Towards Classifying Bootstrap Percolation on Cayley Graphs
Author: Bart Marinissen BSc
Supervisor: prof. dr. Gerard R. Renardel de Lavalette Supervisor: Daniel Rodrigues Valesin, PhD
Abstract
π-threshold bootstrap percolation is a very simple model of infection processes. We study how this model behaves on Cayley-graphs of amenable graphs. This is guided by the conjecture that on these graphs, either complete infection is guaranteed or impossible depending on the threshold π. To this end, we prove this conjecture holds for abelian groups of rank 2. We also find at which π complete infection is guaranteed. These results are essentially an extension of methods used in [Sch92] to prove the conjecture for Cayley-graphs on Z
πusing canonical generators. Finally, we outline an approach that might extend these results to all Cayley graphs of abelian groups. Moreover, we test the conjecture experimentally on Cayley-graphs of Heisenberg and Lamplighter groups. Here we find evidence that the conjecture holds for the Heisenberg groups. The results for the lamplighter groups are inconclusive. However, we do find a very interesting phenomenon with the lamplighter group. The number of final infected nodes in these graphs seems to cluster around tenths of the total number of nodes in the graph.
Sunday 25
thMarch, 2018
Contents
1 Introduction 2
1.1 Aim of this thesis . . . 4
2 Theoretical Work 6 2.1 First theoretical steps . . . 6
2.1.1 Basics of discrete geometry . . . 7
2.1.2 Using the abelian structure theorem . . . 10
2.2 Free abelian case . . . 12
2.2.1 Facet growth . . . 16
2.2.2 An upper bound on ππ for π½π . . . 22
2.3 The seed argument in Z2 . . . 23
2.4 Renormalization . . . 29
2.5 Conclusion for theoretical work . . . 37
2.5.1 Extending result to arbitrary abelian groups . . . 38
3 Experimental Work 41 3.1 Computing bootstrap percolation . . . 42
3.1.1 Implementation . . . 46
3.1.2 Potential distributed algorithm . . . 47
3.2 Tested graphs . . . 48
3.2.1 The Heisenberg group . . . 48
3.2.2 Lamplighter groups . . . 49
3.3 Results . . . 50
3.3.1 Abelian 4D . . . 51
3.3.2 Heisenberg . . . 52
3.3.3 Lamplighter . . . 54
3.4 Performance analysis . . . 58
3.4.1 Profiling results . . . 59
3.4.2 Comparison with an earlier algorithm . . . 59
3.5 Conclusion for experimental work . . . 60
3.5.1 Further work . . . 61
4 Conclusion 62 4.1 Conclusion . . . 62
Acknowledgments . . . 63
Bibliography 64
A C++ code 65
Introduction
Take a locally finite graph πΊ = (π, πΈ). That is, a graph where every node has finitely many neighbors. A vertex in this graph can either be occupied (value 1) or vacant (value 0). We then consider the following increasing discrete time dynamics. At time step π‘, a vacant vertex with more than π occupied neighbors becomes active in step π‘ + 1. This process is called π-threshold bootstrap percolation. Formally, we take π΄ to be the set of occupied points at a given time step. We then define a single step by the following function:
π½π(π΄) = π΄ βͺ {π₯ β π | #πΈ β© ({π₯} Γ π΄) β₯ π} (1.0.1) We are then interested in the behaviour of this system as we let time go to infinity. That is, we are interested in π½πβ(π΄) = βοΈβπ=1π½ππ.
This model was first studied in [CLR79] as a low temperature approximation to the Ising model, which models magnetic spins in a solid. In general, bootstrap percolation is a simple model for any infectious process. For example, the spread of disease through a population, the firing of neurons in a brain, the spread of opinions in a social network, or the seeping of water through coffee grounds against pressure.
We will study what happens when the process is started from a random uniform configuration of occupied vertices π΄0. That is, we pick some density π β [0, 1]. Then, for any vertex π£ β π we have P(π£ β π΄0) = π and this is independent of all other vertices. We say an initial condition percolates if π½πβ(π΄0) = π . In [CLR79] the studied graphs were the 2π-ary trees. Not much later, Van Enter [Ent87] studied this process on the 2D grid Z2 with π = 2. He found that for any positive density, percolation occurs with probability 1. This result sparked our interest, eventually leading to this thesis.
Besides π-threshold bootstrap percolation, there are other processes also referred to as bootstrap percola- tion. These are all increasing processes where vacant vertices become occupied as an increasing function of their neighborhood. We say one percolation process dominates another whenever the growth function of the former dominates the growth function of the latter. If a process π΄ dominates another process π΅ and π΅ percolates, then this guarantees that π΄ also percolates. This will mostly be used by defining a new form of percolation, proving results for the new form, and then using domination to apply those results to π-threshold bootstrap percolation.
Both kinds of graphs we mentioned before (the grids Zπ and the 2π-ary tree) are examples of Cayley graphs. In this paper, we will study bootstrap percolation on Cayley graphs of amenable groups. Before we explain what an amenable group is, we first explain Cayley graphs. They are defined as follows:
Definition 1.0.1 (Cayley graph). Given a group πΊ with operation β and a finite set π β πΊ (called the generating set) we define the Cayley graph on πΊ using generating set π as Ξ(πΊ, π) = (π, πΈ) taking:
π = πΊ
πΈ= {(π₯, π₯ β π ) | π β π, π₯ β πΊ}
One can take a Cayley graph to be directed or undirected. We say a generating set π is a symmetric set if it is closed under inversion. That is if:
π₯ β π =β π₯β1 β π
CHAPTER 1. INTRODUCTION
If the generating set of a Cayley graph is symmetric, the directed and undirected graphs are essentially the same.
Cayley graphs are vertex transitive (defined below). This property together with their constructive nature really helps when studying these graphs. This is part of why we chose to study Cayley graphs.
Definition 1.0.2( Vertex Transitive graphs ). Given a graph πΊ = (π, πΈ) we say πΊ is vertex transitive if π΄π’π‘(πΊ) acts transitively on π . That is, given any π₯, π¦ β π there exists an πΌ β π΄π’π‘(πΊ) such that πΌ(π₯) = π¦. Or equivalently, the image of any vertex π₯ under π΄π’π‘(πΊ) is all nodes π .
