1 Introduction Travellingwave solutions of parabolic equations on the real line arise in a variety of applications. An important issue is their stabi


 Stanley Gibson
 1 years ago
 Views:
Transcription
1 Essential instability of pulses, and bifurcations to modulated travelling waves Bjorn Sandstede Department of Mathematics The Ohio State University 231 West 18th Avenue Columbus, OH 4321, USA Arnd Scheel Institut fur Mathematik I Freie Universitat Berlin Arnimallee Berlin, Germany Abstract Reactiondiusion systems on the real line are considered. Localized travelling waves become unstable when the essential spectrum of the linearization about them crosses the imaginary axis. In this article, it is shown that this transition to instability is accompanied by the bifurcation of a family of large patterns that are a superposition of the primary travelling wave with steady spatiallyperiodic patterns of small amplitude. The bifurcating patterns can be parametrized by the wavelength of the steady patterns; they are timeperiodic in a moving frame. A major diculty in analyzing this bifurcation is its genuinely innitedimensional nature. In particular, nitedimensional LyapunovSchmidt reductions or centermanifold theory do not seem to be applicable to pulses having their essential spectrum touching the imaginary axis. AMS subject classication: 35B32, 35K57 Keywords: pulses, homoclinic bifurcations, essential spectrum
2 1 Introduction Travellingwave solutions of parabolic equations on the real line arise in a variety of applications. An important issue is their stability since it is expected that only stable travelling waves can be observed. Once stability has been proved over a certain range of parameters, the dynamics near a travelling wave is predictable: any solution nearby is attracted to the travelling wave or an appropriate translate of it. In that respect, interesting parameter values are those at which a transition to instability occurs. Near such transitions, other and possibly more complicated patterns may bifurcate from the primary travelling wave. A travelling wave becomes unstable if a subset of the spectrum of the linearization about it crosses the imaginary axis. One possibility is that this subset consists of isolated eigenvalues. The resulting bifurcation problem can be analyzed using standard centermanifold theory. In this article, we focus on a qualitatively dierent mechanism that also leads to instability, namely that of essential spectrum crossing the imaginary axis. We call this instability mechanism an essential instability. This route to instability is considerably more dicult to analyze since it is genuinely innitedimensional. In fact, to our knowledge, this transition has not been investigated previously except for small fronts [2]. Another related article is [2] where essential instabilities of fronts are studied utilizing numerical simulations and the analysis of a series of caricature problems. The waves investigated therein are of a quite dierent nature and we refer to Section 4 for a discussion. We distinguish between several kinds of travelling waves. Pulses are travelling waves that converge to the same asymptotic state as the spatial variable x tends to 1. Fronts, on the other hand, connect dierent asymptotic states at 1. Periodic wave trains are travelling waves that are periodic in the spatial variable. In the following, we focus on pulses. Most of the results presented in this article apply also to fronts and wave trains, and we discuss these generalizations in Section 4. Essential instabilities of pulses are caused by an instability of the asymptotic equilibrium state of the pulse. Suppose that h(x? c t) is a pulse that moves with speed c to the right. Its essential spectrum corresponds to smallamplitude waves of the form e ikx+t that are created at the asymptotic state of the pulse, that is, at its tails. The wavenumber k and the eigenvalue satisfy a certain dispersion relation. At the onset of instability induced by the essential spectrum, there exist then waves of the form e i(kx+!t). There are four qualitatively dierent cases corresponding to all possible combinations of! and k being zero or nonzero. We focus here on the case of a stationary bifurcation, that is! =, for nonzero k. When both the wavenumber k and the eigenvalue! are zero, the situation is actually considerably simpler since it corresponds to a homoclinic bifurcation in an ordinary dierential equation where the equilibrium undergoes a pitchfork bifurcation. On the other hand, the case of nonzero! is similar to the case! =, and we refer to Section 4 for a discussion. 1
3 t?c 2 k Figure 1: The superposition of the localized solitary wave h with small patterns that have a 2 k  periodic spatial structure is shown in a frame moving with speed c >. The smallamplitude patterns move with speed?c relative to the pulse h. The solution shown here has period 2! in time in the moving coordinate frame where! = c k. From now on, we concentrate on situations where! = and k = k are nonzero, that is, we assume that the dispersion relation is satised by = and some k 6=. The associated local bifurcation close to the equilibrium state is known as the Turing instability; see Sections 3.2 and 3.4. It generates small patterns of the form e ik x that are often referred to as Turing patterns. These patterns possess a spatially oscillating structure with period 2 k. We seek timeperiodic modulations of the pulse that are reminiscent of a linear superposition of these small steady patterns and the large localized pulse. In a coordinate system = x? c t moving with the speed c of the pulse, the steady patterns e ik x become travelling waves e ik (+c t) =: e i(k +! t) with! := c k. They move with speed?c relative to the pulse h and have period 2! in time. In this moving frame, the modulated pulse looks roughly like Ae i(k +! t) + h() for small A; see Figure 1. We remark that Turing patterns bifurcate for any wavenumber k close to k. Correspondingly, we expect to nd modulated pulses with asymptotic wavenumber k and temporal frequency! = c k for any k close to k. For the sake of clarity, we rst seek modulated pulses with temporal frequency! = c k. Only at the end of the analysis, in Section 3.7, we show how modulated pulses with other temporal periods can be obtained. To set the scene, consider u t = Du xx + f(u; ); x 2 R; (1.1) where u 2 R n and f(; ) = for all. Casting (1.1) in a frame = x? c t moving with 2
4 i! ik (a) (b) Figure 2: The essential spectrum of the parabolic reactiondiusion system is shown in (a). Picture (b) shows the spectrum of the associated elliptic system. speed c, we obtain u t = Du + c u + f(u; ); 2 R; (1.2) which then has the equilibrium h() for =. We are particularly interested in localized waves satisfying lim jj!1 h() =. The stability of h is determined by the spectrum spec(l) of the linearization Lw := Dw + c w u f(h(); )w (1.3) of (1.2) about h. The essential spectrum of L is the complement in spec(l) of the set of isolated eigenvalues with nite multiplicity. It contains the spectrum of the linearization about the asymptotic state u = that consists of all points in the complex plane such that det(?k 2 D + ikc u f(; )? ) = for some k 2 R. This equation is the aforementioned dispersion relation in a coordinate frame moving with the pulse. We assume that the essential spectrum of the asymptotic state for = has the form depicted in Figure 2(a). With the transformation ~ =?ikc, we recover the dispersion relation det(?k 2 D u f(; )? ) ~ = in the original steady coordinate frame. Note that real solutions ~ of this dispersion relation are double zeroes corresponding to wavenumbers k. Since the critical eigenvalues = i! are not isolated in the spectrum, it is dicult to reduce the dimension by applying LyapunovSchmidt reduction or centermanifold theory. Often, modulation equations have been used to describe the dynamics near homogeneous steady states. In the aforementioned context of bifurcations from equilibria, a partial justi cation of the approximation by modulation equations, that is GinzburgLandau equations, has been achieved in [7]; see also [3, 12] and the references therein. Modulation equations close to a pulse would have to capture both the dynamics close to the asymptotic equilibrium state of the pulse, typically described by a GinzburgLandau equation, and the global 3
5 interaction of the modulation through the pulse. An attempt to derive such a modulation equation, at least formally, has been made in [1]. However, the resulting GinzburgLandau equation is still dicult to analyze. We therefore resort to a completely dierent approach and cast the parabolic equation (1.2) as an elliptic equation on the space of 2! periodic functions, namely d u v = ; t 2 S 1 = R= 2 d v D?1! (u t? c v? f(u; )) Z: (1.4) In other words, we anticipate the temporal period 2! and then reverse the role of time and space by viewing (1.4) as an evolution equation in. The restriction to 2! periodic functions is very ecient since most of the essential spectrum disappears for the initialvalue problem (1.4). In fact, the spectrum of the linearization of (1.4) about (u; v) = has a pair of isolated imaginary eigenvalues ik ; see Figure 2(b). Thus, we expect a Hopf bifurcation leading to spatially periodic solutions with small amplitude. Here, we recover precisely the Turing patterns of the form e ik x = e i(k +! t). On the other hand, the travelling wave h() corresponds to a timeindependent homoclinic solution (h; h )() of (1.4). We then seek solutions close to (h; h )() that are homoclinic to the aforementioned small periodic waves. If (1.4) were an ODE, we could readily investigate the existence of such connections by studying intersections of suitable global invariant manifolds associated with the periodic waves. However, the initialvalue problem for (1.4) is illposed. Indeed, Figure 2(b) indicates that the stable and unstable eigenspaces are both innitedimensional, and semigroup theory fails. Therefore, it is not clear whether global invariant manifolds exist or whether dynamicalsystems techniques can be used at all to investigate elliptic equations such as (1.4). In this article, we construct global stable and unstable manifolds near the given pulse (h; h ) and study their intersections upon changing the parameter. We use exponential dichotomies for elliptic equations to accomplish this construction. Exponential dichotomies are a wellknown technique for ODEs and parabolic PDEs; see, for instance, [4, 6, 13]. For elliptic equations, however, there are major technical obstacles to their global existence that have been resolved only recently in [14]. The idea of using spatial dynamics has been introduced in [9]. Since then it has been used extensively in order to investigate bifurcations from spatially homogeneous equilibria to small steadystate or timeperiodic solutions; see, for instance, [2, 7, 15]. Typically, the resulting elliptic system is reduced to a nitedimensional equation that describes small solutions near the homogeneous steadystate. The reduced equation can then be investigated using bifurcation theory. The problem analyzed in the present article, however, involves a large pulse solution that is not close to the equilibrium state. A nitedimensional reduction to a center manifold for the spatial dynamics is not known in this context. This paper is organized as follows. In the next section, we present a fourdimensional 4
