## Abstract

We review a series of key travelling front problems in reaction–diffusion systems with a time-delayed feedback, appearing in ecology, nonlinear optics and neurobiology. For each problem, we determine asymptotic approximations for the wave shape and its speed. Particular attention is devoted to their validity and all analytical solutions are compared to solutions obtained numerically. We also extend the work by Erneux *et al.* (Erneux *et al.* 2010 *Phil. Trans. R. Soc. A* **368**, 483–493 (doi:10.1098/rsta.2009.0228)) by considering the case of a slowly propagating front subject to a weak delayed feedback. The delay may either speed up the front in the same direction or reverse its direction.

This article is part of the theme issue ‘Nonlinear dynamics of delay systems’.

### 1. Introduction

Time delays naturally appear in our everyday life. The reaction time of a driver is about 2 s, although it depends on various factors such as muscle memory, anticipation and age. The stability of quietly standing up depends on a neural controller that creates corrective actions after a time lag of 70–90 ms. Pharmacodynamic effects of drugs are typically delayed in relation to plasma drug concentrations. If the delay is short (minutes) then the mechanism is probably a distribution process, whereas if the delay is long (hours or longer) then the mechanism is more likely to be physiological [1–9]. In ecology, delays arise from maturation times, and in epidemiology, there is a time lag between host infection and disease. When time-delayed systems are coupled in space, diffusion comes into play, allowing a wide range of spatio-temporal phenomena, such as travelling structures, waves or localized states.

Time-delayed reaction–diffusion problems typically appear in population dynamics where the delay *τ* is the time from birth to maturity [10,11]. They are also used to model infectious diseases where the delay *τ* represents the time needed for a cell that is infected by a virus to release new viruses [12–16]. Often, a solution consists of a travelling front. These fronts are invasion waves where a spatially uniform steady state is propagating with a constant speed into another state.

Two distinct approaches to modelling the delayed reaction–diffusion problem have been proposed. One direction consists of supplementing the classical reaction–diffusion equation by terms proportional to the delay [13–15]. The migration and growth of the density of population *p* is described by

*F*(

*p*) represents the instantaneous reaction or growth term. This equation contains Cattaneo modifications of the Fick diffusion equation as proposed in 1948 to overcome the infinitely fast propagation [17]. Different derivations of this equation are possible. One of them consists of expanding Fick's Law with a time delay: the flux of particles depends on the gradient at a previous time. By then expanding the future flux

*J*(

*t*+

*τ*) around

*τ*= 0, we obtain equation (1.1). Another way of deriving it is by starting from microscopic considerations. A modern derivation of the Cattaneo equation can be found in [18]. By implementing the time delay in this way, the time-delayed problem is reduced to the determination of a front solution of a conventional reaction–diffusion problem. This greatly facilitates comparisons with experimental observations [15].

A different and more radical approach to modelling the effects of a time delay is to deal with the functional equation itself. Fickian diffusion is combined with a nonlinear reaction term that depends on a time delay [10,16]. The delayed reaction–diffusion equation then has the form

*F*(

*p*,

*p*(

*t*−

*τ*)) is typically inspired by reliable growth models such as the delayed logistic equation [19]. In some specific cases, equation (1.2) may be rigorously derived from an age-structured model [10]. Needless to say, the analysis of equation (1.2) is mathematically difficult.

In this work, we limit our review to scalar reaction–diffusion equations and concentrate on monostable and bistable front solutions. More complex spatio-temporal structures have been studied for higher-order reaction–diffusion systems (see, for example, [20–24]). Our main objective is to explore several analytical techniques appropriate for constructing front solutions of equation (1.2).

These methods take advantage of either a particular form of *F*(*p*, *p*(*t* − *τ*)) or limiting values of some parameters. The goal is to derive valuable approximations for both the front shape and its propagation speed. Although these methods have been successfully used for classical reaction–diffusion problems [25], results with time delay remain rare. Here, we consider a series of key problems in ecology, nonlinear optics and neurobiology. For most of these problems, we determine asymptotic approximations and verify their validity by numerical simulations of the original equations.

