Phase ordering and roughening on growing films

Specialization at an expanding front

As a population grows, spreading to new environments may favor specialization. In this paper, we introduce and explore a model for specialization at the front of a colony expanding synchronously into new territory. We show through numerical simulations that, by gaining fitness through accumulating mutations, progeny of the initial seed population can differentiate into distinct specialists. With competition and selection limited to the growth front, the emerging specialists first segregate into sectors, which then expand to dominate the entire population. We quantify the scaling of the fixation time with the size of the population and observe different behaviors corresponding to distinct universality classes: unbounded and bounded gains in fitness lead to superdiffusive (z=3/2) and diffusive (z=2) stochastic wanderings of the sector boundaries, respectively.

Growth instabilities shape morphology and genetic diversity of microbial colonies

Cellular populations assume an incredible variety of shapes ranging from circular molds to irregular tumors. While we understand many of the mechanisms responsible for these spatial patterns, little is known about how the shape of a population influences its ecology and evolution. Here, we investigate this relationship in the context of microbial colonies grown on hard agar plates. This a well-studied system that exhibits a transition from smooth circular disks to more irregular and rugged shapes as either the nutrient concentration or cellular motility is decreased. Starting from a mechanistic model of colony growth, we identify two dimensionless quantities that determine how morphology and genetic diversity of the population depend on the model parameters. Our simulations further reveal that population dynamics cannot be accurately described by the commonly-used surface growth models. Instead, one has to explicitly account for the emergent growth instabilities and demographic fluctuations. Overall, our work links together environmental conditions, colony morphology, and evolution. This link is essential for a rational design of concrete, biophysical perturbations to steer evolution in the desired direction.

Shape of population interfaces as an indicator of mutational instability in coexisting cell populations

Cellular populations such as avascular tumors and microbial biofilms may 'invade' or grow into surrounding populations. The invading population is often comprised of a heterogeneous mixture of cells with varying growth rates. The population may also exhibit mutational instabilities, such as a heavy deleterious mutation load in a cancerous growth. We study the dynamics of a heterogeneous, mutating population competing with a surrounding homogeneous population, as one might find in a cancerous invasion of healthy tissue. We find that the shape of the population interface serves as an indicator for the evolutionary dynamics within the heterogeneous population. In particular, invasion front undulations become enhanced when the invading population is near a mutational meltdown transition or when the surrounding 'bystander' population is barely able to reinvade the mutating population. We characterize these interface undulations and the effective fitness of the heterogeneous population in one- and two-dimensional systems.

Bacterial range expansions on a growing front: Roughness, fixation, and directed percolation

Directed percolation (DP) is a classic model for nonequilibrium phase transitions into a single absorbing state (fixation). It has been extensively studied by analytical and numerical techniques in diverse contexts. Recently, DP has appeared as a generic model for the evolutionary and ecological dynamics of competing bacterial populations. Range expansion-the stochastic reproduction of bacteria competing for space to be occupied by their progeny-leads to a fluctuating and rough growth front, which is known from experiment and simulation to affect the underlying critical behavior of the DP transition. In this work, we employ symmetry arguments to construct a pair of nonlinear stochastic partial differential equations describing the coevolution of surface roughness with the composition field of DP. Macroscopic manifestations (phenomenology) of these equations on growth patterns and genealogical tracks of range expansion are discussed; followed by a renormalization group analysis of possible scaling behaviors at the DP transition.

Chirality provides a direct fitness advantage and facilitates intermixing in cellular aggregates

Chirality in shape and motility can evolve rapidly in microbes and cancer cells. To determine how chirality affects cell fitness, we developed a model of chiral growth in compact aggregates such as microbial colonies and solid tumors. Our model recapitulates previous experimental findings and shows that mutant cells can invade by increasing their chirality or switching their handedness. The invasion results either in a takeover or stable coexistence between the mutant and the ancestor depending on their relative chirality. For large chiralities, the coexistence is accompanied by strong intermixing between the cells, while spatial segregation occurs otherwise. We show that the competition within the aggregate is mediated by bulges in regions where the cells with different chiralities meet. The two-way coupling between aggregate shape and natural selection is described by the chiral Kardar-Parisi-Zhang equation coupled to the Burgers’ equation with multiplicative noise. We solve for the key features of this theory to explain the origin of selection on chirality. Overall, our work suggests that chirality could be an important ecological trait that mediates competition, invasion, and spatial structure in cellular populations.

