Msci Physics Research Project
Dynamics of Domain Wall Systems
Authors: Supervisor: Ciaran O’Hare Dr. Adam Moss James Stevenson
May 31, 2013
Abstract
Domain walls are cosmic topological defects that arise when the potential of a real scalar field has a discrete symmetry broken at a phase transition. We present the results of numerical simulations of domain walls in Minkowski and FRW cosmological backgrounds in box sizes up to 20482 and 1283. We have obtained scaling solutions for wall properties in 2+1 and 3+1 dimensions and in radiation and matter dominated epochs, identifying scale invariant evolution as the asymptotic scaling law. We also study the effects of a conserved charge condensed inside the domain walls in a hybrid domain wall model, a 2+1 dimensional analogue to cosmic vortons. We simulate the formation of domain wall systems in this model and obtain analytic solutions known as “kinky vortons”. We also probe the stability of these solutions under axial and non- axial perturbations and classify the outcomes of collisions between two vortons. Contents
1 Theory of Domain Walls 2 1.1 Introduction ...... 2 1.2 Domain wall formation ...... 3 1.3 Domain walls in FRW ...... 6
2 Numerical Simulations 8 2.1 The PRS algorithm ...... 8 2.2 Evolution algorithm ...... 8 2.3 Tests of the code ...... 10 2.4 Initial conditions ...... 11 2.5 Visualisation ...... 12
3 Cosmological scaling 13 3.1 Motivation ...... 13 3.2 Wall calculations ...... 14 3.3 Parameter choice ...... 16 3.4 Scaling results ...... 18 3.5 Discussion ...... 19
4 Theory of Charged Domain Walls 22 4.1 Introduction ...... 22 4.2 The model ...... 22 4.3 Symmetry breaking conditions ...... 24 4.4 Condensate formation conditions ...... 25
5 Evolution of Charged Domain Walls 27 5.1 Numerical implementation ...... 27 5.2 Dynamical properties and scaling ...... 28 5.2.1 Charged domain walls - Minkowski ...... 28 5.2.2 Charged domain walls - FRW ...... 30 5.2.3 Discussion ...... 31
6 Analytic Solutions and the Construction of Vortons 32 6.1 Infinite planar wall revisited ...... 32 6.2 Kinky vortons ...... 34 6.3 Energy minimising radius ...... 36
7 Vorton Stability 38 7.1 Stability analysis ...... 38 7.2 Vorton candidates via energy minimisation ...... 40 7.3 Vorton collisions ...... 41
1 Chapter 1
Theory of Domain Walls
1.1 Introduction
Phase transitions are ubiquitous in nature. A common feature associated with many phase transitions is the notion of symmetry breaking, a familiar and often quoted example being the symmetry breaking that occurs during the freezing of liquid water to ice. When the rotationally symmetric liquid state has degrees of freedom removed at a phase transition it crystallises into a regular lattice structure. In a macroscopic system of water, the phase transition of freezing will occur independently at many different sites at once each forming a separate crystal structure. As freezing progresses, these “domains” of ice will be forced to meet each other, at which point they are unlikely to be able to mesh together precisely, the result being a crack in the ice - a topological defect. The restoration of broken symmetries at high temperatures is a well known phenomenon, hence topological defects are familiar in many fields of science. A similar domain-domain wall configuration is present in ferromagnets when regions of the solid with aligned magnetic spins lock together. After the superfluid phase transition in Helium, point-like defects known as vortices appear [1] and similar defects have also been observed in superconductors [2] and protein folding [3]. Early Universe phase transitions, too, predict an array of topological defects (see [4] for review) as symmetry breaking plays a central role in grand unified theories (GUTs) which attempt to unify three of the four fundamental interactions of nature [5]. Here the symmetry breaking is associated with quantum fields where the vacuum states in their potentials are not simply connected. The breaking of different symmetry groups can lead to the formation of a range of defects including domain walls, cosmic strings, monopoles, textures and any number of hybrids. Defects have been a fascination in cosmology and particle theory since the description of domain walls and cosmic strings in the 1970s [6, 7]. Offering a unique avenue into the study of early Universe phase transitions, defects can provide valuable constraints on GUTs. They are also of importance in cosmology; for instance, monopoles were famously used in an argument that lead to the development of the theory of inflation [8, 9] and domain walls have also been considered as candidates for dark energy [10] and dark matter [11, 12]. The defects known as cosmic strings have been awarded much of the attention of the literature and have been linked with many areas including gamma-ray bursts [13] and time travel [14], and have recently been shown to provide a potential link to observables in string theory [15, 16]. Defects operate on a principle of “if they can form, they will form” [7]. If symmetry breaking occurs at GUT phase transitions then the formation of the associated defect is inevitable. Al- though there is currently no evidence for any class of defect, the observational search is ongoing. If defects are massive enough they should be evidenced by a number of sources, for example their signatures in the cosmic microwave background [18], primordial density fluctuations [17] or relic gravitational waves emitted under their collapse [19].