The outline of the paper is as follows. The case of monostable fronts is the most documented problem in the population biology literature. In §2, we review that the speed of propagation of these fronts is a decreasing function of the delay in a typical case, where the time delay is due to maturation. When the population dynamics includes an Allee effect, we obtain bistable fronts linking two stable steady states. The interest in bistable fronts is more recent, but it is also very relevant for biochemical reaction–diffusion systems. In §3, we determine the propagation speed for two different problems and contrast their behaviours in the limit of large delays. Finally, we discuss some future directions and possible applications in the field of cell biology.

### 2. Monostable travelling fronts

The ecology of invasions by animals and plants has received more and more attention in recent years, mainly because nearly every ecosystem has been invaded by exotic organisms with potentially drastic consequences for the native fauna or flora [26]. Mathematical reaction–diffusion models are proposed to describe or predict the fate of some particular invasions. Early attempts can be traced back to 1937, when R.A. Fisher was interested in the spatial spread of new genes that appeared in a population. He proposed the now famous FKPP (Fisher–Kolmogoroff–Petrovsky–Piscounoff) equation [25, p. 439]

*u*is the local population density. In this model, population expansion arises from, on the one hand, a balance between the local growth with maximum population density

*α*/

*β*, where

*α*is the linear growth rate, and, on the other hand, the diffusion of individuals with diffusion coefficient

*D*. A linear stability analysis of the homogeneous (non-spatial) system reveals that the zero steady state is unstable, and the state with

*u*=

*α*/

*β*is stable (figure 1

*a*). In the spatial system, a front connects the unstable uniform steady state

*u*= 0 and the uniform non-zero steady state

*u*=

*α*/

*β*(figure 1

*a*,

*b*). This front propagates into the unstable state with a speed

*c*.

A property of this system is that an infinite number of such fronts exist. In the 1970s, a number of authors studied types of initial data that can evolve to fronts with different speeds *c*. If the initial conditions satisfy [27]

*k*$\ge \sqrt{\alpha /D}$, the system evolves to a front with a minimal speed

*c*=

*Dk*+

*α*/

*k*(figure 1

*c*,

*d*).

Al-Omari & Gourley [10] considered the evolution of mature (adult) populations where the immature (juvenile) members of the population do not disperse. For many insect species, the juvenile phase is a larval phase in which the individuals do not move. Using an age structure model, they derive the following delay reaction–diffusion equation:

*u*represents the number of mature members of the population and the delay

*τ*is the time taken from birth to maturity. Compared to equation (2.1), equation (2.3) provides a more realistic model of a single population by recognizing that individuals cannot reproduce right away from birth, but only after they have matured, which takes a time

*τ*. It is assumed that the birth rate at any time is proportional to the number of adults at that time, so that the birth rate at time

*t*is

*αu*(

*t*). The second term on the right-hand side of equation (2.3) represents the rate of recruitment into the adult population. This rate is essentially the birth rate,

*τ*time units ago, decreased by the factor $\mathrm{exp}(-\gamma \tau )$ accounting for mortality during the juvenile phase. The parameter

*γ*thus measures the juvenile death rate while

*β*measures the adult death rate. As for equation (2.1), monotone travelling front solutions connecting the two uniform equilibria exist for any speed

*c*exceeding some critical (

*τ*-dependent) minimum value

*c*=

*c*

_{m}.

The transcendental equation for *c*_{m} is documented by eqn (3.9) in [10] or eqn (2.14) in [28] and is given by

*c*

_{m}=

*c*

_{m}(

*τ*) is shown in figure 2 for two different values of

*γ*together with their local approximation

The expression (2.5) clearly illustrates the fact that the minimal speed is decreasing with the delay. Equation (2.4) is obtained as follows: first, we switch to travelling wave coordinates by substituting *z* = *x* + *ct* in equation (2.3), then we linearize the resulting equation around *u* = 0 and examine the decay to zero of *u*(*z*) as $z\to -\mathrm{\infty}$. A front that decays to zero in an oscillatory fashion is of no interest since *u* is a population and thus needs to remain positive. The interesting result from an ecological point of view is that the minimum speed decreases compared to equation (2.2) as the delay increases.