Is it better to be left- or right-handed? The answer depends on whether the goal is making a handshake or winning a boxing match. The need for coordination favors the handedness of the majority, but being different could also provide an advantage. The same rules could apply to microbial colonies and cancer tumors. Like humans, cells often have handedness (chirality) that reflects the lack of mirror symmetry in their shapes or movement patterns. We find that cells gain a substantial fitness advantage by either increasing the magnitude of their chirality or switching to the opposite handedness. Selection for specific chirality can overcome differences in growth rate and is mediated by the formation of bulges along the colony edge in regions where cells with different chiralities meet.

Response to a small external force and fluctuations of a passive particle in a one-dimensional diffusive environment

We investigate the long-time behavior of a passive particle evolving in a one-dimensional diffusive random environment, with diffusion constant D. We consider two cases: (a) The particle is pulled forward by a small external constant force and (b) there is no systematic bias. Theoretical arguments and numerical simulations provide evidence that the particle is eventually trapped by the environment. This is diagnosed in two ways: The asymptotic speed of the particle scales quadratically with the external force as it goes to zero, and the fluctuations scale diffusively in the unbiased environment, up to possible logarithmic corrections in both cases. Moreover, in the large D limit (homogenized regime), we find an important transient region giving rise to other, finite-size scalings, and we describe the crossover to the true asymptotic behavior.

Time evolution of intermittency in the passive slider problem

How does a steady state with strong intermittency develop in time from an initial state which is statistically random? For passive sliders driven by various fluctuating surfaces, we show that the approach involves an indefinitely growing length scale which governs scaling properties. A simple model of sticky sliders suggests scaling forms for the time-dependent flatness and hyperflatness, both measures of intermittency and these are confirmed numerically for passive sliders driven by a Kardar-Parisi-Zhang surface. Aging properties are studied via a two-time flatness. We predict and verify numerically that the time-dependent flatness is, remarkably, a nonmonotonic function of time with different scaling forms at short and long times. The scaling description remains valid when clustering is more diffuse as for passive sliders evolving through Edwards-Wilkinson driving or under antiadvection, although exponents and scaling functions differ substantially.

Large compact clusters and fast dynamics in coupled nonequilibrium systems

We demonstrate particle clustering on macroscopic scales in a coupled nonequilibrium system where two species of particles are advected by a fluctuating landscape and modify the landscape in the process. The phase diagram generated by varying the particle-landscape coupling, valid for all particle densities and in both one and two dimensions, shows novel nonequilibrium phases. While particle species are completely phase separated, the landscape develops macroscopically ordered regions coexisting with a disordered region, resulting in coarsening and steady state dynamics on time scales which grow algebraically with size, not seen earlier in systems with pure domains.

Nonuniversal effects in mixing correlated-growth processes with randomness: interplay between bulk morphology and surface roughening

To construct continuum stochastic growth equations for competitive nonequilibrium surface-growth processes of the type RD+X that mixes random deposition (RD) with a correlated-growth process X, we use a simplex decomposition of the height field. A distinction between growth processes X that do and do not create voids in the bulk leads to the definition of the effective probability p(eff) of the process X that is a measurable property of the bulk morphology and depends on the activation probability p of X in the competitive process RD+X. The bulk morphology is reflected in the surface roughening via nonuniversal prefactors in the universal scaling of the surface width that scales in p(eff). The equation and the resulting scaling are derived for X in either a Kardar-Parisi-Zhang or Edwards-Wilkinson universality class in (1+1) dimensions and are illustrated by an example of X being a ballistic deposition. We obtain full data collapse on its corresponding universal scaling function for all p∈(0;1]. We outline the generalizations to (1+n) dimensions and to many-component competitive growth processes.

Coupling-induced reorientation phase transitions in ultrathin Fe/Gd films

A phenomenological explanation for the reorientation phase transitions in an Fe/Gd ultrathin film system on the basis of Landau's theory of phase transitions is proposed. We model the film as a strongly coupled bilayer-like system consisting of the surface Fe overlayers and the interfacial Gd layer(s) below them. The total free energy of the system is accordingly obtained and the relevant phases and the order of the phase transitions involved are thus determined. Qualitative accordance between the theory and experiments is obtained. An alternative mechanism is proposed that attributes primarily the observed first-order phase transition in the system to the strong coupling between the Fe and the Gd film and its induced vertical magnetization component of the latter film. Competition between the antiferromagnetic coupling and the anisotropy energy of the Fe-rich ultrathin film is responsible for the other continuous reorientation. The effects of an applied external field including several field-induced first- and second-order phase transitions are predicted for experimental verification of the theory.

