On the Form of Dispersive Shock Waves of the Korteweg-de Vries Equation

We show that the long-time behavior of solutions to the Korteweg{de Vries shock problem can be described as a slowly modulated one-gap solution in the dispersive shock region. The modulus of the elliptic function (i.e., the spectrum of the underlying Schrödinger operator) depends only on the size of...

Full description

Saved in:
Bibliographic Details
Published in:Журнал математической физики, анализа, геометрии
Date:2016
Main Authors: Egorova, I., Gladka, Z., Teschl, G.
Format: Article
Language:English
Published: Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України 2016
Online Access:https://nasplib.isofts.kiev.ua/handle/123456789/140545
Tags: Add Tag
No Tags, Be the first to tag this record!
Journal Title:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Cite this:On the Form of Dispersive Shock Waves of the Korteweg-de Vries Equation / I. Egorova, Z. Gladka , G. Teschl // Журнал математической физики, анализа, геометрии. — 2016. — Т. 12, № 1. — С. 3-16. — Бібліогр.: 23 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id nasplib_isofts_kiev_ua-123456789-140545
record_format dspace
spelling Egorova, I.
Gladka, Z.
Teschl, G.
2018-07-10T10:42:15Z
2018-07-10T10:42:15Z
2016
On the Form of Dispersive Shock Waves of the Korteweg-de Vries Equation / I. Egorova, Z. Gladka , G. Teschl // Журнал математической физики, анализа, геометрии. — 2016. — Т. 12, № 1. — С. 3-16. — Бібліогр.: 23 назв. — англ.
1812-9471
DOI: doi.org/10.15407/mag12.01.003
Mathematics Subject Classification 2000: 37K40, 35Q53 (primary); 33E05, 35Q15 (secondary)
https://nasplib.isofts.kiev.ua/handle/123456789/140545
We show that the long-time behavior of solutions to the Korteweg{de Vries shock problem can be described as a slowly modulated one-gap solution in the dispersive shock region. The modulus of the elliptic function (i.e., the spectrum of the underlying Schrödinger operator) depends only on the size of the step of the initial data and on the direction, x/ t =const, along which we determine the asymptotic behavior of the solution. In turn, the phase shift (i.e., the Dirichlet spectrum) in this elliptic function depends also on the scattering data, and is computed explicitly via the Jacobi inversion problem.
Показано, что поведение при большом времени решений уравнения Кортевега-де Фриза с начальными данными типа ступеньки, соответствующими волне сжатия, в области эллиптической волны может быть описано слабо модулированным двухзонным решением. Модуль этой эллиптической функции, определяемый спектром фонового оператора, зависит от размера ступеньки в начальных данных и от направления, в котором исследуется асимптотическое поведение решения. В свою очередь фазовый сдвиг (то есть спектр задачи Дирихле) в этой эллиптической функции зависит также от данных рассеяния, и он посчитан с помощью проблемы обращения Якоби.
We thank Alexei Rybkin and Johanna Michor for valuable discussions on this topic. I.E. is indebted to the Department of Mathematics at the University of Vienna for its hospitality and support during the fall of 2015, where part of this work was done.
en
Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
Журнал математической физики, анализа, геометрии
On the Form of Dispersive Shock Waves of the Korteweg-de Vries Equation
Article
published earlier
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
title On the Form of Dispersive Shock Waves of the Korteweg-de Vries Equation
spellingShingle On the Form of Dispersive Shock Waves of the Korteweg-de Vries Equation
Egorova, I.
Gladka, Z.
Teschl, G.
title_short On the Form of Dispersive Shock Waves of the Korteweg-de Vries Equation
title_full On the Form of Dispersive Shock Waves of the Korteweg-de Vries Equation
title_fullStr On the Form of Dispersive Shock Waves of the Korteweg-de Vries Equation
title_full_unstemmed On the Form of Dispersive Shock Waves of the Korteweg-de Vries Equation
title_sort on the form of dispersive shock waves of the korteweg-de vries equation
author Egorova, I.
Gladka, Z.
Teschl, G.
author_facet Egorova, I.
Gladka, Z.
