betaFeedback

WorksMathematics

Multiple transition points for an epidemic model

Abstract

This paper is concerned with a spatial epidemic model on the dd-dimensional square lattice that extends the contact process in two ways by assuming that each site is occupied by a social cluster of NN individuals rather than a single individual, and that infected individuals can be either asymptomatic or symptomatic. Newly infected individuals begin as asymptomatic and develop the symptoms at a spontaneous mutation rate γγ, and all infected individuals recover at rate one. The model also distinguishes between external (between adjacent clusters) and internal (within clusters) infection rates for each state. Our analysis shows that, when either asymptomatic or symptomatic individuals only spread the disease externally and at a subcritical rate, while the other class only transmits the disease internally and at a large rate, the system exhibits at least two phase transitions with respect to the mutation rate γγ when social clusters are large. Specifically, the epidemic dies out for extreme values but survives for intermediate values of the mutation rate.

Introduction

This paper shows the lack of monotonicity, resulting in the existence of (at least) two phase transitions, for the epidemic model recently introduced in [1]. This model is a variant of Harris’ contact process [7] on the dd-dimensional square lattice that replaces each lattice point/single individual by a social cluster of NN individuals, like the model in [2], and distinguishes infected individuals depending on whether they are asymptomatic or symptomatic, like the model in [3, 9]. For a review of these two epidemic models, we refer the reader to [8 Chapter 2]. In particular, the state of the process at time tt is a spatial configuration

ξt:Zd→{(i,j)∈Z+×Z+:i+j≤N},\xi_t : \mathbb{Z}^d \to\{(i,j) \in\mathbb{Z}_+ \times\mathbb{Z}_+ : i+j \le N\},

where ξt(x)=(i,j)\xi_t(x) = (i,j) means that social cluster xx contains ii asymptomatic individuals, and jj symptomatic individuals, while the other individuals are healthy. The presence of social clusters motivates the introduction of two infection rates depending on whether the infection originates from a nearby cluster or from the same cluster. Similarly, the distinction between asymptomatic and symptomatic individuals motivates the introduction of two infection rates depending on the state of the individual the infection originates from. Combining the two gives four infection rates: external asymptomatic infections at rate λ1\lambda_1, external symptomatic infections at rate λ2\lambda_2, internal asymptomatic infections at rate ϕ1\phi_1, and internal symptomatic infections rate ϕ2\phi_2. In addition, we assume that newly infected individuals are always asymptomatic and become symptomatic at the mutation rate γ\gamma, that infected individuals recover at rate one, and that external infections are neglected once a cluster is infected. In particular, writing y∼xy \sim x to indicate that cluster yy is one of the 2d2d nearest neighbors of cluster xx, the transition rates at cluster xx are given by

(0,0)→(1,0)at rate ∑y∼x∑i,j(iλ1+jλ2)1{ξt(y)=(i,j)},(i,j)→(i+1,j)at rate iϕ1+jϕ2for i+j<N,(i,j)→(i−1,j+1)at rate iγfor i+j≤N, j<N,(i,j)→(i−1,j)at rate ifor i+j≤N,(i,j)→(i,j−1)at rate jfor i+j≤N.\begin{aligned} (0,0) &\to(1,0) &&\text{at rate } \sum_{y\sim x}\sum_{i,j}(i\lambda_1+j\lambda_2)\mathbf{1}\{\xi_t(y)=(i,j)\},\\ (i,j) &\to(i+1,j) &&\text{at rate } i\phi_1+j\phi_2 \quad\text{for } i+j<N,\\ (i,j) &\to(i-1,j+1) &&\text{at rate } i\gamma\quad\text{for } i+j\le N,\ j<N,\\ (i,j) &\to(i-1,j) &&\text{at rate } i \quad\text{for } i+j\le N,\\ (i,j) &\to(i,j-1) &&\text{at rate } j \quad\text{for } i+j\le N. \end{aligned}

From top to bottom, these transitions are interpreted as external infection of a healthy cluster, additional internal infection of an infected cluster, mutation from asymptomatic to symptomatic, recovery of an asymptomatic individual, and recovery of a symptomatic individual. We assume that the epidemic model starts from a translation-invariant distribution with a positive density of asymptomatic individuals and a positive density of symptomatic individuals. Like the contact process, there are two possible long-term behaviors:

survivalwhenlim inf⁡t→∞P(ξt(x)≠(0,0))>0for all x∈Zd,extinctionwhenlim⁡t→∞P(ξt(x)≠(0,0))=0for all x∈Zd.\begin{aligned} \text{survival} &\quad\text{when}\quad\liminf_{t\to\infty}P(\xi_t(x)\ne(0,0))>0 \quad\text{for all }x\in\mathbb{Z}^d,\\ \text{extinction} &\quad\text{when}\quad\lim_{t\to\infty}P(\xi_t(x)\ne(0,0))=0 \quad\text{for all }x\in\mathbb{Z}^d. \end{aligned}

Our previous work [1] shows the existence of phase transitions between extinction and survival with respect to each of the four infection rates and the cluster size. Even though coupling methods comparing processes with different parameters fail to prove monotonicity, the analysis in [1] suggests monotonicity in the sense that, if there is survival for a given set of parameters, increasing the infection rates and the cluster size still results in survival, which is further supported by numerical simulations of the two-dimensional process. In contrast, increasing the mutation rate γ\gamma can either promote survival or promote extinction depending on the other parameters. More interestingly, based on the analysis of the mean-field model, we conjectured that, for some values of the infection rates and the cluster size, there is extinction for γ\gamma small or large, but survival for intermediate values, thus resulting in the existence of multiple phase transitions with respect to the mutation rate. The objective of this paper is to prove this conjecture.

Mean-field model. To hypothesize how to choose the infection parameters, we first return to the mean-field model, which consists of a system of coupled differential equations for the densities uiju_{ij} of clusters with ii asymptomatic and jj symptomatic individuals, and is obtained by assuming that the population is homogeneously mixing. Defining survival in the context of the mean-field model as the existence of a fixed point with a positive density of infected clusters, [1 Proposition 1] shows that, for clusters of size two, there is survival if and only if

(λ1+γλ2)((1+ϕ1)(1+ϕ2)+(1+ϕ1)+γ(1+ϕ2))−(γϕ1+ϕ2+(1+γ)(2+γ))>0.(1)(\lambda_1+\gamma\lambda_2)((1+\phi_1)(1+\phi_2)+(1+\phi_1)+\gamma(1+\phi_2))-(\gamma\phi_1+\phi_2+(1+\gamma)(2+\gamma))>0. \tag*{(1)}

Using this preliminary result, we also proved that the mean-field model is monotone with respect to the infection rates in the sense that, if there is survival for a given set of parameters, increasing the infection rates still results in survival, and by contraposition, if there is extinction for a given set of parameters, decreasing the infection rates still results in extinction. The mean-field model, however, is not monotone with respect to the mutation rate. Indeed, taking λ2=ϕ1=0\lambda_2=\phi_1=0, the condition for survival (1) reduces to

−γ2+(λ1(1+ϕ2)−3)γ+(λ1−1)(2+ϕ2)>0.-\gamma^2+(\lambda_1(1+\phi_2)-3)\gamma+(\lambda_1-1)(2+\phi_2)>0.

Then, taking for instance λ1=3/4\lambda_1 = 3/4 and ϕ2=11\phi_2 = 11, the left-hand side becomes

−γ2+6γ−13/4=−(γ−γ−)(γ−γ+)whereγ±=3±23/2>0,(2)-\gamma^2 + 6\gamma- 13/4 = -(\gamma-\gamma_-)(\gamma-\gamma_+) \quad\text{where} \quad\gamma_{\pm} = 3 \pm\sqrt{23}/2 > 0, \tag*{(2)}

showing the existence of two phase transitions with

extinction if γ≤γ−\gamma\le\gamma_-, survival if γ−<γ<γ+\gamma_- < \gamma< \gamma_+, extinction if γ≥γ+\gamma\ge\gamma_+.