Dynamic binding of driven interfaces in coupled ultrathin ferromagnetic layers

We demonstrate experimentally dynamic interface binding in a system consisting of two coupled ferromagnetic layers. While domain walls in each layer have different velocity-field responses, for two broad ranges of the driving field H, walls in the two layers are bound and move at a common velocity. The bound states have their own velocity-field response and arise when the isolated wall velocities in each layer are close, a condition which always occurs as H→0. Several features of the bound states are reproduced using a one-dimensional model, illustrating their general nature.

Thin-film growth by random deposition of linear polymers on a square lattice

We present some results of Monte Carlo simulations for the deposition of particles of different sizes on a two-dimensional substrate. The particles are linear, height one, and can be deposited randomly only in the two x and y directions of the substrate and occupy an integer number of cells of the lattice. We show there are three different regimes for the temporal evolution of the interface width. At the initial times we observe an uncorrelated growth, with an exponent β1 characteristic of the random deposition model. At intermediate times, the interface width presents an unusual behavior, described by a growing exponent β2, which depends on the size of the particles added to the substrate. If the linear size of the particle is two we have β2 < β1, otherwise we have β2 > β1, for all other particle sizes. After the growth reaches the saturation regime where the interface width becomes constant and is described by the roughness exponent α, which is nearly independent of the size of the particle. Similar results are found in the surface growth due to the electrophoretic deposition of polymer chains. Contrary to one-dimensional results the growth exponents are nonuniversal.

Random deposition of particles of different sizes

We study the surface growth generated by the random deposition of particles of different sizes. A model is proposed where the particles are aggregated on an initially flat surface, giving rise to a rough interface and a porous bulk. By using Monte Carlo simulations, a surface has grown by adding particles of different sizes, as well as identical particles on the substrate in (1+1) dimensions. In the case of deposition of particles of different sizes, they are selected from a Poisson distribution, where the particle sizes may vary by 1 order of magnitude. For the deposition of identical particles, only particles which are larger than one lattice parameter of the substrate are considered. We calculate the usual scaling exponents: the roughness, growth, and dynamic exponents alpha, beta, and z, respectively, as well as, the porosity in the bulk, determining the porosity as a function of the particle size. The results of our simulations show that the roughness evolves in time following three different behaviors. The roughness in the initial times behaves as in the random deposition model. At intermediate times, the surface roughness grows slowly and finally, at long times, it enters into the saturation regime. The bulk formed by depositing large particles reveals a porosity that increases very fast at the initial times and also reaches a saturation value. Excepting the case where particles have the size of one lattice spacing, we always find that the surface roughness and porosity reach limiting values at long times. Surprisingly, we find that the scaling exponents are the same as those predicted by the Villain-Lai-Das Sarma equation.

Roughness of two nonintersecting one-dimensional interfaces

The dynamics of two spatially discrete one-dimensional single-step model interfaces with a noncrossing constraint is studied in both nonsymmetric propagating and symmetric relaxing cases. We consider possible scaling scenarios and study a few special cases by using continuous-time Monte Carlo simulations. The roughness of the interfaces is observed to be nonmonotonic as a function of time, and in the stationary state it is nonmonotonic also as a function of the strength of the effective force driving the interfaces against each other. This is related on the one hand to the reduction of the available configuration space and on the other hand to the ability of the interfaces to conform to each other.

Dynamic instability transitions in one-dimensional driven diffusive flow with nonlocal hopping

One-dimensional directed driven stochastic flow with competing nonlocal and local hopping events has an instability threshold from a populated phase into an empty-road (ER) phase. We implement this in the context of the asymmetric exclusion process. The nonlocal skids promote strong clustering in the stationary populated phase. Such clusters drive the dynamic phase transition and determine its scaling properties. We numerically establish that the instability transition into the ER phase is second order in the regime where the entry point reservoir controls the current and first order in the regime where the bulk is in control. The first-order transition originates from a turnabout of the cluster drift velocity. At the critical line, the current remains analytic, the road density vanishes linearly, and fluctuations scale as uncorrelated noise. A self-consistent cluster dynamics analysis explains why these scaling properties remain that simple.