Teschl, G.
publishDate 2016
language English
container_title Журнал математической физики, анализа, геометрии
publisher Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
format Article
description We show that the long-time behavior of solutions to the Korteweg{de Vries shock problem can be described as a slowly modulated one-gap solution in the dispersive shock region. The modulus of the elliptic function (i.e., the spectrum of the underlying Schrödinger operator) depends only on the size of the step of the initial data and on the direction, x/ t =const, along which we determine the asymptotic behavior of the solution. In turn, the phase shift (i.e., the Dirichlet spectrum) in this elliptic function depends also on the scattering data, and is computed explicitly via the Jacobi inversion problem. Показано, что поведение при большом времени решений уравнения Кортевега-де Фриза с начальными данными типа ступеньки, соответствующими волне сжатия, в области эллиптической волны может быть описано слабо модулированным двухзонным решением. Модуль этой эллиптической функции, определяемый спектром фонового оператора, зависит от размера ступеньки в начальных данных и от направления, в котором исследуется асимптотическое поведение решения. В свою очередь фазовый сдвиг (то есть спектр задачи Дирихле) в этой эллиптической функции зависит также от данных рассеяния, и он посчитан с помощью проблемы обращения Якоби.
issn 1812-9471
url https://nasplib.isofts.kiev.ua/handle/123456789/140545
citation_txt On the Form of Dispersive Shock Waves of the Korteweg-de Vries Equation / I. Egorova, Z. Gladka , G. Teschl // Журнал математической физики, анализа, геометрии. — 2016. — Т. 12, № 1. — С. 3-16. — Бібліогр.: 23 назв. — англ.
work_keys_str_mv AT egorovai ontheformofdispersiveshockwavesofthekortewegdevriesequation
AT gladkaz ontheformofdispersiveshockwavesofthekortewegdevriesequation
AT teschlg ontheformofdispersiveshockwavesofthekortewegdevriesequation
first_indexed 2025-11-24T02:35:57Z
last_indexed 2025-11-24T02:35:57Z
_version_ 1850409293558317056
fulltext Journal of Mathematical Physics, Analysis, Geometry 2016, vol. 12, No. 1, pp. 3–16 On the Form of Dispersive Shock Waves of the Korteweg–de Vries Equation I. Egorova, Z. Gladka B. Verkin Institute for Low Temperature Physics and Engineering National Academy of Sciences of Ukraine 47 Nauki Ave., Kharkiv, 61103, Ukraine E-mail: iraegorova@gmail.com gladkazoya@gmail.com G. Teschl Faculty of Mathematics University of Vienna Oskar-Morgenstern-Platz 1, 1090 Wien, Austria International Erwin Schrödinger Institute for Mathematical Physics Boltzmanngasse 9, 1090 Wien, Austria E-mail: Gerald.Teschl@univie.ac.at Received October 12, 2015, published online December 4, 2015 We show that the long-time behavior of solutions to the Korteweg–de Vries shock problem can be described as a slowly modulated one-gap solution in the dispersive shock region. The modulus of the elliptic function (i.e., the spectrum of the underlying Schrödinger operator) depends