Fase de desarrollo

Another interesting model is a spatial fluctuating annihilation model with two or more distinct particle types. For the two type case, the particles perform random motion in space Rn _{with n > 0 and annihilate when they meet different types i.e.} A+B → ∅, whereas they do not interact with the same type. An example of such a system in n = 1 spatial dimensions is shown in Figure 7.5 for radially growing and fixed directed structures, where the particles perform Brownian motion with

γ = 1/2. The dynamics leads to the formation of clusters defined to be sets of neighbouring particles of the same type. In the fixed geometry as before, because of the finite size only one type will remain if initial numbers are random or the system will be empty if initial numbers are the same. This will be the same in the growing

radial cluster for γ _≥ 1, however, for γ < 1 despite initial conditions due to the linear expansion we will always have a co-existence between different types.

This is further illustrated by computing several statistics as shown in Fig- ure 7.6 for these Brownian structures with a random number of initial typeAandB. Here_hCN_idenotes the number of clusters, which decreases with particle movement and interaction between different types. hCDi denotes the total distance between neighbouring clusters. For the fixed structure this will be monotonically increasing until there remains only one type. The cluster size denoted as_hCS_iis the number of particles per cluster, which increases due to the independence between same types and annihilation between different types. We can see that by using the mapping (5.4)withγ = 1/2 the growing radial clusters can be mapped to the fixed structure. Inn_≥2 dimensions these clusters are harder to define and it is easier to work with correlation functions.

Although these cluster models are clearly an over simplification, an ambitious application inn= 3 spatial dimensions would be to study cosmological fields such as the early universe during the inflationary period, (see [76, 131] and references within). An interesting question would be, if using such a model, we can make a contribution to the apparent asymmetry of matter and antimatter in the visible universe, which is one of the major unsolved problems in astrophysics, (see [3, 74] and references within). The basic idea is that the expansion of the universe leads to spatial separation of matter and antimatter on characteristic scales, which one might be able to predict using estimates for the expansion rate and the initial conditions, including the total amount of mass. Of course the model and simulations in the current form need significant fine tuning, where we would have to find the value ofγ

and the right geometrical manifold. For instance the sphereS3(r) is a simplification of the actual shape of the universe and it has been proposed in [130] that the shape is a Poincar´e dodecahedral. Another question would be the expansion rate, was it linear, ballistic or exponential? Whilst these simulation parameters are not yet known to us, it is exciting to think that such a simple function as (5.4) could have such wide applications.

100 101 102 101 h, h(r) 〈 CN 〉 〈 CN F(h) 〉 〈 CN R(r) 〉 (a) 100 101 102 101 h, h(r) 〈 CD 〉 〈 CD F(h) 〉 〈 CD_R(r) 〉 (b) 100 101 102 100.5 100.6 100.7 100.8 100.9 h, h(r) 〈 CS 〉 〈 CS_F(h) 〉 〈 CS_R(r) 〉 (c)

Figure 7.6: Statistics from a fixed domain and growing radial domain for a system as illustrated in Figure (7.5). Here we takeL= 100 andr0=L/(2π). The particles

of typesAandB, perform Brownian motion with the reaction conditionA+B _{→ ∅}

and do not interact otherwise. We use the mapping (5.4) withγ = 1/2 to plot the radial statisticshCNR(r)i,hCDR(r)iandhCSR(r)ivsh(r) to obtain a data collapse.

The data is shown for (a) number of clusters, (b) distance between neighbouring clusters and (c) cluster size i.e. number of particles in a cluster.

Appendix A

Elliptic theta functions

Here we give a list of the four elliptic theta functions. Letz, q_∈C_{and |q|<1. The} elliptic theta function of theith kind is denoted asϑi(z, q) and reads

1. ϑ1(z, q) = 2q1/4 ∞ X n=0 (₋1)nqn(n+1)sin((2n+ 1)z), 2. ϑ2(z, q) = 2q1/4 ∞ X n=0 qn(n+1)cos((2n+ 1)z), 3. ϑ3(z, q) = 1 + 2 ∞ X n=1 qn2cos(2nz), 4. ϑ4(z, q) = 1 + 2 ∞ X n=1 (₋1)nqn2cos(2nz).

The functionϑ3(πz, q) is known as the Jacobi theta function and withq fixed

is a Fourier transform for a entire function of z with period 1, and hence satisfies the identityϑ3(π(z+ 1), q) =ϑ3(πz, q). Also, ϑ3(πz, q) is quasi-periodic where

ϑ3(πz−bilogq, q) =q−b

2

exp(₋2πibz)ϑ3(πz, q),

where i = √₋1 and b _∈ Z_{. In Section 5.3 and Section 6.3.1, we use the function} ϑ3(z, q) in the expression of the prediction of hNF(h)i and hD2F(h)i. For further

Appendix B

Stochastic analysis

Here we provide a background on the probability aspects used throughout the thesis. A more in depth review can be found in Sections 3, 4, 5 and 8 of [156].

B.1

Probability space

Let (Ω,F,P_{) denote a probability space with probability measure} P_{. We have} P_{(Ω) = 1 and the σ-algebra F (also sometimes referred to as the σ-field) is a} collection of subsets of Ω such that:

1. _∅,Ω_{∈ F}

2. IfA_{∈ F ⇒}Ac _{∈ F, whereA}c _={ω∈Ω

ω /∈A}

3. _{Ai}i∈N⊂ F ⇒ ∪_i∈NA_i ∈ F. .

A random variableXon a probability space (Ω,_F,P_{) is such thatX : Ω→}R_. Define the Borel-σ-algebraB₍R_{) to be the smallestσ-algebra to contain all the open} subsets of R _{(i.e. contains the topology). A random variable X is a measurable} function satisfying

{ω_∈Ω, X(ω)_∈A_{} ∈ F} for all A_∈B₍R_). Define the_Lp_{(Ω,F,P) vector space with p >0 such that}

Lp(Ω,_F,P_{) :={X: Ω→}R||X||_Lp<∞},

with

where

E_{[X] =}

Z

Ω

X(ω)dP_(ω)

is the expectation operator. Note if _||X_||Lp = 0 this implies X = 0 almost surely (a.s. i.e. P_{(X 6= 0) = 0). The vector spaceL}p_(Ω,F,P_{) will be a Banach space with} the norm (B.1) and if p = 2 then this will be a Hilbert space with inner product E_{[XY] for X, Y ∈ L}2_(Ω,F,P_{). We call σ(X) the σ-algebra generated by X, the} smallest σ-algebra with respect to which X is measurable and contains the sets

{{ω_∈Ω

X(ω)∈A}

for all A∈B(R)}.