6 model problem. The model reects the essential features of the part of the bifurcation we are interested in, though we believe that it is inadequate for a complete description. The analysis of the innitedimensional problem, including all necessary hypotheses, is then carried out in Section 3. The main result, the bifurcation of modulated travelling waves asymptotic to spatiallyperiodic steady patterns, is stated in Section 3.7. We conclude in Section 4 with a discussion and generalizations of the result. 2 A nitedimensional model problem In this section, we outline the bifurcation that occurs in the elliptic problem (1.4) when the essential spectrum of (1.3) crosses the imaginary axis. For the sake of clarity, we utilize a fourdimensional model that mimics precisely the bifurcation we are interested in. Let (u ; u 1 ) 2 R 2 R 2 satisfy the dierential equation d d u = f (u ; c; ) (2.1) d d u 1 = f 1 (u ; u 1 ; c; ): The reader may think of (u ; u 1 )(x) as the zeroth and rst Fourier coecients of the t periodic function (u; u )(t; ) dened in (1.4); see also equation (3.4) below. We assume that the subspace u 1 = is invariant for all values of the parameters, that is f 1 (u ; ; c; ) =. The dynamics in the subspace u 1 = is then governed by the equation d d u = f (u ; c; ): (2.2) Suppose that (2.2) has a homoclinic solution to the hyperbolic equilibrium u = for (c; ) = (c ; ). This homoclinic orbit corresponds to the pulse solution h of equation (1.2). We assume that the homoclinic orbit of (2.2) is transversely unfolded by the parameter c, that is, stable and unstable manifold of the origin cross each other with nonzero speed upon varying c near c = c. As for the second equation in (2.1), we assume an S 1 equivariance with respect to the rotations in R 2. This symmetry represents the time shift of nonzero solutions of (1.2). Finally, upon varying the parameter, suppose that the equilibrium (u ; u 1 ) = undergoes a nondegenerate Hopf bifurcation in R 4 with critical eigenspace f(u ; u 1 ); u = g. Factoring out the S 1 symmetry in the u 1 variable, we are left with a threedimensional ODE having a homoclinic orbit to an equilibrium in a twodimensional owinvariant subspace. Moreover, the equilibrium experiences a pitchfork bifurcation in the direction transverse to this subspace; see Figure 3. Upon changing c, unstable and centerstable manifold of the origin cross each other with nonvanishing speed in the threedimensional space. If the pitchfork bifurcation is supercritical, the bifurcating equilibrium has a onedimensional 5
7 Figure 3: The picture on the left shows the homoclinic orbit (h; h ) of the elliptic system at the bifurcation point. The vertical axis corresponds to the center direction in which a supercritical Hopf bifurcation takes place. The two horizontal directions coincide with the invariant subspace u 1 =. The picture on the right shows the homoclinic solution connecting the bifurcating periodic solution to itself. unstable manifold, which is close to the (strong) unstable manifold of the origin. Therefore, upon changing c, the unstable manifold of the bifurcating equilibrium also crosses the centerstable manifold of the origin due to the persistence of transverse crossings under perturbations. The unique intersection curve corresponds to a homoclinic orbit to the bifurcating equilibrium since the origin is unstable within the center manifold; see Figure 3. The main dierence between the elliptic problem (1.4) and our model problem is that the phase space for the former equation is innitedimensional, and both the unstable and the centerstable manifold of the origin are innitedimensional. Even the existence of these manifolds far away from the equilibrium is not evident as we do not have a ow to propagate local invariant manifolds. 3 Bifurcations of timeperiodic travelling waves 3.1 The parabolic and elliptic equation We consider the semilinear parabolic equation u t = Du xx + f(u; ); x 2 R; (3.1) where u 2 R n, D is a diagonal matrix with positive entries, and f : R n R! R n smooth nonlinearity with f(; ) = for all. is a Hypothesis (TW) Assume that h(x?c t) is a travellingwave solution of (3.1) for = and some c 6= such that h() tends to zero exponentially as jj tends to innity. 6
8 Transforming (3.1) into the moving frame (; t) = (x? ct; t), we obtain u t = Du + cu + f(u; ); 2 R; (3.2) which then admits the equilibrium h() for (c; ) = (c ; ). Equation (3.2) is wellposed on the space X := C unif (R; R n ) of bounded and uniformly continuous functions on R; see, for instance, [6]. Here, we consider strong solutions u(t) of (3.2) that are dierentiable as functions into X, continuous with values in C 2 unif and satisfy (3.2) in X. Next, we cast the parabolic equation (3.2) as an elliptic equation d u d v = v D?1 (u t? cv? f(u; )) ; (3.3) reversing the role of time and space. The functions U = (u; v) are contained in Y := H 1 2 per (; 2! ) L 2 per(; 2! ) for some! > which we specify below. The nonlinearity f maps H 1 2 per into L 2 per provided it has at most polynomial growth. If f has faster growth, we may consider (3.3) on the space H 1 per H 1 2 per. There are then no restrictions on f necessary and the analysis presented below is still valid. We say that (u; v)() is a solution of (3.3) if (u; v)() is dierentiable in as a function into Y, continuous with values in H 1 perh 1 2 per and satises (3.3) in Y. We emphasize that the initialvalue problem for (3.3) is not wellposed on Y. On the space Y, we have the S 1 action ( U)(t) := U(t + ) with 2 R= 2! Z. Note that (h(); h ()) satises (3.3) for (c; ) = (c ; ). We may think of this solution, which is contained in the xedpoint space Fix(S 1 ) of the S 1 action, as a homoclinic orbit to the zero equilibrium. Throughout, we utilize the Fourier series of elements (u; v) 2 Y and identify (u; v) with its Fourier coecients (u`; v`)`2z where Note that j(u; v)j 2 Y = juj 2 H 1 2 (u(t); v(t)) = + jvj 2 L 2 = X`2Z X `2Zu`e i`!t ; X`2Z (1 + j`j)ju`j 2 + jv`j 2 v`e i`! t : (3.4) =: X`2Zj(u`; v`)j 2`: (3.5) Let Y` = span u`;v`;u?`;v?`2r nf(u`; v`)e i`! t ; (u?`; v?`)e?i`! t g equipped with the norm j j`. 7
9 3.2 The linearization about u = Setting (c; ) = (c ; ), we linearize (3.2) about u = and obtain the linear constantcoecient operator L 1 w = Dw + c w u f(; )w: First, we calculate the spectrum of L 1 on X. Dene Owing to [6, Theorem A.2], we have d(; ) := det( 2 D + c u f(; )? ): (3.6) spec(l 1 ) = f 2 C ; d(; ik) = for some k 2 Rg; (3.7) since w() = e ik w is then a bounded eigenfunction associated with the eigenvalue for some nonzero w 2 C n. Hypothesis (P1) Assume that spec(l 1 ) \ ir = fi! g for some! >. Furthermore, we assume that d(; ik) = for close to i! if, and only if, k is close to k =! c = (k) = i! + ic (k? k )? C r (k? k ) 2 + O(jk? k j 3 ); (3.8) where C r > is real and c 6= denotes the wave speed. Finally, we assume d(; )j (i! ;ik ) 6=. Hypothesis (P1) states that the essential spectrum of L 1 touches the imaginary axis at = i!. The corresponding eigenfunction e ik w H is unique, up to constant multiples, and has a nontrivial spatial structure since k 6=. Note that the particular form of the dispersion relation (3.8) follows from a generic assumption on the bifurcation in the steady coordinate frame. Indeed, consider the operator Its dispersion relation for = ik is and L 1 w := Dw u f(; )w: (3.9) det(?k 2 D u f(; )? ) = : (3.1) Eigenvalues of L 1 transform into eigenvalues of L 1 via (k) = (k) + ikc. In Hypothesis (P1), we have assumed that only k = k satisfy (3.1) for =. In addition, we assumed in (P1) that the derivative of (3.1) with respect to k evaluated at (; k) = (; k ) is not zero. Hence, there are unique solutions (k) satisfying (3.1) for k near k with (k ) =. Note that (3.1) is symmetric with respect to k!?k. Therefore, we conclude that (k) are both realvalued. Summarizing, Hypothesis (P1) is satised by an open set of oneparameter families. Many reactiondiusion systems that satisfy (P1) 8
10 are known. One example is the Brusselator; see [5, Ch.VII, x5] or, for the rst reference to Turing instabilities, [21]. Next, we compute the spectrum of the linearization! id A 1 = D?1 u f(; ))?c D?1 of (3.3) at the equilibrium U = considered in the space Y with! chosen as in (P1). We remark that we may consider the space Y for any frequency! close to! ; see Section 3.7. Lemma 3.1 Suppose that (P1) is met. The operator A 1 has then two simple eigenvalues ik on the imaginary axis with eigenfunctions e i! t U H and e?i! t U H, respectively, for some nonzero U H 2 C 2n, while the rest of its spectrum is uniformly bounded away from the imaginary axis. The operator A 1 has compact resolvent. Furthermore, there are constants 6= small and K > such that k(a 1 + (? ik) id)?1 k L(Y ) K 1 + jkj for all k 2 R. Finally, there exist spectral projections P u, P c and P s in L(Y ) corresponding to eigenvalues of A 1 with positive, zero and negative real part, respectively. Proof. Let V = (V 1 ; V 2 ) 2 Y. We have A 1 V = V if, and only if, V 2 = V 1, and ( 2 D + c u f(; t )V 1 = : Upon exploiting the Fourier series (3.4) of V with coecients (a`; b`), we see that 2 spec(a 1 ) if, and only if, det( 2 D + c u f(; )? i`! ) = d(i`! ; ) = for some ` 2 Z. It follows from (P1) and (3.7) that = ik are the only eigenvalues of A 1 on the imaginary axis. These eigenvalues are simple since the algebraic multiplicity of ik coincides with the order of ik as a zero of the determinant d(i! ; ) with respect to. By Hypothesis (P1), this order is equal to one. In particular, A 1 is invertible on Y. It is clear that the inverse is compact since the domain H 1 per H 1 2 per of A 1 is compactly embedded into Y. Next, we consider the eigenvalue problem for A 1. Note that the Fourier subspaces Y` are invariant under A 1. The associated eigenvalue problem for the Fourier coecients (a`; b`) is given by!? id a` = : (3.11) D?1 u f(; ))?? c D?1 b` 9