Here, π΄π’π‘(πΊ) is the group of all graph automorphisms of πΊ. That is:
π΄π’π‘(πΊ) = {πΉ : π β π | (π₯, π¦) β πΈ ββ (πΉ (π₯), πΉ (π¦)) β πΈ}
Lemma 1.0.3. All Cayley graphs are vertex transitive
Proof. Given a Cayley graph Ξ(πΊ, π) we have the following group of automorphisms:
{π₯ β¦β ππ₯ | π β πΊ}
One can see these are automorphisms because given any π β πΊ an edge (π₯, π₯π ) β πΈ with π β π has image:
(π₯, π₯π ) β¦β (ππ₯, ππ₯π )
This pair (ππ₯, ππ₯π ) is again an edge since π β π. Now, given an arbitrary pair π, π β πΊ we have the
automorphism π₯ β¦β ππβ1π₯such that π β¦β π.
So given a graph πΊ and a bootstrap percolation process π½ : 2π β2π we are interested in the behavior of π½ when applied iteratively to some initial condition π΄0 with density π. To this end, we introduce some variables that describe this behavior. Whenever we refer to the variables, it will be clear from context which graph πΊ and bootstrap process π½ are meant. Here, we presume that π½ is an increasing and extensive function.
β’ π , the stopping time. This is a random variable, defined as:
π = inf{π‘ β₯ 0 | π β π½π‘(π΄0)} (1.0.2) Where π is the group identity.
β’ ππ, the critical point for π :
ππ= inf{π β [0, 1] | Pπ(π < β) = 1} (1.0.3)
β’ πΎ(π), the exponential growth rate:
πΎ(π) = sup{πΎ β₯ 0 | βπΆ < β βπ‘ : Pπ(π > π‘) β€ πΆπβπΎπ‘} (1.0.4)
β’ ππ, the critical point for exponential growth:
ππ= inf{π β [0, 1] | πΎ(π) > 0} (1.0.5) Now, there are some simple relations between these variables: πΎ(π) is increasing in π; the expectation of π is decreasing in π; if πΎ(π) > 0 then Pπ(π < β) = 1 and thus ππβ₯ ππ. We say percolation is exponential when πΎ(π) > 0.
Throughout this paper we will often use sequences. For these, we introduce some notation. By [π₯π] we mean a sequence π₯0, π₯1. . .. Such a sequence can be either finite or infinite. The difference will be clear from context. Moreover, given a set π΄, we define [π΄] as the set of sequences with elements in π΄. For example, we would denote the sequence of all sums of squares as:
[π₯π] β [N] : π₯0= 0 π₯π+1= π₯π+ (π + 1)2
We also define the idea of a subset of nodes being internally spanned. Intuitively, this means we take the set of nodes as our full graph, and ask whether the set gets filled. Formally we have:
Definition 1.0.4 (Internally spanned sets). Take a bootstrap process π½ : 2π β 2π and an initial condition π΄0β π.
We then say a set π β π is internally spanned by π΄0 if π β© π΄0 percolates using π½ on the graph (π, πΈ β© (π Γ π)).
1.1 Aim of this thesis
As stated earlier, we mean to study Cayley graphs of amenable groups. An amenable group is defined as follows:
Definition 1.1.1 (Amenable group). A discrete group πΊ is amenable if there exists a FΓΈlner sequence for πΊ.
That is, we have a sequence of finite subsets [πΉπ] β [οΈ2πΊ]οΈsuch that:
βπ β πΊ βπ βπ β₯ π: π β πΉπ πββlim
#(ππΉπM πΉπ)
#πΉπ
= 0 Where M is the symmetric difference operator.
There are many other equivalent definitions as can be seen in [BPP06]. Informally, one could consider amenable groups as those that do not βgrow too quicklyβ.
We are interested in amenable groups due to the following conjecture from [BPP06]:
Conjecture 1.1.2. A group πΊ is amenable if and only if, for every finite symmetric π β πΊ and π β N we have ππβ {0, 1} on Ξ(πΊ, π) for π-threshold bootstrap percolation.
We will seek to make progress towards this conjecture in two ways. The first is a theoretical approach, contained in Chapter 2. The second approach is experimental, covered in Chapter 3.
In the theoretical approach we will look at abelian groups, these are guaranteed to be amenable. There, we will work towards the following conjecture:
Conjecture 1.1.3. Let π» be a finitely generated abelian group π» and π β π» be a symmetric set with 0 /β π. Then set ππ equal to half the number of non-periodic elements in π. That is, we set:
ππ= #{π β π | βπ β N+: π Γ π ΜΈ= 0}
2 Then for π½π on Ξ(π», π) we have:
ππ= ππ={οΈ0 if π β€ ππ
1 if π > ππ
The main result of this paper is Theorem 2.5.1 which states the above conjecture holds when π» has rank 2 (the rank of an abelian group is defined in Theorem 2.1.8). We also have Theorem 2.2.12 which states the above conjecture holds for arbitrary abelian groups in the case π > ππ. In Section 2.5.1 we outline an idea for extending the methods of the proof of Theorem 2.5.1 to work for all finitely generated abelian groups. This would then prove Conjecture 1.1.3.
Most important to our theoretical work is [Sch92]. That paper proves a sub case of Conjecture 1.1.3.
Specifically, it proves the case where π» = Zπ and π = {π1. . . ππ}where [ππ] are the canonical generators for Zπ. Our method uses the proof form that paper as a template. That paper was preceded by [AL88]
and dealt with finite subsets (hypercubes) of Zπ rather than looking at the full space Zπ.
Important to the conjecture itself is the paper in which it was posed: [BPP06]. That paper looks explicitly at non-amenable groups. For these groups, it seems that ππ takes intermediate values for π-threshold bootstrap percolation regardless of π. This contrasts sharply with the results known for specific amenable groups. Moreover, their results depend explicitly on groups being non-amenable.
1.1. AIM OF THIS THESIS CHAPTER 1. INTRODUCTION
The second approach is experimental. Here we intend to compute π½πon finite subgraphs of Cayley graphs of amenable groups. Specifically, we will be looking at the Heisenberg and lamplighter groups. Note that for any finite subgraph, we have ππ = 1 since the probability that π΄0 = β is technically positive.
However, we hope to see (and do see) some form of critical behavior in these subgraphs. That is, at certain densities we see almost no growth from the initial condition. Then, for densities higher than some
βcritical densityβ, we see growth continues until almost the entire graph is occupied. Next, we study how this βcritical densityβ changes when we take ever larger subgraphs. If this βcritical densityβ converges to 0 that suggests that ππ= 0 for the full infinite graph.