Similarly, when λ1=ϕ2=0\lambda_1 = \phi_2 = 0, the condition for survival (1) reduces to

(λ2−1)γ2+(2λ2(1+ϕ1)−ϕ1−3)γ−2>0.(\lambda_2 - 1)\gamma^2 + (2\lambda_2(1+\phi_1)-\phi_1-3)\gamma- 2 > 0.

Taking λ2=3/4\lambda_2 = 3/4 and ϕ1=11\phi_1 = 11, the left-hand side becomes

−γ2/4+4γ−2=−(γ−γ−′)(γ−γ+′)/4whereγ±′=8±214>0,(3)-\gamma^2/4 + 4\gamma- 2 = -(\gamma-\gamma'_-)(\gamma-\gamma'_+)/4 \quad\text{where} \quad\gamma'_{\pm} = 8 \pm2\sqrt{14} > 0, \tag*{(3)}

showing again the existence of two phase transitions with

extinction if γ≤γ−′\gamma\le\gamma'_-, survival if γ−′<γ<γ+′\gamma'_- < \gamma< \gamma'_+, extinction if γ≥γ+′\gamma\ge\gamma'_+.

In conclusion, at least in the mean-field model with clusters of size two, we have two phase transitions with respect to the mutation rate when asymptomatic individuals infect nearby clusters occasionally but not their own cluster, while symptomatic individuals cannot infect nearby clusters but are highly infectious in their own cluster, leading to scenario (2). Similarly, there are also two phase transitions when asymptomatic individuals cannot infect nearby clusters but are highly infectious in their own cluster, while symptomatic individuals infect nearby clusters occasionally but not their own cluster, leading to scenario (3).

Stochastic spatial model. Following the results for the mean-field model, to mimic scenario (2) above but in the context of the interacting particle system, we again take λ2=ϕ1=0\lambda_2 = \phi_1 = 0. To ensure extinction for γ\gamma small in the mean-field model, we need λ1<1\lambda_1 < 1. For the interacting particle system, this condition can be relaxed taking λ1\lambda_1 smaller than the critical value λc\lambda_c of the basic contact process. Finally, instead of simply taking ϕ2\phi_2 large, we set ϕ2=N\phi_2 = N and study the process when the cluster size NN is large. This choice of the four infection rates and the cluster size leads to the existence of two phase transitions with respect to γ\gamma when NN is sufficiently large.

Theorem 1. Assuming that the infection rates are given by

λ1∈(0,λc),λ2=0,ϕ1=0,ϕ2=N=cluster size,\lambda_1 \in(0,\lambda_c), \quad\lambda_2 = 0, \quad\phi_1 = 0, \quad\phi_2 = N = \textit{cluster size},

there are at least two phase transitions with respect to γ\gamma in the sense that

a. ∀N>1\forall N > 1, ∃ 0<γ−<γ+<∞:γ∉(γ−,γ+)⇒\exists\ 0 < \gamma_- < \gamma_+ < \infty: \gamma\notin(\gamma_-,\gamma_+) \Rightarrow extinction,

b. ∀γ∈(0,∞)\forall\gamma\in(0,\infty), ∃N+<∞:N>N+⇒\exists N_+ < \infty: N > N_+ \Rightarrow survival.

Note that the theorem indeed implies the presence of at least two phase transitions for some range of the infection parameters and cluster size, with extinction for extreme values and survival for intermediate values of γ\gamma. To see this, fix γ0∈(0,∞)\gamma_0 \in(0,\infty), and apply the second part of the theorem to get an N0N_0 finite such that the process survives, then apply the first part of the theorem to get two values γ−<γ0<γ+\gamma_- < \gamma_0 < \gamma_+ such that the process dies out for γ<γ−\gamma< \gamma_- and γ>γ+\gamma> \gamma_+. This shows that, when the cluster size is N0N_0, there is a phase transition from extinction to survival between γ−\gamma_- and γ0\gamma_0, and another phase transition from survival to extinction between γ0\gamma_0 and γ+\gamma_+. Figure 1 illustrates this result for clusters with only three individuals. Scenario (3) above can also be reproduced for the interacting particle system by simply reversing the roles of the asymptomatic and symptomatic infection rates, which gives the following theorem.

Figure 1. Snapshots of the process on the 300×300300 \times300 torus at various times and for various values of γ\gamma. In all three pictures, the parameters are λ1=1/4<λc\lambda_1=1/4<\lambda_c, λ2=ϕ1=0\lambda_2=\phi_1=0 and ϕ2=N=3\phi_2=N=3, and the process starts with one healthy, one asymptomatic, and one symptomatic individuals in each cluster. The two simulations at the top illustrate the first part of Theorem 1, while the simulation at the bottom illustrates the second part of the theorem, and the three simulations together show the presence of at least two phase transitions with respect to γ\gamma.

Theorem 2. Assuming that the infection rates are given by

λ1=0,λ2∈(0,λc),ϕ1=N=cluster size,ϕ2=0,\lambda_1=0,\qquad\lambda_2\in(0,\lambda_c),\qquad\phi_1=N=\text{cluster size},\qquad\phi_2=0,

there are at least two phase transitions with respect to γ\gamma in the sense that

a.  ∀N>1, ∃ 0<γ−<γ+<∞: γ∉(γ−,γ+) ⇒ extinction,b.  ∀γ∈(0,∞), ∃ N+<∞: N>N+ ⇒ survival.\begin{aligned} &\text{a. }\ \forall N>1,\ \exists\,0<\gamma_-<\gamma_+<\infty:\ \gamma\notin(\gamma_-,\gamma_+)\ \Rightarrow\ \text{extinction},\\ &\text{b. }\ \forall\gamma\in(0,\infty),\ \exists\,N_+<\infty:\ N>N_+\ \Rightarrow\ \text{survival}. \end{aligned}

Even though both theorems give the same conclusion, with very similar assumptions, the proofs are significantly different. This is due to the fact that newly infected individuals are always asymptomatic, while symptomatic individuals can only appear later as the result of a mutation, which completely breaks down symmetry. In particular, reversing the values of the asymptomatic and symptomatic infection rates does not result in general in the same behavior. In fact, survival for intermediate values of γ\gamma follows from the quick spread of the disease across space in Theorem 1, but from the long persistence of the disease within clusters in Theorem 2.

Graphical representation

The first step to prove the theorems is to construct the process graphically. As pointed out in [1], it is convenient to view each social cluster as a pile of NN sites corresponding to the individuals living in that cluster, and to think of the process as keeping track of the state of each individual, rather than the state of each cluster. In particular, the state at time tt becomes

ξt:Δ⟶{0,1,2}whereΔ=Zd×{1,2,…,N},\xi_t:\Delta\longrightarrow\{0,1,2\} \quad\text{where} \quad\Delta= \mathbb{Z}^d \times\{1,2,\ldots,N\},

and where 0 means healthy, 1 means asymptomatic, and 2 means symptomatic. From now on, we call each element (x,i)∈Δ(x,i) \in\Delta a site, or a pair cluster-level. Using an idea of Harris [6] adapted to our model in [1 Section 3], the process starting from any initial configuration can be constructed from the following collections of independent Poisson processes:

  • External infections. For j=1,2j=1,2, and all (x,i)∈Δ(x,i) \in\Delta and y∼xy \sim x, place an exponential clock with rate λj\lambda_j along the edge (x,i)→(y,1)(x,i) \to(y,1), and draw

a type j arrow (x,i,t)→j(y,1,t) at the times t the clock rings\text{a type }j\text{ arrow }(x,i,t) \xrightarrow{j} (y,1,t)\text{ at the times }t\text{ the clock rings}

to indicate that if the tail of the arrow is occupied by a jj and each of the levels of yy are occupied by a 0, then the head of the arrow becomes occupied by a 1.