only on the size of the step of the initial data and on the direction, x t =const, along which we determine the asymptotic behavior of the solution. In turn, the phase shift (i.e., the Dirichlet spectrum) in this elliptic function depends also on the scattering data, and is computed explicitly via the Jacobi inversion problem. Key words: KdV equation, steplike, dispersive shock wave. Mathematics Subject Classification 2010: 37K40, 35Q53 (primary); 33E05, 35Q15 (secondary). 1. Introduction The Korteweg–de Vries (KdV) shock problem is concerned with the long-time behavior of the solution of the KdV equation qt(x, t) = 6q(x, t)qx(x, t)− qxxx(x, t), (x, t) ∈ R× R, Research supported by the Austrian Science Fund (FWF) under Grant V120. c© I. Egorova, Z. Gladka, and G. Teschl, 2016 I. Egorova, Z. Gladka, and G. Teschl with the steplike initial profile{ q(x, 0)→ 0, as x→ +∞, q(x, 0)→ −c2, as x→ −∞. (1.1) This behavior is well understood on a physical level of rigor for the pure-step ini- tial data (i.e., when q(x, 0) = 0 as x > 0, q(x, 0) = −c2 as x < 0), and was studied using a quasi-classical Whitham approach in [13, 15, 16] with further treatments using the matched-asymptotic method in [1, 21]. Extensions to steplike finite-gap backgrounds were given in [3–6] and [23]. This led to the following three main regions with different asymptotical behavior of the solution: 1. Region x < −6c2t, where the solution is asymptotically close to the back- ground −c2 up to a decaying dispersive tail. 2. Middle region −6c2t < x < 4c2t, also known as dispersive shock or elliptic region, where the solution can asymptotically be described by a modulated elliptic wave. 3. Soliton region 4c2t < x, where the solution is asymptotically given by a sum of solitons. We refer to our paper [10] for further details and more on the history of this problem. In this note we want to revisit the middle region which is the most interesting and challenging one from a mathematical point of view. In particular, in the case of pure-step initial data, Gurevich and Pitaevskii ([15, 16], see also [21]) derived the following large-time asymptotical formula in the elliptic region: q(x, t) ∼ qGP(x, t) = −2c2dn2 ( 2tc(6ξ − c2(1 + m2(ξ)), m(ξ) ) + c2(1−m2(ξ)), (1.2) where ξ = x 12t , and the modulus m(ξ) is determined implicitly by c2 6 ( 1 + m2(ξ)− 2m2(ξ)(1−m2(ξ))K(m(ξ)) [E(m(ξ))− (1−m2(ξ))K(m(ξ))] ) = ξ. (1.3) Here dn(s,m) is the Jacobi elliptic function and K(m), E(m) are the standard complete elliptic integrals. The function qGP(x, t) is a stationary running wave of the KdV equation if the parameter ξ is a constant. On the other hand, in [10], under the assumption +∞∫ 0 eC0x(|q(x)|+ |q(−x) + c2|dx <∞, C0 > c > 0, (1.4) 4 Journal of Mathematical Physics, Analysis, Geometry, 2016, vol. 12, No. 1 On the Form of Dispersive Shock Waves of the Korteweg–de Vries Equation we derived the formula (for precise conditions, see our paper) q(x, t) ∼ qRH(x, t) = Γ(ξ)B′(ξ) 6π d2 dv2 log θ3 ( tB(ξ) + ∆(ξ) 2π + v ) ∣∣∣ v=0 − z′(ξ) 