Local Minimizers of the Magnetic Ginzburg-Landau Functional with S¹-valued Order Parameter on the Boundary

It was shown in [L. Berlyand and V. Rybalko, Solution with Vortices of a Semi-Stiff Boundary Value Problem for the Ginzburg-Landau Equation, J. Eur. Math. Soc. 12 (2010), 1497{1531] that in doubly connected domains there exist local minimizers of the simplified Ginzburg-Landau functional with modulu...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Журнал математической физики, анализа, геометрии
Datum:2014
1. Verfasser: Rybalko, V.
Format: Artikel
Sprache:English
Veröffentlicht: Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України 2014
Online Zugang:https://nasplib.isofts.kiev.ua/handle/123456789/106788
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Назва журналу:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Zitieren:Local Minimizers of the Magnetic Ginzburg-Landau Functional with S¹-valued Order Parameter on the Boundary / V. Rybalko // Журнал математической физики, анализа, геометрии. — 2014. — Т. 10, № 1. — С. 134-151. — Бібліогр.: 23 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id nasplib_isofts_kiev_ua-123456789-106788
record_format dspace
spelling Rybalko, V.
2016-10-05T18:58:37Z
2016-10-05T18:58:37Z
2014
Local Minimizers of the Magnetic Ginzburg-Landau Functional with S¹-valued Order Parameter on the Boundary / V. Rybalko // Журнал математической физики, анализа, геометрии. — 2014. — Т. 10, № 1. — С. 134-151. — Бібліогр.: 23 назв. — англ.
1812-9471
DOI: 10.15407/mag10.01.134
MSC2000: 35A01, 35J20, 35Q56
https://nasplib.isofts.kiev.ua/handle/123456789/106788
It was shown in [L. Berlyand and V. Rybalko, Solution with Vortices of a Semi-Stiff Boundary Value Problem for the Ginzburg-Landau Equation, J. Eur. Math. Soc. 12 (2010), 1497{1531] that in doubly connected domains there exist local minimizers of the simplified Ginzburg-Landau functional with modulus one and prescribed degrees on the boundary, unlike global minimizers that typically do not exist. We generalize the results and techniques of the aforementioned paper to the case of the magnetic Ginzburg-Landau functional.
В работе [L. Berlyand and V. Rybalko, Solution with Vortices of a Semi-Stiff Boundary Value Problem for the Ginzburg-Landau Equation, J. Eur. Math. Soc. 12 (2010), 1497-1531] было показано, что в двусвязных областях существуют локальные минимизанты упрощенного функционала Гинзбурга-Ландау, имеющие модуль один и заданные степени отображения на границе, в отличие от глобальных минимизантов, которые обычно не существуют. Результаты и методы упомянутой выше статьи обобщаются на случай "магнитного" функционала Гинзбурга-Ландау.
en
Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
Журнал математической физики, анализа, геометрии
Local Minimizers of the Magnetic Ginzburg-Landau Functional with S¹-valued Order Parameter on the Boundary
Article
published earlier
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
title Local Minimizers of the Magnetic Ginzburg-Landau Functional with S¹-valued Order Parameter on the Boundary
spellingShingle Local Minimizers of the Magnetic Ginzburg-Landau Functional with S¹-valued Order Parameter on the Boundary
Rybalko, V.
title_short Local Minimizers of the Magnetic Ginzburg-Landau Functional with S¹-valued Order Parameter on the Boundary
title_full Local Minimizers of the Magnetic Ginzburg-Landau Functional with S¹-valued Order Parameter on the Boundary
title_fullStr Local Minimizers of the Magnetic Ginzburg-Landau Functional with S¹-valued Order Parameter on the Boundary
title_full_unstemmed Local Minimizers of the Magnetic Ginzburg-Landau Functional with S¹-valued Order Parameter on the Boundary
title_sort local minimizers of the magnetic ginzburg-landau functional with s¹-valued order parameter on the boundary
author Rybalko, V.
author_facet Rybalko, V.
publishDate 2014
language English
container_title Журнал математической физики, анализа, геометрии
publisher Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
format Article
description It was shown in [L. Berlyand and V. Rybalko, Solution with Vortices of a Semi-Stiff Boundary Value Problem for the Ginzburg-Landau Equation, J. Eur. Math. Soc. 12 (2010), 1497{1531] that in doubly connected domains there exist local minimizers of the simplified Ginzburg-Landau functional with modulus one and prescribed degrees on the boundary, unlike global minimizers that typically do not exist. We generalize the results and techniques of the aforementioned paper to the case of the magnetic Ginzburg-Landau functional. В работе [L. Berlyand and V. Rybalko, Solution with Vortices of a Semi-Stiff Boundary Value Problem for the Ginzburg-Landau Equation, J. Eur. Math. Soc. 12 (2010), 1497-1531] было показано, что в двусвязных областях существуют локальные минимизанты упрощенного функционала Гинзбурга-Ландау, имеющие модуль один и заданные степени отображения на границе, в отличие от глобальных минимизантов, которые обычно не существуют. Результаты и методы упомянутой выше статьи обобщаются на случай "магнитного" функционала Гинзбурга-Ландау.
issn 1812-9471
url https://nasplib.isofts.kiev.ua/handle/123456789/106788
citation_txt Local Minimizers of the Magnetic Ginzburg-Landau Functional with S¹-valued Order Parameter on the Boundary / V. Rybalko // Журнал математической физики, анализа, геометрии. — 2014. — Т. 10, № 1. — С. 134-151. — Бібліогр.: 23 назв. — англ.