  • Internal infections. Similarly, for j=1,2j=1,2, and all (x,i)∈Δ(x,i) \in\Delta, place an exponential clock with rate ϕj\phi_j at site (x,i)(x,i), and put

a type j bullet j-∙ at (x,i,t) at the times t the clock rings\text{a type }j\text{ bullet }j\text{-}\bullet\text{ at }(x,i,t)\text{ at the times }t\text{ the clock rings}

to indicate that if the bullet is occupied by a jj and there is at least one 0 in cluster xx, then one of the 0s, say the lowest 0 to fix the ideas, turns into a 1.

  • Mutations. Place an exponential clock with rate γ\gamma at each (x,i)∈Δ(x,i) \in\Delta, and put

a circle ∘ at (x,i,t) at the times t the clock rings\text{a circle }\circ\text{ at }(x,i,t)\text{ at the times }t\text{ the clock rings}

to indicate that if the circle is occupied by a 1, then this 1 turns into a 2.

  • Recoveries. Place an exponential clock with rate one at each (x,i)∈Δ(x,i) \in\Delta, and put

a cross × at (x,i,t) at the times t the clock rings\text{a cross }\times\text{ at }(x,i,t)\text{ at the times }t\text{ the clock rings}

to indicate that if the cross is occupied by a 1 or a 2, then it turns into a 0.

Proof of Theorem 1

Theorem 1.a, which states extinction when the mutation rate is either very small or very large, follows directly from results in [1]. Indeed,

  • Because λ1<λc\lambda_1 < \lambda_c and ϕ1=0\phi_1 = 0, for all values of λ2\lambda_2, ϕ2\phi_2 and NN, [1 Theorem 5.b] implies that there exists γ−>0\gamma_- > 0 such that the process dies out for all γ<γ−\gamma< \gamma_-.

  • Because λ2=0\lambda_2 = 0, for all values of λ1\lambda_1, ϕ1\phi_1, ϕ2\phi_2 and NN, [1 Theorem 3.c] implies that there exists γ+<∞\gamma_+ < \infty such that the process dies out for all γ>γ+\gamma> \gamma_+.

In fact, these two theorems imply more generally extinction of the stochastic process when the infection rates λ2\lambda_2 and ϕ1\phi_1 are positive but small.

The proof of Theorem 1.b, which states survival for intermediate values of γ\gamma, and so confirms the presence of at least two phase transitions and the lack of monotonicity with respect to the mutation rate, is more complicated. We first define the space-time block

B={−e1,0,e1}×{1,2,…,N}×[0,2T]where T=ln⁡(ln⁡(N)),\mathcal{B}=\{-e_1,0,e_1\}\times\{1,2,\ldots,N\}\times[0,2T]\qquad\text{where }T=\ln(\ln(N)),

and where e1e_1 refers to the first unit vector in Zd\mathbb{Z}^d. The goal is to prove that, if there are initially many asymptomatic individuals in cluster 0, then there will be many asymptomatic individuals in cluster e1e_1 at time 2T2T with probability close to one and regardless of the graphical representation and configuration of the process outside B\mathcal{B}. Due to the lack of attractiveness of the process, there is no “worse case scenario” for the initial configuration in cluster e1e_1, so we need to deal with all the possible cases. More precisely, we let

H(x,t)=∑i1{ξt(x,i)=0}=# healthy individuals in cluster x,A(x,t)=∑i1{ξt(x,i)=1}=# asymptomatic individuals in cluster x,S(x,t)=∑i1{ξt(x,i)=2}=# symptomatic individuals in cluster x,\begin{aligned} H(x,t)&=\sum_i1\{\xi_t(x,i)=0\}=\#\text{ healthy individuals in cluster }x,\\ A(x,t)&=\sum_i1\{\xi_t(x,i)=1\}=\#\text{ asymptomatic individuals in cluster }x,\\ S(x,t)&=\sum_i1\{\xi_t(x,i)=2\}=\#\text{ symptomatic individuals in cluster }x, \end{aligned}

and introduce the following ten events:

A−={A(e1,0)<N0.6},E0={A(0,0)≥N0.4},A+={A(e1,0)≥N0.6},E1={∀t≤T, A(0,t)≥N0.2},S0={S(e1,0)=0},E2={∃t≤T: 0<S(e1,t)<N−N0.8},S−={0<S(e1,0)<N−N0.8},E3={∃t≤2T: A(e1,t)≥N0.6},S+={S(e1,0)≥N−N0.8}.E4={A(e1,2T)≥N0.4}.\begin{aligned} A_-&=\{A(e_1,0)<N^{0.6}\},& E_0&=\{A(0,0)\ge N^{0.4}\},\\ A_+&=\{A(e_1,0)\ge N^{0.6}\},& E_1&=\{\forall t\le T,\ A(0,t)\ge N^{0.2}\},\\ S_0&=\{S(e_1,0)=0\},& E_2&=\{\exists t\le T:\ 0<S(e_1,t)<N-N^{0.8}\},\\ S_-&=\{0<S(e_1,0)<N-N^{0.8}\},& E_3&=\{\exists t\le2T:\ A(e_1,t)\ge N^{0.6}\},\\ S_+&=\{S(e_1,0)\ge N-N^{0.8}\}.& E_4&=\{A(e_1,2T)\ge N^{0.4}\}. \end{aligned}

The events on the left are the possible initial configurations in cluster e1e_1. To prove the theorem, we will show that the following implications occur with probability close to one. Starting with many 1s in cluster 0 (event E0E_0), the number of 1s will remain large until time TT (event E1E_1), so if cluster e1e_1 is initially healthy, it will get infected during this time interval and contain at least one 2 by time TT (event E2E_2). If on the contrary cluster e1e_1 starts with many 2s, some of these 2s will die leaving some space in the cluster for the 1s to spread (event E2E_2). In any case, there will be again many 1s in cluster e1e_1 by time 2T2T (event E3E_3), and the number of 1s will remain large until that time (event E4E_4). The structure of the proof is summarized in Figure 2. Before proving the properties listed in the figure, we start with a preliminary result used repeatedly later.

Figure 2. Structure of the proof. In the picture, A⟶BA \longrightarrow B means that AA implies BB, while A→nBA \xrightarrow{n} B means that there is a good event GG that only depends on the graphical representation in B\mathcal{B}, with probability close to one when NN is large, and such that A∩GA \cap G implies BB, where nn refers to the lemma where this property is proved.

Lemma 3. Let X=Binomial⁡(K,p)X=\operatorname{Binomial}(K,p). Then,

P(X≤M)≤exp⁡(M−Kp(1−e−1))for all M>0.P(X\le M)\le\exp\left(M-Kp(1-e^{-1})\right)\qquad\text{for all }M>0.

Proof. Writing X=X1+⋯+XKX=X_1+\cdots+X_K as the sum of KK independent Bernoulli random variables with parameter pp, the moment generating function is given by

MX(s)=E(esX)=E(es(X1+X2+⋯+XK))=E(esX1)E(esX2)⋯E(esXK)=(P(Xi=0)+esP(Xi=1))K=(1−p(1−es))K≤exp⁡(−Kp(1−es)).\begin{aligned} M_X(s)&=E(e^{sX})=E(e^{s(X_1+X_2+\cdots+X_K)})=E(e^{sX_1})E(e^{sX_2})\cdots E(e^{sX_K})\\ &=(P(X_i=0)+e^sP(X_i=1))^K=(1-p(1-e^s))^K\\ &\le\exp(-Kp(1-e^s)). \end{aligned}

Then, taking s=−1s=-1 and using Markov’s inequality, we get

P(X≤M)=P(e−X≥e−M)≤eME(e−X)=eMMX(−1)≤exp⁡(M−Kp(1−e−1))P(X \le M)=P(e^{-X}\ge e^{-M})\le e^{M}E(e^{-X}) = e^{M}M_X(-1)\le\exp\left(M-Kp(1-e^{-1})\right)

for all M>0M>0. This completes the proof. □

We can now prove the five lemmas listed in Figure 2.