6 , (1.5) where all quantities are associated with the elliptic Riemann surface y2 = λ(λ+ c2)(λ+ a2(ξ)) depending on the parameter a(ξ) > 0 for ξ ∈ (−c2/2, c2/3) implic- itly determined by 0∫ a(ξ) ( ξ + c2 − a2(ξ) 2 − s2 )√ s2 − a2(ξ) c2 − s2 ds = 0. (1.6) As is shown in [18, 19], the point a(ξ) increases monotonically from 0 to c > 0 as ξ changes from −c2/2 to c2/3. The quantities B, Γ, z and ∆ are defined in (2.9)–(2.12) below. Formula (1.5) is very reminiscent of the usual Its–Matveev formula for the one-gap solution of the KdV equation. Hence this raises two natural questions, namely, whether (1.5) agrees with (1.2) and whether they reduce to the Its– Matveev formula for the constant ξ, that is, whether (1.5) can be viewed as a slowly modulated one-gap solution of the KdV equation. The purpose of the present note is to give a positive answer to both questions. 2. Comparison between qRH(x, t) and the Its–Matveev Formula We begin by recalling the well-known Its–Matveev formula for the finite- gap solution of the KdV equation in the case of one gap in the spectrum (see, for example, [14, 20, 22, 17]). To facilitate further comparison of this formula with qRH(x, t), we suppose that the spectrum of this one-gap solution is the set σ = [−c2,−a2(ξ)] ∪ [0,∞). Let M = M(ξ) be the elliptic Riemann surface associated with the function R(λ) := R(λ, ξ) = √ λ(λ+ c2)(λ+ a2(ξ)), where the cuts are taken along the spectrum σ, and R(λ) takes positive values on the upper side of the cut along the interval [0,∞). A point on M is denoted by p = (λ,±), λ ∈ C, and the projection onto C ∪ {∞} is denoted by π(p) = λ. The sheet exchange map is given by p∗ = (λ,∓) for p = (λ,±). The sets ΠU = {(λ,+) | λ ∈ C \ σ} ⊂M, ΠL = {p∗ | p ∈ ΠU}, are called the upper, the lower sheets, respectively. Introduce the canonical bases of a and b cycles. The cycle b surrounds the interval [−c2,−a2] clockwise on the Journal of Mathematical Physics, Analysis, Geometry, 2016, vol. 12, No. 1 5 I. Egorova, Z. Gladka, and G. Teschl upper sheet, and the cycle a supplements b passing along the gap from −a2 to 0 in the positive direction on the lower sheet and back from 0 to −a2 on the upper sheet. Next, let dω be a holomorphic differential on M normalized by ∫ a dω = 2πi. Evidently, dω = Γ̃ dλ R(λ) , Γ̃ = 2πi ∫ a dλ R(λ) −1 > 0, (2.1) and τ̃ = ∫ b dω < 0. (2.2) Define now the theta function of M by θ(z) = θ(z | τ̃) = ∑ m∈Z exp{1 2 τ̃m2 + zm}. Recall that this function is even and takes real values for z ∈ iR ∪ R. Next, following [20], introduce two meromorphic differentials of the second kind dΩ1 and dΩ3 with the vanishing a periods and with the only pole at infinity of the form dΩ1 = i 2 λ− h R(λ) dλ, dΩ3 = −3i 2 (λ− ν1)(λ− ν2) R(λ) dλ, (2.3) where h = ∫ a λdλ R(λ) ∫ a dλ R(λ) −1 ∈ (−a2, 0), (2.4) and the points νj ∈ R are chosen such that ∫ a dΩ3 = 0, and 2ν1 + 2ν2 + c2 + a2 = 0. (2.5) The last equality guarantees the absence of the term of order λ−1/2dλ in repre- sentation for dΩ3. Note that at least one of the