work_keys_str_mv AT rybalkov localminimizersofthemagneticginzburglandaufunctionalwiths1valuedorderparameterontheboundary
first_indexed 2025-11-25T06:28:37Z
last_indexed 2025-11-25T06:28:37Z
_version_ 1850509511078445056
fulltext Journal of Mathematical Physics, Analysis, Geometry 2014, vol. 10, No. 1, pp. 134–151 Local Minimizers of the Magnetic Ginzburg–Landau Functional with S1-valued Order Parameter on the Boundary V. Rybalko B. Verkin Institute for Low Temperature Physics and Engineering National Academy of Sciences of Ukraine 47 Lenin Ave., Kharkiv, 61103, Ukraine E-mail: vrybalko@ilt.kharkov.ua Received February 15, 2013, revised July 17, 2013 It was shown in [L. Berlyand and V. Rybalko, Solution with Vortices of a Semi-Stiff Boundary Value Problem for the Ginzburg–Landau Equation, J. Eur. Math. Soc. 12 (2010), 1497–1531] that in doubly connected domains there exist local minimizers of the simplified Ginzburg–Landau functional with modulus one and prescribed degrees on the boundary, unlike global mi- nimizers that typically do not exist. We generalize the results and techniques of the aforementioned paper to the case of the magnetic Ginzburg–Landau functional. Key words: superconductivity, Ginzburg–Landau functional, variational problems with lack of compactness. Mathematics Subject Classification 2010: 35A01, 35J20, 35Q56. 1. Introduction The purpose of the paper is to generalize the results and methods of [6] (see also [13]) to the magnetic Ginzburg–Landau model. We consider the functional Fλ[u, A] = 1 2 ∫ G ( |∇u− iAu|2 + λ 4 (|u|2 − 1)2 ) dx + 1 2 ∫ Ω (curlA)2 dx, (1.1) where G ⊂ R2 is a bounded multiply connected domain and Ω is the smallest simply connected domain containing G, u ∈ H1(G;C) is the order parameter, A ∈ H1(Ω;R2) is the vector potential of the induced magnetic field, and λ > 0 is the coupling constant ( √ λ/2 is the Ginzburg–Landau parameter). We assume c© V. Rybalko, 2014 Local Minimizers of the Magnetic Ginzburg–Landau Functional that the order parameter u takes values in S1 on the boundary ∂G and study critical points of the functional Fλ[u,A] in the space (u,A) ∈ J = {u ∈ H1(G;C); |u| = 1 a.e. on ∂G} ×H1(Ω;R2). (1.2) For the sake of simplicity, we will consider doubly connected domains G, G = Ω \ ω, where ω, Ω are smooth simply connected domains (ω ⊂ Ω). However, the results of the paper can be easily extended for general multiply connected domains. Since the external magnetic field is zero in (1.1), global minimizers of the functional are trivial (up to a gauge transformation u ≡ const ∈ S1, A ≡ 0), and we are interested in finding nontrivial local minimizers. A way to produce these nontrivial local minimizers is to minimize the functional in the class of pairs (u,A) with prescribed topological degree of u on the connected components of the boundary. Recall that, given a simple closed curve γ, the topological degree (winding number) of a map u ∈ H1/2(γ;S1) is an integer given by the classical formula (cf., e.g., [8]) deg(u, γ) = 1 2π ∫ γ u ∧ ∂u ∂τ ds, where the integral is understood via H1/2 − H−1/2 duality, and ∂ ∂τ is the tan- gential derivative with respect to the counterclockwise orientation of γ. The functional deg( · , γ) : H1/2(γ; S1) → R is continuous with respect to the strong H1/2-convergence. Therefore, for any prescribed p, q ∈ Z minimizers of Fλ[u,A] over the set Jpq = {(u, A) ∈ J ; deg(u, ∂ω) = p, deg(u, ∂Ω) = q} are local minimizers of Fλ[u,A] in J . The problem however is that, in general, global minimizers of Fλ[u,A] in Jpq do not exist ([4], see also [1],[3],[6]) because of the lack of continuity of deg( · , γ) : H1/2(γ;S1) → R with respect to the weak H1/2-convergence. To construct local minimizers of Fλ[u, A] in J , we consider the constrained minimization problem mλ(p, q, d) := inf{Fλ[u,A]; (u,A) ∈ J (d) pq }, (1.3) where p, q and d are given integers, J (d) pq = {(u,A) ∈ Jpq; d− 1/2 ≤ Φ(u,A, V0) ≤ d + 1/2}, (1.4) Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 135 V. Rybalko and Φ(u,A, V0) is given by Φ(u,A, V0) = 1 2π ∫ G ( u∧ (( ∂u ∂x2 − iA2u )∂V0 ∂x1 −( ∂u ∂x1 − iA1u )∂V0 ∂x2 ) +A ·∇⊥V0 ) dx, (1.5) with V0 being the unique solution of the boundary value problem    −∆V0 + V0 = 1 in G V0 = 1 on ∂Ω V0 = 0 on ∂ω. (1.6) The functional Φ(u,A, V ) (with harmonic V and A ≡ 0) was introduced in [6] to find nontrivial local minimizers of the simplified Ginzburg–Landau functional with prescribed degrees. Here we will make use of Φ(u,A, V ) for various smooth functions V . Note that the functional Φ(u,A, V0) has the following properties (cf. [6], Section 3): (a) for every u ∈ H1(G; S1) and A ∈ H1(Ω;R2) we have