### 3. Bistable travelling fronts

#### (a) Population dynamics

The situation is different if the population growth is damped by a strong Allee effect (after the American zoologist W.C. Allee). The Allee effect has to do with the fact that the fitness of small populations is sometimes negative, i.e. if the population density is too small, the species or group of individuals will not survive. In other words, the zero steady state is stable (figure 3*a*). To model this effect, an extra factor is added to the reaction term of the FKPP equation. Let us consider a population whose growth is described by a cubic polynomial

*a*< 1 and −1 <

*a*< 0 correspond to the strong and weak Allee effects, respectively. For the strong Allee effect, there is a front connecting the two stable uniform steady states

*u*= 0 and

*u*= 1 (figure 3

*a*,

*b*). In contrast to monostable fronts, where an infinite number of speeds is possible for a front depending on the initial condition, this bistable front has a unique speed given by [29]

This expression is obtained by inserting $u={(1+\mathrm{exp}(Az))}^{-1}$ into equation (3.1) reformulated in terms of the travelling wave coordinate *z* = *x* + *ct*. By solving this equation using as boundary conditions *u*( − ∞) = 0 and *u*(∞) = 1, we obtain *A* and *c* (equation (3.2)). By reversing the boundary conditions, *A* and *c* change sign.

If *a* < 1/2, then *c* > 0, which implies that the invasion is successful: the populated steady state invades the unpopulated state. If instead *a* > 1/2, then *c* < 0. This means that the front travels in the other direction, and the low population density state wins out in the end. Interpreted in ecological terms, this means that an invasion fails.

When there is a weak Allee effect, which can be modelled by taking −1 < *a* < 0, the zero steady state is unstable, but has a growth rate which is less than that at intermediate population densities. This case is a crossover between the strong Allee effect and logistic growth. There now is a front connecting the unstable zero steady state to the stable populated state, but determining the final propagation speed is more subtle. See [29, Sec. 3.3] for a discussion.

#### (b) Optics

The obvious question now is how these results for bistable fronts change when a time delay is introduced. We will now broaden our selection of topics, since travelling wave solutions of simple bistable reaction–diffusion equations subject to a time-delayed feedback are attracting the interest of many scientists with different objectives and experiences. In optics, the control of localized structures appearing in broad-area lasers is realized by optical feedback from a distant mirror. Small-area bistable lasing spots (10 μm) were first observed within the aperture of a broad-area vertical-cavity surface emitting laser (VCSEL) (80 μm) with frequency-selective feedback [30,31] and have immediately prompted theoretical studies [32]. The impact of time-delayed feedback on the dynamics of localized structures or cavity solitons has been further investigated theoretically for the cases of a driven nonlinear optical resonator [33,34] and broad-area VCSELs [35–37]. Such patterns may be stationary or oscillatory, static or moving (see, for example, figure 4*b*). In [33], the following delayed Swift–Hohenberg (SH) equation

*a*,

*b*,

*c*are constant parameters, and

*η*and

*τ*are defined as the gain and the delay of the feedback, respectively. The time-delayed feedback does not modify the stationary time-independent state, but instead changes its stability properties [38,39]. An important result is that localized stationary solutions (spots) can be set into motion by the delayed feedback (figure 4

*b*). It is the first delay-induced instability for the SH equation but other bifurcation scenarios are possible for higher-order reaction–diffusion systems [23]. The authors emphasized a necessary condition for moving spots given by the inequality

A necessary condition for stability is obtained by first determining the real roots. Keeping *τ* fixed, we analyse λ = λ(*η*) from equation (3.5) in the implicit form $\eta \tau =\mathrm{\lambda}\tau {(\mathrm{exp}(-\mathrm{\lambda}\tau )-1)}^{-1}$. We find that λ is unique and negative if

#### (c) Neurobiology

Bistable fronts appear in the neurosciences when one needs to describe the spreading of neuronal activities. One model that has been studied in this context is described by the following equation:

*u*

_{τ}. In [41], the feedback is a delayed spatial average defined by

*L*= 1 and the constants

*a*,

*η*and

*τ*are parameters. For

*ε*= 0, this system admits a front solution

The influence of the time-delayed feedback on the shape and speed of the front given by equation (3.10) can be studied by taking advantage of the small parameter *ε*. In this way, Boubendir *et al.* found that the steady front may undergo a Hopf bifurcation to oscillatory travelling fronts [41].