Experiments like these were run previously for Z2. In these experiments they took squares and increased their side length. They ran simulations up to 350 Γ 350 squares. The critical point then seemed to converge to some small but positive value. Later, in [Hol03] we got theoretical results showing this is actually not the case. This was an expansion of the work done in [AL88]. The results show that the critical point converges to 0 as the side length πΏ goes to infinity. However, convergence is very slow. In the case of Z2 it is π(1/log πΏ) and in the case of Z3 it converges at a rate of π(1/log log πΏ). In fact, from [Bal+12] we know that in Zπ the critical point converges to 0 with order:
π (οΈ 1
logπβ1πΏ )οΈ
(1.1.1)
where log1 = log and logπ = log logπβ1. That is, logπ is log iterated π times. As such, finding results congruent with such convergence would suggest we have similar convergence.
For the Heisenberg group, our experiments found evidence for π(οΈ
1
log3πΏ)οΈ convergence. For the lamp- lighter group, our experiments were inconclusive. We did find a rather interesting phenomenon with the lamplighter group. On the binary lamplighter group, the size of the stable sets π½πβ(π΄0) seems to cluster around tenths of the full subgraph.
Theoretical Work
2.1 First theoretical steps
In the section, we will lie the foundation for the remainder of this chapter. First, we define the π-fort and introduce the 0-1 law. After that, we define some basic concepts of discrete geometry and derive some propositions regarding these concepts; this will be instrumental in Section 2.4. Finally, we make use of the abelian structure theorem (2.1.8). This yields Theorem 2.1.13, which allows us to focus solely on free-abelian groups.
A very useful tool in analyzing π-threshold bootstrap percolation is the π-fort, introduced by [BPP06]:
Definition 2.1.1 (π-fort). Given a graph πΊ = (π, πΈ), we call a subset π΄ β π a π-fort if it is connected and each vertex in π΄ has fewer than π connections outside of π΄. Formally:
π΄ is connected and:
βπ β π΄: #(οΈπΈ β© ({π} Γ (π β π΄)))οΈ < π
Note that other papers often use β€ π instead of < π in the above definition. These are useful due to the following lemma
Lemma 2.1.2. π-threshold bootstrap percolation fills the entire graph if and only if no π-fort is entirely vacant in the initial condition.
Proof. Take π΄0 to be an initial condition with density π > 0 and take π to be the set of vertices of our Cayley graph.
Now suppose that π΄β = π½πβ(π΄0) ΜΈ= π . Then we can find a connected component in π΄βπΆ. This connected component must be a π-fort, as any vertex with π or more connections to the outside would not remain vacant. Now, since bootstrap percolation is an increasing process, this vacant π-fort must have been vacant in π΄0.
What remains to show is that percolation implies that all π-forts were entirely vacant. We do this via contraposition. So, suppose we start with a vacant π-fort. Then, by definition of π½π, that π-fort will
remain vacant.
Now consider what happens when we have a finite π-fort. If we have a density π < 1 then this finite π-fort has positive probability of being vacant. Thus, the probability of percolation is less then 1. The converse need not be true. If all π-forts are infinite, the probability that a given π-fort is vacant is 0, but if there are uncountably many π-forts, the probability that we could find a vacant one might still be positive.
We also have the following theorem, called the 0-1 law. This theorem follows from: [LP16, Prop. 7.3];
Cayley graphs being vertex transitive; the probability of complete percolation being invariant under the automorphisms of our Cayley graphs; and the event of complete occupation being translation invariant.
2.1. FIRST THEORETICAL STEPS CHAPTER 2. THEORETICAL WORK
Theorem 2.1.3 (0-1 Law). For any infinite Cayley graph π-threshold bootstrap percolation we have Pπ(Complete occupation) β {0, 1}
We can use this theorem to deduce the following result about the existence of finite π-forts.
Lemma 2.1.4. If a Cayley graph Ξ generated by a finite π has finite π-forts then, for π-threshold bootstrap percolation, ππ = 1. Otherwise, it has ππβ€1 β π where π > 0 is the critical probability for site percolation on Ξ.
Proof. These results are due to [BPP06].
First, we consider the case where finite π-forts exist. In this case, unless the initial density π = 1, the probability that a given finite π-fort is vacant is positive. Therefore, the probability of complete occupation is lower than 1. By the 0-1 law, it must then be 0.
For the case of no finite π-forts existing, if we have no complete occupation, there must exist a vacant π-fort. Since this π-fort is connected and infinite, this can only occur when the density of vacant sites is at least π. This follows directly from the definition of π. Moreover, as π is finite, π > 0. Having looked at π-forts, we now consider the effect of growing π.
Lemma 2.1.5. Let π» be a finitely generated abelian group, and let π, π β π» be finite symmetric subsets of π».
Now, if π β π it follows that π½π on Ξ(π», π) dominates π½π on Ξ(π», π ).
Proof. Let π½π,π be π½π on Ξ(π», π) and let π½π,π be π½π on Ξ(π», π ). Both π½π,π and π½π,π are of the type:
2πΊ β2πΊ.
We need to show that π½π,π(π΄) β π½π,π(π΄) for any π΄ β πΊ. Now, let πΈπ be all edges of Ξ(π», π) and πΈπ
be all edges of Ξ(π», π ). It follows from π β π that πΈπ β πΈπ. Thus, given any π₯ β π΄ we have:
πΈπβ©({π₯} Γ π΄) β πΈπ β©({π₯} Γ π΄)
We will use this lemma to ignore periodic generators in π. Specifically, to prove Conjecture 1.1.3 in the case π β€ ππ it now suffices to prove the conjecture holds when π contains no periodic elements. In this case we have ππ =#π2 .
2.1.1 Basics of discrete geometry
Our proofs will often concern convex polytopes and other discrete geometry. Here, we introduce the discrete geometry we need for our proves.
We start with the most basic operations. Given a set π΄ β Rπ a point π₯ β Rπ and a scalar πΌ β R we define the following operations:
π₯+ π΄ = {π₯ + π | π β π΄} = π΄ + π₯ πΌπ΄= {πΌπ | π β π΄}
βπ΄= β1π΄ π΄ β π₯= π΄ + (βπ₯) π₯ β π΄= π₯ + (βπ΄)
The main subject of this section is the convex polytope. Convex polytopes have many different descrip- tions. We will use the following three:
β’ A set of the form
{π₯ β Rπ| π π₯ < π}
for some π β RπΓπ and π β Rπ. Here, comparison of vectors is component-wise and we require all components to satisfy the inequality.