Φ(u,A, V0) = deg(u, ∂ω) = deg(u, ∂Ω); (b) Φ( · , V0) : H1(G;C) × H1(Ω;R2) → R is continuous with respect to the weak convergence. Moreover, if we fix Λ > 0 and consider pairs (u,A) from the sublevel set Fλ[u,A] ≤ Λ, then Φ(u,A, V0) is never half-integer for sufficiently large λ > 0. More precisely, we prove (see Section ) Proposition 1. Fix Λ > 0. There exists λ0 = λ0(Λ) > 0 such that if λ ≥ λ0, then for every integer d and every (u,A) ∈ H1(G;C) × H1(Ω;R2) satisfying Fλ[u,A] ≤ Λ the closed constraint Φ(u,A, V0) ∈ [d− 1/2, d + 1/2] is equivalent to an open one, that is, d− 1/2 ≤ Φ(u, A, V0) ≤ d + 1/2 ⇐⇒ d− 1/2 < Φ(u, A, V0) < d + 1/2. (1.7) Actually, it will be shown that Φ(u,A, V0) is close to integers uniformly in (u,A) satisfying Fλ[u,A] ≤ Λ when λ is sufficiently large. It follows from Proposition 1 that if • the infimum mλ(p, q, d) in (1.3) is attained and (u,A) is a minimizer, • mλ(p, q, d) < Λ and λ ≥ λ0, where λ0(= λ0(Λ)) is as in Proposition 1, 136 Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 Local Minimizers of the Magnetic Ginzburg–Landau Functional then (u,A) is a local minimizer of Fλ[u, A] in J . However the attainability of mλ(p, q, d) is still a nontrivial question due to the lack of continuity of deg(u, ∂Ω) and deg(u, ∂ω) with respect to the weak H1-convergence. The main result of the paper is Theorem 2. For any integers p, q and d > 0 (d < 0) with d ≥ max{p, q} (d ≤ min{p, q}) there exists λ1 = λ1(p, q, d) > 0 such that the infimum in (1.3) is always attained when λ ≥ λ1 and any minimizer of (1.3) is a local minimizer of Fλ[u,A] in J . Throughout the paper we will assume d > 0 (the case d < 0 follows by taking (u,−A) instead of (u, A)). As already mentioned, the principal part of Theorem 2 is to prove the attainability of the infimum in (1.3). To show this we follow essentially the same scheme as in [6]; namely, we first prove the attainability of the infimum in (1.3) for p = q = d and argue by induction, that is, pass to p = d, q = d − 1 and p = d − 1, q = d, etc. The main technical result we use to pass from the prescribed degrees p, q to p, q − 1 and p− 1, q is the following strict inequalities: mλ(p, q − 1, d) < mλ(p, q, d) + π and mλ(p− 1, q, d) < mλ(p, q, d) + π (1.8) that hold provided that mλ(p, q, d) is attained and λ is sufficiently large. Bounds (1.8) are proved by constructing testing pairs (u,A) ∈ J (d) p(q−1) or (u,A) ∈ J (d) (p−1)q with Fλ[u,A] < mλ(p, q, d) + π, and this is the key point of the present paper. The construction of testing pairs completely differs from that of [6] and makes use of Bogomol’nyi’s representation of the functional [8] and the factorization idea of C. Taubes [23]. There is a vast body of literature on 2D Ginzburg–Landau type problems. Local minimizers of the simplified Ginzburg–Landau functional with the Dirichlet condition on the boundary were studied in [11, 16], more general results were obtained in [12]. Aforementioned works deal with large values of the Ginzburg– Landau parameter and essentially rely on the results of the pioneering work [7]; more specifically, they use the reduction to the renormalized energy functional introduced in [7]. Local minimizers with the Neumann boundary condition in multiply con- nected domains are related to the phenomenon of permanent currents, see, e.g., [18, 14]. However, it is not known whether these local minimizers can have zeros (vortices) in both simplified 2D model and magnetic Ginzburg–Landau 2D model with zero external field. One can find some results on nonexistence of local min- imizers with vortices in [15] and [21]. In the case of nonzero external magnetic field, global minimizers and local ones can have vortices ([10, 19, 20, 22], see also references therein). Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 137 V. Rybalko While the idea of local minimization is general, its implementation depends strongly on the concrete problem one is dealing with. The distinguishing feature of the problem studied in the present paper is its being a variational problem with a possible lack of compactness. Also, the local minimizers obtained in Theorem 2 have nonstandard behavior, their zeros are situated near the boundary and approach it as λ → +∞ (this can be shown quite similarly to [6]). In the paper, we use the following notation and conventions: • Every closed curve is counterclockwise oriented. For such a curve, τ and ν stand for the unit tangent and unit normal vectors, respectively, that agree with the orientation ((ν, τ) is direct); • The complex plane C is identified with R2 such that if x, y ∈ C, then (x, y) = 1 2(xȳ + yx̄) and x ∧ y = i 2(xȳ − yx̄) are the scalar and the wedge products, respectively; • Given a fixed orthonormal frame (x1, x2) in R2, ∂ ∂z = 1 2 ( ∂ ∂x1 − i ∂ ∂x2 ) and ∂ ∂z̄ = 1 2 ( ∂ ∂x1 + i ∂ ∂x2 ) denote the classical Cauchy operators. For a scalar (real-valued) function f , ∇⊥f is the vector field given by ∇⊥f = (−∂f/∂x2, ∂f/∂x1). 