Lemma 4 (Keep some 1s alive at 0). For all ϵ>0\epsilon>0, there exists an event G1G_1 measurable with respect to the graphical representation in B\mathcal{B} such that

E0∩G1⊂E1andP(G1)≥1−ϵ/10 for all N large.E_0\cap G_1\subset E_1 \quad\text{and}\quad P(G_1)\ge1-\epsilon/10\text{ for all }N\text{ large}.

Proof. Note that a 1 at a given site can only be removed due to a mutation mark ∘\circ or a death mark ×\times occurring together at rate γ+1\gamma+1. In particular, on the event E0E_0, the number of 1s initially present at cluster 0 that remain 1 until at least time TT dominates

X=Binomial⁡(N0.4,p)wherep=e−(γ+1)T=e−(γ+1)ln⁡(ln⁡(N))=(ln⁡(N))−(γ+1).X=\operatorname{Binomial}(N^{0.4},p)\quad\text{where}\quad p=e^{-(\gamma+1)T}=e^{-(\gamma+1)\ln(\ln(N))}=(\ln(N))^{-(\gamma+1)}.

In particular, applying Lemma 3 with M=N0.2M=N^{0.2}, for all NN large, we get

P(E1c∣E0)=P(∃ t≤T:A(0,t)<N0.2∣E0)≤P(X≤N0.2)≤exp⁡(N0.2−N0.4(1−e−1)(ln⁡(N))−(γ+1))≤ϵ/10.P(E_1^{\mathrm{c}\mid E_0})=P(\exists\,t\le T:A(0,t)<N^{0.2}\mid E_0) \le P(X\le N^{0.2})\le\exp\left(N^{0.2}-N^{0.4}(1-e^{-1})(\ln(N))^{-(\gamma+1)}\right)\le\epsilon/10.

Because these estimates only depend on the graphical representation in the block B\mathcal{B}, this implies the existence of a good event G1G_1 that satisfies the lemma. □

Lemma 5 (Send a 2 onto e1e_1). For all ϵ>0\epsilon>0, there exists an event G2G_2 measurable with respect to the graphical representation in B\mathcal{B} such that

E1∩S0∩G2⊂E2andP(G2)≥1−ϵ/10 for all N large.E_1\cap S_0\cap G_2\subset E_2 \quad\text{and}\quad P(G_2)\ge1-\epsilon/10\text{ for all }N\text{ large}.

Proof. To begin with, we denote respectively by

H=∫0T1{H(e1,s)=N} dsandY=Poisson⁡(γT/2)H=\int_0^T \mathbf{1}\{H(e_1,s)=N\}\,\mathrm{d}s \quad\text{and}\quad Y=\operatorname{Poisson}(\gamma T/2)

the amount of time cluster e1e_1 is fully healthy by time TT and the Poisson random variable with parameter γT/2\gamma T/2. On the event E1E_1, cluster e1e_1 gets infected at rate at least λ1N0.2\lambda_1N^{0.2}, while infected individuals recover at rate one, therefore E(H)E(H) is at most T/(1+λ1N0.2)T/(1+\lambda_1N^{0.2}). In addition, each 1 turns into a 2 at rate γ\gamma. In particular, conditioning on whether HH is larger or smaller than T/2T/2, and using Markov’s inequality, we deduce that, for all NN large,

P(E2c∣E1∩S0)=P(∀t≤T,S(e1,t)=0∣E1∩S0)≤P(H≥T/2)+P(Y=0)≤2E(H)/T+e−γT/2\begin{aligned} P(E_2^c\mid E_1\cap S_0)=P(\forall t\leq T,S(e_1,t)=0\mid E_1\cap S_0) \\ &\leq P(H\geq T/2)+P(Y=0)\leq2E(H)/T+e^{-\gamma T/2} \end{aligned}
≤2/(1+λ1N0.2)+(ln⁡(N))−γ/2≤ϵ/10.\leq2/(1+\lambda_1N^{0.2})+(\ln(N))^{-\gamma/2}\leq\epsilon/10.

As previously, because these estimates only depend on the graphical representation of the process in the space-time block B\mathcal{B}, the lemma follows. □\square

Lemma 6 (Kill some 2s at e1e_1). For all ϵ>0\epsilon>0, there exists an event G3G_3 measurable with respect to the graphical representation in B\mathcal{B} such that

S+∩G3⊂E2andP(G3)≥1−ϵ/10 for all N large.S_+\cap G_3\subset E_2\quad\text{and}\quad P(G_3)\geq1-\epsilon/10\text{ for all }N\text{ large}.

Proof. Note that a site initially occupied by a 2 at which there is at least one death mark ×\times but no mutation mark ∘\circ between time 0 and time 1 is not occupied by a 2 at time 1. In particular, on the event S+S_+, the number of 2s at cluster e1e_1 replaced by a 0 or a 1 at time 1 dominates

X=Binomial⁡(N−N0.8,p)wherep=(1−e−1)e−γ.X=\operatorname{Binomial}(N-N^{0.8},p)\quad\text{where}\quad p=(1-e^{-1})e^{-\gamma}.

Applying again Lemma 3 but with M=N0.8M=N^{0.8}, we get

P(E2c∣S+)≤P(S(e1,1)≥N−N0.8∣S+)P(E_2^c\mid S_+)\leq P(S(e_1,1)\geq N-N^{0.8}\mid S_+)
≤P(X≤N0.8)≤exp⁡(N0.8−(N−N0.8)(1−e−1)2e−γ)≤ϵ/10\leq P(X\leq N^{0.8})\leq\exp\left(N^{0.8}-(N-N^{0.8})(1-e^{-1})^2e^{-\gamma}\right)\leq\epsilon/10

for all NN large. This completes the proof. □\square

Lemma 7 (Produce many 1s at e1e_1). For all ϵ>0\epsilon>0, there exists an event G4G_4 measurable with respect to the graphical representation in B\mathcal{B} such that

E2∩A−∩G4⊂E3andP(G4)≥1−ϵ/10 for all N large.E_2\cap A_-\cap G_4\subset E_3\quad\text{and}\quad P(G_4)\geq1-\epsilon/10\text{ for all }N\text{ large}.

Proof. On the event E2E_2, with probability one,

τ=inf⁡{t:0<S(e1,t)<N−N0.8}<Tandσ=τ+N−0.2<2T.\tau=\inf\{t:0<S(e_1,t)<N-N^{0.8}\}<T\quad\text{and}\quad\sigma=\tau+N^{-0.2}<2T.

If there are more than N0.6N^{0.6} asymptomatic individuals at e1e_1 at time τ\tau, we are done. Otherwise, we will show that, at time σ\sigma, the number of 1s at e1e_1 exceeds N0.6N^{0.6}, and so the event E3E_3 occurs, with probability close to one. By our assumption and the definition of time τ\tau,

  • S(e1,τ)>0⇒ξτ(e1,i0)=2S(e_1,\tau)>0\Rightarrow\xi_\tau(e_1,i_0)=2 for some level i0∈{1,2,…,N}i_0\in\{1,2,\ldots,N\},

  • S(e1,τ)<N−N0.8S(e_1,\tau)<N-N^{0.8} and A(e1,τ)<N0.6⇒H(e1,τ)≥N0.8−N0.6≥N0.7A(e_1,\tau)<N^{0.6}\Rightarrow H(e_1,\tau)\geq N^{0.8}-N^{0.6}\geq N^{0.7}.

To prove that many infections originate from the 2 at level i0i_0, we introduce the event

Bi0={no × and at least N0.7 type 2 ∙ in (e1,i0)×[τ,σ]}.B_{i_0}=\{\text{no }\times\text{ and at least }N^{0.7}\text{ type 2 }\bullet\text{ in }(e_1,i_0)\times[\tau,\sigma]\}.