β’ A finite intersection of shifted half-spaces
βοΈ
π
π»(π’π) + ππ
Here, we define a half-space as:
π»(π’) = {π₯ β Rπ| π₯ Β· π’ β€0}
where π’ ΜΈ= 0. We do not define π»(0) as that would not be a half-space. This later saves us from needing to exclude the special case π’ = 0.
For the shifts we take: ππ β Rπ. So we take a finite intersection of half spaces with normal π’π
shifted so their border contains the point ππ.
β’ Bounded polytopes can be written as: Conv(π ) for some finite set π β Rπ. That is, the convex hull of some finite set of points. Here, the convex hull is defined as follows:
Conv(π΄) = {οΈ
βοΈ
πβπ΄
πΌ(π)π
β
β
β
β
β
(βπ β π΄ : πΌ(π) β₯ 0) β§βοΈ
πβπ΄
πΌ(π) = 1 }οΈ
Besides the half-space π»(π’) we also define the border of that half space, also known as the normal complement:
π(π’) = {π₯ β Rπ| π₯ Β· π’= 0}
where π’ ΜΈ= 0. Again, we exclude π’ = 0 to prevent needing to treat it as a special case every time we use π(π’) or π»(π’). It should be noted that for shifted half-spaces of the form π»(π’) + π different values of π can give the same shifted half-space. If we take some π β π + π(π’) then π»(π’) + π = π»(π’) + π.
We then define the faces of a polytope π as follows. A set πΉ β Rπ is a face of π if there exist some π’ and π such that
πΉ = π β© (π»(π’) + π) = π β© (π(π’) + π)
Every such face πΉ has a dimension, this is defined as dim(span(πΉ β π)) where π is the base of the shifted half-space that defines πΉ . We call the faces with dimension π β 1 the facets of π . The union of all facets forms the boundary of π .
We then also consider the polytope π and β to be faces with dimension π and β1 respectively. In this case, the faces form a lattice. That is, they form a partial order (under set inclusion) where any two elements have a unique least upper bound and greatest lower bound.
There is a special case of an unbounded polytope called the conical sum. It is defined as:
Coni(π΄) = {οΈ
βοΈ
πβπ΄
πΌππ | βπ β π΄: πΌπ β₯0 }οΈ
One can see this as the union of all scaled versions of Conv(π΄). The conical sum is closed under addition and positive scaling. Note that unlike the convex hull, the conical sum is not invariant under affine transformations. If a conical sum is not equal to the full space, it can be written in either of the following forms:
β’ {π₯ β Rπ | π π₯ β€0} for some π β RπΓπ
β’ βοΈππ»(π’π)
These are convex polytopes with a single vertex at the origin.
Finally, we have the Minkowski sum and difference. These are also known as dilation and erosion. The Minkowski sum β has three equivalent definitions.
π΄ β π΅= {π + π | π β π΄ β§ π β π΅}
= βοΈ
πβπ΅
π΄+ π
= βοΈ
πβπ΄
π+ π΅
2.1. FIRST THEORETICAL STEPS CHAPTER 2. THEORETICAL WORK
One key property of the Minkowski sum is that it commutes with taking the convex hull. That is:
Conv(π΄ β π΅) = Conv(π΄) β Conv(π΅). Moreover, the Minkowski sum of two convex sets is itself convex.
The Minkowski difference β is defined as:
π΄ β π΅= {π₯ β π΄ | π + π΅ β π΄}
Note that β is not the inverse of β. Moreover, if π΄ and π΅ are convex, so is π΄ β π΅. In fact, this holds even if π΅ is not convex. To see this, consider that π + π΅ β π΄ implies that Conv(π + π΅) β π΄ as π΄ is convex.
Now, let π΄ and π΅ be convex polytopes with 0 β π΅. Then any π₯ β π΄βπ΅ has many different decompositions π₯= π + π with π β π΄ and π β π΅. However, we will define a unique one:
π₯β²= arg min βπβ : π β (π₯ β π΄) β© π΅
Λπ₯ = π₯ β π₯β²
π₯= π₯β²+ Λπ₯, π₯β²β π΅, Λπ₯ β π΄
(2.1.1)
To see this is unique, (π₯ β π΄) β© π΅ is clearly a convex polytope. Then, the following proposition proves π₯β² is unique, from which Λπ₯ must be unique.
Proposition 2.1.6. Given a convex polytope π , the following is uniquely defined:
arg min
π₯βπ
βπ₯β
Proof. Suppose we have two distinct points π₯, π¦ β π with βπ₯β = βπ¦β. Then, by the triangle inequality and π₯ ΜΈ= π¦ we have:
β¦
β¦
β¦ π₯ 2 +π¦
2
β¦
β¦
β¦< βπ₯β
and by convexity we have:
π₯ 2 +π¦
2 β π
Thus, the above arg min must be unique.
Now, given any polytope π΅ with a facet πΉ we define
π’(πΉ ) = π’ β R2 such that βπ β πΉ : πΉ = π + π»(π’) β© π΅
This is unique up to scaling by positive real numbers. With this, we can introduce the following lemma:
Lemma 2.1.7. Given π΄, π΅ convex polytopes with 0 β π΅, let β± be the set of facets of π΄ then:
π΄ β π΅= π΄ βͺβοΈ
{οΈ
πΉ β π΅
β
β
β
β
β
πΉ β β± β§ π’(πΉ ) β βοΈ
π£βπ΅
π»(π£)πΆ }οΈ
Proof. It is trivial to see that the right-hand side is a subset of the left-hand side. This follows from 0 β π΅ and πΉ β π΄ for all πΉ β β±.
Given any π₯ β π΄ β π΅, take the unique decomposition π₯ = Λπ₯ + π₯β² from Equation 2.1.1. This has Λπ₯ β π΄, π₯β² β π΅ and π₯β² is the smallest element in π that allows such a decomposition. Clearly, if π₯β² = 0 then π₯ β π΄ β π΄ β π΅.
Now, suppose π₯β²ΜΈ= 0. Then π₯ = Λπ₯ + π₯β²β π΄/ . Moreover we know that:
Conv(Λπ₯, Λπ₯ + π₯β²) β© π΄ = {Λπ₯}
For otherwise, π₯β² would not be minimal. But then Λπ₯ must lie on the border of π΄, and thus lies in some facet πΉ β β±. Thus if π₯β² ΜΈ= 0 we have:
π₯= Λπ₯ + π₯β²β πΉ β π΅
Moreover if Λπ₯ β πΉ then π’(πΉ ) β π»(π₯β²)πΆ. This is again a consequence of the minimality of π₯β².