11 In order to prove the remaining claims on the resolvent and spectral splittings, it suces to investigate (3.11) for ` 2 Z with j`j large. We then scale a` = p 1 b` = j`j^a`; ^b`: (3.12) This rescaling accounts for the norm on Y`; see (3.5). In particular, j(a`; b`)j 2` = j`j ja`j 2 + jb`j 2 = j^a`j 2 + j^b`j 2 =: j(^a`; ^b`)j 2 : p We also rescale the eigenvalue = j`j^. The eigenvalue problem then reads 1?^ A ^a` D?1 (i! sign `? j`j uf(; ))?^? p = 1 ; (3.13) c D?1 j`j ^b` which has a nontrivial solution if, and only if, det ^ 2 D + ^c p + 1 j`j uf(; )? i! sign ` = : (3.14) Taking the limit j`j! 1 gives?^ D?1 i! sign ` id?^!^a` ^b` = (3.15) and det(^ 2 D? i! sign `) = ; respectively. The last equation has 2n solutions ^ j which are not imaginary and independent of `. By Rouche's p Theorem, there are then 2n zeroes of (3.14) near the set f^ j g. The rescaling = j`j^ shows that the real parts of the corresponding eigenvalues are actually unbounded as j`j! 1. Similarly, the spectral projections associated with the limiting problem (3.15) perturb to spectral projections of (3.13) in Y` that are bounded uniformly in `. Due to the denition of the norms on Y and the rescaling (3.12), the lemma is proved. 3.3 The linearization about the travelling wave We consider the linearizations of (3.2) and (3.3) about the pulse h() for (c; ) = (c ; ). For the parabolic equation, dene Lw = Dw + c w u f(h(); )w for w 2 X. The variational equation about the homoclinic solution (h; h )() of the elliptic equation (3.3) is given by V = A()V =! id V (3.16) D?1 u f(h(); ))?c D?1 1
12 with V 2 Y. Note that the Fourier subspaces Y` are invariant under A() since h() does not depend on t. In Y`, equation (3.16) reads d d a` b` =! id a` D?1 u f(h(); ))?c D?1 b` The next lemma characterizes the set of bounded solutions of (3.16). : (3.17) Lemma 3.2 Assume that Hypothesis (P1) is met. We then have = i`! 2 spec(l) for some ` 2 Z if, and only if, there exists a bounded solution V (; t) = e i`! t V () of (3.16) dened for 2 R. Proof. If V (; t) = e i`! t V () is a bounded solution of (3.16) on R, then V () = (w; w )(), and w() lies in the null space of L? i`! : Moreover, w 2 X. Therefore, i`! 2 spec(l). Dw + cw u f(h(); )w = i`! w: (3.18) Next, suppose that = i`! 2 spec(l). If j`j 6= 1, then is not contained in the essential spectrum by (P1). Hence, the eigenfunction associated with i`! is localized, and therefore corresponds to a bounded solution of (3.18). It remains to consider the case = i!. We seek bounded solutions of (3.17) with ` = 1, that is, d a1 = d b 1 id D?1 u f(h(); ))?c D?1! a1 b 1 : (3.19) Due to Hypothesis (P1), we have 2 spec(l 1 ). Moreover, Lemma 3.1 shows that the spectrum of the asymptotic matrix A 1 :=! id D?1 u f(; ))?c D?1 dened on Y 1 has two simple imaginary eigenvalues ik, while the other eigenvalues have nonzero real part. Since the function h() converges to zero exponentially as jj! 1, we can now apply ODE results on exponential dichotomies [4, 13]. Hence, there are two subspaces E cs 1 () and Ecu () of Y 1 1 such that solutions of (3.19) with initial values in E cs() 1 or E cu 1 () are bounded for! 1 or!?1, respectively. Furthermore, dim E cs 1 () = #f 2 spec(a 1 ); Re g; dim E cu 1 () = #f 2 spec(a 1 ); Re g; counted with multiplicity; see [4]. In particular, dim E cs 1 () + dim Ecu () = dim Y Therefore, any solution of (3.19) with initial value in E c 1 () := Ecs 1 () \ Ecu 1 () is bounded on R. Moreover, dim E c() 2, and therefore 1 Ec 1 () contains nontrivial initial values. 11
13 Note that the lemma would be wrong if the limiting matrix A 1 were containing a nontrivial Jordan block corresponding to the eigenvalue = i!. In this situation, even though 2 spec(l), there would in general be no bounded solution of (3.16) since solutions are expected to grow linearly in. Actually, we have proved much more. Using the notation introduced in the proof above, the set E c 1 () of bounded solutions of (3.17) with j`j = 1 is at least twodimensional. If we modify the nonlinearity f(u; ) by adding a small rotation normal to the homoclinic orbit h, we can arrange that E c 1 () is twodimensional. Furthermore, by the same argument, solutions associated with initial values in E c 1 () do generically not decay exponentially as jj! 1 but oscillate. In other words, generically in f(; ), we have E c() \ 1 Es 1 () = fg and E c() \ 1 Eu() = fg where 1 Es() and 1 Eu() are subspaces of Y 1 1 such that solutions of (3.19) with initial values in E s() or 1 Eu 1 () decay exponentially for! 1 or!?1, respectively. Using similar arguments, = i`! is generically not in the spectrum of L for j`j > 1. Note that = 2 spec(l) with eigenfunction h by translation invariance. This eigenvalue is typically simple. For generic nonlinearities f(u; ), the following hypothesis is therefore met. Hypothesis (P2) (i) = 2 spec(l) is a simple eigenvalue. (ii) (L? i! )w = has a unique, up to constant complex multiples, nonzero bounded solution w c (), and we have jw c ()? e ik w j! as! 1 for appropriate H nonzero vectors w 2 C n. H (iii) = i`! is not in spec(l) for ` > 1. On account of Hypothesis (P2) and Lemma 3.2, the subspace of initial values in Y associated with bounded solutions of (3.16) is given by E c () = spanf(h; h ) (); w c ()e i! t ; w c ()e?i! t g: (3.2) Our next goal is to solve (3.3) using the information gathered so far. Unfortunately, the initialvalue problem for (3.3) is not wellposed on Y. Under certain circumstances, however, (3.3) can be solved in forward or backward direction for initial values in certain depending subspaces of Y. We say that (3.3) has an exponential dichotomy on R + if there are projections P + () dened for with the following property: for any V 2 R(P + ()), there exists a unique solution V () of (3.3) which is dened for such that V () = V. Moreover, V () tends to zero exponentially as! 1, and V () 2 R(P + ()) for all. Similarly, for any V in the null space of P + ( ), there is a unique solution V () of (3.3) 12
14 which is dened for such that V ( ) = V ; furthermore, V () decays exponentially for decreasing with. In other words, for, there are two complementary subspaces, R(P + ()) and R(id?P + ()), such that we can solve the elliptic equation forward and backward in for initial values in R(P + ()) and R(id?P + ()), respectively. Exponential dichotomies on R? are dened analogously; solutions in R(P? ()) decay exponentially as!?1. In the following lemma, we show that equation (3.3) has dichotomies so that we can solve it forward and backward in provided the initial values are contained in appropriate subspaces. The only dierence to the situation described right above is that solutions do not necessarily decay. Lemma 3.3 Assume that Hypothesis (P1) is met. There are bounded operators s +(; ), c +(; ) and u +(; ) dened on Y for, ; and, respectively, such that s +(; )V, c +(; )V and u +(; )V satisfy (3.16) for >, any and <, respectively, and are continuous in (; ) for any V 2 Y. Furthermore, s + satises the evolution property s +(; ) s +(; ) = s +(; ) for any. Analogous properties hold for c +(; ) and u +(; ). Moreover, s +(; ) + c +(; ) + u +(; ) = id; i +(; ) j + (; ) = for i 6= j; where i; j 2 fs; c; ug. Finally, there are constants K > and > such that k s +(; )k L(Y ) e?(?) ; k c +(; )k L(Y ) K; k u +(; )k L(Y ) Ke?(?) (3.21) for any. Similar properties hold for operators s?(; ), c?(; ) and u?(; ) dened for negative and. Proof. The statement of the lemma follows from [14, Theorem 1]. We give another simpler proof that works for the particular case studied here. As mentioned above, the Fourier subspaces Y` are invariant under A() since h() does not depend on t. The Fourier coecients (a`; b`) satisfy equation (3.17) d d a` b` =! id a` D?1 u f(h(); ))?c D?1 b` We can readily solve this equation for any ` 2 Z. Lemma 3.1 shows that the spectrum of the asymptotic operator A 1 j Y` is strictly hyperbolic except when j`j = 1 where it contains two simple imaginary eigenvalues. The case j`j = 1 has been discussed in Lemma 3.2. Hence, we conclude the existence of evolution operators s +;`, c +;` and u +;` in each subspace Y`. In fact, c +;` = except when j`j = 1. Furthermore, the estimates (3.21) are true in Y` for some independent of ` due to Lemma 3.1. : 13
15 It is, however, not clear whether the constant K is independent of ` and whether the resulting evolution operators are bounded on Y. To prove this, it suces to estimate the norm of the evolution operators on the space Y` for large `. Thus, let j`j > 1. Using the scaling (3.12), that is, a` = p 1 and b` = ^b`, we obtain j`j^a` d d ^a` ^b` = p Rescaling the variable by p j`j = ^, we get d d^ ^a` ^b` 1 id D?1 (i! sign `? j`j uf(h(); ))? p 1 c D?1 j`j 1 p id D?1 (i! sign `? j`j uf(h(^= j`j); ))? p 1 c D?1 j`j A ^a` ^b` A ^a` ^b` : : (3.22) Taking the limit j`j! 1, we obtain the equation d d^^a^b id = D?1 i! sign `!^a ; ^b which is independent of ^. The matrix on the righthand side is hyperbolic; see Lemma 3.1. A perturbation argument shows that the evolution operators ^ s +;` and ^ u +;` of (3.22) satisfy k^ s +;`(^; ^)k Ke?( ^?^) ; k^ u +;`(^; ^)k?( ^?^) Ke for ^ ^, where K and are independent of `. Due to the denition of the norms on Y and the rescaling of the variable, the lemma is proved. With Lemma 3.3 at hand, we can dene the subspaces E cs + () = R( s +(; ) + c +(; )); E+() s = R( s +(; )); E? cu () = R( u?(; ) + c?(; )); E?() u = R( u?(; )): For any initial value in E cs + () or E s +(), there exists a solution of (3.16), and it is bounded or exponentially decaying, respectively, as! 1. An analogous characterization is true for E? cu () or E?() u as!?1. Note that the subspace E c () dened in (3.2) is given by E c () = E+ cs () \ E? cu (). Lemma 3.4 Assume that Hypotheses (P1) and (P2) are true. There exists then a nonzero element 2 Y such that (E s +() + E u?()) spanfw c ()e i! t ; w c ()e?i! t g spanf g = Y; and E s +() \ E u?() = spanf(h; h ) ()g. 14