Let ρ\rho be the time it takes starting from τ\tau to have N0.7N^{0.7} type 2 ∙\bullet at site (e1,i0)(e_1,i_0). Using that the death marks ×\times occur at rate one, and that, because the internal infections 2-∙\bullet occur at rate ϕ2=N\phi_2=N, time ρ\rho has expected value N0.7/N=N−0.3N^{0.7}/N=N^{-0.3}, it follows from Markov’s inequality that

P(Bi0c)≤P(at least one × in (e1,i0)×[τ,σ])+P(ρ≥σ−τ)≤(1−e−N−0.2)+P(ρ≥N−0.2)≤N−0.2+N0.2E(ρ)=N−0.2+N0.2N−0.3=N−0.2+N−0.1.(4)\begin{aligned} P(B_{i_0}^{c}) &\le P(\text{at least one }\times\text{ in }(e_1,i_0)\times[\tau,\sigma])+P(\rho\ge\sigma-\tau)\\ &\le(1-e^{-N^{-0.2}})+P(\rho\ge N^{-0.2})\\ &\le N^{-0.2}+N^{0.2}E(\rho)=N^{-0.2}+N^{0.2}N^{-0.3}=N^{-0.2}+N^{-0.1}. \tag*{(4)} \end{aligned}

Now, on the event Bi0B_{i_0}, there are at least N0.7N^{0.7} potential infections in the time interval [τ,σ][\tau,\sigma], and because there are at least N0.7N^{0.7} healthy individuals at time τ\tau, there must be N0.7N^{0.7} successful infections in the time interval indeed turning a 0 into a 1. These successful infections occur in a set Λ\Lambda of at most N0.7N^{0.7} sites. Because mutations ∘\circ and deaths ×\times occur together at rate γ+1\gamma+1, the number of mutations and deaths in Λ×[τ,σ]\Lambda\times[\tau,\sigma] is dominated by

Y=Poisson⁡(μ)whereμ=(γ+1)(σ−τ)N0.7=(γ+1)N0.5.Y=\operatorname{Poisson}(\mu)\quad\text{where}\quad\mu=(\gamma+1)(\sigma-\tau)N^{0.7}=(\gamma+1)N^{0.5}.

In particular, using again Markov’s inequality shows that the number of mutations and deaths in the space-time region Λ×[τ,σ]\Lambda\times[\tau,\sigma] exceeds N0.6N^{0.6} with probability at most

P(Y≥N0.6)≤N−0.6E(Y)=N−0.6(γ+1)N0.5=(γ+1)N−0.1.P(Y\ge N^{0.6})\le N^{-0.6}E(Y)=N^{-0.6}(\gamma+1)N^{0.5}=(\gamma+1)N^{-0.1}.

while if there are at most N0.6N^{0.6} mutations and deaths, then

A(e1,σ)≥(#1s created−#1s removed) in Λ×[τ,σ]≥N0.7−N0.6≥N0.6,(5)A(e_1,\sigma)\ge(\#1\text{s created}-\#1\text{s removed})\text{ in }\Lambda\times[\tau,\sigma]\ge N^{0.7}-N^{0.6}\ge N^{0.6}, \tag*{(5)}

and so the event E3E_3 occurs. Combining (4)–(5), we conclude that

P(E3c∣E2∩A−)≤P(Bi0c)+P(Y≥N0.6)≤N−0.2+(γ+2)N−0.1≤ϵ/10P(E_3^c\mid E_2\cap A_-)\le P(B_{i_0}^c)+P(Y\ge N^{0.6})\le N^{-0.2}+(\gamma+2)N^{-0.1}\le\epsilon/10

for all NN large. This completes the proof. □\square

Lemma 8 (Keep some 1s alive at e1e_1). For all ϵ>0\epsilon>0, there exists an event G5G_5 measurable with respect to the graphical representation in B\mathcal{B} such that

A+∩G5⊂E3∩G5⊂E4andP(G5)≥1−ϵ/10 for all N large.A_+\cap G_5\subset E_3\cap G_5\subset E_4\quad\text{and}\quad P(G_5)\ge1-\epsilon/10\text{ for all }N\text{ large}.

Proof. Because A+⊂E3A_+\subset E_3, the first inclusion is obvious. To prove the second inclusion, note that, on the event E3E_3, we can find a set of N0.6N^{0.6} asymptomatic individuals at cluster e1e_1 at some time t≤2Tt\le2T, and if at least N0.4N^{0.4} of these individuals remain asymptomatic for 2T−t≤2T2T-t\le2T units of time, then the event E4E_4 occurs. Now, using the same argument as in the proof of Lemma 4, the number of 1s at time tt that remain 1 until at least time 2T2T dominates

X=Binomial⁡(N0.6,p)wherep=e−2(γ+1)T=e−2(γ+1)ln⁡(ln⁡(N))=(ln⁡(N))−2(γ+1).X=\operatorname{Binomial}(N^{0.6},p)\quad\text{where}\quad p=e^{-2(\gamma+1)T}=e^{-2(\gamma+1)\ln(\ln(N))}=(\ln(N))^{-2(\gamma+1)}.

Then, applying Lemma 3 with M=N0.4M=N^{0.4}, we get

P(E4c∣E3)≤P(X≤N0.4)≤exp⁡(N0.4−N0.6(1−e−1)(ln⁡(N))−2(γ+1))≤ϵ/10P(E_4^c\mid E_3)\le P(X\le N^{0.4})\le\exp\left(N^{0.4}-N^{0.6}(1-e^{-1})(\ln(N))^{-2(\gamma+1)}\right)\le\epsilon/10

for all NN large. This completes the proof. □\square

Using the previous lemmas, we can now prove the theorem.

Proof of Theorem 1.b. To deduce survival, we now use a block construction and couple the process with a 2-dependent oriented site percolation model. Define the lattice

L={(m,n)∈Z×Z+:m+n is even},\mathcal{L}=\{(m,n)\in\mathbb{Z}\times\mathbb{Z}_+ : m+n\text{ is even}\},

which we turn into a directed graph L⃗\vec{\mathcal{L}} by placing arrows

(m,n)→(m′,n′)⟺∣m−m′∣=1 and n′=n+1.(m,n) \to(m',n') \Longleftrightarrow|m-m'|=1 \text{ and } n'=n+1.

Oriented site percolation with parameter pp and range RR is obtained by assuming that sites are open with probability pp and closed with probability 1−p1-p, and that sites that are more than graph distance RR apart are independent. Using a contour argument [4 Section 10] shows that

p>1−3−4(2R+1)2⟹P(∣C0∣=∞)>0,p > 1-3^{-4(2R+1)^2} \quad\Longrightarrow\quad P(|C_0|=\infty)>0,

where C0C_0 refers to the cluster of sites that can be reached from the origin by a directed path of open sites. Returning to our process, we let

Bm,n=(me1,2nT)+B∀ (m,n)∈L,\mathcal{B}_{m,n}=(me_1,2nT)+\mathcal{B} \qquad\forall\ (m,n)\in\mathcal{L},

take ϵ<3−4(2R+1)2\epsilon<3^{-4(2R+1)^2} with R=2R=2 in order to have

Bm,n∩Bm′,n′=∅whenever∣m−m′∣∨∣n−n′∣>R,\mathcal{B}_{m,n}\cap\mathcal{B}_{m',n'}=\varnothing\qquad\text{whenever}\quad|m-m'|\vee|n-n'|>R,

and say that (m,n)∈L(m,n)\in\mathcal{L} is a good site if the event

Em,n={cluster me1 has at least N0.4 asymptomatic individuals at time 2nT}={A(me1,2nT)≥N0.4}\begin{aligned} E_{m,n}&=\{\text{cluster }me_1\text{ has at least }N^{0.4}\text{ asymptomatic individuals at time }2nT\}\\ &=\{A(me_1,2nT)\geq N^{0.4}\} \end{aligned}

occurs. Note that Em,nE_{m,n} is the space-time translation of E0E_0 by vector (me1,2nT)(me_1,2nT). Now, define the good event G=G1∩⋯∩G5G=G_1\cap\cdots\cap G_5. Then, GG is measurable with respect to the graphical representation in the block space-time B\mathcal{B}. In addition, it follows from Lemmas 4–8 that