points νi lies in the gap (−a2, 0). Denote iV = ∫ b dΩ1, and iW = ∫ b dΩ3. (2.6) The values V and W are called the wave number and the frequency. Evidently, V,W ∈ R. Next recall the Abel map A(p) = ∫ p ∞ dω. Let p0 be a point on M with the projection in the gap, π(p0) ∈ [−a2, 0]. For this point, the value A(p0) + K is pure imaginary (cf. [8]), where K = − τ̃ 2 + πi is the Riemann constant. Then 6 Journal of Mathematical Physics, Analysis, Geometry, 2016, vol. 12, No. 1 On the Form of Dispersive Shock Waves of the Korteweg–de Vries Equation the Its–Matveev formula for the one-gap solution with initial Dirichlet divisor p0 reads (cf. [20, 22]): qIM(x, t) = −2 d2 dx2 log θ (iV x− 4iWt−A(p0)−K)− a2 − c2 − 2h, (2.7) where h is defined by (2.4). Now we will study in more details formula (1.5), where the quantities are defined as follows: τ ∈ iR+ is the period given by τ := τ(ξ) = − −a2(ξ)∫ −c2 dλ√ λ(λ+ c2)(λ+ a2(ξ))  0∫ −a2(ξ) dλ√ λ(λ+ c2)(λ+ a2(ξ))  −1 , (2.8) and θ3(v) = θ3(v | τ) = ∑ m∈Z exp{(m2τ + 2mv)πi} is the associated theta function. Furthermore, B(ξ) = 24 c∫ a(ξ) ( ξ + c2 − a2(ξ) 2 − s2 )√ s2 − a2(ξ) c2 − s2 ds; (2.9) Γ(ξ) = −1 2  a(ξ)∫ −a(ξ) ( (c2 − s2)(a2(ξ)− s2) )−1/2 ds  −1 ; (2.10) z(ξ) = 12ξ(c2 − a(ξ)2) + 3c4 + 9a(ξ)4 − 6a(ξ)2c2 2 ; (2.11) ∆(ξ) = 2 c∫ a(ξ) log |(T (is)T1(is)|√ (c2 − s2)(s2 − a(ξ)2) ds  a(ξ)∫ −a(ξ) ds√ (c2 − s2)(a2(ξ)− s2)  −1 ; (2.12) T and T1 are the left and the right transmission coefficients of the initial data (1.1). In formulas (2.9), (2.10), and (2.12) the positive value of square root is taken. Denote −a2(ξ) = γ(ξ) := γ , µ = −ξ − c2 + γ 2 , (2.13) Journal of Mathematical Physics, Analysis, Geometry, 2016, vol. 12, No. 1 7 I. Egorova, Z. Gladka, and G. Teschl and put λ = −s2 in (1.6). Then (1.6) is equivalent to 0∫ γ G(λ)dλ = 0, G(λ, ξ) := (λ− µ)(λ− γ) R(λ) , (2.14) or F (γ, ξ) := 0∫ γ ( λ+ ξ + γ + c2 2 )√ λ− γ λ(λ+ c2) dλ = 0. (2.15) Equation (2.15) determines the function γ(ξ) implicitly, and the function µ(ξ) by (2.13). Lemma 2.1. For ξ ∈ (−c2/2, c2/3) the function γ(ξ) decays monotonically from 0 to −c2, and µ(ξ) ∈ (γ(ξ), 0). Moreover, d dξ γ(ξ) = 4 h(ξ)− γ(ξ) 3γ(ξ) + 2ξ + c2 , (2.16) where h(ξ) is defined by formula (2.4) with R(λ) = √ λ(λ− γ(ξ))(λ+ c2). P r o o f. The existence and uniqueness of γ(ξ) is proved in [18]. Formally, differentiating (2.15) with respect to ξ gives dγ dξ = −∂F ∂ξ (∂F∂γ )−1. In turn, this implies dγ dξ = 4S 3γ + 2ξ + c2 , S = 0∫ γ λ− γ R(λ) dλ  0∫ γ dλ R(λ) −1 , which implies (2.16). Evidently, S > 0 for γ < 0. Therefore, to prove monotonic- ity of γ(ξ), it is sufficient to prove that f(ξ) = 3γ(ξ) + 2ξ + c2 is negative for ξ ∈ (−c2/2, c2/3), and that γ(ξ) is also negative there. We observe that F (−c2, c2/3) = 0, therefore γ(c2/3) = −c2. Since f(c2/3) = −4c2/3 < 0, then γ′(c2/3) < 0. Thus γ(ξ) grows continuously starting from −c2 as ξ decreases starting from c2/3. The function f is a continuous function of ξ, and monotonicity of γ can only stop if there is a change of sign of f . Let ξ0 be a point where f(ξ0) = 0. Then ξ0 satisfies the system{ 3γ(ξ0) + 2ξ0 + c2 = 0 2µ(ξ0)− 2ξ0 − c2 − γ(ξ0) = 0. Thus, µ(ξ0) = γ(ξ0). But for γ < 0, formula (2.14) holds iff µ ∈ (γ, 0). This means that µ(ξ0) = γ(ξ0) = 0, that is, ξ0 = −c2/2. 