright ^T C.

Let (U,{\cal D} , G^* ;{\cal B}) be a nested standard configuration and let C,S\in G^* \cdot {\cal B}. We will write C \blacktriangleright S instead of C\blacktriangleright ^{S^\dag } S, and we will write C\matrix{\blacktriangleright \vspace*{-.1in} \cr \blacktriangleleft \cr } S, iff C \blacktriangleright S and S \blacktriangleright C. Moreover, we will write C\vartriangleright S iff there is a conjugate C' of C such that C'\blacktriangleright S. We define C\matrix{\vartriangleright \vspace*{-.1in} \cr \vartriangleleft \cr } S iff there is a conjugate C' of C such that C'\matrix{\blacktriangleright \vspace*{-.1in} \cr \blacktriangleleft \cr } S.

Let (U,{\cal D} , G^* ;{\cal B} ) be a nested standard configuration and let {\cal B} _0 := \left\{ B_0 ^D : D\in {\cal D} \right\} \subset {\cal B}. We define \vartriangleright _0 := \vartriangleright |_{{\cal B} _0 \times {\cal B} _0 } to be the reduction of the natural control relation to {\cal B} _0 \times {\cal B} _0 and call it the depth 0 control relation. The transitive closure of \vartriangleright _0 is denoted \vartriangleright \hspace*{-.08in} \vartriangleright _0. For every B_0 ^D \in {\cal B} _0, we define \left[ B_0 ^D \vartriangleright \hspace*{-.08in} \vartriangleright _0 \right] :=\{ S\in {\cal B} _0 :B_0 ^D \vartriangleright \hspace*{-.08in} \vartriangleright _0 S\} to be the set of all S\in {\cal B} _0 that are (directly or indirectly) depth 0 controlled by B_0 ^D.

Let (U,{\cal D} , G^* ;{\cal B} ) be a nested standard configuration and let B_0 ^{F } be a depth 0 block. Then we will say that B_0 ^{F } is controlled by deeper nestings iff there are a B_\ell ^{E} \in {\cal B} such that \ell \geq 1 and a B_0 ^{D } \in {\cal B} _0 such that B_\ell ^{E} \vartriangleright B_0 ^{D} \vartriangleright \hspace*{-.08in} \vartriangleright _0 B_0 ^{F}.

Let (U,{\cal D} , G^* ) be a standard configuration and let D,E\in {\cal D}. We say that the action of G^* on E controls the action of G^* on D iff, for all \Phi \in G^*, if \Phi |_E ={\rm id} |_E, then \Phi |_D ={\rm id} |_D.

For an orbit E\in {\cal D}, we define the controlled tail of E to be the largest set of orbits \{ D_0 , \ldots , D_m\} \in {\cal D} such that E=D_0, and, for all i\in \{ 1, \ldots , m\}, there is a j\in \{ 0,\ldots , i-1\} such that D_{j} controls D_i.

Let \left( U,{\cal D} , G^* ; {\cal B}\right) be a nested standard configuration and let E\in {\cal D}. Then we say that E is cooperative iff \left( B_0 ^{E } , B_1 ^{E } \right) is not a factorial pair of width w\geq 5, and there is a D\in {\cal D} such that |E|\leq |D| and such that the action of G^* on E controls the action of G^* on D. If the action of G^* on D controls the action of G^* on E, too, we call (E,D) a cooperating pair.

Let \left( U,{\cal D} , G^* ; {\cal B}\right) be a nested standard configuration. We define {\cal B} _0 ^r to be the set of all S\in {\cal B} _0 that are neither controlled by deeper nestings, nor in the union of the \vartriangleright \hspace*{-.08in} \vartriangleright _0-closures of the depth 0 blocks of the orbits in the controlled tail of a cooperating pair.

Let \left( U,{\cal D} , G^* ; {\cal B}\right) be a nested standard configuration. We define {\cal R} to be the union of all \matrix{\vartriangleright \vspace*{-.1in} \cr \vartriangleleft \cr }-components W of \left( {\cal B} _0 ^r , \matrix{\vartriangleright \vspace*{-.1in} \cr \vartriangleleft \cr } \right) such that, for any two S,C\in W such that C\matrix{\vartriangleright \vspace*{-.1in} \cr \vartriangleleft \cr } S, the following hold.

1. At least one of \left( S, S^\dag \right) and \left( C, C^\dag \right) is factorial.

2. If \left( C, C^\dag \right) is not factorial, then \left| D_C \right| > \left| D_S \right|.

Finally, we define {\cal G} := {\cal B} _0 ^r \setminus {\cal R}.

Let \left( U,{\cal D} , G^* ; {\cal B}, \overrightarrow{{\cal K}} \right) be a partially enumerated nested standard configuration. Then \overrightarrow{\cal K} is called opportunistic with control function \gamma _{\cal K} ^* iff the following hold.

1. For every cooperating pair (E,D), all blocks of E\cup D in {\cal B} are contained in {\cal K}.

2. For all S\in {\cal K} _u ^\gamma \cap {\cal K} _f, if S^\dag is the component B_j of \overrightarrow{\cal K}, then S is the immediately following component B_{j+1}.