E0∩G⊂E1∩Gby Lemma 4⊂(E1∩S0∩G)∪(S−∩G)∪(S+∩G)by {S0,S−,S+} partition⊂(E2∩G)∪(S−∩G)∪(S+∩G)by Lemma 5⊂E2∩Gby Lemma 6⊂(E2∩A−∩G)∪(A+∩G)by {A−,A+} partition⊂(E3∩G)∪(A+∩G)by Lemma 7⊂E4=E1,1by Lemma 8.(6)\begin{aligned} E_0\cap G&\subset E_1\cap G &&\text{by Lemma 4}\\ &\subset(E_1\cap S_0\cap G)\cup(S_-\cap G)\cup(S_+\cap G) &&\text{by }\{S_0,S_-,S_+\}\text{ partition}\\ &\subset(E_2\cap G)\cup(S_-\cap G)\cup(S_+\cap G) &&\text{by Lemma 5}\\ &\subset E_2\cap G &&\text{by Lemma 6}\\ &\subset(E_2\cap A_-\cap G)\cup(A_+\cap G) &&\text{by }\{A_-,A_+\}\text{ partition}\\ &\subset(E_3\cap G)\cup(A_+\cap G) &&\text{by Lemma 7}\\ &\subset E_4=E_{1,1} &&\text{by Lemma 8.} \tag*{(6)} \end{aligned}

Taking the complement, using the union bound, and applying the second part of Lemmas 4–8 also imply that, for all NN sufficiently large,

P(G)=1−P(⋃i=15Gic)≥1−∑i=15P(Gic)≥1−5(ϵ/10)=1−ϵ/2.(7)P(G)=1-P\left(\bigcup_{i=1}^{5}G_i^c\right)\geq1-\sum_{i=1}^{5}P(G_i^c)\geq1-5(\epsilon/10)=1-\epsilon/2. \tag*{(7)}

Finally, using symmetry and translation invariance in space and time of the graphical representation, and combining (6)–(7), we deduce the existence of a collection of good events Gm,nG_{m,n}, measurable with respect to the graphical representation in the block Bm,n\mathcal{B}_{m,n}, such that

Em,n∩Gm,n⊂Em−1,n+1∩Em+1,n+1andP(Gm,n)≥1−ϵ for all N large.E_{m,n}\cap G_{m,n}\subset E_{m-1,n+1}\cap E_{m+1,n+1}\quad\text{and}\quad P(G_{m,n})\geq1-\epsilon\text{ for all }N\text{ large}.

This and [5 Theorem 4.3] imply that the good sites dominate the open sites in a 2-dependent oriented site percolation process with parameter 1−ϵ1-\epsilon. In particular, by our choice of ϵ\epsilon, the open sites percolate, so the good sites percolate as well, which proves survival. □

Proof of Theorem 2

As previously, the first part of the theorem follows directly from [1].

  • Because λ1=0\lambda_1=0, for all values of λ2\lambda_2, ϕ1\phi_1, ϕ2\phi_2 and NN, [1 Theorem 3.b] implies that there exists γ−>0\gamma_->0 such that the process dies out for all γ<γ−\gamma<\gamma_-.

  • Because λ2<λc\lambda_2<\lambda_c and ϕ2=0\phi_2=0, for all values of λ1\lambda_1, ϕ1\phi_1 and NN, [1 Theorem 5.c] implies that there exists γ+<∞\gamma_+<\infty such that the process dies out for all γ>γ+\gamma>\gamma_+.

These two theorems again imply more generally extinction of the stochastic process when the infection rates λ1\lambda_1 and ϕ2\phi_2 are positive but small.

Theorem 2.b, which states survival for intermediate values of the mutation rate and the presence of at least two phase transitions, but for other values of the infection rates, again relies on a block construction, though the general strategy strongly differs from Theorem 1.b. More precisely, the space-time block B\mathcal{B} is now defined using a much larger time scale:

B={−e1,0,e1}×{1,2,…,N}×[0,T]whereT=N0.4,\mathcal{B}=\{-e_1,0,e_1\}\times\{1,2,\ldots,N\}\times[0,T]\quad\text{where}\quad T=N^{0.4},

and the goal is to prove that, if we start with at least one asymptomatic individual in cluster 0, then there will be at least one asymptomatic individual in cluster e1e_1 at time TT with probability close to one when the cluster size NN is large, regardless of the graphical representation and configuration of the process outside B\mathcal{B}. As previously, there is no “worst-case scenario” for the initial configuration in cluster e1e_1 so we must deal with all the possible cases. Define the seven events

A0={A(e1,0)=0},E0={A(0,0)≠0},A_0=\{A(e_1,0)=0\},\qquad E_0=\{A(0,0)\ne0\},
A+={A(e1,0)≠0},E1={∀t≤T, A(0,t)≠0},A_+=\{A(e_1,0)\ne0\},\qquad E_1=\{\forall t\le T,\ A(0,t)\ne0\},
E2={∃t≤T:H(e1,t)=N},E_2=\{\exists t\le\sqrt{T}:H(e_1,t)=N\},
E3={∃t≤T:A(e1,t)≠0},E_3=\{\exists t\le T:A(e_1,t)\ne0\},
E4={A(e1,T)≠0}.E_4=\{A(e_1,T)\ne0\}.

To prove the theorem, we will first show that, with probability close to one when the cluster size is large, starting with at least one 1 in cluster 0 (event E0E_0), there will be at least one 1 in the cluster at all times until time TT (event E1E_1). In addition, if cluster e1e_1 starts with no 1s, it will become healthy by time T\sqrt{T} (event E2E_2). Then, either cluster e1e_1 starts with at least one 1 (event E3E_3), or it starts with no 1s, in which case many 1s in cluster 0 turn into a 2, and one of these 2s will infect cluster e1e_1 between time T\sqrt{T} and time TT (event E3E_3). Once we have a 1 in cluster e1e_1, there will be at least one 1 in the cluster at all times until time TT (event E4E_4). The structure of the proof is summarized in Figure 3, and consists of the following four lemmas.

Figure 3. Structure of the proof. In the picture, A⟶BA \longrightarrow B means that AA implies BB, while A⟶nBA \mathrel{\mathop{\longrightarrow}^{n}} B means that there is a good event GG that only depends on the graphical representation in B\mathcal{B}, with probability close to one when NN is large, and such that A∩GA \cap G implies BB, where nn refers to the lemma where this property is proved.

Lemma 9 (Keep a 1 alive at 0). For all ϵ>0\epsilon>0, there exists an event G1G_1 measurable with respect to the graphical representation in B\mathcal{B} such that

E0∩G1⊂E1andP(G1)≥1−ϵ/8for all N large.E_0\cap G_1\subset E_1\quad\text{and}\quad P(G_1)\ge1-\epsilon/8\quad\text{for all }N\text{ large}.

Proof. The idea is to define a set-valued process Ξt\Xi_t that keeps track of a subset of asymptomatic individuals as follows. The process starts at

Ξ0={i1}wherei1=min⁡{i:ξ0(0,i)=1}.(8)\Xi_0=\{i_1\}\quad\text{where}\quad i_1=\min\{i:\xi_0(0,i)=1\}. \tag*{(8)}

Note that such a level i1i_1 exists because A(0,0)≠0A(0,0) \ne0. Then, define

σ1=time of the first death × in the cluster at a level ≠i0,\sigma_1 = \text{time of the first death } \times\text{ in the cluster at a level } \ne i_0,
σ2=time of the first internal infection 1−∙ after time σ1 at site (0,i1),\sigma_2 = \text{time of the first internal infection } 1\mathbin{-}\bullet\text{ after time } \sigma_1 \text{ at site } (0,i_1),
σ3=time of the first mutation ∘ or death mark × at site (0,i1).\sigma_3 = \text{time of the first mutation } \circ\text{ or death mark } \times\text{ at site } (0,i_1).