16 Proof. It suces to construct a complement of the stable and unstable subspaces in Y. Note that Y = Fix() is invariant under the nonlinear elliptic equation (3.3). In fact, on Y, (3.3) coincides with the travellingwave equation d u d v = v?d?1 (cv + f(u; )) (3.23) for u 2 R n, which is satised by the wave (h; h )(). The equilibrium u = is hyperbolic, and the intersection T (h;h )()W s () \ T (h;h )()W u () of tangent spaces of the stable and unstable manifolds of (3.23) is onedimensional by Hypothesis (P2)(i). Otherwise, the geometric multiplicity of = would be bigger than one. We may choose as the unit vector in the onedimensional orthogonal complement of T (h;h )()W s () + T (h;h )()W u (). 3.4 Hopf bifurcations near U = in Y We return to the nonlinear reactiondiusion system, rst considered in the original coordinate frame u t = Du xx + f(u; ): Under the assumptions on the linearization L 1 in the steady coordinate frame, see (3.9), spatiallyperiodic steady patterns with wavelength 2 bifurcate typically from the zero solution for k close to the critical wavelength k. This is usually proved using k centermanifold theory or LyapunovSchmidt reduction in a function space of 2 k periodic functions. Next, consider the nonlinear parabolic equation (3.2) u t = Du + cu + f(u; ); 2 R: In a coordinate frame moving with speed c, the aforementioned spatiallyperiodic steady patterns become timeperiodic travelling wavetrains with frequency! = ck. We assume from now on that the wave speed c of the pulse is negative, i.e. c <. If c >, we change 7!? and obtain c < in the new spatial variable. Since the pulse is not spatially periodic, we introduced spatial dynamics on timeperiodic functions. In the next step, we rephrase the aforementioned result on bifurcation to wave trains in terms of the spatial dynamics. Consider the nonlinear elliptic problem (3.3) d u d v = v D?1 (u t? cv? f(u; )) with (u; v) 2 Y. The linearization of (3.3) at U = is given by! id A 1 (c; ) = : D?1 u f(; ))?cd?1 ; 15
17 The operator A 1 (c ; ) has a pair of simple eigenvalues ik with eigenfunctions e i! t U H and e?i! t U H ; see Lemma 3.1. As in (3.6), we dene Note that we have the relation d(; ; c; ) := det( 2 D + c u f(; )? ): which follows immediately from the denition. d(; ; c; ) = d(? (c? c ); ; c ; ); (3.24) For a generic Hopf bifurcation, the eigenvalues ik should cross the imaginary axis with nonzero velocity. We assume the following: Hypothesis (P3) Assume that C 1 =? d(i! ; ik ; c ; d(i! ; ik ; c ; ) >. The reader might check, using (3.24), that the condition C 1 6= is equivalent to the transverse crossing of eigenvalues when considering the temporal dynamics of 2 k periodic functions. If C 1 <, we can transform the parameter 7!? to achieve C 1 > ; see also Remark 3.1 below. Note that the d(i! ; ik ; c ; ) is not equal to zero d 6= by (P1). Upon dierentiating (3.8) with respect to k and using = ik, it follows = c. The eigenvalue ik persists as a simple eigenvalue () of the operator A 1 (c ; ) considered in Y. d = c, we d d = d d : Hence, owing to (P3), the real part of () is given approximately by?c 1 =c. Therefore, with c < and the sign of C 1 as in (P3), the eigenvalues ik cross the imaginary axis from left to right as becomes positive. Furthermore, exploiting (P1), (P3) and (3.24), d(; ; c; ) vanishes for (; ) close to (i! ; ik ) if, and only if, = ik(c? c ) + i! + ic (k? k )? C r (k? k ) 2 + C 1 + O(jj 2 + jk? k j 3 ) = i! + i(ck?! )? C r (k? k ) 2 + C 1 + O(jj 2 + jk? k j 3 ): According to Lemma 3.1, eigenvalues of the linearization A 1 (c; ) are on the imaginary axis precisely when Im =! and Re =, that is = ck?! (3.25) =?C r (k? k ) 2 + C 1 + O(jj 2 + jk? k j 3 ): This equation can be solved for (; k). Thus, A 1 (c; ) has a pair of imaginary eigenvalues whenever k? k =? k c (c? c ) (3.26) = C rk 2 (c? c C 1 c 2 ) 2 + O(jc? c j 3 ): 16
18 Denoting the function in the last equation by = (c), we see that (3.3) has a simple pair of imaginary eigenvalues for (c; ) = (c; (c)) for any c close to c. We introduce new parameters by (c; ) = (c; (c) + ^): (3.27) The Jacobian of this transformation is equal to the identity at (c; ) = (c ; ). Also, imaginary eigenvalues occur precisely for ^ =. Alternatively, we may solve the rst equation in (3.26) with respect to k and obtain (c; ) = (! =k; (k) + ^); where we again use with a slight abuse of notation. Recall that S 1 acts on Y via ( U)(t) = U(t + ). We say that a manifold W is invariant under equation (3.3) for ( ) if, for any U 2 W, there is a solution U() of (3.3) dened for ( ) with U() = U and U() 2 W for suciently small. Lemma 3.5 Assume that Hypothesis (P1) is met. For any (c; ^) close to (c ; ), there exists then a twodimensional, smooth and S 1 invariant centermanifold Wc;^ c () Y that contains U = and is tangent to spanfe i! t U H ; e?i! t U H g at U = for (c; ^) = (c ; ). Furthermore, Wc;^ c () is invariant under (3.3) and smooth in (c; ^). Proof. The lemma follows from results of Mielke [11]; see also [22]. The assumptions in these references are satised due to Lemma 3.1 and 3.3. Hence, the elliptic PDE (3.3) near U = is essentially reduced to an S 1 equivariant ODE on Wc;^ c (). We assume that the Hopf bifurcation is supercritical. Hypothesis (H) Assume that the vector eld on Wc c ;(), projected onto the center eigenspace and in polar coordinates, is given by r =?C 2 r 3, ' = k up to terms of fourth order for some C 2 >. Note that the vector eld on the center eigenspace takes this particularly simple form due to the equivariance with respect to the isometric action of S 1 on the center eigenspace. We remark that the sign of C 2 is not important. The arguments given below work also in the case where C 2 < ; see Remark 3.1 below. The coecient C 2 may be computed explicitly following standard procedures. The linear part of the vector eld on the center manifold is given by the restriction of the linearization, A 1, to the invariant center eigenspace spanfe i! t U H ; e?i! t U H g. By S 1 equivariance, the quadratic terms of the Taylor expansion of the vector eld projected onto this subspace vanish. The computation of the cubic term requires in general the quadratic approximation of the center manifold. However, if the nonlinearity f is cubic, the computation of the cubic 17
19 term simplies greatly: in this situation, the thirdorder term of the vector eld is obtained by simply evaluating and then projecting the nonlinearity onto the center eigenspace. A vector in the center eigenspace is of the general form ze i! t U H + c:c: with z 2 C. We write U H = (u H ; v H ) t where, due to the secondorder structure of the equation, v H = ik u H. Evaluating f and projecting onto the subspace spanfe i! t U; U 2 R 2n g, we obtain 3 uf(; )(ze i! t u H ; ze i! t u H ; ze i! t u H ) + c:c: up to fourth order. In order to compute the equation on the center manifold, we have to multiply with the left eigenvector U = H (u ; H v H ) which satises! ik U H = U id H : D?1 u f(; ))?c D?1 For cubic nonlinearities, the cubic coecient C 2 is therefore given by C 2 = 1 2 v 3 u f(; )(u H; u H ; u H ) + c:c:): Hypotheses (P3) and (H) are related to the signs of the coecients in the GinzburgLandau equation A t = A xx + 1 A? 2 jaj 2 A associated with (1.1) near u =. Indeed, Hypothesis (P3) implies 1 >, while (H) enforces 2 >. We may now apply the S 1 equivariant Hopf bifurcation theorem and obtain the following lemma. Lemma 3.6 Assume that Hypotheses (P1){(P3) and (H) are satised. There is then a family? k;^ () 2 Y of periodic solutions of (3.3) with (c; ) = (! =k; (k) + ^) dened for k close to k and ^ small. These solutions are C 2 in k uniformly in ^. Moreover, they are relative equilibria, that is,? k;^ (; t) = k!? k;^ () (t) =? k;^ ; t + k :! In particular,? k;^ has period 2 2 in and k! in t. Furthermore,? k;^ is stable with respect to the dynamics on Wc;^ c (). Finally, we have the expansion for some A H 6=.? k;^ (; t) = A H p^ e i! t U H + O(jk? k j p^ + j^j) (3.28) Proof. The lemma follows from the standard S 1 equivariant Hopfbifurcation theorem. We obtain a family? c;^ of periodic solutions parametrized by (c; ^). Using the relations (3.25) and (3.27), it is easy to see that we can parametrize the periodic solutions also by the spatial wavenumber k. The relation c =! =k follows since we deal with steady patterns of spatial period 2 k considered in a coordinate frame moving with speed c. 18
20 Of course, the family of solutions? k;^ is precisely the family of Turing patterns that we would have obtained via standard LyapunovSchmidt reduction for the temporal dynamics. The periodic solutions? k;^ () of (3.3) correspond to solutions k; (; t) of (3.2) with = (k) + ^. The Turing patterns k; have period 2! in t and 2 in. Furthermore, they k satisfy k; (; t) = k; (? ct; ). In the original frame (x; t), their wave speed is zero. For any ~c, the function ~? ~c;k;^ (; t) :=? k;^ (? ~ct; t) =? k;^ ; (1? k~c! )t + k! satises (3.3) with (c; ) = (! k + ~c; (k)+ ^) and has frequency! =!?k~c in t. However,? k;^;! () is not contained in Y but in the space of 2! periodic functions. Solving the equation for!, we obtain ~c =!?! and we set? k k;^;! () = ~? ~c;k;^ (). Remark 3.7 The rst component k;;! (; t) of? k;?(k);! satises (3.2) for (k) and c =! k. It has period 2! in t and 2 k in. 3.5 Existence of invariant manifolds We state existence results for the global centerstable manifold W cs;+ () of the equilibrium c;^ U = and the local unstable manifold W u;loc c;^ (? c;^()) of the periodic solution? c;^ (). The key to obtain these manifolds are the exponential dichotomies derived in Lemma 3.3. In this section, we parametrize the periodic waves by (c; ^) rather than using (k; ^). Proposition 1 Assume that Hypothesis (P1) is satised. Equation (3.3) has then a C 2  smooth, locally invariant centerstable manifold W cs;+ c;^ () which is tangent to Ecs + () at (h; h )() for (c; ^) = (c ; ). It contains all solutions that stay close to (h; h )() for all >. Moreover, W cs;+ c;^ () is C 2 smooth in (c; ^). Proof. If we parametrize a neighborhood of (h; h )() by U = (h; h ) + V, we obtain the equation! id V = V D?1 u f(h; ))?c D?1 + D?1 u f(h; )V 1 + f(h; (c) + ^)? f(h + V 1 ; (c) + ^)? (c? c )V 2 ) for V = (V 1 ; V 2 ) 2 Y. Since Y is a Hilbert space, there exists a smooth cuto function (hv; V i Y ). We dene the modied nonlinearity G(; V; c; ^) := (hv; V i Y ) D?1 u f(h(); )V 1 + f(h(); (c) + ^)? f(h() + V 1 ; (c) + ^)? (c? c )V 2 ) 19 :
21 The linear equation V = A()V has been solved in Lemma 3.3. For the constant appearing in Lemma 3.3 and any number with < <, we dene Z + := fv 2 C (R + ; Y ); sup e? jv ()j Y =: jv j < 1g: We seek the solutions in the centerstable manifold as xed points of the equation V () = cs +(; ) V cs + Z 1 Z + cs +(; )G(; V (); c; ^) d (3.29) u +(; )G(; V (); c; ^) d; where V cs 2 E cs + () and V 2 Z +. It follows from the estimates obtained in Lemmata 3.1 and 3.3 that the hypotheses in [22] are met. The proposition is then a consequence of the results presented in [22]. Note that any solution of the integral equation is actually a smooth solution; see [14, Lemma 3.1]. Similarly, we obtain the global centerunstable manifold W cu;? () of U = that enjoys c;^ the analogous properties for!?1. Finally, we construct the local unstable manifold of? c;^ (). Proposition 2 Assume that Hypotheses (P1){(P3) and (H) are satised. For any (c; ^) with jc? c j and ^ small, equation (3.3) has a C 2 smooth, locally invariant unstable manifold W u;loc c;^ (? c;^()) which is tangent to R(P u ) at U = for (c; ^) = (c ; ). It consists precisely of those solutions U that stay in a small neighborhood of U = for and satisfy ju()?? c;^ ()j Ke as!?1. smooth in c. Moreover, W u;loc c;^ (? c;^()) is continuous in ^ in the C 2 topology and C 2  Here, > and the projection P u have been dened in Lemma 3.3 and 3.1, respectively. Proof. V = We use the parametrization U() =? c;^ () + V () and obtain the equation! id V D?1 u f(? c;^ ; (c) + ^))?cd?1 + D?1 u f(? c;^ ; (c) + ^)V 1 + f(? c;^ ; (c) + ^)? f(? c;^ + V 1 ; (c) + ^)) for V = (V 1 ; V 2 ) 2 Y. As before, we dene the modied nonlinearity G(; V; c; ^) := (hv; V i Y ) D?1 u f(? c;^ (); (c) + ^)V 1 + f(? c;^ (); (c) + ^)? f(? c;^ () + V 1 ; (c) + ^)) 2 :