The exact form of the feedback term *g*(*u*_{τ}) greatly influences the dynamics. A different global delayed feedback for a Ginzburg–Landau equation exhibiting a subcritical bifurcation is considered in [42]. They noted that, for a large delay, localized solutions can start oscillating. Equation (3.8) has been used as an early model for brain depression waves appearing for both migraine and stroke [43,44]. In those models, the delayed feedback was of the form

*a*= 0. Note that, in our present study, the feedback strength is given by

*εη*, whereas in equation (3.3) it is

*η*. In the presence of a time-delayed feedback (

*ε*≠0), the steady front (3.10) starts to slowly move provided that

*εη*< − 1/

*τ*, the same conditions as for the Swift–Hohenberg case (equation (3.7), figure 4

*a*). In this case, a strong enough feedback will set the front into motion. Since the system is symmetric, the front can travel in either direction with the same speed. We now generalize the results of [45] and explore the asymmetric case

*a*≠0. Specifically, we seek a solution

*u*=

*u*(

*z*) where

*z*=

*x*−

*ct*that satisfies

*z*. By taking advantage of the small parameter

*ε*, we find that the leading solution is again given by (3.10). As in [45], we formulate the

*O*(

*ε*) problem and apply a solvability condition. The latter provides an equation for the speed

*c*=

*c*(

*τ*) given by

*a*,

*b*shows the solutions of equation (3.13) with

*η*= − 1. The pitchfork bifurcation in orange corresponds to the case

*a*= 0. The imperfect bifurcation in blue is for

*a*= 0.01. In the language of the theory of singularities, the effect of

*εa*≠0 is to unfold the pitchfork bifurcation. A local analysis near the bifurcation point at

*τ*=

*ε*

^{−1}similar to the one in [45] indicates that the upper and lower branches are stable while the middle branch is unstable for

*τ*> 1/

*ε*. From (3.13), we note that the speed

*c*approaches constant values as $\tau \to \mathrm{\infty}$ given by

*η*= − 1 here, the two upper branches (

*c*> 0) are coming closer together as $a\to 1$ (figure 6

*c*). Furthermore, their left limit point moves to large

*τ*in the same limit $a\to 1$. Consequently, if

*a*> 1, the lower branch (

*c*< 0) is the only branch of solutions. We conclude that a front that is too asymmetric may only be observed to be propagating in one direction (i.e. one specific sign of

*c*). The delayed feedback then speeds up the front in that same direction. Second, the fact that the speed approaches constant values as $\tau \to \mathrm{\infty}$ implies that we may replace

*u*(

*z*+

*cτ*) in equation (3.12) by its asymptotic limit for

*z*large, i.e.

*u*= 1. The reduced problem then becomes an ordinary differential equation which is easier to solve. Using (3.12) with

*u*(

*z*+

*cτ*) = 1 we found analytically the correct speed limits.

The final example we discuss is again motivated by simple bistable models in neurobiology. The following equation exhibits a delayed threshold nonlinearity:

*H*(

*u*) denotes the Heaviside function, and the function

*H*(

*u*−

*a*) −

*u*can be seen as a piecewise linear version of a cubic polynomial with zeros at 0 and 1 (figure 7

*a*). The advantage of using piecewise linear functions is that this approach allows one to find an exact solution for a stable front connecting

*u*= 0 and

*u*= 1. The speed

*c*=

*c*(

*τ*) is again unique and satisfies the implicit equation

where ${\mathrm{\lambda}}_{\pm}=(c\pm \sqrt{{c}^{2}+4})/2$. In the limit of very small delays ($\tau \to 0$) and very large delays ($\tau \to \mathrm{\infty}$), the front speeds go to $c\to {c}_{0}$ and $c\to {c}_{\mathrm{\infty}},$ respectively. These speeds are given by the following expressions:

*b*shows

*c*as a function of

*τ*. We note that the speed is decaying to zero as $\tau \to \mathrm{\infty}$, which contrasts with our previous result using (3.11) as the feedback. There, the speed approaches constant values in the limit of large delays.

### 4. Discussion

Since the 1970s, there have been systematic studies of the impact of spatial diffusion on population dynamics when time lags are taken into account. More recently, new reaction–diffusion problems have appeared in neurobiology and nonlinear optics with the aim to understand how delays may control the progression of travelling waves. In this review, we emphasized the need for simple one-variable models in order to have an alternative to extensive numerical simulations. Key problems have been selected and various analytical techniques have been used to obtain results of physical significance.