As such, every π₯ β π΄ β π΅ either lies in π΄ or lies in πΉ β π΅ for some πΉ β β± with π’(πΉ ) β βοΈπ£βπ π»(π£)πΆ.
The above lemma means that when taking the Minkowski sum, we only need to βdilateβ at the facets.
Note that in some cases, our extra requirement on the facet π’(πΉ ) β βοΈπ£βπ΅π»(π£)πΆ means we can leave out some of the dilated facets.
2.1.2 Using the abelian structure theorem
As we focus on finitely generated abelian groups, the abelian structure theorem will be used often. We state it here for the sake of completeness:
Theorem 2.1.8 (Abelian structure theorem). Any finitely generated abelian group is isomorphic to a product of the form ZπΓ π where π is a finite abelian group. π is called the βtoroidal componentβ of the group and π is called the βrankβ.
Note that in this decomposition all periodic elements lie in π β© {0} Γ π . Furthermore, as each subgroup of an abelian group is normal, we have the quotient group (ZπΓ π)/({0} Γ π ) β Zπ. This means there exists a surjective group homomorphism π : ZπΓ π β Zπ. Such homomorphisms can be seen as just discarding the part in π .
One application of the abelian structure theorem is the following result.
Theorem 2.1.9. Given a finitely generated abelian group π» and a finite symmetric set π β π» β {0} we set:
ππ= #{π β π | βπ β N+: π Γ π ΜΈ= 0}
2 Then, all ππ-forts in Ξ(π», π) are infinite.
Proof. First, we name the set of all non-periodic elements in π π= {π β π | βπ β N+: π Γ π ΜΈ= 0}
Note that #π = 2ππ. By the abelian structure theorem there exists a surjective homomorphism:
π: π» β Zπ
We know that for any π₯ β π we have π(π₯) ΜΈ= 0 because π₯ is not periodic.
Now, we can create a weak ordering (total order allowing for ties) on π» by simply extending a lexico- graphical ordering on Zπ. That is, if π₯, π¦ β π» then π₯ β₯ π¦ if and only if π(π₯) β₯ π(π¦). Where we write π₯ β‘ π¦ if π₯ β₯ π¦ β§ π¦ β₯ π₯. This only occurs if π₯ β π¦ β π . To see this, consider how the generators are ordered. Notably, our ordering is well behaved w.r.t. addition. That is, if π > 0 then π₯ + π > π₯. This is because addition on Zπ also has this property under the lexicographical ordering.
As π is symmetric, π is also symmetric . This, combined with the fact that no elements in π are ordered equally with 0, means that exactly half of π is larger than 0. We shall call this half π+. Note that
#π+ = ππ. Now, consider any finite subset π΄ β π». We wish to show that such an π΄ cannot be a ππ-fort.
As π΄ is finite, it has βmaximalβ points with respect to our order. Formally, there exist π β π΄ such that for all π β π΄ π β₯ π. Now, take such an π and consider the set of points π + π+. These are all neighbours of π, because π+ β π. Moreover, all of π + π+ is larger than π in our ordering and thus lies outside π΄. Therefore π β π΄ has at least #π+= ππ neighbors outside of π΄. As such π΄ cannot be a
ππ-fort.
This immediately allows us to apply Lemma 2.1.4 to get the following corollary:
Corollary 2.1.10. Given a finitely generated abelian group π» and a symmetric set of generators π pick:
π= ππ = #{π β π | βπ β N+: π Γ π ΜΈ= 0}
2
Then π-threshold bootstrap percolation on Ξ(π», π) has ππ β€1 β π where π > 0 is the critical probability for site percolation.
2.1. FIRST THEORETICAL STEPS CHAPTER 2. THEORETICAL WORK
To better make use of the abelian structure theorem, we define a new form of bootstrap percolation called modified bootstrap percolation.
Definition 2.1.11 (Modified bootstrap percolation). Let πΊ be an abelian group and take π β πΊ β {0}
to be a finite symmetric subset. We call a set π β π a symmetric half of π if π βͺ βπ = π. Modified bootstrap percolation is then defined by the following discrete step, based on symmetric halves:
ππ(π΄) = π΄ βͺ {π₯ β π΄ | βπ β π : π₯ + π β π΄ β§ π βͺ βπ = π}
Simply stated, in order for π₯ to become occupied by ππ, each symmetric pair π₯ + π , π₯ β π given an π β π has to have at least one occupied node. Now compare modified bootstrap percolation to π-threshold bootstrap percolation with π = #π2 . We see that modified bootstrap percolation is a lot like π-threshold bootstrap percolation, only we have some more geometric requirements. Instead of allowing any half of the neighbors of a point to be occupied, we require a symmetric half to be occupied. Note that the 0-1 law (Theorem 2.1.3) still holds for modified bootstrap percolation. The underlying probability distribution remains ergodic, and the event of percolation under modified bootstrap percolation is also translation invariant.
For now, we wish to show that modified bootstrap percolation is dominated by π-threshold bootstrap percolation when π = #π/2. This follows quite readily from the fact that 0 /β π. For then, if π is a symmetric half of π then #π β₯#π/2. This yields:
ππ(π΄) β π½#π/2(π΄) (2.1.2)
Thus if we prove that modified percolation leads to complete percolation, then we also prove that π- threshold bootstrap percolation leads to complete percolation at π =#π/2.
We can combine the above with Lemma 2.1.5 to get the following proposition:
Proposition 2.1.12. Let π» be a finitely generated abelian group, and π β π» β {0} be a finite symmetric subset. Moreover, set π to be all non-periodic elements of π.
Then it follows that ππ = #π2 and:
ππ(π΄) β π½ππ(π΄)
Next, we introduce a theorem that takes modified bootstrap percolation without periodic generators on any abelian group and reduces it to modified bootstrap percolation on a free abelian group (i.e. a group with no toroidal part).
Theorem 2.1.13. Let π» be any abelian group isomorphic to ZπΓ π where π is finite. Let π β π» be a finite symmetric set that does not include periodic elements. Finally, let π : π» β Zπ be a surjective homomorphism.
In that case, if ππ = 0 for modified bootstrap percolation on Ξ(Zπ, π(π)) then ππ= 0 for modified bootstrap percolation on Ξ(π», π). The same holds for ππ.