22 It follows from the roughness theorem for exponential dichotomies proved in [14] that the linear equation V =! id V D?1 u f(? c;^ (); (c) + ^))?cd?1 has evolution operators cs c;^ (; ) and u c;^ (; ) dened for provided jc? c j and ^ > are small. The evolution operators satisfy the estimates k cs c;^(; )k L(Y ) K; k u c;^(; )k L(Y ) Ke?(?) and k cs c;^ (; )? cs c ;(; )k L(Y ) + k u c;^ (; )? u c ;(; )k L(Y ) K(jc? c j + p^) for ; see [14]. For any with < <, we dene Z? := fv 2 C (R? ; Y ); sup e? jv ()j Y =: jv j < 1g: We seek the unstable manifold as a xed point of the equation V () = u c;^(; ) V u + Z?1 Z + u c;^(; )G(; V (); c; ^) d (3.3) cs c;^(; )G(; V (); c; ^) d; where V u 2 Eu and V 2 Z?. Since G(; V; c; ^) = O(jV j2 Y ) uniformly in c and ^, the nonlinearity G is C 2 as a map from Z? into itself. By the uniformcontraction theorem, there exists a unique xed point of (3.3) that depends smoothly on V u and c, and continuously on ^. This proves the proposition. By the above proof and equation (3.28), W u;loc c;^ (? c;^()) is given by where V u 2 Ecu. U =? c;^ () + u c;^(; ) V u + O(jV u j 2 Y ) (3.31) = A H p^ e i! t U H + V u + O(j^j + p^(jc? c j + jv u j Y ) + jv u j2 Y ); 3.6 Transversality We seek solutions of (3.3) connecting the bifurcating periodic solution? c;^ with itself. Therefore, we are interested in intersections of the local unstable manifold W u;loc c;^ (? c;^()) with the global centerstable manifold W cs;+ c;^ (). For (c; ^) = (c ; ), the former manifold coincides with W u;loc c (). We may then shift the variable such that (h; h ; )() is contained 21
23 in the local unstable manifold W u;loc u;loc cs;+ (). In particular, W () and W () intersect c ; c ; along the homoclinic solution U() = (h; h )(). In order to nd intersections for ^ 6=, we consider the suspended manifolds ~W cs;+ ^ := f(u; c); jc? c j < ; U 2 W cs;+ c;^ ()g ~W u;? ^ := f(u; c); jc? c j < ; U 2 W u;loc c;^ (? c;^())g as manifolds in Y R. Note that they are indeed C 2 due to the propositions proved above. For ^ =, these manifolds intersect along (U; c) = ((h; h ); c ). Lemma 3.8 For ^ =, we have T ((h;h )();c ) ~ W cs;+ \ T ((h;h )();c ) ~ W u;? = spanf((h; h ) (); )g; T ((h;h )();c ) ~ W cs;+ + T ((h;h )();c ) ~ W u;? = Y R: In other words, the suspended manifolds intersect transversely in the extended phase space Y R. Proof. We observe that? c ;() = vanishes identically for all. The tangent spaces of ~W cs;+ and W ~ u;? are given by T ~ ((h;h )();c ) W cs;+ = (E+() s fg) + (E c () fg) + spanf( V ~ cs;+ c ; (); 1)g; T ~ ((h;h )();c ) W u;? = (E?() u fg) + spanf((h; h ) (); )g + spanf( V ~ u;? (); 1)g: The tangent vector V ~ cs;+ c ; () of the centerstable manifold in the cdirection can be calculated by taking the derivative of (3.29) with respect to c at V cs =. Similarly, V ~ u;? c ; () is the derivative of (3.3) with respect to c at V u = (h; h )(). Computing these derivatives, we obtain the expressions ~V u;? c () ; = Z ~V cs;+?1 c ; () = Z 1 u?(; ) cs +(; )?D?1 h ()?D?1 h () On account of Lemma 3.4, it suces to prove that D E D ; V ~ cs;+ c () ; 6= ; V ~ u;? c () ; E; that is, D Z ; u?(; )(;?D?1 h ()) t d??1 Z Note that the integrands are actually contained in Y. lefthand side in (3.32) is given by M := Z 1?1 1 d; d: c ; c ; E cs +(; )(;?D?1 h ()) t d 6= : (3.32) In particular, the term on the h (); (;?D?1 h ()) t i d (3.33) 22
24 where () is the unique, up to constant multiples, bounded solution of the adjoint variational equation d d u =? v! t id u f(h(); )?c D?1 v for (u; v) 2 R n R n. Since zero is a simple eigenvalue by (P2)(i), we can conclude that M, dened in (3.33), is nonzero; see [16, Lemma 5.5]. A similar argument, and more details, can be found in [17, Section 5]. Therefore, for any ^ >, the manifolds W ~ cs;+ ^ and W ~ u;? ^ intersect along a unique line (U^ (); c(^)) that depends on ^. The associated solution U^ () of (3.3) with c = c(^) converges exponentially to? c(^);^ as!?1 by denition. It is also contained in the centerstable manifold W cs;+ c(^);^ (). Lemma 3.9 We have the estimate jc(^)? c j Kj^j. Proof. We consider the suspended local centerunstable manifold ~W cu;? ^ := f(u; c); jc? c j < ; U 2 W cu;loc c;^ ()g; see the comment after Proposition 1. Since the manifolds W ~ cu;? ^ and W ~ cs;+ ^ ^, we can parametrize them locally near ((h; h )(); c ) according to are smooth in ~W cu;? ^ = ((h; h )(); c ) + (V? cu )( V ~ cu c ;(); 1) + O(jc? c j 2 + jv? cu Y + j^j) ~W cs;+ ^ = ((h; h )(); c ) + (V cs + ; ) + (c? c )( V ~ cs c ;(); 1) + O(jc? c j 2 + jv cs + j 2 Y + j^j); where V cu? 2 E cu? () and V cs + 2 E cs + (). Projecting the dierence of elements (U cu (U cs + ; c) in ~ W cu;? ^ and ~ W cs;+ ^, respectively, onto spanf( ; )g, we obtain h( ; ); (U cu? ; c)? (U cs + ; c)i = (c? c )M + O(jc? c j 2 + jv cu? j 2 Y + jv cs + j 2 Y + j^j):? ; c) and Upon inserting the intersection point, the lefthand side vanishes. On the other hand, the distance between the intersection point and (h; h )() is of the order p^. This proves the lemma. It remains to show that U^ () cannot converge to zero as! 1 but approaches the periodic solution? c(^);^. Since it already converges to? c(^);^ for!?1 by construction, it is then a homoclinic orbit to? c(^);^. On account of Lemma 3.6, it suces to show that U^ () is not contained in the stable manifold of the origin W s;+ (). Firstly, we shift time c(^);^ such that (h; h )() is contained in the local unstable manifold of zero and has distance r > small from zero. We shall now estimate the distance between W u;loc c(^);^ (? c;^()) and 23
25 W u;loc c(^);^ (), measured near (h; h )(). Using the expansion (3.31) and Lemma 3.9, we can estimate this distance from below by p^ jah j ju H j? K p^(jv u j Y + jc(^)? c j) p^ ja H j ju H j? K p^(r + ^): Note that we do not have to account for the quadratic terms O(jV u j Y ) in (3.31) since they correspond to the local unstable manifold for ^ =. Hence, they disappear when computing the distance. After choosing r suciently small, we conclude from the above estimate that the aforementioned distance between W u;loc (? c(^);^ c;^()) and W u;loc () is bigger than p^ c(^);^ for some >. On the other hand, the distance between W s;+ u;loc () and W () is of c(^);^ c(^);^ the order ^ since both are smooth in ^. Therefore, W s;+ u;loc () and W (? c(^);^ c(^);^ c;^()) cannot intersect near (h; h )(). 3.7 The homoclinic bifurcation We summarize our ndings in the following existence theorem. Theorem 1 Assume that Hypotheses (H), (P1){(P3) and (TW) are satised. There is then a smooth function (!) with (! ) = (! ) = and (! ) > such that, for any! close to! and any small > (!), the following is true. For a unique wave speed c = c (;!) close to c, equation (3.2) has a solution h ;! (; t) with the following properties. (i) h ;! (; t) is periodic in t with period 2!. In other words, the bifurcating pulse is timeperiodic in an appropriate moving frame. The family h ;! (; ) is continuous in (;!) with values in C (R 2 ; R n ) provided with the local topology. (ii) We have c (;! ) = c and h ;! (; t) = h(). (iii) There exists a constant > such that, for c <, jh ;! (; t)? k(;!);;!( + ' + ; t)j Ke?! 1 jh ;! (; t)? k(;!);;!( + '? ; t)j Ke?jj!?1 for some ' = ' (;!) independent of and t, where k ( (!);!) =! c (;!). If c >, replace by? in the above expressions. (iv) The functions k(;!);;!(; t) have amplitude of the order p? (!), spatial period 2 k in and temporal period 2 ( (!);!)! ; see Remark 3.7. Proof. For! =!, the claims in the theorem have been proved in the previous sections. It is straightforward to see that these proofs remain valid for any xed! close to!. Indeed, 24
Institut fur Mathematik I, Freie Universitat Berlin, Arnimallee 26, Berlin, Germany,
BIFURCATION TO SPIRAL WAVES IN REACTIONDIFFUSION SYSTEMS ARND SCHEEL Abstract. For a large class of reactiondiusion systems on the plane, we show rigorously that marmed spiral waves bifurcate from a
More information8.1 Bifurcations of Equilibria
1 81 Bifurcations of Equilibria Bifurcation theory studies qualitative changes in solutions as a parameter varies In general one could study the bifurcation theory of ODEs PDEs integrodifferential equations
More informationNumerical Computation of Solitary Waves on Innite. Cylinders
Numerical Computation of Solitary Waves on Innite Cylinders Gabriel J. Lord CISE National Physical Laboratory Queens Road Teddington TW11 LW, U.K. Daniela Peterhof WeierstraInstitut fur Angewandte Analysis
More informationCongurations of periodic orbits for equations with delayed positive feedback
Congurations of periodic orbits for equations with delayed positive feedback Dedicated to Professor Tibor Krisztin on the occasion of his 60th birthday Gabriella Vas 1 MTASZTE Analysis and Stochastics
More informationII. Systems of viscous hyperbolic balance laws. Bernold Fiedler, Stefan Liebscher. Freie Universitat Berlin. Arnimallee 26, Berlin, Germany
Generic Hopf bifurcation from lines of equilibria without parameters: II. Systems of viscous hyperbolic balance laws Bernold Fiedler, Stefan Liebscher Institut fur Mathematik I Freie Universitat Berlin
More informationVector Space Basics. 1 Abstract Vector Spaces. 1. (commutativity of vector addition) u + v = v + u. 2. (associativity of vector addition)
Vector Space Basics (Remark: these notes are highly formal and may be a useful reference to some students however I am also posting Ray Heitmann's notes to Canvas for students interested in a direct computational
More informationLinear Algebra (part 1) : Vector Spaces (by Evan Dummit, 2017, v. 1.07) 1.1 The Formal Denition of a Vector Space
Linear Algebra (part 1) : Vector Spaces (by Evan Dummit, 2017, v. 1.07) Contents 1 Vector Spaces 1 1.1 The Formal Denition of a Vector Space.................................. 1 1.2 Subspaces...................................................