For the population model (2.3) and the delayed threshold model (3.15), the delay always slows down moving fronts. If the delayed feedback *η*(*u*(*t* − *τ*) − *u*) is added to the reaction–diffusion equation as for the Swift–Hohenberg equation (3.3) and the bistable equation (3.8) with (3.11) and *a* = 0, different responses are possible. For negative feedback strength (*η* < 0) and sufficiently large time delays *τ*, stationary solutions are set into motion. If the basic reaction–diffusion equation admits a slowly propagating front as equation (3.8) with (3.11) and *a*≠0, the delay may either speed up the front in the same direction or, interestingly, reverse its direction.

While looking for the linear stability properties of the unstable uniform solution provides the propagation speed in the case of the monostable front, we need to solve the nonlinear problem for the bistable case. Preference has been given to asymptotic techniques. One novel observation is the unfolding of the pitchfork bifurcation illustrated in figure 6 for the asymmetric bistable equation (3.8) with (3.11) and *a*≠0. A second novel contribution of our work are the comparisons between analytical and numerical solutions which allow us to evaluate the quantitative validity of our approximations.

Fronts are the building blocks of more complex structures such as pulses and wave trains, and asymptotic studies on the effects of a time delay on propagation speeds are currently being investigated. As explained by Gourley *et al.* [10], diffusion and time delays are not independent of each other since individuals have not been at the same point in space at previous times. One way to correct this is to introduce a non-local term in the reaction–diffusion equation, i.e. a weighted average in space. The effects of spatial non-locality without delay have recently been studied in different settings [47–51]. Non-locality allows one to enhance or decrease front interaction and to create new structures.

Besides population dynamics, epidemiology and neurobiology, there are systems in cell biology where it is relevant to study time-delayed reaction–diffusion equations. In these systems, a time delay can be used to model a chain of intermediate, and usually unknown, chemical reactions. One example of such a system is the early embryonic cell cycle of the frog *Xenopus laevis*. In these cells, travelling waves have been observed experimentally [52]. Similar to models in population biology, these waves could be interpreted as invasion waves, where one state (the mitotic state of the cell in which the cell divides) invades another state (interphase). This early cell cycle is controlled by a time-delayed negative feedback loop [53,54]. In future work, we plan to use and extend the techniques described here to study how such waves organize cell division in space and time.

### Data accessibility

This article has no additional data.

### Authors' contributions

J.R. performed the numerical simulations, helped with calculations and participated in writing. L.G. participated in interpreting results, structuring the content and writing the manuscript. T.E. initiated the research, did the literature review, made calculations and drafted the manuscript.

### Competing interests

We have no competing interests.

### Funding

The authors acknowledge support by the Research-Foundation Flanders (FWO-Vlaanderen) for individual support (J.R.) and project support (L.G. grant G0A5217N).