Let (U,{\cal D} , G^* ) be a standard configuration, let D,E\in {\cal D}, let w\geq 5, let (S ,B ) be a primitive pair of width w for \Lambda ^* (D), and let (C,T) be a primitive pair of width w for \Lambda ^* (E). We say that (C,T) and (S,B) are max-locked and write (S,B) \blacksquare (C,T) iff T[\Lambda ^* (E)\cdot C]=\{ C_1 , \ldots , C_w \}, B[\Lambda ^* (D)\cdot S]=\{ S_1 , \ldots , S_w \} and there are \bigstar ,\blacktriangledown \in \{ \diamondsuit , \| , \curlyvee _1 , \ldots , \curlyvee _p \} such that, for every j\in \{ 1,\ldots , w\}, we have S_j \bigstar C_j, and, for every i\in \{ 1,\ldots , w\} \setminus \{ j\}, we have S_i \blacktriangledown C_j. Moreover, we define \Theta _{B,T} : B[\Lambda ^* (D)\cdot S] \to T[\Lambda ^* (E)\cdot C] by \Theta _{B,T} (S_j ):=C_j and we define \Theta _{T,B} : T[\Lambda ^* (E)\cdot C] \to B[\Lambda ^* (D)\cdot S] by \Theta _{T,B} (C_j ):= S_j.

Let (U,{\cal D} , G^* ;{\cal B} , {\cal F}) be a nested standard configuration. The graph ({\cal F},\blacksquare ) is called the ({\cal B}-)factorial max-lock graph. A \blacksquare-connected subset {\cal S}\subseteq {\cal F} is called synchronized iff we have that, for all max-lock cycles (S_1 , B_1 )\blacksquare (S_2 , B_2 )\blacksquare \cdots \blacksquare (S_n , B_n )\blacksquare (S_1 , B_1 ) with repeated vertices allowed, we have that \Theta _{B_{n} , B_1 } \circ \Theta _{B_{n-1} , B_{n} } \circ \cdots \circ \Theta _{B_{1} , B_2 } ={\rm id} \! \Rsh ^ {\Lambda ^* (D_{S_1 } )\cdot S_1 } _{B_1}. A \blacksquare-connected subset {\cal U}\subseteq {\cal F} that is not synchronized is called unsynchronized.

Let (U,{\cal D} , G^* ;{\cal B} , {\cal F}) be a nested standard configuration. A subset {\cal P}\subseteq {\cal F} is called saturated iff it is a component of G^* \cdot {\cal P} with adjacency relation \blacksquare. Two saturated subsets {\cal L} ,{\cal P} of ({\cal F},\blacksquare ) will be called conjugate and we will write {\cal L}\asymp {\cal P}, iff {\cal L}\subseteq G^* \cdot {\cal P}.

Let (U,{\cal D} , G^* ;{\cal B} , {\cal F}) be a nested standard configuration. Let the saturated subset {\cal S}\subseteq {\cal F} be synchronized. For any two (C,T), (S, B )\in {\cal S}, we define \Theta _{T,B} :=\Theta _{B_{m-1} , B_m } \circ \cdots \circ \Theta _{B_{0} , B_1 }, where (C,T)=(S_0 , B_0 )\blacksquare \cdots \blacksquare (S_m , B_m )= (S , B ) is a path from (C,T) to (S,B).

Let (U,{\cal D} , G^* ;{\cal B} , {\cal F}) be a nested standard configuration. For any subset {\cal Q} \subseteq {\cal F}, the lower level of {\cal Q} is L({\cal Q} ):=\left\{ S: (S,B)\in {\cal Q} \right\}, and the upper level of {\cal Q} is U({\cal Q} ):=\left\{ B: (S,B)\in {\cal Q} \right\}. For any \blacksquare-connected {\cal C} \subseteq {\cal F} the vertex width of {\cal C} is the width of any factorial pair (S,B)\in {\cal C}.

Let \left( U,{\cal D} , G^* ; \overrightarrow{\cal B}, {\cal F}\right) be an opportunistically enumerated nested standard configuration. Let {\cal P} be a saturated subset of ({\cal F},\blacksquare ) of vertex width w. We say that {\cal P} is \overrightarrow{\cal B}-outsized iff there is an n\in {\mathbb{N}} such that |U({\cal P})|\geq n\lg(w) and such that \prod _{S\in {\cal B} \cap G^* \cdot L({\cal P}) } \left| \Lambda ^* \left( \overleftarrow{\cal B} \not\searrow S \right) \right| \leq (w!)^{n|\{ {\cal L}:{\cal L}\asymp {\cal P} \} |}. Otherwise, {\cal P} is called \overrightarrow{\cal B}-undersized.

Let (U,{\cal D} , G^* ;{\cal B},{\cal F} ) be a nested standard configuration and let {\cal P} be a component of ({\cal F},\blacksquare ). We call {\cal P} thin iff L({\cal P}) consists of singletons.