Strong clustering of noninteracting, sliding passive scalars driven by fluctuating surfaces

We study the clustering of passive, noninteracting particles moving under the influence of a fluctuating field and random noise, in one and two dimensions. The fluctuating field in our case is provided by surfaces governed by the Kardar-Parisi-Zhang (KPZ) and the Edwards-Wilkinson (EW) equations, and the sliding particles follow the local surface slope. As the KPZ equation can be mapped to the noisy Burgers equation, the problem translates to that of passive scalars in a Burgers fluid. Monte Carlo simulations on discrete lattice models reveal very strong clustering of the passive particles for all sorts of dynamics under consideration. The resulting strong clustering state is defined using the scaling properties of the two point density-density correlation function. Our simulations show that the state is robust against changing the ratio of update speeds of the surface and particles. We also solve the related equilibrium problem of a stationary surface and finite noise, well known as the Sinai model for random walkers on a random landscape. For this problem, we obtain analytic results which allow closed form expressions to be found for the quantities of interest. Surprisingly, these results for the equilibrium problem show good agreement with the nonequilibrium KPZ problem.

Dynamic properties in a family of competitive growing models

The properties of a wide variety of growing models, generically called X-RD, involving the deposition of particles according to competitive processes, such that a particle is attached to the aggregate with probability p following the mechanisms of a generic model X that provides the correlations and at random [random deposition (RD)] with probability (1-p), are studied by means of numerical simulations and analytic developments. The study comprises the following X models: Ballistic deposition, random deposition with surface relaxation, Das Sarma-Tamboronea, Kim-Kosterlitz, Lai-Das Sarma, Wolf-Villain, large curvature, and three additional models that are variants of the ballistic deposition model. It is shown that after a growing regime, the interface width becomes saturated at a crossover time (tx2) that, by fixing the sample size, scales with p according to tx2(p) proportional variant p-y (P>0), where is an exponent. Also, the interface width at saturation (Wsat) scales as Wsat(p) proportional variant p-delta (p>0), where delta is another exponent. It is proved that, in any dimension, the exponents delta and y obey the following relationship: delta=y beta RD, where beta RD=1/2 is the growing exponent for RD. Furthermore, both exponents exhibit universality in the p --> 0 limit. By mapping the behavior of the average height difference of two neighboring sites in discrete models of type X-RD and two kinds of random walks, we have determined the exact value of the exponent delta. When the height difference between two neighbouring sites corresponds to a random walk that after walking <n> steps returns to a distance from its initial position that is proportional to the maximum distance reached (random walk of type A), one has delta=1/2. On the other hand, when the height difference between two neighboring sites corresponds to a random walk that after <n> steps moves <l> steps towards the initial position (random walk of type B), one has delta=1. Finally, by linking four well-established universality classes (namely Edwards-Wilkinson, Kardar-Parisi-Zhang, linear [molecular beam epitaxy (MBE)] and nonlinear MBE) with the properties of type A and B of random walks, eight different stochastic equations for all the competitive models studied are derived.

Universal scaling in mixing correlated growth with randomness

We study two-component growth that mixes random deposition (RD) with a correlated growth process that occurs with probability p. We find that these composite systems are in the universality class of the correlated growth process. For RD blends with either Edwards-Wilkinson or Kardar-Parisi-Zhang processes, we identify a nonuniversal exponent in the universal scaling in p.

Properties of the interfaces generated by the competition between stable and unstable growth models

Two different growing mechanisms, given by the Eden model (EM) and the unstable Eden model (UEM), are used to numerically explore the properties of the interface generated by a competitive dynamic process in which particles are aggregated according to the rules of the EM with probability (1-p) and following the UEM with probability p . Based on extensive numerical simulations, it is shown that the interface width exhibits a growing regime that at time t(x2) crosses over to a saturation state such that the width (Wsat) remains stationary. It is shown that Wsat and t(x2) depend on both the lattice size L and the probability p . This behavior can be rationalized by proposing new scaling relationships, which are tested numerically. Furthermore, the relevant exponents are determined showing that the instabilities of the UEM dominate the dynamics of the growing process.