More informationCoherent structures near the boundary between excitable and oscillatory media
Coherent structures near the boundary between excitable and oscillatory media Jeremy Bellay University of Minnesota Department of Computer Science 200 Union St. S.E. Minneapolis, MN 55455, USA Arnd Scheel
More informationPropagating terraces and the dynamics of frontlike solutions of reactiondiffusion equations on R
Propagating terraces and the dynamics of frontlike solutions of reactiondiffusion equations on R P. Poláčik School of Mathematics, University of Minnesota Minneapolis, MN 55455 Abstract We consider semilinear
More informationStability of Feedback Solutions for Infinite Horizon Noncooperative Differential Games
Stability of Feedback Solutions for Infinite Horizon Noncooperative Differential Games Alberto Bressan ) and Khai T. Nguyen ) *) Department of Mathematics, Penn State University **) Department of Mathematics,
More informationLECTURE VII: THE JORDAN CANONICAL FORM MAT FALL 2006 PRINCETON UNIVERSITY. [See also Appendix B in the book]
LECTURE VII: THE JORDAN CANONICAL FORM MAT 204  FALL 2006 PRINCETON UNIVERSITY ALFONSO SORRENTINO [See also Appendix B in the book] 1 Introduction In Lecture IV we have introduced the concept of eigenvalue
More informationNonlinear convective stability of travelling fronts near Turing and Hopf instabilities
Nonlinear convective stability of travelling fronts near Turing and Hopf instabilities Margaret Beck Joint work with Anna Ghazaryan, University of Kansas and Björn Sandstede, Brown University September
More informationQUASIUNIFORMLY POSITIVE OPERATORS IN KREIN SPACE. Denitizable operators in Krein spaces have spectral properties similar to those
QUASIUNIFORMLY POSITIVE OPERATORS IN KREIN SPACE BRANKO CURGUS and BRANKO NAJMAN Denitizable operators in Krein spaces have spectral properties similar to those of selfadjoint operators in Hilbert spaces.
More informationSpurious Chaotic Solutions of Dierential. Equations. Sigitas Keras. September Department of Applied Mathematics and Theoretical Physics
UNIVERSITY OF CAMBRIDGE Numerical Analysis Reports Spurious Chaotic Solutions of Dierential Equations Sigitas Keras DAMTP 994/NA6 September 994 Department of Applied Mathematics and Theoretical Physics
More information4.1 Eigenvalues, Eigenvectors, and The Characteristic Polynomial
Linear Algebra (part 4): Eigenvalues, Diagonalization, and the Jordan Form (by Evan Dummit, 27, v ) Contents 4 Eigenvalues, Diagonalization, and the Jordan Canonical Form 4 Eigenvalues, Eigenvectors, and
More informationMath 350 Fall 2011 Notes about inner product spaces. In this notes we state and prove some important properties of inner product spaces.
Math 350 Fall 2011 Notes about inner product spaces In this notes we state and prove some important properties of inner product spaces. First, recall the dot product on R n : if x, y R n, say x = (x 1,...,
More informationThe Erwin Schrodinger International Boltzmanngasse 9. Institute for Mathematical Physics A1090 Wien, Austria
ESI The Erwin Schrodinger International Boltzmanngasse 9 Institute for Mathematical Physics A1090 Wien, Austria Common Homoclinic Points of Commuting Toral Automorphisms Anthony Manning Klaus Schmidt
More informationSpectral stability of periodic waves in dispersive models
in dispersive models Collaborators: Th. Gallay, E. Lombardi T. Kapitula, A. Scheel in dispersive models Onedimensional nonlinear waves Standing and travelling waves u(x ct) with c = 0 or c 0 periodic
More informationvestigated. It is well known that the initial transition to convection is accompanied by the appearance of (almost) regular patterns. s the temperatur
Symmetry of ttractors and the KarhunenLoeve Decomposition Michael Dellnitz Department of Mathematics University of Hamburg D2046 Hamburg Martin Golubitsky Department of Mathematics University of Houston
More informationOne Dimensional Dynamical Systems
16 CHAPTER 2 One Dimensional Dynamical Systems We begin by analyzing some dynamical systems with onedimensional phase spaces, and in particular their bifurcations. All equations in this Chapter are scalar
More informationMath 413/513 Chapter 6 (from Friedberg, Insel, & Spence)
Math 413/513 Chapter 6 (from Friedberg, Insel, & Spence) David Glickenstein December 7, 2015 1 Inner product spaces In this chapter, we will only consider the elds R and C. De nition 1 Let V be a vector
More informationB5.6 Nonlinear Systems
B5.6 Nonlinear Systems 4. Bifurcations Alain Goriely 2018 Mathematical Institute, University of Oxford Table of contents 1. Local bifurcations for vector fields 1.1 The problem 1.2 The extended centre
More informationGarrett: `Bernstein's analytic continuation of complex powers' 2 Let f be a polynomial in x 1 ; : : : ; x n with real coecients. For complex s, let f
1 Bernstein's analytic continuation of complex powers c1995, Paul Garrett, garrettmath.umn.edu version January 27, 1998 Analytic continuation of distributions Statement of the theorems on analytic continuation
More informationThe Rationality of Certain Moduli Spaces of Curves of Genus 3
The Rationality of Certain Moduli Spaces of Curves of Genus 3 Ingrid Bauer and Fabrizio Catanese Mathematisches Institut Universität Bayreuth, NW II D95440 Bayreuth, Germany Ingrid.Bauer@unibayreuth.de,
More informationIm + α α. β + I 1 I 1< 0 I 1= 0 I 1 > 0
ON THE HAMILTONIAN ANDRONOVHOPF BIFURCATION M. Olle, J. Villanueva 2 and J. R. Pacha 3 2 3 Departament de Matematica Aplicada I (UPC), Barcelona, Spain In this contribution, we consider an specic type
More information2 Garrett: `A Good Spectral Theorem' 1. von Neumann algebras, density theorem The commutant of a subring S of a ring R is S 0 = fr 2 R : rs = sr; 8s 2
1 A Good Spectral Theorem c1996, Paul Garrett, garrett@math.umn.edu version February 12, 1996 1 Measurable Hilbert bundles Measurable Banach bundles Direct integrals of Hilbert spaces Trivializing Hilbert
More information2 G. D. DASKALOPOULOS AND R. A. WENTWORTH general, is not true. Thus, unlike the case of divisors, there are situations where k?1 0 and W k?1 = ;. r;d
ON THE BRILLNOETHER PROBLEM FOR VECTOR BUNDLES GEORGIOS D. DASKALOPOULOS AND RICHARD A. WENTWORTH Abstract. On an arbitrary compact Riemann surface, necessary and sucient conditions are found for the
More informationBlowup for Hyperbolic Equations. Helge Kristian Jenssen and Carlo Sinestrari
Blowup for Hyperbolic Equations Helge Kristian Jenssen and Carlo Sinestrari Abstract. We consider dierent situations of blowup in supnorm for hyperbolic equations. For scalar conservation laws with a
More informationOn low speed travelling waves of the KuramotoSivashinsky equation.
On low speed travelling waves of the KuramotoSivashinsky equation. Jeroen S.W. Lamb Joint with Jürgen Knobloch (Ilmenau, Germany) MarcoAntonio Teixeira (Campinas, Brazil) Kevin Webster (Imperial College
More informationA NOTE ON MATRIX REFINEMENT EQUATIONS. Abstract. Renement equations involving matrix masks are receiving a lot of attention these days.