2. Critical Points of Fλ[u,A] and Bogomol’nyi’s Representation One of the main properties of the functional Fλ[u,A] is its invariance under the gauge transformations u 7→ eiφu, A 7→ A +∇φ (where φ ∈ H2(Ω)). It is easy to see that Φ(u,A, V0) also has the aforementioned property as well as deg(u, ∂Ω) and deg(u, ∂ω). Thus, without loss of generality, we can assume that A is in the Coulomb gauge, i.e., { divA = 0 in Ω A · ν = 0 on ∂Ω. (2.1) The critical points of Fλ[u,A] in J , in particular, the local minimizers, are the solutions of the system of Euler–Lagrange equations −(∇− iA)2u + λ 2 u(|u|2 − 1) = 0 in G, (2.2) −∇⊥h = { j in G 0 in ω, (2.3) where h = curl A is the magnetic field (scalar function), and j = (iu,∇u− iAu) 138 Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 Local Minimizers of the Magnetic Ginzburg–Landau Functional is the current (note that h and j are gauge invariant). Additionally, h ∈ H1(Ω) and the boundary conditions |u| = 1, j · ν = 0 on ∂G, h = 0 on ∂Ω, ∂h ∂τ = 0 on ∂ω (2.4) are satisfied. We will assume that ∂G ∈ C∞, then every solution (u,A) of (2.2)– (2.4) satisfies u ∈ C∞(Ḡ;C) and A ∈ C∞(Ḡ;R2)(see [5]). We also have the pointwise inequality |u| ≤ 1 in G which is a consequence of the maximum principle applied to the equation for |u|2, ∆|u|2 = λ|u|2(|u|2 − 1) + 2|∇u− iAu|2 in G. (2.5) Note that the equation −∇⊥h = j can be written either in the form ∂ ∂z ( h− 1 2 (|u|2 − 1) ) = −u (∂u ∂z − A2 + iA1 2 u ) or ∂ ∂z̄ ( h + 1 2 (|u|2 − 1) ) = u (∂u ∂z̄ + A2 − iA1 2 u ) . Then, taking ∂/∂z̄ (or ∂/∂z) and using (2.5), we get ∆ ( h− 1 2 (|u|2− 1) )− |u|2h = −4 ∣∣∂u ∂z − A2 + iA1 2 u ∣∣2− λ 2 |u|2(|u|2− 1) in G (2.6) and ∆ ( h + 1 2 (|u|2 − 1) )− |u|2h = 4 ∣∣∂u ∂z̄ + A2 − iA1 2 u ∣∣2 + λ 2 |u|2(|u|2 − 1) in G. (2.7) The representation valid for every (u, A) ∈ J , Fλ[u,A] = ±π(deg(u, ∂Ω)− deg(u, ∂ω)) + F±[u,A] + 1 2 ∫ ω |curlA|2dx + λ− 1 8 ∫ G (|u|2 − 1)2dx, (2.8) where F+[u,A] = 2 ∫ G ∣∣∂u ∂z̄ + A2 − iA1 2 u ∣∣2dx + 1 2 ∫ G ∣∣curlA + |u|2 − 1 2 ∣∣2dx, (2.9) F−[u, A] = 2 ∫ G ∣∣∂u ∂z − A2 + iA1 2 u ∣∣2dx + 1 2 ∫ G ∣∣curlA− |u|2 − 1 2 ∣∣2dx, (2.10) Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 139 V. Rybalko plays an important role in the analysis of the functional Fλ[u,A]. This represen- tation is due to a remarkable observation of E.B. Bogomol’nyi [8], for a detailed derivation of (2.8) we refer to [9]. 3. Properties of Functional Φ(u,A, V ) Let us rewrite Φ(u, A, V ) as the sum of two terms Φ(u, A, V ) = 1 2π ∫ G u ∧ ( ∂u ∂x2 ∂V ∂x1 − ∂u ∂x1 ∂V ∂x2 ) dx + 1 2π ∫ G A · ∇⊥V (1− |u|2) dx. (3.1) Then, using the results of [6] (Section 3) for the first term, we get that for every fixed V ∈ C1(G) the functional Φ( · , V ) is continuous with respect to the weak convergence in H1(G;C) × H1(Ω;R2). If, in addition, V is such that V = 0 on ∂ω and V = 1 on ∂Ω, then Φ(u,A, V ) = deg(u, ∂ω) = deg(u, ∂Ω) for every u ∈ H1(G; S1). Fix now Λ > 0. Let us show that sup {dist(Φ(u,A, V0),Z); Fλ[u,A] ≤ Λ} → 0 as λ →∞. (3.2) Since Φ(u, A, V0) is invariant under gauge transformations, we can always assume (2.1). Then, given a sequence of pairs (uλ, Aλ), λ → ∞, satisfying the bound Fλ[uλ, Aλ] ≤ Λ, we can extract a subsequence converging weakly to a limit (u,A). Moreover, since ∫ G (1− |uλ|2)2 dx ≤ 8Λ/λ → 0, we have u ∈ H1(G;S1). Thus Φ(u, A, V0) ∈ Z, at the same time |Φ(u,A, V0) − Φ(uλ, Aλ, V0)| → 0 as λ → ∞, thanks to the continuity of Φ( · , V0) with respect to the weak convergence. So far we have proven (3.2), which in turn implies Proposition 1. 4. Construction of Testing Pairs As mentioned in Introduction, the main technical point in the proof of The- orem 2 is to show strict inequalities (1.8). In this Section we provide a detailed construction of testing pairs with energy control and prove (1.8) with their help. Given a local minimizer (u,A) ∈ J (d) pq of Fλ[u,A] in J , we construct (w(ξ), B(ξ)) ∈ J (d) p(q−1) in the form w(ξ) = uaeφ/2, B(ξ) = { A + 1 2∇⊥φ, in G, A +∇⊥θ +∇χ in ω, (4.1) 140 Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 Local Minimizers of the Magnetic Ginzburg–Landau Functional where a is a conformal map from Ω onto the unit disk with zero at ξ ∈ G, and φ ∈ H2(G), θ, χ ∈ H2(ω) are scalar functions. We assume that φ, θ, χ depend on the parameter ξ omitted for brevity. In order to satisfy |w(ξ)| = 1 on ∂G, we have the following conditions: φ = 0 on ∂Ω, φ = −2 log |a| on ∂ω. (4.2) Since the (exactly one) zero