2 Classical field theory simulation of defect producing models has been an active topic of research since the late 1980s. Domain walls present the simplest case and therefore make a useful testbed to compare numerical results with analytic predictions. The most intriguing aspect of domain walls and one which is important for understanding their cosmology is the dynamical scaling they exhibit. Systems of domain walls can be characterised by a scaling law for their energy density and the initial effort of this study was to find this for two domain wall models. There has been some speculation in the literature [23, 24] as to the artificial effect of the simulation size on the scaling law and this is something that has been investigated here. This report is made up of two distinct parts which cover the work we have carried out on two domain wall models. The first part extends some previous research of the Z2 domain wall model (e.g. [20–22]). We describe the analytic steps taken in developing a code to simulate and analyse domain walls in 2+1 and 3+1 dimensions and the measures employed to ensure accuracy. The second domain wall model that has been investigated is a hybrid Z2 U(1) model [25] which × involves introducing a conserved charge condensing inside the walls. We fully outline the analytic approach taken in creating a simulation of this model. We then present results of work done on this model which extends the research carried out by Battye et al. (2009, [26]) to wall formation in a Friedmann-Robertson-Walker (FRW) cosmological background (expanding frame). Finally we describe the search for analytic solutions and the tests performed on their stability under perturbations and collisions. This report makes use of standard conventions used in the field. Firstly relativistic indexing, where positions in D+1 dimensional spacetime are written with an index µ running from 0 to µ ∂ D, x = (ct, x). Their respective partial derivatives are written as a µ ∂µ, note that the ∂x ≡ Einstein summation convention places an implicit summation sign wherever there are repeated spacetime indices. Finally, all equations are written in a system of natural units in which ~ = c = kB = 1.
1.2 Domain wall formation
Domain walls occur at the breaking of a discrete symmetry associated with the potential of a real scalar field φ(t, x). The temperature dependent symmetry breaking potential considered here has the form, λ 2 (φ, T ) = φ2 η(T )2 . (1.1) V 4 − (shown in Figure 1.1) where η(T ) and λ are model dependent parameters. The Lagrangian density is written as the difference of kinetic and potential terms and packages up the necessary dynamics of the field, 1 µν = η ∂µφ∂νφ (φ, T ). (1.2) L −2 − V Where ηµν = diag( 1, 1, 1, 1) is the Minkowski metric which describes the geometry of flat, − non-expanding space. The symmetry group that is broken in this potential is known in group theory jargon as Z2, or symmetry under reflection φ φ. Above some critical phase transition temperature, Tc, → − the potential possesses a single minimum at φ = 0, known as vacuum state of the potential. With this form, the potential is symmetric under reflections about its vacuum state. As the temperature cools and the phase transition progresses, the potential acquires two separated minima at φ = η, and the previous minimum acquires a non-zero energy value. There are now ± two new vacuum states at the two minima and the state at φ = 0 which is now a maximum in potential becomes known as the “false” vacuum state. The field will then naturally tend to fall down into one of the two minima with equal probability. The set of vacuum states that the field may occupy will be referred to as the vacuum manifold. The vacuum manifold before the phase transition possesses a Z2 symmetry because fluctuations in the field about the vacuum
3 (φ = 0 + δφ) are invariant under reflections φ φ. However expansions around the vacua → − that appear after the phase transition φ = η + δφ do not possess this symmetry. The process ± of changing the vacuum state of the field from one which possesses a symmetry to one which does not is known as spontaneous symmetry breaking.