Numerical simulation of a continuum model of growth of thin composite films

We present the results of extensive numerical integration in ( 1+1 ) dimensions of a set of equations that couple the Kardar-Parisi-Zhang (KPZ) equation to the time-dependent Ginzburg-Landau (TDGL) equation, recently proposed for modeling the growth of thin composite solid films. We find that for times t shorter than a crossover time t(c) the mean domain size L(t) grows logarithmically with the time, whereas for t>> t(c) L(t) grows as t(1/ z(m) ), with z(m) being nonuniversal and depending on the parameters of the model. The roughness exponent is also found to be nonuniversal. Thus, neither the dynamics of the domains' growth is governed by the TDGL equation, nor is the scaling of the surface roughness described by the KPZ equation.

Condensation transitions in a two-species zero-range process

We study condensation transitions in the steady state of a zero-range process with two species of particles. The steady state is exactly soluble-it is given by a factorized form provided the dynamics satisfy certain constraints-and we exploit this to derive the phase diagram for a quite general choice of dynamics. This phase diagram contains a variety of mechanisms of condensate formation, and a phase in which the condensate of one of the particle species is sustained by a "weak" condensate of particles of the other species. We also demonstrate how a single particle of one of the species (which plays the role of a defect particle) can induce Bose condensation above a critical density of particles of the other species.

Dynamics of a passive sliding particle on a randomly fluctuating surface

We study the motion of a particle sliding under the action of an external field on a stochastically fluctuating one-dimensional Edwards-Wilkinson surface. Numerical simulations using the single-step model shows that the mean-square displacement of the sliding particle shows distinct dynamic scaling behavior, depending on whether the surface fluctuates faster or slower than the motion of the particle. When the surface fluctuations occur on a time scale much smaller than the particle motion, we find that the characteristic length scale shows anomalous diffusion with xi(t) approximately t(2phi), where phi approximately 0.67 from numerical data. On the other hand, when the particle moves faster than the surface, its dynamics is controlled by the surface fluctuations and xi(t) approximately t(1/2). A self-consistent approximation predicts that the anomalous diffusion exponent is phi=2/3, in good agreement with simulation results. We also discuss the possibility of a slow crossover toward asymptotic diffusive behavior. The probability distribution of the displacement has a Gaussian form in both the cases.

Nonlinear stochastic equations with calculable steady states

We consider generalizations of the Kardar-Parisi-Zhang equation that accommodate spatial anisotropies and the coupled evolution of several fields, and focus on their symmetries and nonperturbative properties. In particular, we derive generalized fluctuation-dissipation conditions on the form of the (nonlinear) equations for the realization of a Gaussian probability density of the fields in the steady state. For the amorphous growth of a single height field in one dimension we give a general class of equations with exactly calculable (Gaussian and more complicated) steady states. In two dimensions, we show that any anisotropic system evolves in long time and length scales either to the usual isotropic strong coupling regime or to a linearlike fixed point associated with a hidden symmetry. Similar results are derived for textural growth equations that couple the height field with additional order parameters which fluctuate on the growing surface. In this context, we propose phenomenological equations for the growth of a crystalline material, where the height field interacts with lattice distortions, and identify two special cases that obtain Gaussian steady states. In the first case compression modes influence growth and are advected by height fluctuations, while in the second case it is the density of dislocations that couples with the height.

Passive random walkers and riverlike networks on growing surfaces

Passive random walker dynamics is introduced on a growing surface. The walker is designed to drift upward or downward and then follow specific topological features, such as hill tops or valley bottoms, of the fluctuating surface. The passive random walker can thus be used to directly explore scaling properties of otherwise somewhat hidden topological features. For example, the walker allows us to directly measure the dynamical exponent of the underlying growth dynamics. We use the Kardar-Parisi-Zhang (KPZ) -type surface growth as an example. The world lines of a set of merging passive walkers show nontrivial coalescence behaviors and display the riverlike network structures of surface ridges in space-time. In other dynamics, such as Edwards-Wilkinson growth, this does not happen. The passive random walkers in KPZ-type surface growth are closely related to the shock waves in the noiseless Burgers equation. We also briefly discuss their relations to the passive scalar dynamics in turbulence.