ξ of a lies in G, we have deg(w(ξ), ∂ω) = p, deg(w(ξ), ∂Ω) = q − 1. (4.3) We calculate Fλ[w(ξ), B(ξ)] using (2.8), (2.10), Fλ[w(ξ), B(ξ)] =π + Fλ[u,A] + 2 ∫ G ∣∣∂u ∂z − A2 + iA1 2 u ∣∣2(|a|2eφ − 1) dx + 1 2 ∫ G ( v ( ∆φ− |u|2(|a|2eφ − 1) ) + 1 4 ( ∆φ− |u|2(|a|2eφ − 1) )2 ) dx + λ− 1 8 ∫ G ( (|u|2|a|2eφ − 1)2 − (|u|2 − 1)2 ) dx + 1 2 ∫ ω ( (∆θ + curl A)2 − (curlA)2 ) dx, (4.4) where v = curl A− (|u|2 − 1)/2. Then we expand the integrands in the two last terms of (4.4) and use the pointwise equality 4 ∣∣∂u ∂z − A2 + iA1 2 u ∣∣2(|a|2eφ − 1) = (−∆v + |u|2v)(|a|2eφ − 1)− λ− 1 2 |u|2(|u|2 − 1)(|a|2eφ − 1) (cf. (2.6)) to get Fλ[w(ξ), B(ξ)] =π + Fλ[u,A] + 1 2 ∫ G (−∆v + |u|2v)(|a|2eφ − 1) dx + 1 2 ∫ G ( v ( ∆φ− |u|2(|a|2eφ − 1) ) + 1 4 (∆φ− |u|2(|a|2eφ − 1) )2 ) dx + λ− 1 8 ∫ G |u|4(|a|2eφ − 1 )2 dx + 1 2 ∫ ω ( (∆θ)2 + 2∆θ curlA ) dx. (4.5) Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 141 V. Rybalko Set φ ∈ H2(G) to be the unique solution of the equation −∆φ + |u|2(|a|2eφ − 1) = 0 in G (4.6) subject to the boundary conditions (4.2) (this problem has the unique solution φ ∈ H2(G), see, e.g., [9], Theorem 4.3), then (4.5) simplifies to Fλ[w(ξ), B(ξ)] =π + Fλ[u,A] + 1 2 ∫ G (−∆v + |u|2v)(|a|2eφ − 1) dx + λ− 1 8 ∫ G |u|4(|a|2eφ − 1)2 dx + 1 2 ∫ ω ( (∆θ)2 + 2∆θ curlA ) dx. (4.7) We set the requirement that ∂θ ∂ν = 1 2 ∂φ ∂ν on ∂ω, (4.8) which leads, after integrating by parts in (4.7), to Fλ[w(ξ), B(ξ)] =π + Fλ[u,A] + 1 2 ∫ G v(−∆(|a|2eφ) + |u|2(|a|2eφ − 1)) dx + λ− 1 8 ∫ G |u|4(|a|2eφ − 1)2 dx + 1 2 ∫ ω (∆θ)2 dx, (4.9) where we have also used the facts that v = curl A on ∂ω, curl A = const in ω and ∫ ∂ω ∂|a|2 ∂ν ds |a|2 = ∫ ω ∆log |a|2 dx = 0 (a is a holomorphic function without zeros in ω). Now set θ to be a solution of the equation ∆θ = 1 2|ω| ∫ ∂ω ∂φ ∂ν ds in ω subject to the boundary condition (4.8). In order to have B(ξ) ∈ H1(Ω;R2), we define χ ∈ H2(G) as a function satisfying the boundary conditions χ = 0 on ∂ω, ∂χ ∂ν = −1 2 ∂φ ∂τ + ∂θ ∂τ on ∂ω (for definiteness, we can assume ∆2χ = 0 in ω). Thus, for every ξ ∈ G we have (w(ξ), B(ξ)) ∈ Jp(q−1), and (4.9) yields Fλ[w(ξ), B(ξ)] = π + Fλ[u,A] + 1 2 ∫ G v(−∆(|a|2eφ) + |u|2(|a|2eφ − 1)) dx + λ− 1 8 ∫ G |u|4(|a|2eφ − 1)2 dx + 1 8|ω| (∫ ∂ω ∂φ ∂ν ds )2 . (4.10) 142 Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 Local Minimizers of the Magnetic Ginzburg–Landau Functional Next we show that Fλ[w(ξ), B(ξ)] < π +Fλ[u,A] when ξ is sufficiently close to ∂Ω. To this end, we first study the asymptotic behavior of the conformal map a and the solution of φ of problem (4.2), (4.6) as ξ → ∂Ω. Lemma 3. Let ξ̃ be the nearest point projection of ξ ∈ G on ∂Ω and let δ = |ξ − ξ̃| be sufficiently small. Then (i) |∂a ∂z (x)| ≤ Cδ/(δ + |x− ξ̃|)2 and ‖1− |a|2‖L2(G) ≤ Cδ| log δ|1/2; (ii) ‖φ‖W 2,r(G) ≤ C(r)δ for every 1 < r < 2 and ‖φ‖W 2,2(G) ≤ Cδ| log δ|1/2, where a is the unique (up to a constant factor with modulus one) conformal map from Ω onto the unit disk with prescribed zero at the point ξ, and φ is the unique solution of problem (4.2), (4.6). P r o o f. To see (i), we should note that a can be written as a = σ(F(x)− F(ξ))/(1 − F(ξ)F(x)), where σ ∈ S1 is a constant and F is a fixed conformal map from Ω onto the unit disk. Then the proof of the first bound in (i) is straightforward. The second bound in (i) is shown in Section 8 of [4]. In the proof of (ii) we also follow the lines of Section 8 from [4] (the main ingredient here is the chain of pointwise inequalities 0 ≤ 1− |a|2eφ ≤ 1− |a|2 that follow by the maximum principle; details are left to the reader). It follows from Lemma 3 that Fλ[w(ξ), B(ξ)] ≤ π+Fλ[u,A]+ 1 2 ∫ G v(−∆(|a|2eφ)+|u|2(|a|2eφ−1)) dx+O(δ2| log δ|). (4.11) Observe that ∆(|a|2eφ)− |u|2(|a|2eφ − 1) = |u|2(|a|2eφ − 1)2 + 4 ∣∣∣∂a ∂z ∣∣∣ 2 eφ + 4a ∂a ∂z eφ ∂φ ∂z + 4a ∂a ∂z eφ ∂φ ∂z + |a|2eφ|∇φ|2, where we have used (4.6). Then we make use of Lemma 3 and the fact that v = 0 on ∂Ω to get, after routine calculations, ∫ G v(∆(|a|2eφ)− |u|2(|a|2eφ − 1))dx = 4 ∫ G v ∣∣∣∂a ∂z ∣∣∣ 2 dx + O(δ2| log δ|). (4.12) Taking ṽ = ∂v ∂ν (ξ̃)ν(ξ̃)(x − ξ̃) in place of v in the right-hand side of (4.12) (note that |v − ṽ| ≤ C|x− ξ̃|2), we obtain 4 ∫ G v ∣∣∣∂a ∂z ∣∣∣ 2 dx = 4 ∫ Ω ṽ ∣∣∣∂a ∂z ∣∣∣ 2 dx + O(δ2| log δ|). (4.13) Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 143 V. Rybalko Finally, since ∆ log |a|2 = 4πδξ(x) in Ω (where δξ(x) stands for the Dirac delta centered at ξ) and |a|2 = 1 on ∂Ω while ∆ṽ = 0 in Ω, we have 4 ∫ Ω ṽ ∣∣∣∂a ∂z ∣∣∣ 2 dx = ∫ Ω ṽ∆|a|2 dx = ∫ ∂Ω ṽ ∂|a|2 ∂ν ds = ∫ ∂Ω ṽ ∂ log |a|2 ∂ν ds = 4πṽ(ξ). (4.14) We combine (4.11)–(4.14) to get Fλ[w(ξ), B(ξ)] ≤ π + Fλ[u,A] + 2π ∂v ∂ν (ξ̃)δ + O(δ2| log δ|), (4.15) where ξ̃ is the nearest