On the event that σ3<σ2\sigma_3 < \sigma_2, we define

Ξt={i1}  ∀t∈[0,σ3),Ξt=∅  ∀t∈[σ3,∞),\Xi_t = \{i_1\}\ \ \forall t \in[0,\sigma_3), \qquad\Xi_t = \varnothing\ \ \forall t \in[\sigma_3,\infty),

to indicate that there might be no 1s left in the cluster starting from time σ3\sigma_3. In contrast, on the event that σ3>σ2\sigma_3 > \sigma_2, there is at least one healthy level at time σ1\sigma_1 and the 1 at level i1i_1 attempts to infect a healthy level at time σ2>σ1\sigma_2 > \sigma_1. This forces the existence of a healthy level, say j1j_1, becoming infected at some time σ4\sigma_4 in the time interval (σ1,σ2](\sigma_1,\sigma_2], and we put

σ5=time of the first ∘ or × at level i1 or j1 after time σ4.\sigma_5 = \text{time of the first } \circ\text{ or } \times\text{ at level } i_1 \text{ or } j_1 \text{ after time } \sigma_4.

Letting i2∈{i1,j1}i_2 \in\{i_1,j_1\} be the level where the ∘\circ or ×\times occurs, we define

Ξt={i1}  ∀t∈[0,σ4),Ξt={i1,j1}  ∀t∈[σ4,σ5),Ξσ5={i2}.\Xi_t = \{i_1\}\ \ \forall t \in[0,\sigma_4), \qquad\Xi_t = \{i_1,j_1\}\ \ \forall t \in[\sigma_4,\sigma_5), \qquad\Xi_{\sigma_5} = \{i_2\}.

Then, the process is defined by reiterating the same procedure but starting from {i2}\{i_2\} at time σ5\sigma_5 rather than from {i1}\{i_1\} at time 0, and so on. This continues until the asymptomatic individual at level iKi_K mutates or dies before trying to infect another site, at which time the process gets trapped in state ∅\varnothing. By construction, each level in Ξt\Xi_t is in state 1 at time tt, therefore

A(0,t)≥card⁡(Ξt)andτ=inf⁡{t:A(0,t)=0}≥inf⁡{t:Ξt=∅}=σ.(9)A(0,t) \ge\operatorname{card}(\Xi_t) \qquad\text{and} \qquad\tau= \inf\{t : A(0,t)=0\} \ge\inf\{t : \Xi_t=\varnothing\} = \sigma. \tag*{(9)}

Now, by the memoryless and superposition properties,

P(σ3>σ2)=P(σ3>σ2∣σ3>σ1)P(σ3>σ1)=(NN+γ+1)(N−1N+γ)≥1−N−0.8P(\sigma_3 > \sigma_2) = P(\sigma_3 > \sigma_2 \mid\sigma_3 > \sigma_1)P(\sigma_3 > \sigma_1) = \left(\frac{N}{N+\gamma+1}\right)\left(\frac{N-1}{N+\gamma}\right) \ge1-N^{-0.8}

for all NN large, so the number of steps KK until the process hits ∅\varnothing dominates stochastically the geometric random variable Y=Geometric⁡(N−0.8)Y = \operatorname{Geometric}(N^{-0.8}), and we have

P(K≤N0.6)≤P(Y≤N0.6)=1−P(Y>N0.6)=1−(1−N−0.8)N0.6≤1−exp⁡(−2N−0.2)≤2N−0.2.(10)\begin{aligned} P(K \le N^{0.6}) \le P(Y \le N^{0.6}) = 1-P(Y>N^{0.6}) \\ &= 1-(1-N^{-0.8})^{N^{0.6}} \le1-\exp(-2N^{-0.2}) \le2N^{-0.2}. \tag*{(10)} \end{aligned}

Also, because σ5≥σ5−σ4=Exponential⁡(2(γ+1))\sigma_5 \ge\sigma_5-\sigma_4 = \operatorname{Exponential}(2(\gamma+1)), the time to extinction σ\sigma dominates

Z1+Z2+⋯+ZKwithZi=Exponential⁡(2(γ+1)) independent,Z_1+Z_2+\cdots+Z_K \quad\text{with} \quad Z_i=\operatorname{Exponential}(2(\gamma+1))\text{ independent},

from which it follows that, for all s>0s>0,

E(e−sσ∣K>n)≤E(e−s(Z1+Z2+⋯+Zn))=∏i=1nE(e−sZi)=(1+s2(γ+1))−n.E(e^{-s\sigma}\mid K>n)\le E(e^{-s(Z_1+Z_2+\cdots+Z_n)})=\prod_{i=1}^{n}E(e^{-sZ_i})=\left(1+\frac{s}{2(\gamma+1)}\right)^{-n}.

In particular, taking s=2(γ+1)(e−1)>0s=2(\gamma+1)(e-1)>0, we get

E(e−2(γ+1)(e−1)σ∣K>n)≤(1+2(γ+1)(e−1)2(γ+1))−n=e−n.(11)E(e^{-2(\gamma+1)(e-1)\sigma}\mid K>n)\le\left(1+\frac{2(\gamma+1)(e-1)}{2(\gamma+1)}\right)^{-n}=e^{-n}. \tag*{(11)}

Using (8)–(10) and applying Markov’s inequality, we conclude that

P(E2c∣E1)=P(τ≤N0.4)≤P(σ≤N0.4)≤P(K≤N0.6)+P(σ≤N0.4∣K>N0.6)≤2N−0.2+P(e−2(γ+1)(e−1)σ≥e−2(γ+1)(e−1)N0.4∣K>N0.6)≤2N−0.2+eN0.4E(e−2(γ+1)(e−1)σ∣K>N0.6)≤2N−0.2+exp⁡(N0.4−N0.6)≤ϵ/8\begin{aligned} P(E_2^c\mid E_1)&=P(\tau\le N^{0.4})\le P(\sigma\le N^{0.4})\\ &\le P(K\le N^{0.6})+P(\sigma\le N^{0.4}\mid K>N^{0.6})\\ &\le2N^{-0.2}+P(e^{-2(\gamma+1)(e-1)\sigma}\ge e^{-2(\gamma+1)(e-1)N^{0.4}}\mid K>N^{0.6})\\ &\le2N^{-0.2}+e^{N^{0.4}}E(e^{-2(\gamma+1)(e-1)\sigma}\mid K>N^{0.6})\\ &\le2N^{-0.2}+\exp(N^{0.4}-N^{0.6})\le\epsilon/8 \end{aligned}

for all NN large. This completes the proof. □

Lemma 10 (Kill all the 2s at e1e_1). For all ϵ>0\epsilon>0, there exists an event G2G_2 measurable with respect to the graphical representation in B\mathcal{B} such that

A0∩G2⊂E2andP(G2)≥1−ϵ/8 for all N large.A_0\cap G_2\subset E_2\quad\text{and}\quad P(G_2)\ge1-\epsilon/8\text{ for all }N\text{ large}.

Proof. The idea is to study the time it takes to have at least one death mark ×\times at each level of the cluster. To this end, we let τ0=0\tau_0=0 and

τ1<τ2<⋯<τN\tau_1<\tau_2<\cdots<\tau_N

be the times of the first ×\times at the NN levels of cluster e1e_1, ordered from the smallest to the largest. Notice that the cluster becomes healthy by time τN\tau_N in general, and exactly at time τN\tau_N when starting from NN symptomatic individuals. Now, by the memoryless and superposition properties for the exponential distribution,

E(τN)=∑i=0N−1E(τi+1−τi)=11+12+13+⋯+1N≤1+ln⁡(N).E(\tau_N)=\sum_{i=0}^{N-1}E(\tau_{i+1}-\tau_i)=\frac{1}{1}+\frac{1}{2}+\frac{1}{3}+\cdots+\frac{1}{N}\le1+\ln(N).