### References

- 1.
- 2.
Stepan G (ed.) 2009 Delay effects in brain dynamics.**Phil. Trans. R. Soc. A**, 1059–1212. (doi:10.1098/rsta.2008.0279) Link, ISI, Google Scholar**367** - 3.
Lakshmanan M, Senthilkumar DV . 2010**Dynamics of nonlinear time-delay systems**.Springer Series in Synergetics, Springer Complexity . Berlin, Germany: Springer. Google Scholar - 4.
Smith H . 2011**An introduction to delay differential equations with applications to the life sciences**. Berlin, Germany: Springer. Crossref, Google Scholar - 5.
Sipahi R, Niculescu S-I, Abdallah CT, Michiels W, Gu K . 2011 Stability and stabilization of systems with time delay.**IEEE Control Syst. Mag.**, 38–65. (doi:10.1109/MCS.2010.939135) Crossref, ISI, Google Scholar**31** - 6.
Milton J, Ohira T . 2014**Mathematics as a laboratory tool**. Berlin, Germany: Springer. Crossref, Google Scholar - 7.
Koch G, Krzyzanski W, Pérez-Ruixo JJ, Schropp J . 2014 Modeling of delays in PKPD: classical approaches and a tutorial for delay differential equations.**J. Pharmacokinet. Pharmacodyn.**, 291–318. (doi:10.1007/s10928-014-9368-y) Crossref, PubMed, ISI, Google Scholar**41** - 8.
Bélair J, Rimbu A . 2015 Time delays in drug administration: effect, transit, tricks and oscillations.**IFAC-PapersOnLine**, 111–116. (doi:10.1016/j.ifacol.2015.09.362) Google Scholar**48** - 9.
Mackey MC, Santillán M, Tyran-Kamińska M, Zeron ES . 2016**Simple mathematical models of gene regulatory dynamics**.Lecture Notes on Mathematical Modelling in the Life Sciences . Berlin, Germany: Springer. Crossref, Google Scholar - 10.
Al-Omari J, Gourley SA . 2002 Monotone traveling fronts in an age-structured reaction–diffusion model of a single species.**J. Math. Biol.**, 294–312. (doi:10.1007/s002850200159) Crossref, PubMed, ISI, Google Scholar**45** - 11.
Gourley SA, So JW-H, Wu JH . 2004 Nonlocality of reaction–diffusion equations induced by delay: biological modeling and nonlinear dynamics.**J. Math. Sci.**, 5119–5153. (doi:10.1023/B:JOTH.0000047249.39572.6d) Crossref, Google Scholar**124** - 12.
Gourley SA, Kuang Y . 2005 A delay reaction–diffusion model of the spread of bacteriophage infection.**SIAM J. Appl. Math.**, 550–566. (doi:10.1137/S0036139903436613) Crossref, ISI, Google Scholar**65** - 13.
Fort J, Méndez V . 2002 Time-delayed spread of viruses in growing plaques.**Phys. Rev. Lett.**,**89**178101 . (doi:10.1103/PhysRevLett.89.178101) Crossref, PubMed, ISI, Google Scholar - 14.
Fort J, Méndez V . 2002 Wavefronts in time-delayed reaction–diffusion systems. Theory and comparison to experiment.**Rep. Prog. Phys.**, 895–954. (doi:10.1088/0034-4885/65/6/201) Crossref, ISI, Google Scholar**65** - 15.
Amor DR, Fort J . 2010 Virus infection speeds: theory versus experiment.**Phys. Rev. E**,**82**061905 . (doi:10.1103/PhysRevE.82.061905) Crossref, ISI, Google Scholar - 16.
Jones DA, Smith HL, Thieme HR, Röst G . 2012 On spread of phage infection of bacteria in a petri dish.**SIAM J. Appl. Math.**, 670–688. (doi:10.1137/110848360) Crossref, ISI, Google Scholar**72** - 17.
Cattaneo C . 1948 Sulla conduzione del calore.**Atti Semin. Mat. Fis. Univ. Modena**, 83–101. (doi:10.1007/978-3-642-11051-1_5) Google Scholar**3** - 18.
Keller JB . 2004 Diffusion at finite speed and random walks.**Proc. Natl Acad. Sci. USA**, 1120–1122. (doi:10.1073/pnas.0307052101) Crossref, PubMed, ISI, Google Scholar**101** - 19.
Gourley SA, So JW-H . 2002 Dynamics of a food-limited population model incorporating nonlocal delays on a finite domain.**J. Math. Biol.**, 49–78. (doi:10.1007/s002850100109) Crossref, PubMed, ISI, Google Scholar**44** - 20.
Bestehorn M, Grigorieva EV, Kaschenko SA . 2004 Spatiotemporal structures in a model with delay and diffusion.**Phys. Rev. E**,**70**026202 . (doi:10.1103/PhysRevE.70.026202) Crossref, ISI, Google Scholar - 21.