point projection of ξ on ∂Ω. It is clear now that (4.15) yields the required inequality Fλ[w(ξ), B(ξ)] < π + Fλ[u,A] when δ is sufficiently small, provided that there is ξ̃ on ∂Ω where ∂v ∂ν < 0. The following result establishes the existence of the point ξ̃ for large λ. Lemma 4. Let (u(λ), A(λ)) be a local minimizer of Fλ[u,A] in J . Assume that (u(λ), A(λ)) ∈ J (d) = ⋃ p,q∈Z J (d) p,q (where d is a fixed positive integer number) and Fλ[u(λ), A(λ)] ≤ Λ for some fixed Λ > 0. Then h(λ) = curlA(λ) satisfies ∂h(λ) ∂ν (ξ(λ)) < 0 for some ξ(λ) ∈ ∂Ω when λ is sufficiently large, λ ≥ λ2(Λ). P r o o f. Assume by contradiction that ∂h(λ) ∂ν ≥ 0 on ∂Ω for a sequence λ = λk, λk →∞. Since divA(λ) = 0 in Ω and A(λ) · ν = 0 on ∂Ω, we have ‖A(λ)‖H1(Ω;R2) ≤ C‖h‖L2(Ω). It follows that ‖A(λ)‖H1(Ω;R2) ≤ C thanks to the bound Fλ[u(λ), A(λ)] ≤ Λ. These two bounds imply that ‖u(λ)‖H1(G;C) ≤ C, where C is independent of λ. Therefore, up to extracting a subsequence, (u(λ), A(λ)) → (u,A) weakly in H1(G;C)×H1(Ω;R2) as λ →∞, and |u| = 1 a.e. in G (since ‖|u(λ)|2−1‖2 L2(G) ≤ 8Λ/λ → 0). It follows from (2.3) that −∆h(λ) + h(λ) = 2 ∂u(λ) ∂x1 ∧ ∂u(λ) ∂x2 + curl ( (1− |u(λ)|2)A(λ) ) in G. (4.16) We also have h(λ) = 0 on ∂Ω and h(λ) = const in ω. Let V ∈ C1(Ω) be the unique solution of the equation { −∆V + V = 0 in G, V = g on ∂Ω, V = 0 on ∂ω, (4.17) 144 Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 Local Minimizers of the Magnetic Ginzburg–Landau Functional where g ∈ C1(∂Ω) is some nonnegative function. Multiply (4.16) by V and integrate on G to get, after integrating by parts, − ∫ ∂Ω ∂h(λ) ∂ν g ds = ∫ ∂ω ∂V ∂ν h(λ) ds + ∫ ∂Ω u(λ) ∧ ∂u(λ) ∂τ g ds− 2πΦ(u(λ), A(λ), V ). We know that u(λ) → u weakly in H1/2(∂Ω; S1). Due to the result from [17] (see Lemma 3.2 therein) there is a subsequence such that for every g ∈ C1(∂Ω), ∫ ∂Ω u(λ) ∧ ∂u(λ) ∂τ g ds → ∫ ∂Ω u ∧ ∂u ∂τ g ds + 2π ∑ finite Dkg(αk), where the points αk ∈ ∂Ω and integers Dk are independent of g. Now choose a nonnegative function g 6≡ 0 such that g(αk) = 0 for every αk. Then we have ∫ ∂Ω u(λ) ∧ ∂u(λ) ∂τ g ds− 2πΦ(u(λ), A(λ), V ) → ∫ ∂Ω u ∧ ∂u ∂τ g ds− 2πΦ(u,A, V ) = 2 ∫ G ∂u ∂x1 ∧ ∂u ∂x2 V dx = 0, where we have used the continuity of Φ( · , V ) with respect to the weak conver- gence in H1(G;C)×H1(Ω;R2) and the pointwise equality ∂u ∂x1 ∧ ∂u ∂x2 = 0 a.e. in G (which holds for every u ∈ H1(G;S1)). Thus, 0 ≤ ∫ ∂Ω ∂h(λ) ∂ν g ds = − ∫ ∂ω ∂V ∂ν h(λ) ds + o(1) when λ →∞. On the other hand, since g ≥ 0 and g 6≡ 0, by the strong maximum principle and Hopf’s boundary lemma applied to (4.17) we have ∂V ∂ν > 0 on ∂ω. Therefore we will have a contradiction, if we show that ∫ ∂ω h(λ) ds ≥ c > 0 for sufficiently large λ. To this end, note that in view of (2.3), 2πΦ(u(λ), A(λ), V0) = − ∫ G ∇h(λ) · ∇V0 dx + ∫ G A(λ) · ∇⊥V0 dx = ∫ ∂ω ∂V0 ∂ν h(λ) ds + ∫ G h(λ)∆V0 dx + ∫ G A(λ) · ∇⊥V0 dx. Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 145 V. Rybalko Since V0 solves (1.6), we have 2πΦ(u(λ), A(λ), V0) = ∫ ∂ω ∂V0 ∂ν h(λ) ds + ∫ G h(λ)(V0 − 1) dx + ∫ G A(λ) · ∇⊥V0 dx = ∫ ∂ω ∂V0 ∂ν h(λ) ds + ∫ G curl ( (V0 − 1)A(λ) ) dx. (4.18) Using Stokes’ theorem twice, we get ∫ G curl ( (V0 − 1)A(λ) ) dx = ∫ ∂ω A(λ) · τ ds = ∫ ω h(λ) dx = |ω| |∂ω| ∫ ∂ω h(λ) ds, while ∫ ∂ω ∂V0 ∂ν h(λ) ds = ∫ ∂ω ∂V0 ∂ν ds ∫ ∂ω h(λ) ds |∂ω| = ∫ G (|∇V0|2 + (V0 − 1)2 ) dx ∫ ∂ω h(λ) ds |∂ω| . Thus, passing to the limit in (4.18) yields lim λ→∞ ∫ ∂ω h(λ) ds = 2πd|∂ω|/ ( |ω|+ ∫ G (|∇V0|2 + (V0 − 1)2 ) dx ) > 0, where we have used the fact that Φ(u(λ), A(λ), V0) → d as λ →∞. R e m a r k 5. In exactly the same way as in Lemma 4, one can show that ∂h(λ) ∂ν > 0 at a point on ∂ω for sufficiently large λ. Corollary 6. If ξ(λ) is as in Lemma 4, then ∂v(λ) ∂ν (ξ(λ)) < 0, where v(λ) = curlA(λ) − (|u(λ)|2 − 1)/2. P r o o f. Since |u(λ)| = 1 on ∂G and |u(λ)| ≤ 1 in G, we have ∂|u(λ)|2 ∂ν (ξ(λ)) ≥ 0. Now, if we take ξ ∈ G sufficiently close to the point ξ(λ), where ∂v(λ) ∂ν (ξ(λ)) < 0, by (4.15) and Corollary 6, we have Fλ[w(ξ), B(ξ)] < π + Fλ[u, A]. On the other hand, (w(ξ), B(ξ)) ∈ Jp(q−1) and (w(ξ), B(ξ)) converges weakly to (γu, A) (γ = const ∈ S1) as ξ → ∂Ω, up to a subsequence. Consequently, Φ(w(ξ), B(ξ), V0) → Φ(u,A, V0). Thus, if d− 1/2 < Φ(u,A, V0) < d + 1/2, then (w(ξ), B(ξ)) ∈ J (d) p(q−1), when ξ is sufficiently close to ξ(λ). Quite similarly, we can show that there exists a testing pair from J (d) (p−1)q whose Ginzburg–Landau energy is strictly less than π + Fλ[u,A]. These results are summarized in 146 Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 Local Minimizers of the Magnetic Ginzburg–Landau Functional Lemma 7. Given integers d > 0, p and q, if mλ(p, q, d) is attained and λ is sufficiently large, λ ≥ λ3(p, q, d), then mλ(p − 1, q, d) < mλ(p, q, d) + π and mλ(p, q − 1, d) < mλ(p, q, d) + π. P r o o f. It is not hard to prove that the bound mλ(p, q, d) ≤ Λ(p, q, d) holds for some Λ(p, q, d) independent of λ (see, e.g., [6]). Then the above results in conjunction with Proposition 1 yield the statement of the lemma. 