Static phase and dynamic scaling in a deposition model with an inactive species

We extend a previously proposed deposition model with two kinds of particle, considering the restricted solid-on-solid condition. The probability of incidence of particle C (A) is p (1-p). Aggregation is possible if the top of the column of incidence has a nearest neighbor A and if the difference in the heights of neighboring columns does not exceed 1. For any value of p>0, the deposit attains some static configuration, in which no deposition attempt is accepted. In 1+1 dimensions, the interface width has a limiting value W(s) approximately p(-eta), with eta=3/2, which is confirmed by numerical simulations. The dynamic scaling relation W(s)=p(-eta)f(tp(z)) is obtained in very large substrates, with z=eta.

Fluctuation-dominated phase ordering driven by stochastically evolving surfaces: depth models and sliding particles

We study an unconventional phase ordering phenomenon in coarse-grained depth models of the hill-valley profile of fluctuating surfaces with zero overall tilt, and for hard-core particles sliding on such surfaces under gravity. We find that several such systems approach an ordered state with large scale fluctuations which make them qualitatively different from conventional phase ordered states. We consider surfaces in the Edwards-Wilkinson (EW), Kardar-Parisi-Zhang (KPZ) and noisy surface-diffusion (NSD) universality classes. For EW and KPZ surfaces, coarse-grained depth models of the surface profile exhibit coarsening to an ordered steady state in which the order parameter has a broad distribution even in the thermodynamic limit, the distribution of particle cluster sizes decays as a power-law (with an exponent straight theta), and the scaled two-point spatial correlation function has a cusp (with an exponent alpha=1/2) at small values of the argument. The latter feature indicates a deviation from the Porod law which holds customarily, in coarsening with scalar order parameters. We present several numerical and exact analytical results for the coarsening process and the steady state. For linear surface models with a dynamical exponent z, we show that alpha=(z-1)/2 for z<3 and alpha=1 for z>3, and there are logarithmic corrections for z=3, implying alpha=1/2 for the EW surface and 1 for the NSD surface. Within the independent interval approximation we show that alpha+straight theta=2. We also study the dynamics of hard-core particles sliding locally downward on these fluctuating one-dimensional surfaces, and find that the surface fluctuations lead to large-scale clustering of the particles. We find a surface-fluctuation driven coarsening of initially randomly arranged particles; the coarsening length scale grows as approximately t(1/z). The scaled density-density correlation function of the sliding particles shows a cusp with exponents alpha approximately 0.5 and 0.25 for the EW and KPZ surfaces. The particles on the NSD surface show conventional coarsening (Porod) behavior with alpha approximately 1.

Reconstructed rough growing interfaces: ridge-line trapping of domain walls

We investigate whether surface reconstruction order exists in stationary growing states at all length scales or only below a crossover length l(rec). The latter behavior would be similar to surface roughness in growing crystal surfaces; below the equilibrium roughening temperature they evolve in a layer-by-layer mode within a crossover length scale l(R), but are always rough at large length scales. We investigate this issue in the context of Kardar-Parisi-Zhang (KPZ) type dynamics and a checkerboard type reconstruction, using the restricted solid-on-solid model with negative monatomic step energies. This is a topology where surface reconstruction order is compatible with surface roughness and where a so-called reconstructed rough phase exists in equilibrium. We find that during growth reconstruction order is absent in the thermodynamic limit, but exists below a crossover length l(rec)>l(R), and that this local order fluctuates critically. Domain walls become trapped at the ridge lines of the rough surface, and thus the reconstruction order fluctuations are slaved to the KPZ dynamics.

Competitive growth model involving random deposition and random deposition with surface relaxation

A deposition model that considers a mixture of random deposition with surface relaxation and a pure random deposition is proposed and studied. As the system evolves, random deposition with surface relaxation (pure random deposition) take place with probability p and (1-p), respectively. The discrete (microscopic) approach to the model is studied by means of extensive numerical simulations, while continuous equations are used in order to investigate the mesoscopic properties of the model. A dynamic scaling ansatz for the interface width W(L,t,p) as a function of the lattice side L, the time t and p is formulated and tested. Three exponents, which can be linked to the standard growth exponent of random deposition with surface relaxation by means of a scaling relation, are identified. In the continuous limit, the model can be well described by means of a phenomenological stochastic growth equation with a p-dependent effective surface tension.