8 Journal of Mathematical Physics, Analysis, Geometry, 2016, vol. 12, No. 1 On the Form of Dispersive Shock Waves of the Korteweg–de Vries Equation Lemma 2.2. Let γ(ξ) and µ(ξ) be as in Lemma 2.1 and h(ξ) as in (2.4). Let G(λ, ξ) be defined by (2.14) and dΩj, j = 1, 3 by (2.3)–(2.5). Then, for any ξ ∈ ( − c2 2 , c2 3 ) , the following representation is valid: Gdλ = 2i 3 dΩ3 − 2iξ dΩ1. (2.17) Moreover, the following formula holds: ∂ ∂ξ G(λ, ξ) = λ− h(ξ) R(λ, ξ) . (2.18) P r o o f. By (2.14), the differential Gdλ has the vanishing a period. Moreover, using (2.5), (2.13), one checks that Gdλ− 2i 3 dΩ3 + 2iξ dΩ1 has no pole at ∞ and hence must vanish. This proves (2.17). To get (2.18), we evaluate ∂ ∂ξ G(λ, ξ)− λ− h(ξ) R(λ, ξ) = ∂ ∂ξ ( (λ− µ) √ λ− γ√ λ(λ+ c2) ) − λ− h R(λ) = (−2µ′ − γ′)λ+ 2µ′γ + µγ′ − 2λ+ 2h 2R(λ) . Formula (2.13) implies 2µ′ + γ′ + 2 = 0. (2.19) Thus, to justify (2.18), one has to prove the equality 2µ′γ + µγ′ + 2h = 0. By virtue of (2.16), (2.19), and (2.13), we get 2µ′γ + µγ′ + 2h = (−γ′ − 2)γ − γ′ ( ξ + c2 + γ 2 ) + 2h = −1 2 γ′(3γ + c2 + 2ξ)− 2γ + 2h = 0, which proves (2.18). Lemma 2.3. Let B(ξ), Γ(ξ), and z(ξ) be defined by (2.9), (2.10), and (2.11), respectively. Let V (ξ) and W (ξ) be the wave number and the frequency, defined by (2.6), and let h(ξ) be as in (2.4). Then the following identities hold: tB(ξ) = −4W (ξ)t+ V (ξ)x, (2.20) (a) d dξ B(ξ) = 12V (ξ), (b) 4πΓ(ξ) = −V (ξ), (2.21) 1 6 d dξ z(ξ) = c2 + a2(ξ) + 2h(ξ). (2.22) Journal of Mathematical Physics, Analysis, Geometry, 2016, vol. 12, No. 1 9 I. Egorova, Z. Gladka, and G. Teschl P r o o f. To get (2.20), we make change of variables λ = −s2 in (2.9), and take into account (2.17), (2.13), ξ = x 12t , and the definition of the b period on the Riemann surface M(ξ). Then tB(ξ) = −12t −c2∫ −a2 (λ− µ) √ −a2 − λ −λ(λ+ c2) dλ = 12t γ∫ −c2 G(λ, ξ)dλ = 4it ∫ b dΩ3 − ix ∫ b dΩ1 = −4tW (ξ) + xV (ξ). This proves (2.20). Formula (2.21), (a) follows from (2.17) and (2.18): d dξ B(ξ) = 12 d dξ γ∫ −c2 G(λ, ξ)dλ = 12γ′(ξ)G(γ(ξ), ξ)− 12i ∫ b dΩ1 = 12V (ξ). Next, by definition of the a period, formula (2.10) reads: Γ(ξ) = 2 −a2∫ 0 dλ√ −λ(λ+ c2)(λ+ a2)  −1 = i∫ a dλ R(λ) = Γ̃ 2π , (2.23) where Γ̃ is the normalization constant from (2.1). On the other side, formulas (2.1), (2.3), and the residue theorem [12] yield ∫ a dω ∫ b dΩ1 = 2πi ∫ b dΩ1 = 2πi Res∞ dΩ1(p) p∫ ∞ dω  , p = (λ,+). Since in the local parameter z = λ−1/2, z → 0, we have dΩ1 = ( i 2 √ λ +O ( 1 λ3/2 )) dλ = − i z2 (1 + o(1))dz, and λ∫ ∞ dω = − 2Γ̃√ λ +O ( 1 λ3/2 ) = −2Γ̃z(1 + o(1)), then, by (2.23), Res∞ dΩ1(p) p∫ ∞ dω  = −2iΓ̃ = −4πiΓ. 10 Journal of Mathematical Physics, Analysis, Geometry, 2016, vol. 12, No. 1 On the Form of Dispersive Shock Waves of the Korteweg–de Vries Equation Together with (2.6), this proves (2.21), (b). To