Fixation results for threshold voter systems (Q2365745)

From MaRDI portal





scientific article; zbMATH DE number 222708
Language Label Description Also known as
default for all languages
No label defined
    English
    Fixation results for threshold voter systems
    scientific article; zbMATH DE number 222708

      Statements

      Fixation results for threshold voter systems (English)
      0 references
      0 references
      0 references
      29 June 1993
      0 references
      Two processes \((\xi_ t)\) are studied, in which the state at time \(t\) is given by \(\xi_ t:Z^ d\to\{0,1\}\) (here, \(\xi_ t(x)\) is interpreted as the opinion of the voter at \(x\) at time \(t)\). The first (stochastic!) process (called the threshold voter model) has continuous time. For each site \(x\in Z^ d\) there is an independent rate one Poisson process \(\{T^ x_ n,n\geq 1\}\). At time \(T^ x_ n\), the voter at \(x\) examines the sites in his neighbourhood \(\{y:y-x\in {\mathcal N}\}\). The voter's opinion changes if and only if at least \(\tau\) neighbours have the opposite opinion. It is shown that if \(0\in{\mathcal N}\) and \(\tau>(|{\mathcal N}|-1)/2\), then starting from any initial configuration, \(\lim_{t\to\infty}\xi_ t(x)\) exists (i.e., the system ``fixates'' at \(x)\) almost surely for all \(x\). The second (deterministic!) process (called the threshold voter automaton) has discrete time. Here, at each time \(n\), the voter at \(x\) examines the opinions of his neighbours \(x+{\mathcal N}\) and changes his opinion if and only if at least \(\tau\) neighbours have the opposite opinion. Let \(d=1\) and let \(A_ k\) denote the event that all voters at all sites \(x\) with \(| x|\leq k\) fixate in the same state. Let \(B_ k\) denote the event that the voters at all sites \(x\) with \(| x|\leq k\) never change. Suppose \({\mathcal N}=\{y:| y|\leq r\}\), \(\tau=\theta|{\mathcal N}|\), and let the system start from product measure with density 1/2. Let \(1/2<\theta_ c<3/4\) denote the unique solution of the equation \(2c(\theta)=c(2 \theta-1)\) where \(c(a)=\log 2+a\log a+(1-a)\log(1-a)\), \(0<a<1\). Then it is shown: (i) If \(\theta>\theta_ c\), then for all \(k\), \(P(B_ k)\to 1\) as \(r\to\infty\). (ii) If \(1/2<\theta<\theta_ c\), then for all \(k\), \(P(A_ k)\to 1\) as \(r\to\infty\).
      0 references
      voter model
      0 references
      Poisson process
      0 references
      opposite opinion
      0 references
      product measure
      0 references

      Identifiers