A NOTE ON MATRI REFINEMENT EQUATIONS THOMAS A. HOGAN y Abstract. Renement equations involving matrix masks are receiving a lot of attention these days. They can play a central role in the study of renable
More informationHomework 6 Solutions. Solution. Note {e t, te t, t 2 e t, e 2t } is linearly independent. If β = {e t, te t, t 2 e t, e 2t }, then
Homework 6 Solutions 1 Let V be the real vector space spanned by the functions e t, te t, t 2 e t, e 2t Find a Jordan canonical basis and a Jordan canonical form of T on V dened by T (f) = f Solution Note
More informationDRIFT OF SPECTRALLY STABLE SHIFTED STATES ON STAR GRAPHS
DRIFT OF SPECTRALLY STABLE SHIFTED STATES ON STAR GRAPHS ADILBEK KAIRZHAN, DMITRY E. PELINOVSKY, AND ROY H. GOODMAN Abstract. When the coefficients of the cubic terms match the coefficients in the boundary
More information294 Meinolf Geck In 1992, Lusztig [16] addressed this problem in the framework of his theory of character sheaves and its application to Kawanaka's th
Doc. Math. J. DMV 293 On the Average Values of the Irreducible Characters of Finite Groups of Lie Type on Geometric Unipotent Classes Meinolf Geck Received: August 16, 1996 Communicated by Wolfgang Soergel
More informationNondegeneracy of perturbed solutions of semilinear partial differential equations
Nondegeneracy of perturbed solutions of semilinear partial differential equations Robert Magnus, Olivier Moschetta Abstract The equation u + F(V (εx, u = 0 is considered in R n. For small ε > 0 it is
More informationTravelling waves. Chapter 8. 1 Introduction
Chapter 8 Travelling waves 1 Introduction One of the cornerstones in the study of both linear and nonlinear PDEs is the wave propagation. A wave is a recognizable signal which is transferred from one part
More informationMET Workshop: Exercises
MET Workshop: Exercises Alex Blumenthal and Anthony Quas May 7, 206 Notation. R d is endowed with the standard inner product (, ) and Euclidean norm. M d d (R) denotes the space of n n real matrices. When
More informationNonlinear Modulational Instability of Dispersive PDE Models
Nonlinear Modulational Instability of Dispersive PDE Models Jiayin Jin, Shasha Liao, and Zhiwu Lin Georgia Tech ICERM workshop on water waves, 4/28/2017 Jiayin Jin, Shasha Liao, and Zhiwu Lin Georgia Tech
More information1 Introduction Selfreplicating spots and pulses have been observed in excitable reactiondiusion systems [22, 17, 24, 23, 16, 9, 2, 3, 4, 25, 21, 19,
Slowlymodulated two pulse solutions and pulse splitting bifurcations Arjen Doelman KortewegdeVries Instituut Universiteit van Amsterdam Plantage Muidergracht 24 1018TV Amsterdam, The Netherlands Wiktor
More information2 1. Introduction. Neuronal networks often exhibit a rich variety of oscillatory behavior. The dynamics of even a single cell may be quite complicated
GEOMETRIC ANALYSIS OF POPULATION RHYTHMS IN SYNAPTICALLY COUPLED NEURONAL NETWORKS J. Rubin and D. Terman Dept. of Mathematics; Ohio State University; Columbus, Ohio 43210 Abstract We develop geometric
More informationSpectrum for compact operators on Banach spaces
Submitted to Journal of the Mathematical Society of Japan Spectrum for compact operators on Banach spaces By Luis Barreira, Davor Dragičević Claudia Valls (Received Nov. 22, 203) (Revised Jan. 23, 204)
More informationSymmetry Properties of Confined Convective States
Symmetry Properties of Confined Convective States John Guckenheimer Cornell University 1 Introduction This paper is a commentary on the experimental observation observations of Bensimon et al. [1] of convection
More informationStudy Guide for Linear Algebra Exam 2
Study Guide for Linear Algebra Exam 2 Term Vector Space Definition A Vector Space is a nonempty set V of objects, on which are defined two operations, called addition and multiplication by scalars (real
More informationSecond Order Elliptic PDE
Second Order Elliptic PDE T. Muthukumar tmk@iitk.ac.in December 16, 2014 Contents 1 A Quick Introduction to PDE 1 2 Classification of Second Order PDE 3 3 Linear Second Order Elliptic Operators 4 4 Periodic
More informationAbstract. Jacobi curves are far going generalizations of the spaces of \Jacobi
Principal Invariants of Jacobi Curves Andrei Agrachev 1 and Igor Zelenko 2 1 S.I.S.S.A., Via Beirut 24, 34013 Trieste, Italy and Steklov Mathematical Institute, ul. Gubkina 8, 117966 Moscow, Russia; email:
More informationFoliations of hyperbolic space by constant mean curvature surfaces sharing ideal boundary
Foliations of hyperbolic space by constant mean curvature surfaces sharing ideal boundary David Chopp and John A. Velling December 1, 2003 Abstract Let γ be a Jordan curve in S 2, considered as the ideal
More informationMATH 205C: STATIONARY PHASE LEMMA
MATH 205C: STATIONARY PHASE LEMMA For ω, consider an integral of the form I(ω) = e iωf(x) u(x) dx, where u Cc (R n ) complex valued, with support in a compact set K, and f C (R n ) real valued. Thus, I(ω)
More informationAbsolute and convective instabilities of waves on unbounded and large bounded domains
Physica D 145 (2000) 233 277 Absolute and convective instabilities of waves on unbounded and large bounded domains Björn Sandstede a, Arnd Scheel b, a Department of Mathematics, Ohio State University,
More informationNotes by Maksim Maydanskiy.
SPECTRAL FLOW IN MORSE THEORY. 1 Introduction Notes by Maksim Maydanskiy. Spectral flow is a general formula or computing the Fredholm index of an operator d ds +A(s) : L1,2 (R, H) L 2 (R, H) for a family
More informationHalf of Final Exam Name: Practice Problems October 28, 2014
Math 54. Treibergs Half of Final Exam Name: Practice Problems October 28, 24 Half of the final will be over material since the last midterm exam, such as the practice problems given here. The other half
More information7 Planar systems of linear ODE
7 Planar systems of linear ODE Here I restrict my attention to a very special class of autonomous ODE: linear ODE with constant coefficients This is arguably the only class of ODE for which explicit solution
More informationSTABILITY OF INVARIANT SUBSPACES OF COMMUTING MATRICES We obtain some further results for pairs of commuting matrices. We show that a pair of commutin
On the stability of invariant subspaces of commuting matrices Tomaz Kosir and Bor Plestenjak September 18, 001 Abstract We study the stability of (joint) invariant subspaces of a nite set of commuting
More informationIn these chapter 2A notes write vectors in boldface to reduce the ambiguity of the notation.
1 2 Linear Systems In these chapter 2A notes write vectors in boldface to reduce the ambiguity of the notation 21 Matrix ODEs Let and is a scalar A linear function satisfies Linear superposition ) Linear
More informationCENTER MANIFOLD AND NORMAL FORM THEORIES
3 rd Sperlonga Summer School on Mechanics and Engineering Sciences 37 September 013 SPERLONGA CENTER MANIFOLD AND NORMAL FORM THEORIES ANGELO LUONGO 1 THE CENTER MANIFOLD METHOD Existence of an invariant
More informationPARAMETER IDENTIFICATION IN THE FREQUENCY DOMAIN. H.T. Banks and Yun Wang. Center for Research in Scientic Computation
PARAMETER IDENTIFICATION IN THE FREQUENCY DOMAIN H.T. Banks and Yun Wang Center for Research in Scientic Computation North Carolina State University Raleigh, NC 7695805 Revised: March 1993 Abstract In
More informationOctober 7, :8 WSPC/WSIJWMIP paper. Polynomial functions are renable
International Journal of Wavelets, Multiresolution and Information Processing c World Scientic Publishing Company Polynomial functions are renable Henning Thielemann Institut für Informatik MartinLutherUniversität
More informationInvariant Manifolds of Dynamical Systems and an application to Space Exploration
Invariant Manifolds of Dynamical Systems and an application to Space Exploration Mateo Wirth January 13, 2014 1 Abstract In this paper we go over the basics of stable and unstable manifolds associated
More informationStability and instability of nonlinear waves:
Stability and instability of nonlinear waves: Introduction 1. Nonlinear Waves 2. Stability problems 3. Stability of pulses and fronts 4. Stability of periodic waves 1 Nonlinear waves particular solutions
More informationInstability, index theorem, and exponential trichotomy for Linear Hamiltonian PDEs
Instability, index theorem, and exponential trichotomy for Linear Hamiltonian PDEs Zhiwu Lin and Chongchun Zeng School of Mathematics Georgia Institute of Technology Atlanta, GA 30332, USA Abstract Consider
More informationAPPPHYS217 Tuesday 25 May 2010
APPPHYS7 Tuesday 5 May Our aim today is to take a brief tour of some topics in nonlinear dynamics. Some good references include: [Perko] Lawrence Perko Differential Equations and Dynamical Systems (SpringerVerlag
More information1 Introduction We study innitedimensional systems that can formally be written in the familiar form x 0 (t) = Ax(t) + Bu(t); y(t) = Cx(t) + Du(t); t
Quadratic Optimal Control of Regular WellPosed Linear Systems, with Applications to Parabolic Equations Olof J. Staans Abo Akademi University Department of Mathematics FIN20500 Abo, Finland Olof.Staans@abo.
More informationB5.6 Nonlinear Systems
B5.6 Nonlinear Systems 5. Global Bifurcations, Homoclinic chaos, Melnikov s method Alain Goriely 2018 Mathematical Institute, University of Oxford Table of contents 1. Motivation 1.1 The problem 1.2 A
More informationThe Best Circulant Preconditioners for Hermitian Toeplitz Systems II: The MultipleZero Case Raymond H. Chan Michael K. Ng y Andy M. Yip z Abstract In
The Best Circulant Preconditioners for Hermitian Toeplitz Systems II: The Multipleero Case Raymond H. Chan Michael K. Ng y Andy M. Yip z Abstract In [0, 4], circulanttype preconditioners have been proposed
More informationMath113: Linear Algebra. Beifang Chen
Math3: Linear Algebra Beifang Chen Spring 26 Contents Systems of Linear Equations 3 Systems of Linear Equations 3 Linear Systems 3 2 Geometric Interpretation 3 3 Matrices of Linear Systems 4 4 Elementary
More information1 Lyapunov theory of stability
M.Kawski, APM 581 Diff Equns Intro to Lyapunov theory. November 15, 29 1 1 Lyapunov theory of stability Introduction. Lyapunov s second (or direct) method provides tools for studying (asymptotic) stability
More informationIntroduction LECTURE 1
LECTURE 1 Introduction The source of all great mathematics is the special case, the concrete example. It is frequent in mathematics that every instance of a concept of seemingly great generality is in
More informationPart V. 17 Introduction: What are measures and why measurable sets. Lebesgue Integration Theory
Part V 7 Introduction: What are measures and why measurable sets Lebesgue Integration Theory Definition 7. (Preliminary). A measure on a set is a function :2 [ ] such that. () = 2. If { } = is a finite
More informationSPECTRAL PROPERTIES OF THE LAPLACIAN ON BOUNDED DOMAINS
SPECTRAL PROPERTIES OF THE LAPLACIAN ON BOUNDED DOMAINS TSOGTGEREL GANTUMUR Abstract. After establishing discrete spectra for a large class of elliptic operators, we present some fundamental spectral properties
More informationSpectral Theorem for Selfadjoint Linear Operators
Notes for the undergraduate lecture by David Adams. (These are the notes I would write if I was teaching a course on this topic. I have included more material than I will cover in the 45 minute lecture;
More informationLinear Regression and Its Applications
Linear Regression and Its Applications Predrag Radivojac October 13, 2014 Given a data set D = {(x i, y i )} n the objective is to learn the relationship between features and the target. We usually start
More informationLecture 4. Chapter 4: Lyapunov Stability. Eugenio Schuster. Mechanical Engineering and Mechanics Lehigh University.