In particular, using Markov’s inequality, we conclude that

P(E2c∣A0)≤P(τN≥N0.2)≤N−0.2E(τN)≤N−0.2(1+ln⁡(N))≤ϵ/8P(E_2^c\mid A_0)\le P(\tau_N\ge N^{0.2})\le N^{-0.2}E(\tau_N)\le N^{-0.2}(1+\ln(N))\le\epsilon/8

for all NN large. This completes the proof. □

Lemma 11 (Send a 1 onto e1e_1). For all ϵ>0\epsilon>0, there exists an event G3G_3 measurable with respect to the graphical representation in B\mathcal{B} such that

E1∩E2∩G3⊂E3andP(G3)≥1−ϵ/8 for all N large.E_1\cap E_2\cap G_3\subset E_3\quad\text{and}\quad P(G_3)\ge1-\epsilon/8\text{ for all }N\text{ large}.

Proof. Because, on the event E1E_1, cluster 0 is occupied by at least one 1 until time TT, the number of times a 1 turns into a 2 between time T\sqrt{T} and time TT dominates

Y1=Poisson⁡(γ(T−T)).Y_1 = \operatorname{Poisson}\left(\gamma(T-\sqrt{T})\right).

Moreover, each such 2 attempts to infect cluster e1e_1 at rate λ2\lambda_2 and recovers at rate one, so the probability of at least one attempt is λ2/(λ2+1)\lambda_2/(\lambda_2+1). In particular, by the thinning property, the number of attempts to infect cluster e1e_1 during the time window [T,T][\sqrt{T},T] dominates

Y2=Poisson⁡(λ2λ2+1γ(T−T))=Poisson⁡(λ2λ2+1γ(N0.4−N0.2)).Y_2 = \operatorname{Poisson}\left(\frac{\lambda_2}{\lambda_2+1}\gamma(T-\sqrt{T})\right) = \operatorname{Poisson}\left(\frac{\lambda_2}{\lambda_2+1}\gamma(N^{0.4}-N^{0.2})\right).

Now, on the event E2E_2 that cluster e1e_1 is healthy by time T=N0.2\sqrt{T}=N^{0.2}, the cluster remains healthy until time TT only if no 2s attempt to infect the cluster until that time, so

P(E3c∣E1∩E2)≤P(Y2=0)=exp⁡(−λ2λ2+1γ(N0.4−N0.2))≤ϵ/8P(E_3^c \mid E_1\cap E_2) \le P(Y_2=0)=\exp\left(-\frac{\lambda_2}{\lambda_2+1}\gamma(N^{0.4}-N^{0.2})\right)\le\epsilon/8

for all NN large. This completes the proof. □

Lemma 12 (Keep a 1 alive at e1e_1). For all ϵ>0\epsilon>0, there exists an event G4G_4 measurable with respect to the graphical representation in B\mathcal{B} such that

A+∩G4⊂E3∩G4⊂E4andP(G4)≥1−ϵ/8 for all N large.A_+\cap G_4\subset E_3\cap G_4\subset E_4 \quad\text{and}\quad P(G_4)\ge1-\epsilon/8\text{ for all }N\text{ large}.

Proof. Because A+⊂E3A_+\subset E_3, the first inclusion is obvious, while Lemma 9, together with the translation invariance of the graphical representation, gives the second inclusion. □

Using the previous lemmas, we can now prove the theorem.

Proof of Theorem 2.b. As previously, we couple the process with 2-dependent oriented site percolation on the directed graph L⃗\vec{\mathcal{L}}. We define

Bm,n=(me1,nT)+B∀ (m,n)∈L\mathcal{B}_{m,n}=(me_1,nT)+\mathcal{B}\quad\forall\,(m,n)\in\mathcal{L}

and take ϵ<3−4(2R+1)2\epsilon<3^{-4(2R+1)^2} with R=2R=2, but now say that (m,n)∈L(m,n)\in\mathcal{L} is a good site if

Em,n={cluster me1 has at least one asymptomatic individual at time nT}={A(me1,nT)≠0}E_{m,n}=\{\text{cluster }me_1\text{ has at least one asymptomatic individual at time }nT\} =\{A(me_1,nT)\ne0\}

occurs. As previously, the event G=G1∩⋯∩G4G=G_1\cap\cdots\cap G_4 is measurable with respect to the graphical representation in the block B\mathcal{B}, and it follows from Lemmas 9–12 that

E0∩G⊂E1∩Gby Lemma 9⊂(E0∩A0∩G)∪(A+∩G)by {A0,A+} partition⊂(E1∩E2∩G)∪(A+∩G)by Lemma 10⊂(E3∩G)∪(A+∩G)by Lemma 11⊂E3∩G⊂E4=E1,1by Lemma 12,(12)\begin{aligned} E_0\cap G &\subset E_1\cap G &&\text{by Lemma 9}\\ &\subset(E_0\cap A_0\cap G)\cup(A_+\cap G) &&\text{by }\{A_0,A_+\}\text{ partition}\\ &\subset(E_1\cap E_2\cap G)\cup(A_+\cap G) &&\text{by Lemma 10}\\ &\subset(E_3\cap G)\cup(A_+\cap G) &&\text{by Lemma 11}\\ &\subset E_3\cap G\subset E_4=E_{1,1} &&\text{by Lemma 12,} \tag*{(12)} \end{aligned}

and that, for all NN sufficiently large,

P(G)=1−P(⋃i=14Gic)≥1−∑i=14P(Gic)≥1−4(ϵ/8)=1−ϵ/2.(13)P(G)=1-P\left(\bigcup_{i=1}^{4}G_i^c\right)\ge1-\sum_{i=1}^{4}P(G_i^c)\ge1-4(\epsilon/8)=1-\epsilon/2. \tag*{(13)}

Using again symmetry and translation invariance in space and time of the graphical representation, and combining (12)–(13), we deduce the existence of a collection of good events Gm,nG_{m,n}, measurable with respect to the graphical representation in the block Bm,n\mathcal{B}_{m,n}, such that

Em,n∩Gm,n⊂Em−1,n+1∩Em+1,n+1andP(Gm,n)≥1−ϵ for all N large.E_{m,n} \cap G_{m,n} \subset E_{m-1,n+1} \cap E_{m+1,n+1} \quad\text{and} \quad P(G_{m,n}) \ge1-\epsilon\text{ for all } N \text{ large.}

As in the proof of the Theorem 1.b, survival follows from [5 Theorem 4.3]. □

Faculty of Exact Sciences and Computer Science University of Abdelhamid Ibn Badis Mostaganem, Algeria [email protected]

School of Mathematical and Statistical Sciences Arizona State University Tempe, AZ 85287, USA. [email protected]

References

  1. [1]K. Belatreche, L. Belhadji and N. Lanchier. The contact process with social clusters and asymptomatic states. Available as arXiv:2610.04323.arxiv.org/abs/2610.04323
  2. [2]L. Belhadji and N. Lanchier. Individual versus cluster recoveries within a spatially structured population. Ann. Appl. Probab. 16 403–422, 2006.
  3. [3]L. Belhadji, N. Lanchier and M. Mercer. The contact process with an asymptomatic state. Stochastic Process. Appl. 176 104417, 2024.arxiv.org/abs/2308.00731
  4. [4]R. Durrett. Oriented percolation in two dimensions. Ann. Probab., 12(4):999–1040, 1984.DOI
  5. [5]R. Durrett. Ten lectures on particle systems. In Lectures on probability theory (Saint-Flour, 1993), volume 1608 of Lecture Notes in Math., pages 97–201. Springer, Berlin, 1995.DOI
  6. [6]T. E. Harris. Nearest-neighbor Markov interaction processes on multidimensional lattices. Adv. Math., 9:66–89, 1972.DOI
  7. [7]T. E. Harris. Contact interactions on a lattice. Ann. Probab., 2:969–988, 1974.DOI
  8. [8]N. Lanchier. Stochastic interacting systems in life and social sciences. De Gruyter Series in Probability and Stochastics, 2024. xviii + 468 pp.DOI
  9. [9]N. Lanchier. On the extinction phase of the contact process with an asymptomatic state. Electron. Commun. Probab. 31, no. 23, 11 pp., 2026.

Paper details

vlnc:W_3u7a6ykv
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 5

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents