Documentation

Plurality.MeanField

Binary 3-Majority: mean field, and conformist transmission on counts #

Binary 3-Majority on the complete graph with self-loops (binKernel n: in every round each of the n nodes samples three nodes uniformly at random, with replacement and possibly itself, and adopts the majority opinion of its samples) maps the fraction x of nodes holding opinion 1 to 3x² - 2x³ in expectation (binKernel_meanField, ThreeMajority.avg_card_step stated as a Dynamics.Kernel.MeanField). This map is Dynamics.majorityMap, which is Galam's map with groups of three (Dynamics.galamMap_eq) and conformist transmission with D = 1 (Dynamics.conformistMap_one).

Conformist transmission (Boyd and Richerson 1985) in a population of n with non-overlapping generations, in which every new individual samples its three models with replacement from the whole previous generation, is a chain on the number of carriers of a variant: from k carriers, the three models of a new individual include a number of carriers of law Bin(3, k/n), so each new individual acquires the variant independently with probability p + D p (1 - p)(2p - 1), p = k/n (Dynamics.conformistMap_eq_expect), and the next generation has Bin(n, p + D p (1 - p)(2p - 1)) carriers (conformistCountKernel). Its mean-field map is Boyd and Richerson's recursion (conformistCountKernel_meanField). With D = 1, binary 3-Majority lumps to it under the number of nodes holding opinion 1 (binKernel_lumps): the two are the same chain on counts (row three-majority of the cross-field dictionary).

The mean-field map of binary 3-Majority #

theorem Plurality.binKernel_meanField (n : ℕ) [NeZero n] :
(binKernel n).MeanField (fun (I : Finset (Fin n)) (x : Unit) => ↑I.card / ↑n) fun (u : Unit → ℝ) (x : Unit) => Dynamics.majorityMap (u ())

The mean-field map of binary 3-Majority is the majority map. One round maps the fraction x = |I|/n of nodes holding opinion 1 to 3x² - 2x³ in expectation.

Conformist transmission on counts #

theorem Plurality.count_div_nonneg (n : ℕ) (k : Fin (n + 1)) :
0 ≤ ↑↑k / ↑n
theorem Plurality.count_div_le_one (n : ℕ) (k : Fin (n + 1)) :
↑↑k / ↑n ≤ 1
noncomputable def Plurality.conformistCountKernel (n : ℕ) (D : ℝ) (hD0 : 0 ≤ D) (hD1 : D ≤ 1) :

Conformist transmission on counts (Boyd and Richerson 1985, in its individual-based form). In a population of n with non-overlapping generations, every new individual samples three models with replacement from the whole previous generation and acquires a variant held by j of them with probability Dynamics.conformistTable D j. From k carriers of the variant, the next generation has Bin(n, p + D p (1 - p)(2p - 1)) carriers, p = k/n.

Equations
Instances For
    theorem Plurality.conformistCountKernel_meanField {n : ℕ} (hn : 1 ≤ n) {D : ℝ} (hD0 : 0 ≤ D) (hD1 : D ≤ 1) :
    (conformistCountKernel n D hD0 hD1).MeanField (fun (k : Fin (n + 1)) (x : Unit) => ↑↑k / ↑n) fun (u : Unit → ℝ) (x : Unit) => Dynamics.conformistMap D (u ())

    The mean-field map of conformist transmission is Boyd and Richerson's recursion: one generation maps the frequency p of the variant to p + D p (1 - p)(2p - 1) in expectation.

    def Plurality.cardFin {n : ℕ} (I : Finset (Fin n)) :
    Fin (n + 1)

    The number of nodes in I, in {0, …, n}.

    Equations
    Instances For

      A round of 3-Majority draws the sample triples of the n nodes independently and uniformly.

      Binary 3-Majority is conformist transmission with D = 1 on counts. Under binary 3-Majority on K_n with self-loops, the number of nodes holding opinion 1 is the chain of conformist transmission with D = 1, in which every new individual samples three models with replacement from the whole previous generation: one round from a configuration with k such nodes leads to Bin(n, 3p² - 2p³) of them, p = k/n.