5. Existence of Minimizers To begin with, let us quote the following result which is an important tool in the proof of Theorem 2. Lemma 8. ([2]) Let (u(n), A(n)) ∈ Jpq be a sequence converging weakly to (u,A) in H1(G;C)×H1(Ω;R2). Then lim inf 1 2 ∫ G |∇u(n) − iA(n)u(n)|2 dx ≥π(|p− deg(u, ∂ω)|+ |q − deg(u, ∂Ω)|) + 1 2 ∫ G |∇u− iAu|2 dx. Now consider the auxiliary minimization problem Mλ(d) := inf{Fλ[u,A]; (u,A) ∈ J (d)}, (5.1) where J (d) = ⋃ p,q∈Z J (d) pq = {(u,A) ∈ J ; d− 1/2 ≤ Φ(u,A, V0) ≤ d + 1/2}. Note that Mλ(d) is always attained. Indeed, Φ( · , V0) is continuous with respect to the weak convergence (in H1(G;C) ×H1(Ω;R2)), and J (d) 6= ∅ (J (d) contains, in particular, pairs (u, 0) with u ∈ H1(G; S1) and deg(u, ∂Ω) = d). Therefore every minimizing sequence (u(n), A(n)) contains a subsequence converging weakly in H1(G;C)×H1(Ω;R2) to a minimizer (u, A) ∈ J (d). Lemma 9. For sufficiently large λ, λ ≥ λ4(d), Mλ(d) = mλ(d, d, d), and minimizers of (1.3) (with p = q = d) and (5.1) coincide. P r o o f. Clearly, Mλ(d) ≤ mλ(d, d, d). Now assume by contradiction that for a sequence λ = λk, λk →∞, we have Mλ(d) = Fλ[u(λ), A(λ)] and (u(λ), A(λ)) ∈ J (d) \ J (d) dd . (5.2) In other words, either deg(u(λ), ∂Ω) 6= d or deg(u(λ), ∂ω) 6= d. We assume that divA(λ) = 0 in Ω and A(λ) ·ν = 0 on ∂Ω (Coulomb gauge). Thanks to the obvious Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 147 V. Rybalko bound Mλ(d) ≤ M∞(d) := inf {1 2 ∫ G |∇u− iAu|2 dx + 1 2 ∫ Ω (curlA)2 dx; (u,A) ∈ J (d), u ∈ H1(G; S1) } < ∞ (5.3) we can extract a subsequence such that u(λ) → u weakly in H1(G;C), A(λ) → A weakly in H1(Ω;R2) as λ → ∞, and deg(u(λ), ∂Ω) = q, deg(u(λ), ∂ω) = p with integers p, q independent of λ. Note that (u, A) ∈ J (d) and u ∈ H1(G;S1). Therefore, Fλ[u(λ), A(λ)] ≤ Mλ(d) ≤ Fλ[u,A] = 1 2 ∫ G ( |∇u− iAu|2 dx + 1 2 ∫ Ω (curlA)2 dx, while lim inf λ→∞ Fλ[u(λ), A(λ)] ≥ π(|d−p|+ |d−q|)+ 1 2 ∫ G |∇u− iAu|2 dx+ 1 2 ∫ Ω (curlA)2 dx by virtue of Lemma 8. Thus p = q = d for sufficiently large λ, i.e., u(λ) ∈ J (d) dd , which contradicts (5.2). Thus, Lemma 9 shows that mλ(d, d, d) is always attained for λ ≥ λ4(d). Moreover, thanks to Proposition 1 every minimizer of (1.3) with p = q = d is a local minimizer of Fλ[u,A] in J × H1(Ω;R2) when λ ≥ λ0(M∞(d) + 1), where M∞(d) is defined in (5.3) (note that mλ(d, d, d) < M∞(d) + 1). We next prove the attainability of the infimum (1.3) in Proposition 10. For every K = 0, 1, 2, . . . there is λ5 = λ5(K) > 0 such that for all λ ≥ λ5 and all integers p and q satisfying p ≤ d, q ≤ d, and |q − d|+ |p− d| ≤ K: (i) the infimum mλ(p, q, d) is attained, (ii) if p ≤ p′ ≤ d, q ≤ q′ ≤ d and either p 6= p′ or q 6= q′, then mλ(p, q, d) < mλ(p′, q′, d) + π(|p− p′|+ |q − q′|). P r o o f. By Lemma 9, Proposition 10 holds for K = 0 (induction base). Now assume that (i) and (ii) hold for given K ≥ 0. Then by Lemma 7, (ii) holds for K +1 in place of K when λ ≥ max { λ5(K), max{λ3(p, q, d); |q−d|+ |p−d| = K, p ≤ d, q ≤ d}}. 148 Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 Local Minimizers of the Magnetic Ginzburg–Landau Functional To show ( i), we consider a minimizing sequence (u(n), A(n)) for problem (1.3). This minimizing sequence exists since mλ(p, q, d) < mλ(d, d, d) + π(|p− d|+ |q − d|) ≤ M∞(d)+π(K +1). Thanks to this bound, up to extracting a subsequence, u(n) → u ∈ J weakly in H1(G;C), and A(n) → A weakly in H1(Ω;R2). By virtue of Lemma 8, we have mλ(p, q, d) = lim n→∞Fλ[u(n), A(n)] ≥ Fλ[u,A]+π(|q−deg(u, ∂Ω)|+|p−deg(u, ∂ω)|). (5.4) Let us show that deg(u, ∂Ω) = q and deg(u, ∂ω) = p. To this end, we need the following. Lemma 11. For every Λ > 0 there is λ6 = λ6(Λ) such that mλ(p, q, d) ≥ M∞(d)+π(|p−d|+|q−d|−1/2) when λ ≥ λ6 and M∞(d)+π(|p−d|+|q−d|) ≤ Λ. This lemma and (5.4) imply that if λ ≥ λ6, then mλ(p, q, d) ≥ mλ(deg(u, ∂ω),deg(u, ∂Ω), d)+π(|q−deg(u, ∂Ω)|+|p−deg(u, ∂ω)|) ≥ M∞(d) + π(|p− deg(u, ∂ω)|) + |deg(u, ∂ω)− d|) + π(|q − deg(u, ∂Ω)|+ |deg(u, ∂Ω)− d|))− π/2. On the other hand, (ii) guaranties that mλ(p, q, d) < M∞(d)+π(|p−d|+ |q−d|), therefore p ≤ deg(u, ∂ω) ≤ d and q ≤ deg(u, ∂Ω) ≤ d. Furthermore, if we assume that either deg(u, ∂Ω) 6= q or deg(u, ∂ω) 6= p, then (5.4) implies that mλ(p, q, d) ≥ mλ(deg(u, ∂ω),deg(u, ∂Ω), d)+π(|q−deg(u, ∂Ω)|+|p−deg(u, ∂ω)|), but this contradicts (ii). Thus mλ(p, q, d) is always attained for sufficiently large λ when p ≤ d, q ≤ d, and |q − d|+ |p− d| ≤ K + 1. Proposition 10 is proved. P r o o f of Lemma 11. Similarly to Lemma 9, we argue by contradiction. Namely, assume that mλ(p, q, d) < M∞(d) + π(|p − d| + |q − d| − 1/2) for some integers p, q, d and a sequence λ = λk, λk → ∞. In other words, there are (u(λ), A(λ)) ∈ J (d) pq such that Fλ[u(λ), A(λ)] < M∞(d) + π(|p− d|+ |q − d| − 1/2) for λ = λk. Since Fλ[u(λ), A(λ)] is bounded, up to extracting a subsequence, u(λ) → u weakly in H1(G;C), A(λ) → A weakly in H1(Ω;R2). Besides, u ∈ J (d) ∩H1(G; S1). Therefore, 1 2 ∫ G |∇u− iAu|2 dx + 1 2 ∫ Ω (curlA)2 dx ≥ M∞(d). On the other hand, by virtue of Lemma 8, we have lim inf λ→∞ Fλ[u(λ), A(λ)] ≥ 1 2 ∫ G |∇u− iAu|2 dx+ 1 2 ∫ Ω (curlA)2 dx+π(|p−d|+ |q−d|), and thus we arrive at a contradiction with the bound Fλ[u(λ), A(λ)] < M∞(d) + π(|p− d|+ |q − d| − 1/2). Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 149 V. Rybalko References [1] L. Berlyand and P. Mironescu, Ginzburg–Landau Minimizers with Prescribed De- grees. Capacity of the Domain and Emergence of Vortices. Preprint, available at http://desargues.univ-lyon1.fr. [2] L. Berlyand and P. Mironescu, Ginzburg–Landau Minimizers with Prescribed De- grees. Capacity of the Domain and Emergence of Vortices. — J. Funct. Anal. 239 (2006), 76–99. [3] L. Berlyand, D. Golovaty, and V. Rybalko, Nonexistence of Ginzburg–Landau Min- imizers with Prescribed Degree on the Boundary of a Doubly Connected Domain. — C. R. Math. Acad. Sci. Paris 343 (2006), 63–68. [4] L. Berlyand, O. Misiats, and V. Rybalko, Near Boundary Vortices in a Magnetic Ginzburg–Landau Model: their Locations Via Tight Energy Bounds. — J. Func. Analysis 258 (2010), 1728–1762. [5] L. Berlyand, O. Misiats, and V. Rybalko, Minimizers of the Magnetic Ginzburg– Landau Functional in Simply Connected Domain with Prescribed Degree on the Boundary. — Commun. Contemp. Math. 13 (2011), 53–66. [6] L. Berlyand and V. Rybalko, Solutions with Vortices of a Semi-Stiff Boundary Value Problem for the Ginzburg–Landau Equation. — J. Eur. Math. Soc. 12 (2010), 1497– 1531. [7] F. Bethuel, H. Brezis, and F. Helein, Ginzburg–Landau Vortices. Birkhauser, 1994. [8] E.B. Bogomol’nyi, The Stability of Classical Solutions. — Sov. J. Nuclear Phys. 24 (1976), 449–454. [9] A. Boutet de Monvel-Berthier, V. Georgescu, and R. Purice, A Boundary Value Problem Related to the Ginzburg–Landau Model. — Comm. Math. Phys. 142 (1991), 1–23. [10] A. Contreras and S. Serfaty, Large Vorticity Stable Solutions to the Ginzburg– Landau Equations. to appear in Indiana Univ. Math. J. [11] M. Del Pino and P.-L. Felmer, Local Minimizers for the Ginzburg–Landau Energy. — Math. Zh. 225 (1997), 671–684. [12] M. Del Pino, M. Kowalczyk, and M. Musso, Variational Reduction for Ginzburg– Landau Vortices. — J. Funct. Anal. 239 (2006), 497–541. [13] M. Dos Santos, Local Minimizers of the Ginzburg–Landau Functional with Pre- scribed Degrees. — J. Funct. Anal. 257 (2009), 1053–1091. [14] S. Jimbo, Y. Morita, and J. Zhai, Ginzburg–Landau Equation and Stable Steady State Solutions in a Nontrivial Domain. — Comm. Part. Diff. Eq. 20 (1995), 2093– 2112. [15] S. Jimbo and P. Sternberg Nonexistence of Permanent Currents in Convex Planar Samples. — SIAM J. Math. Anal. 33 (2002), 1379–1392. 150 Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 Local Minimizers of the Magnetic Ginzburg–Landau Functional [16] F.-H. Lin and T.-Ch. Lin, Minimax Solutions of the Ginzburg–Landau Equations. — Selecta Math. (N.S.) 3 (1997), 99-113. [17] P. Mironescu and A. Pisante, A Variational Problem with Lack of Compactness for H1/2(S1;S1) Maps of Prescribed Degree. — J. Funct. Anal. 217 (2004), 249–279. [18] J. Rubinstein and P. Sternberg, Homotopy Classification of Minimizers of the Ginzburg–Landau Energy and the Existence of Permanent Currents. — Comm. Math. Phys. 179 (1996), 257–263. [19] E. Sandier and S. Serfaty, Vortices in the Magnetic Ginzburg–Landau Model. Progress in Nonlinear Differential Equations and their Applications. Birkhauser, 2007. [20] S. Serfaty, Local Minimizers for the Ginzburg–Landau Energy near Critical Mag- netic Field. I, II. — Commun. Contemp. Math. 1 (1999), 213–254, ibid 1 (1999), 295–333. [21] S. Serfaty, Stability in 2D Ginzburg–Landau Passes to the Limit. — Indiana Univ. Math. J. 54 (2005), 199–221. [22] S. Serfaty, Stable Configurations in Superconductivity: Uniqueness, Multiplicity and Vortex-Nucleation. — Arch. Ration. Mech. Anal. 149 (1999), 329–365. [23] C. Taubes, Arbitrary N -vortex Solutions to the First Order Ginzburg–Landau Equa- tions. — Comm. Math. Phys. 72 (1980), 277–292. Journal of Mathematical Physics, Analysis, Geometry, 2014, vol. 10, No. 1 151