Li QS, Hu HX . 2007 Pattern transitions induced by delay feedback.**J. Chem. Phys.**,**127**154510 . (doi:10.1063/1.2792877) Crossref, PubMed, ISI, Google Scholar - 22.
Tian C, Zhang L . 2013 Delay-driven irregular spatiotemporal patterns in a plankton system.**Phys. Rev. E**,**88**012713 . (doi:10.1103/PhysRevE.88.012713) Crossref, ISI, Google Scholar - 23.
Gurevich SV . 2013 Dynamics of localized structures in reaction–diffusion systems induced by delayed feedback.**Phys. Rev. E**,**87**052922 . (doi:10.1103/PhysRevE.87.052922) Crossref, ISI, Google Scholar - 24.
Iri R, Tonosaki Y, Shitara K, Ohta T . 2014 Dynamics of Turing pattern under time-delayed feedback.**J. Phys. Soc. Jpn.**,**83**024001 . (doi:10.7566/JPSJ.83.024001) Crossref, Google Scholar - 25.
Murray JD . 2002**Mathematical biology I: an introduction**, 3rd edn.Interdisciplinary Applied Mathematics 17 . Berlin, Germany: Springer. Crossref, Google Scholar - 26.
Lewis MA, Kareiva P . 1993 Allee dynamics and the spread of invading organisms.**Theor. Popul. Biol.**, 141–158. (doi:10.1006/tpbi.1993.1007) Crossref, ISI, Google Scholar**43** - 27.
McKean HP . 1975 Application of Brownian motion to the equation of Kolmogorov–Petrovskii–Piskunov.**Commun. Pure Appl. Math.**, 323–331. (doi:10.1002/(ISSN)1097-0312) Crossref, ISI, Google Scholar**28** - 28.
Gourley SA . 2005 Linear stability of traveling fronts in an age-structured reaction–diffusion population model.**Q. J. Mech. Appl. Math.**, 257–268. (doi:10.1093/qjmamj/hbi012) Crossref, ISI, Google Scholar**58** - 29.
Lewis MA, Petrovskii SV, Potts JR . 2016**The mathematics behind biological invasions**.Interdisciplinary Applied Mathematics 44 . Berlin, Germany: Springer. Crossref, Google Scholar - 30.
Tanguy Y, Ackemann T, Jäger R . 2006 Characteristics of bistable localized emission states in broad-area vertical-cavity surface-emitting lasers with frequency-selective feedback.**Phys. Rev. A**,**74**053824 . (doi:10.1103/PhysRevA.74.053824) Crossref, ISI, Google Scholar - 31.
Tanguy Y, Radwell N, Ackemann T, Jäger R . 2008 Characteristics of cavity solitons and drifting excitations in broad-area vertical-cavity surface-emitting lasers with frequency-selective feedback.**Phys. Rev. A**,**78**023810 . (doi:10.1103/PhysRevA.78.023810) Crossref, ISI, Google Scholar - 32.
Paulau PV, Gomila D, Ackemann T, Loiko NA, Firth WJ . 2008 Self-localized structures in vertical-cavity surface-emitting lasers with external feedback.**Phys. Rev. E**,**78**016212 . (doi:10.1103/PhysRevE.78.016212) Crossref, ISI, Google Scholar - 33.
Tlidi M, Vladimirov AG, Pieroux D, Turaev D . 2009 Spontaneous motion of cavity solitons induced by a delayed feedback.**Phys. Rev. Lett.**,**103**103904 . (doi:10.1103/PhysRevLett.103.103904) Crossref, PubMed, ISI, Google Scholar - 34.
Gurevich SV, Friedrich R . 2013 Instabilities of localized structures in dissipative systems with delayed feedback.**Phys. Rev. Lett.**,**110**014101 . (doi:10.1103/PhysRevLett.110.014101) Crossref, PubMed, ISI, Google Scholar - 35.
Panajotov K, Tlidi M . 2010 Spontaneous motion of cavity solitons in vertical-cavity lasers subject to optical injection and to delayed feedback.**Eur. Phys. J. D**, 67–72. (doi:10.1140/epjd/e2010-00111-y) Crossref, ISI, Google Scholar**59** - 36.
Tlidi M, Averlant E, Vladimirov AG, Panajotov K . 2012 Delay feedback induces a spontaneous motion of two-dimensional cavity solitons in driven semiconductor microcavities.**Phys. Rev. A**,**86**033822 . (doi:10.1103/PhysRevA.86.033822) Crossref, ISI, Google Scholar - 37.
Pimenov A, Vladimirov AG, Gurevich SV, Panajotov K, Huyet G, Tlidi M . 2013 Delayed feedback control of self-mobile cavity solitons.**Phys. Rev. A**,**88**053830 . (doi:10.1103/PhysRevA.88.053830) Crossref, ISI, Google Scholar - 38.
Pyragas K . 1992 Continuous control of chaos by self-controlling feedback.**Phys. Lett. A**, 421–428. (doi:10.1016/0375-9601(92)90745-8) Crossref, ISI, Google Scholar**170** - 39.