Lecture 4 Chapter 4: Lyapunov Stability Eugenio Schuster schuster@lehigh.edu Mechanical Engineering and Mechanics Lehigh University Lecture 4 p. 1/86 Autonomous Systems Consider the autonomous system ẋ
More information5 Eigenvalues and Diagonalization
Linear Algebra (part 5): Eigenvalues and Diagonalization (by Evan Dummit, 27, v 5) Contents 5 Eigenvalues and Diagonalization 5 Eigenvalues, Eigenvectors, and The Characteristic Polynomial 5 Eigenvalues
More informationChampneys, AR., Haerterich, J., & Sandstede, B. (1994). A nontransverse homoclinic orbit to a saddlenode equilibrium.
Champneys, AR., Haerterich, J., & Sandstede, B. (994). A nontransverse homoclinic orbit to a saddlenode equilibrium. Early version, also known as preprint Link to publication record in Explore Bristol
More information. HILBERT POLYNOMIALS
CLASSICAL ALGEBRAIC GEOMETRY Daniel Plaumann Universität Konstan Summer HILBERT POLYNOMIALS T An ane variety V A n with vanishing ideal I(V) K[,, n ] is completely determined by its coordinate ring A(V)
More informationCHAPTER 6 HOPFBIFURCATION IN A TWO DIMENSIONAL NONLINEAR DIFFERENTIAL EQUATION
CHAPTER 6 HOPFBIFURCATION IN A TWO DIMENSIONAL NONLINEAR DIFFERENTIAL EQUATION [Discussion on this chapter is based on our paper entitled HopfBifurcation Ina Two Dimensional Nonlinear Differential Equation,
More informationRichard DiSalvo. Dr. Elmer. Mathematical Foundations of Economics. Fall/Spring,
The Finite Dimensional Normed Linear Space Theorem Richard DiSalvo Dr. Elmer Mathematical Foundations of Economics Fall/Spring, 20202 The claim that follows, which I have called the nitedimensional normed
More informationIMPORTANT DEFINITIONS AND THEOREMS REFERENCE SHEET
IMPORTANT DEFINITIONS AND THEOREMS REFERENCE SHEET This is a (not quite comprehensive) list of definitions and theorems given in Math 1553. Pay particular attention to the ones in red. Study Tip For each
More informationRobustly transitive diffeomorphisms
Robustly transitive diffeomorphisms Todd Fisher tfisher@math.byu.edu Department of Mathematics, Brigham Young University Summer School, Chengdu, China 2009 Dynamical systems The setting for a dynamical
More informationGlobal Attractors in PDE
CHAPTER 14 Global Attractors in PDE A.V. Babin Department of Mathematics, University of California, Irvine, CA 926973875, USA Email: ababine@math.uci.edu Contents 0. Introduction.............. 985 1.
More informationATTRACTORS FOR SEMILINEAR PARABOLIC PROBLEMS WITH DIRICHLET BOUNDARY CONDITIONS IN VARYING DOMAINS. Emerson A. M. de Abreu Alexandre N.
ATTRACTORS FOR SEMILINEAR PARABOLIC PROBLEMS WITH DIRICHLET BOUNDARY CONDITIONS IN VARYING DOMAINS Emerson A. M. de Abreu Alexandre N. Carvalho Abstract Under fairly general conditions one can prove that
More information290 J.M. Carnicer, J.M. Pe~na basis (u 1 ; : : : ; u n ) consisting of minimally supported elements, yet also has a basis (v 1 ; : : : ; v n ) which f
Numer. Math. 67: 289{301 (1994) Numerische Mathematik c SpringerVerlag 1994 Electronic Edition Least supported bases and local linear independence J.M. Carnicer, J.M. Pe~na? Departamento de Matematica
More informationApril 13, We now extend the structure of the horseshoe to more general kinds of invariant. x (v) λ n v.
April 3, 005  Hyperbolic Sets We now extend the structure of the horseshoe to more general kinds of invariant sets. Let r, and let f D r (M) where M is a Riemannian manifold. A compact f invariant set
More informationLECTURE VI: SELFADJOINT AND UNITARY OPERATORS MAT FALL 2006 PRINCETON UNIVERSITY
LECTURE VI: SELFADJOINT AND UNITARY OPERATORS MAT 204  FALL 2006 PRINCETON UNIVERSITY ALFONSO SORRENTINO 1 Adjoint of a linear operator Note: In these notes, V will denote a ndimensional euclidean vector
More informationDispersion relations, stability and linearization
Dispersion relations, stability and linearization 1 Dispersion relations Suppose that u(x, t) is a function with domain { < x 0}, and it satisfies a linear, constant coefficient partial differential
More informationTOEPLITZ OPERATORS. Toeplitz studied infinite matrices with NWSE diagonals constant. f e C :
TOEPLITZ OPERATORS EFTON PARK 1. Introduction to Toeplitz Operators Otto Toeplitz lived from 18811940 in Goettingen, and it was pretty rough there, so he eventually went to Palestine and eventually contracted
More informationBifurcations of MultiVortex Configurations in Rotating Bose Einstein Condensates
Milan J. Math. Vol. 85 (2017) 331 367 DOI 10.1007/s0003201702758 Published online November 17, 2017 2017 Springer International Publishing AG, part of Springer Nature Milan Journal of Mathematics Bifurcations
More informationFree Boundary Minimal Surfaces in the Unit 3Ball
Free Boundary Minimal Surfaces in the Unit 3Ball T. Zolotareva (joint work with A. Folha and F. Pacard) CMLS, Ecole polytechnique December 15 2015 Free boundary minimal surfaces in B 3 Denition : minimal
More informationCAAM 336 DIFFERENTIAL EQUATIONS IN SCI AND ENG Examination 1
CAAM 6 DIFFERENTIAL EQUATIONS IN SCI AND ENG Examination Instructions: Time limit: uninterrupted hours There are four questions worth a total of 5 points Please do not look at the questions until you begin
More informationDiscreteness of Transmission Eigenvalues via Upper Triangular Compact Operators
Discreteness of Transmission Eigenvalues via Upper Triangular Compact Operators John Sylvester Department of Mathematics University of Washington Seattle, Washington 98195 U.S.A. June 3, 2011 This research
More information(1.) For any subset P S we denote by L(P ) the abelian group of integral relations between elements of P, i.e. L(P ) := ker Z P! span Z P S S : For ea
Torsion of dierentials on toric varieties Klaus Altmann Institut fur reine Mathematik, HumboldtUniversitat zu Berlin Ziegelstr. 13a, D10099 Berlin, Germany. Email: altmann@mathematik.huberlin.de Abstract
More information5.2.2 Planar AndronovHopf bifurcation
138 CHAPTER 5. LOCAL BIFURCATION THEORY 5.. Planar AndronovHopf bifurcation What happens if a planar system has an equilibrium x = x 0 at some parameter value α = α 0 with eigenvalues λ 1, = ±iω 0, ω
More informationPROOF OF TWO MATRIX THEOREMS VIA TRIANGULAR FACTORIZATIONS ROY MATHIAS
PROOF OF TWO MATRIX THEOREMS VIA TRIANGULAR FACTORIZATIONS ROY MATHIAS Abstract. We present elementary proofs of the CauchyBinet Theorem on determinants and of the fact that the eigenvalues of a matrix
More informationCHAPTER 3. Gauss map. In this chapter we will study the Gauss map of surfaces in R 3.
CHAPTER 3 Gauss map In this chapter we will study the Gauss map of surfaces in R 3. 3.1. Surfaces in R 3 Let S R 3 be a submanifold of dimension 2. Let {U i, ϕ i } be a DS on S. For any p U i we have a
More informationON WEAKLY NONLINEAR BACKWARD PARABOLIC PROBLEM
ON WEAKLY NONLINEAR BACKWARD PARABOLIC PROBLEM OLEG ZUBELEVICH DEPARTMENT OF MATHEMATICS THE BUDGET AND TREASURY ACADEMY OF THE MINISTRY OF FINANCE OF THE RUSSIAN FEDERATION 7, ZLATOUSTINSKY MALIY PER.,
More informationTHEODORE VORONOV DIFFERENTIABLE MANIFOLDS. Fall Last updated: November 26, (Under construction.)
4 Vector fields Last updated: November 26, 2009. (Under construction.) 4.1 Tangent vectors as derivations After we have introduced topological notions, we can come back to analysis on manifolds. Let M
More informationDefinition 5.1. A vector field v on a manifold M is map M T M such that for all x M, v(x) T x M.
5 Vector fields Last updated: March 12, 2012. 5.1 Definition and general properties We first need to define what a vector field is. Definition 5.1. A vector field v on a manifold M is map M T M such that
More informationContents. 6 Systems of FirstOrder Linear Dierential Equations. 6.1 General Theory of (FirstOrder) Linear Systems
Dierential Equations (part 3): Systems of FirstOrder Dierential Equations (by Evan Dummit, 26, v 2) Contents 6 Systems of FirstOrder Linear Dierential Equations 6 General Theory of (FirstOrder) Linear
More informationTopic 1: Matrix diagonalization
Topic : Matrix diagonalization Review of Matrices and Determinants Definition A matrix is a rectangular array of real numbers a a a m a A = a a m a n a n a nm The matrix is said to be of order n m if it
More informationgrowth rates of perturbed timevarying linear systems, [14]. For this setup it is also necessary to study discretetime systems with a transition map
Remarks on universal nonsingular controls for discretetime systems Eduardo D. Sontag a and Fabian R. Wirth b;1 a Department of Mathematics, Rutgers University, New Brunswick, NJ 08903, b sontag@hilbert.rutgers.edu
More informationThe Erwin Schrodinger International Pasteurgasse 6/7. Institute for Mathematical Physics A1090 Wien, Austria
ESI The Erwin Schrodinger International Pasteurgasse 6/7 Institute for Mathematical Physics A090 Wien, Austria On the Rigid Body with Two Linear Controls Mircea Craioveanu Mircea Puta Vienna, Preprint
More informationTopic 1: Matrix diagonalization
Topic : Matrix diagonalization Review of Matrices and Determinants Definition A matrix is a rectangular array of real numbers a a a m a A = a a m a n a n a nm The matrix is said to be of order n m if it
More information