Proof. The basis of this proof is to show that modified bootstrap percolation on π» dominates modified bootstrap percolation on Zπin some sense.
Note that, if π» = ZπΓ π we can simply take:
π: (π§, π‘) β¦β π§
Any other surjective homomorphism π : π» β¦β Zπcan be factored as π = π β π β π where π : π» β ZπΓ π and π : Zπ β Zπ are isomorphisms. Nowhere in the proof do these isomorphisms affect the reasoning.
As such, once can essentially presume the above form of π for the entire proof.
Now, given an initial condition π΄0 β π» we will transform it to an initial condition π΅0β Zπ as follows:
π΅0 = {π₯ β Zπ | πβ1(π₯) β π΄0}. That is, π΅0 corresponds to all totally occupied translates of π in π΄0. Now for any π₯ β Zπ we have ππ(π₯ β π΅0) = π#π and these are fully independent of any other points.
Thus π΅0 can be seen as an initial condition with density π#π.
We then define a second generating set πβ² = π(π). Note that the requirement that no generators be periodic means that no generator lies in π so 0 /β π(π). We then define two sequences [π΄π] and [π΅π]:
π΄π= (ππ)π(π΄0) π΅π= (ππβ²)π(π΅0)
Now, we want to show that [π΄π] dominates [π΅π] in some sense, but they are sequences in different spaces.
So instead we will prove the following:
πβ1(π΅π) β π΄π
We still call this βdominationβ because as π΅π grows, this also require π΄π to grow. Moreover, if π΅π = Zπ then it follows that π΄π = π». We will prove this using induction on π. By definition of π΅0this holds for π= 0. What remains is to show that:
πβ1(π΅π) β π΄π =β πβ1(π΅π+1) β π΄π
To this end, it suffices to show that:
π₯ β π΅π+1β π΅π =β πβ1(π₯) β π΄π+1 (2.1.3)
By definition of modified bootstrap percolation, the left hand side of (2.1.3) implies there exists a symmetric half πβ² of πβ² such that:
π₯+ πβ² β π΅π
Now take π¦ to be any point in πβ1(π₯). Moreover, take π = π β© πβ1(πβ²). Note that π is a symmetric half of π. Then the above implies:
π¦+ π β πβ1(π΅π) β π΄π
Thus, by definition of ππ we have π¦ β ππ(π΄π) = π΄π+1.
Now, suppose that modified bootstrap percolation on Ξ(Zπ, π) has ππ = 0 for any finite symmetric π.
Then the above theorem means we have ππ = 0 for modified bootstrap percolation on any Ξ(π», π) as long as π» has rank π and π does not contain periodic generators. Now, by Proposition 2.1.12 we know that modified bootstrap percolation on Ξ(π», π) dominates ππ-threshold bootstrap percolation on the same graph. Moreover, if we take π = π βͺ π where all elements in π are periodic, we have ππ = ππ. Moreover, the graph Ξ(π», π) is an extension of Ξ(π», π). Thus π½ππ on Ξ(π»π) dominates π½ππon Ξ(π», π).
Now, suppose we have some abelian group π» with rank π. Next take π β π» to be a finite symmetric subset and set π β π to contain all non-periodic elements in π. This means that ππ = ππ. Moreover, we know that π½ππ on Ξ(π», π) dominates π½ππ on Ξ(π», π).
Then by the above theorem, if ππ = 0 for modified bootstrap percolation on all Cayley graphs Ξ(Zπ, πβ²) for finite symmetric πβ², then we also have ππ= 0 for ππ-threshold bootstrap percolation on Ξ(π», π). By the domination we derived above, this also means ππ= 0 for ππ-threshold percolation on Ξ(π», π).
Thus, to prove the case π β€ ππ in Conjecture 1.1.3 for groups of rank π it suffices to only deal with modified bootstrap percolation on the free abelian group Zπ. As such, we now focus on modified bootstrap percolation on free abelian groups.
2.2 Free abelian case
Theorem 2.1.13 shows that to prove Conjecture 1.1.3 for π β€ ππ we need only consider Cayley graphs over the free abelian groups Zπ. In this section, we get some general results for such graphs. These represent first steps towards proving we have ππ= 0 for modified bootstrap percolation. Sadly, later we need steps that only work in Z2but in this section.
Our results here are inspired by the proofs in [Ent87] and [Sch92]. These proofs are for Cayley graphs of Z2and Zπusing a minimal generating set, we will call these the canonical graphs. We also take some inspiration from [GG93], though we consider the proofs from that paper to be suspect. First, we sketch the ideas behind these proofs.
2.2. FREE ABELIAN CASE CHAPTER 2. THEORETICAL WORK
(2.1-a)the first step of growth (2.1-b)final state Figure 2.1: How an edge grows by adding one point adjacent to the edge.
The proofs for the canonical graphs depend on rectangles or more generally, boxes. These are stable shapes at π = π β₯ 2 (π is the dimension). That is they are sets π΅ that satisfy:
π΅= π½π(π΅)
For any point outside a box has at most 1 connection to a point inside the box, and π > 1.
However, in some sense such boxes are only barely stable. To see this, first consider the 2D case of a rectangle of occupied points. Then, add a single occupied point along one of the edges. This activates the points along the edge adjacent to this one extra point. These added points then repeatedly activate their neighboring points until we reach the end of the edge. Once this process has finished, the edge has grown by one line. This is illustrated in Figure 2.1 The proof from [Sch92] extends this reasoning to π-dimensional boxes in their lemma 3.1. Summarized, the reasoning is as follows: consider again a box of occupied points. We now take a facet of this box and ask, βwhat does it take for this facet to growβ.
Here, by growing we mean occupying all points adjacent to this facet. This set of vacant points forms a π β1 dimensional box. Moreover, these points all already have 1 occupied neighbor. As such, the facet will grow when this π β 1 dimensional box would fill itself under (π β 1)-threshold percolation.
Our idea is to apply the proof ideas sketched above to modified bootstrap percolation. To this end, we introduce even more geometry by viewing Zπ as a subset of Rπ. This embedding plays a key role in this paper. It allows us to use norms (euclidean unless stated otherwise), inner-products, and discrete geometry as described in Section 2.1.1.