Schöll E, Hiller G, Hövel P, Dahlem MA . 2009 Time-delayed feedback in neurosystems.**Phil. Trans. R. Soc. A**, 1079–1096. (doi:10.1098/rsta.2008.0258) Link, ISI, Google Scholar**367** - 40.
Michiels W, Niculescu S-I . 2007**Stability and stabilization of time-delay systems: an eigenvalue-based approach**.Advances in Design and Control, no. 12 . Philadelphia, PA: SIAM. Crossref, Google Scholar - 41.
Boubendir Y, Méndez V, Rotstein HG . 2010 Dynamics of one- and two-dimensional fronts in a bistable equation with time-delayed global feedback: propagation failure and control mechanisms.**Phys. Rev. E**,**82**036601 . (doi:10.1103/PhysRevE.82.036601) Crossref, ISI, Google Scholar - 42.
Rubinstein BY, Nepomnyashchy AA, Golovin AA . 2007 Stability of localized solutions in a subcritically unstable pattern-forming system under a global delayed control.**Phys. Rev. E**,**75**046213 . (doi:10.1103/PhysRevE.75.046213) Crossref, ISI, Google Scholar - 43.
Schneider FM, Schöll E, Dahlem MA . 2009 Controlling the onset of traveling pulses in excitable media by nonlocal spatial coupling and time-delayed feedback.**Chaos**,**19**015110 . (doi:10.1063/1.3096411) Crossref, PubMed, ISI, Google Scholar - 44.
Dahlem MA, Graf R, Strong AJ, Dreier JP, Dahlem YA, Sieber M, Podoll HWK, Scholl E . 2010 Two-dimensional wave patterns of spreading depolarization: retracting, re-entrant, and stationary waves.**Physica D**, 889–903. (doi:10.1016/j.physd.2009.08.009) Crossref, ISI, Google Scholar**239** - 45.
Erneux T, Kozyreff G, Tlidi M . 2010 Bifurcation to fronts due to delay.**Phil. Trans. R. Soc. A**, 483–493. (doi:10.1098/rsta.2009.0228) Link, ISI, Google Scholar**368** - 46.
Coombes S, Laing CR . 2009 Instabilities in threshold-diffusion equations with delay.**Physica D**, 264–272. (doi:10.1016/j.physd.2008.10.014) Crossref, ISI, Google Scholar**238** - 47.
Gelens L, Gomila D, Van der Sande G, Matías MA, Colet P . 2010 Nonlocality-induced front-interaction enhancement.**Phys. Rev. Lett.**,**104**154101 . (doi:10.1103/PhysRevLett.104.154101) Crossref, PubMed, ISI, Google Scholar - 48.
Fernandez-Oto C, Clerc MG, Escaff D, Tlidi M . 2013 Strong nonlocal coupling stabilizes localized structures: an analysis based on front dynamics.**Phys. Rev. Lett.**,**110**174101 . (doi:10.1103/PhysRevLett.110.174101) Crossref, PubMed, ISI, Google Scholar - 49.
Colet P, Matías MA, Gelens L, Gomila D . 2014 Formation of localized structures in bistable systems through nonlocal spatial coupling. I. General framework.**Phys. Rev. E**,**89**012914 . (doi:10.1103/PhysRevE.89.012914) Crossref, ISI, Google Scholar - 50.
Gelens L, Matías MA, Gomila D, Dorissen T, Colet P . 2014 Formation of localized structures in bistable systems through nonlocal spatial coupling. II. The nonlocal Ginzburg–Landau equation.**Phys. Rev. E**,**89**012915 . (doi:10.1103/PhysRevE.89.012915) Crossref, ISI, Google Scholar - 51.
Clerc MG, Escaff D, Kenkre VM . 2010 Analytical studies of fronts, colonies, and patterns: combination of the Allee effect and nonlocal competition interactions.**Phys. Rev. E**,**82**036210 . (doi:10.1103/PhysRevE.82.036210) Crossref, ISI, Google Scholar - 52.
Chang J, Ferrell JE Jr. 2013 Mitotic trigger waves and the spatial coordination of the*Xenopus*cell cycle.**Nature**, 603–607. (doi:10.1038/nature12321) Crossref, PubMed, ISI, Google Scholar**500** - 53.
Yang Q, Ferrell JE Jr. 2013 The Cdk1-APC/C cell cycle oscillator circuit functions as a time-delayed, ultrasensitive switch.**Nat. Cell Biol.**, 519–525. (doi:10.1038/ncb2737) Crossref, PubMed, ISI, Google Scholar**15** - 54.
Rombouts J, Vandervelde A, Gelens L . 2018 Delay models for the early embryonic cell cycle oscillator.**PLoS ONE**, e0194769. (doi:10.1371/journal.pone.0194769) Crossref, PubMed, ISI, Google Scholar**13**