V (φ) T > Tc T < Tc
φ
Figure 1.1: Z2 symmetry breaking potential, (φ) above and below the phase transition tem- V perature, Tc. The initial vacuum state at φ = 0 rises and becomes an unstable maximum after the phase transition. The field must then choose to fall down into one of the two new vacuum states, spontaneously breaking the Z2 symmetry.
For the purposes of this study the dynamics prior to the phase transition are not needed, so from here on the temperature dependence in the potential will be turned off and only temper- atures T < Tc will be considered. This restricts the parameter η to be constant and fixes the locations of the vacuum states. To describe the appearance of domain walls it is necessary to obtain the equation of motion for φ. To find this one must vary the action, Z S = d4x (1.3) L under small perturbations in the field. This eventually leads to the Euler-Lagrange equation for φ, ∂2φ ∂ = 2φ V . (1.4) ∂t2 5 − ∂φ Where t is cosmic (physical) time, and is the Laplacian operator in physical co-ordinates. 5 This equation of motion when placed under the fluctuations present after the phase transition describes the subsequent evolution of the field. The tendency of the field to fall randomly into one of the minima and the appearance of the Laplacian operator which acts to correlate the field locally, leads to the partitioning of space into “domains” in which the field is aligned into one of the two minima, (Figure 1.2). Regions that have had enough time to talk to one another, i.e. are causally correlated, will tend to arrange themselves into the same vacuum state, however regions that are not correlated will be randomly aligned in either state. Walls appear when oppositely aligned domains meet and the field is forced to leave the vacuum manifold and traverse the potential barrier separating the two vacuum states.
4 Figure 1.2: Partitioning of the Universe into domains soon after the phase transition. Blue/white colours correspond to each vacuum state. The domain walls that appear here in 2D are shown as black lines (in 3D they appear as 2 dimensional surfaces).
Domain walls are described by a mathematical function known as the kink solution [6], displayed in Figure 1.3. In the simple one dimensional case this solution describes how the field interpolates between the vacua, φ(x ) = η, but is a general form that also profiles walls → ±∞ ± in 2 and 3 dimensions. This solution can be obtained by considering the energy functional which must be minimised for a stable field configuration, Z E = dx . (1.5) H Where is the Hamiltonian density for the field (the sum of kinetic and potential terms). H Minimising this functional under the boundary conditions for a kink one can obtain the exact solution (see [19, 27] for full derivation), r λ φ(x) = ηtanh( ηx). (1.6) 2 The important feature of this solution is that it is stable. It cannot be simply smoothed out to a single field value, as taking one half of the field that resides in one domain to the other would require an infinite amount of energy. For a system of these kinks, the only possible evolution that can take place is for the domains to organise themselves so as to minimise their total surface area. In the description of domain walls we will frequently refer to them as ‘objects’ moving around in space, but this is perhaps a somewhat misleading interpretation. For cosmological purposes they can be interpreted in this way, as physical surfaces that possess mass which contributes to the energy density of the Universe and evolve according to relativistic equations for elastic surfaces. However another equally valid interpretation, and one which is more appropriate when thinking about their appearance in simulations, is that they are the necessary solutions to the non-linear field equation (1.4) and that domain walls themselves are simply regions of the field with particularly high spatial gradients.
5 φ
+η
x
η −
Figure 1.3: The kink solution which interpolates between the two vacua, η, and describes the ± profile of a domain wall. The width is determined by the parameters, λ and η.
1.3 Domain walls in FRW
For the purposes of understanding the impact of walls in a physical context the domain wall forming field equation must now be derived in a Friedmann-Robertson-Walker (FRW) cosmo- logical background, essentially an expanding Universe. This involves using the line element,
ds2 = dt2 + a2(t) dx2 + dy2 + dz2 . (1.7) − In which a(t) is the dimensionless scale factor, parametrising the expansion of space. The co- ordinates x, y, z are referred to as “co-moving” as they remain fixed as the Universe expands. The metric and inverse metric for this line element are,