We then use this embedding in Rπ to introduce a new form of bootstrap percolation on our free abelian Cayley graphs: convex bootstrap percolation. Recall that in modified bootstrap percolation, a point became occupied when a symmetric half of itβs neighbors were occupied. In convex bootstrap percolation, we replace the βsymmetric halfβ with a βconvex halfβ.
Definition 2.2.1 (Convex Half). Given a finite symmetric set π β Zπβ {0} we say πΆ is a convex half of π if:
βπ’ β Rπ: πΆ = π β© π»(π’)
A single step of convex bootstrap percolation is then defined as follows:
π½π(π΄) = π΄ βͺ {π₯ β Zπ| βπ’ β Rπ: π₯ + (π β© π»(π’)) β π΄} (2.2.1) Note that it is possible that a convex half contains more than half of all points from π. This will be resolved when we define βminimal convex halvesβ (Definition 2.2.4).
Convex halves of π are all symmetric halves of π. Thus, ππ dominates π½π. In the case of canonical generators, convex and modified bootstrap percolation coincide. For convex bootstrap percolation, like modified bootstrap percolation, a necessary condition for growth of a point π₯ is as follows. For every π β π either π₯ + π or π₯ β π needs to be occupied.
Now, we will only use generating sets with a few reasonable properties. To prevent needlessly repeating these properties, we define a generating set as a set that has the properties we want.
Definition 2.2.2 (Generating set). We say a set π β Zπ is a generating set if it has all the following properties:
β’ π is finite
β’ π = βπ (i.e. π is symmetric)
β’ 0 /β π
We call a half-space π»(π’) stable if:
π½π(π»(π’)) = π»(π’)
We then ask which half-spaces are stable. In the canonical case for Z2under 2-threshold bootstrap per- colation, the only half-spaces that are stable are π» ((1, 0)) and π» ((0, 1)) and their negative counterparts.
This is why rectangles are stable shapes in the canonical case. They are intersections of such (shifted) half-spaces. In arbitrary dimension, the same holds when π consists only of the canonical basis vectors and their negative counterparts. This is why, in that case, boxes are stable shapes.
We will see later that stable half-spaces are only barely stable in a sense similar to how the facets of boxes are barely stable. This is the subject of Section 2.2.1 which forms the basis of Section 2.3.
First though, we want to classify which half spaces are stable. To this end, we define ππ as the set of all stable directions:
ππ= {π’ β Rπ | π½π(π»(π’)) = π»(π’)}
We then have the following alternate description for ππ. Proposition 2.2.3. Given a generating set π β Zπ, we have:
ππ = {π’ β Rπ| π(π’) β© π ΜΈ= β }
For the proof of this proposition we need a definition we will use throughout this section:
ππ(π’) = min{π₯ Β· π’ | π₯ β (π β π»(π’))} (2.2.2) This is the βshiftβ of a direction π’. Note that by the symmetry of π we have ππ(π’) = ππ(βπ’). Moreover, as π is finite we know that ππ(π’) > 0.
Proof. We will prove the set equality by proving mutual inclusion.
First, we prove the inclusion β.
To this end, suppose we have a direction π’ β Rπ such that:
π(π’) β© π ΜΈ= β
Then, we can find a pair points: π , βπ β π(π’) β© π. Moreover, if πΆ is a convex half of π then either π β πΆ or βπ β πΆ (or both). Now, take any point π₯ /β π»(π’), then by definition of π’ it follows that:
π₯+ π /β π»(π’) β§ π₯ β π /β π»(π’) So, for any convex half πΆ β π we have:
π₯+ πΆ * π»(π’)
Thus, we can conclude that π₯ /β π½π(π»(π’); which means that π’ β ππ. Next we prove the inclusion β.
This will be done using contraposition, so suppose that we have a direction π’ β Rπ for which:
π(π’) β© π = β
Then note that π»(π’) βͺ π»(βπ’) = Rπ and π»(π’) β© π»(βπ’) = π(π’). We can combine that with our assumption on π’ to get the following convex half πΆ:
πΆ= π β© π»(π’) = π β π»(βπ’)
2.2. FREE ABELIAN CASE CHAPTER 2. THEORETICAL WORK
Now, take π₯ = ππ(π’) π’ noting that π₯ /β π»(π’). Further, pick any π¦ β πΆ Then, by the above equation for πΆ and the definition of ππ we know that:
0 < π₯ Β· π’ β€ βπ¦ Β· π’ (π₯ + π¦) Β· π’ β€ 0 (π₯ + π¦) β π»(π’)
Therefore, π₯+πΆ β π»(π’) and thus π₯ β π½π(π»(π’)) whilst π₯ /β π»(π’). This means π’ is not a stable direction:
π(π’) β© π = β =β π’ /β ππ
ππ β {π’ β Rπ| π(π’) β© π ΜΈ= β }
Thus, we can view ππ as the union of all orthogonal complements to some π β π. In this sense, ππ has dimension π β 1.
Our set ππ finds another use in the following definition of minimal convex halves:
Definition 2.2.4(Minimal convex half). Given a generating set π we call πΆ β π a minimal convex half of π if:
βπ’ /β ππ : πΆ = π β© π»(π’) We define the family of all minimal convex halves of π:
ππ = {πΆ β π | βπ’ /β ππ : πΆ = π β© π»(π’)}
The following proposition explains why these specific convex halves are called minimal:
Proposition 2.2.5. Given a generating set π and any convex half πΆ = π β©π»(π’), there exists a minimal convex half π β ππ such that π β πΆ.
Moreover, given any minimal convex half π β πΆπ there exists no other convex half πΆ = π β© π»(π’) such that πΆ ( π.
Proof. We first prove the second claim. Let π = π β© π»(π’) β ππ be a minimal convex half. So π’ /β ππ. Then, by Proposition 2.2.3 we have: π β© βπ = π β© π(π’) = β . Therefore, #π = #π2 . This means that if we leave out any point from π we have too few elements to be a convex half. So π cannot have a convex half as a proper subset.
Now, for the first claim, consider an arbitrary convex half πΆ = π β© π»(π’). If π’ /β ππ then πΆ is a minimal convex half and we are done. So we only need to consider the case where π’ β ππ. In that case, we know that π β© π(π’) ΜΈ= β . Now set:
π΅ = (π β© π»(π’)) β π(π’)
We will extend π΅ to a minimal half of π by adding a minimal convex half of π β© π(π’). To this end, pick some
π’β² β π(π’) β ππβ©π (π’)
Then π β© π(π’) β© π»(π’β²) is a minimal convex half of π β© π(π’).
Next, as π is finite, there exists some π > 0 such that π β π»(βπ’) = π β π»(βπ£) whenever βπ’ β π£β < π.
We can then find a factor 0 < πΌ < π such that π£ = π’ + πΌ π’β²β π/ π. It then follows that π β© π»(π£) = π΅ βͺ (π β© π(π’) β© π»(π’β²)) β πΆ
Thus, πΆ contains a minimal convex half.
Corollary 2.2.6. We have the following equivalent definition for convex bootstrap percolation:
π½π(π΄) = π΄ βͺ {π₯ β Zπ| β πΆ β ππ : π₯ + πΆ β π΄} (2.2.3)
Next, we define a finite set related to ππ that contains the βmost stableβ directions:
ππ= {π’ | βπ’β = 1 β§ span(π(π’) β© π) = π(π’)}
That is, ππ contains all unit vectors π’ for which π(π’) β© π spans a hyper-plane. Given a π’ β ππ, consider the half space π»(π’). We say this half-space is βmost stableβ because if π΄0 = π»(π’) then any point π₯ /β π»(π’) has vacant neighbours in a set of directions that span a full hyperplane.
Now, as π is finite, it can only span finitely many hyper-planes. Therefore, ππ is finite. Like ππ was the union of all points that are orthogonal to some π β π, ππ consists of all points that are orthogonal to π β1 linearly independent elements from π.
Below we give a quick recap of newly introduced names. Here, we presume that π is a generating set (Definition 2.2.2).
β’ The set of all stable directions:
ππ = {π’ β Rπ| π½π(π»(π’)) = π»(π’)}
= {π’ β Rπ| π(π’) β© π ΜΈ= β }
β’ The set of the most stable directions:
ππ = {π’ | βπ’β = 1 β§ span(π(π’) β© π) = π(π’)}
β’ The shift of a direction π’ β Rπ:
ππ(π’) = min{π₯ Β· π’ | π₯ β (π β π»(π’))}
β’ The set of all minimal convex halves:
ππ = {πΆ β π | βπ’ /β ππ : πΆ = π β© π»(π’)}
β’ The convex bootstrap percolation process in terms of minimal convex halves:
π½π(π΄) = π΄ βͺ {π₯ β Zπ| βπΆ β ππ : π₯ + πΆ β π΄}
β’ We also introduce a new definition, the radius of π:
π π= max{βπ₯β | π₯ β π}
2.2.1 Facet growth
Now that we have a nice overview of what constitutes a stable direction, we start looking at what it takes for growth to occur in such a stable direction. Specifically, we look at this in the context of facets of polytopes. Note that half-spaces are just a special case of polytopes. The idea here is that we might replaces boxes in the proof of the canonical case with polytopes.
To proceed, we want a nice way to refer to polytopes with facet normals in ππ. To this end, we take some enumeration of ππ, which yields:
ππ = {π’1. . . π’π} (2.2.4)
We then use this to define a matrix ππ. Considering each π’π to be a column vector we define:
ππ = (π(π’1) π’1| π(π’2) π’2| . . . | π(π’π) π’π)
Where the vertical bars are just visual separation of the columns. Our polytopes can then by defined as solutions to πππ₯ β€ πfor some π. Our scaling by π(π’π) for ππmeans that integer choices for π correspond to polytopes whose bounding hyperplanes are supported by points from Zπ. We then identify vectors π with polytopes as follows:
Polyπ(π) = {π₯ β Rπ| ππ π₯ β€ π}
As these polytopes are the intersection of stable sets and π½π is increasing, these polytopes are stable as well.
Now, to generalize the proof ideas from the canonical case, we need some way of describing a βgrownβ
facet. We then define the growing function πΊ and the βfacetβ that gets added πΉ . These functions take a
2.2. FREE ABELIAN CASE CHAPTER 2. THEORETICAL WORK
(2.2-a)An example of a polygon that is degenerate because it has a corner that is βtoo sharpβ. Thus, that vertex will never fit a convex half
(2.2-b)With this polygon, every vertex fits a convex half. How- ever, it is degenerate due to the mid-point of the edge. The cho- sen minimal convex half does not fit, neither does its vertical mirror image.
(2.2-c) Finally, this polygon is non-degenerate. It is essentially a scaled-up version of the previous case.
Figure 2.2: Various polygons and whether or not they are degenerate. We take π containing all points within a 1 Γ 1 square (excluding 0). The polytopes are shown in light-gray. The minimal convex half is shown in (translucent) black. The origin of the convex half is the open circle, whilst the actual elements are shown as full circles. Note that all minimal convex halves of this π are rotated and mirrored versions of the shown minimal convex half.
polytope π = Polyπ(π) and a direction π’π β ππ (we take the index of π’ so we can find the corresponding canonical basis vector ππ).
πΊ(π, π’π) = Polyπ(π + ππ)
πΉ(π, π’π) = πΊ(π, π’π) β π (2.2.5)
We call πΉ (π, π’) the facet in direction π’ β ππ. Note that technically, πΉ has volume and thus is not a facet of π .
Next, we introduce the concept of degeneracy.
Definition 2.2.7 (Degenerate Polytope). Given a generating set π, we say a polytope π is degenerate (with respect to S) if
βπ₯ β π : β πΆ β ππ : π₯ + πΆ ΜΈβ π If a polytope is not degenerate, we say it is non-degenerate.
The issue with degenerate polytopes is that they contain points that cannot be activated without help from outside the polytope. The concept is illustrated in Figure 2.2.
Using the growth functions from (2.2.5) and the concept of degeneracy, we can introduce the following lemma:
Lemma 2.2.8. Take π = Polyπ(π) to be fully occupied. Take π’ β ππ and presume πΊ(π, π’) is non- degenerate.
Then πΊ(π, π’) is internally spanned by π if πΉ (π, π’) is internally spanned by π.
Proof. Our proof is based on the decomposition πΊ(π, π’) = π βͺ πΉ (π, π’). We set π΄ β Zπ to be the set of active points and then define Β―π΄= π΄ β© πΉ (π, π’), which is the set of active points inside the facet.
Now, suppose we have a point π₯ β πΉ (π, π’) that gets activated by the dynamics using generating set π β© π(π’). That is,
π₯ β π½πβ©π (π’)(οΈπ΄Β―)οΈ
If we then show that this implies π₯ β π½π(π΄) the proof can be finished with a trivial inductive argument, as πΉ (π, π’) was internally spanned.