prove the remaining formula (2.22), we represent z(ξ) via γ(ξ) and apply (2.16). Then we get d dξ z(ξ) = 1 2 d dξ ( 12ξ(c2 + γ(ξ)) + 3c4 + 9γ(ξ)2 + 6γ(ξ)c2 ) = 6(c2 + γ(ξ)) + ( 6ξ + 9γ(ξ) + 3c2 ) γ′(ξ) = 6(c2 + γ(ξ)) + 12(h(ξ)− γ(ξ)) = 6(c2 + a2(ξ)) + 12h(ξ). Now we are ready to compare formula (1.5) with (2.7). Comparing the periods τ defined by (2.8) and τ̃ defined by (2.1), (2.2), we observe that 2πiτ = τ̃ , and therefore θ(z | τ̃) = θ3( z 2πi | τ). Put B(ξ) = Γ(ξ) 6π ( d dξ B(ξ) ) , C(ξ) = 1 6 d dξ z(ξ). (2.24) Then qRH(x, t) can be represented as qRH(x, t) = B(ξ) d2 dv2 log θ3 ( itB(ξ) + i∆(ξ) + 2iπv 2πi ) |v=0 +C(ξ), = −4π2B(ξ) d2 dv2 log θ (itB(ξ) + i∆(ξ) + v) |v=0 +C(ξ). (2.25) On the other hand, by (2.22), we have qIM(x, t)− C(ξ) = −2(iV (ξ))2 d2 dv2 log θ (iV (ξ)x− 4iW (ξ)t−A(p0, ξ)−K(ξ) + v) |v=0 . Substituting (2.20) and (2.21) into (2.24) and then into (2.25), we get B(ξ) = −V 2(ξ) and itB(ξ) = iV (ξ)x − 4iW (ξ)t. We conclude that qRH(x, t) = qIM(x, t) iff there exists a point p0, π(p0) ∈ [−a2, 0] such that the following equality is fulfilled: −A(p0, ξ)−K(ξ) = i∆(ξ) (mod 2πi), where ∆(ξ) is defined by (2.12). Since A(−c2) = πi (mod 2πi), A(−a2)−A(−c2) = τ̃ 2 , K = − τ̃ 2 + πi, and ∆(ξ) is a real value, then the point p0(ξ) can be found as the unique solution of the Jacobi inversion problem (cf. [12]): p0(ξ)∫ −a2(ξ) dω = −i∆(ξ). (2.26) Journal of Mathematical Physics, Analysis, Geometry, 2016, vol. 12, No. 1 11 I. Egorova, Z. Gladka, and G. Teschl In summary we have proved Theorem 2.4. For any fixed ξ = x 12t ∈ (−c2/2, c2/3) the function qRH(x, t) is the usual one-gap solution of the KdV equation: qRH(x, t) = −2 d2 dx2 log θ (iV (ξ)x− 4iW (ξ)t−A(p0, ξ)−K(ξ)) − a(ξ)2 − c2 − 2h(ξ), associated with the spectrum [−c2,−a2(ξ)]∪ [0,∞) and the Dirichlet divisor p0(ξ) defined via the Jacobi inversion problem (2.26). Note that since θ(z+ 2πi) = θ(z), we see that qRH(x, t) is a periodic function with respect to x and t of the periods 2π V (ξ) and π 2W (ξ) , respectively. 3. Copmarison between qRH(x, t) and the Gurevich–Pitaevskii Formula Our next aim is to find a connection between the two functions m(ξ) and a(ξ) implicitly given by (1.3) and (1.6). Lemma 3.1. The following is true: m2(ξ) = a2(ξ) c2 . P r o o f. Represent (1.6) as 0 = ( −c 2 + a2(ξ) 2 + ξ ) a(ξ)∫ 0 √ a2(ξ)− s2 c2 − s2 ds + a(ξ)∫ 0 √ (a2(ξ)− s2)(c2 − s2)ds = ( ξ − c2 + a2(ξ) 2 ) I1(ξ) + I2(ξ). (3.1) Put m := m(ξ) = a(ξ) c . Then (cf. [9], formulas 781.61 and 781.22) I1 = c(E(m)− (1−m2)K(m)), I2 = c3 3 {(m2 − 1)K(m) + (m2 + 1)E(m)}, 12 Journal of Mathematical Physics, Analysis, Geometry, 2016, vol. 12, No. 1 On the Form of Dispersive Shock Waves of the Korteweg–de Vries Equation where K(m) and E(m) are the standard complete elliptic integrals for 0 < m < 1. Substituting this into (3.1), we obtain ξ = c2 { m2 + 1 2 − 1 3 (m2 − 1)K(m) + (m2 + 1)E(m) E(m)− (1−m2)K(m) } = c2 6 E(m)(m2 + 1)− (1−m2)(m2 + 1)K(m) + 2m2(m2 − 1)K(m) E(m)− (1−m2)K(m) = c2 6 ( (m2 + 1) + 2m2(m2 − 1)K(m) E(m)− (1−m2)K(m) ) . Thus, (1.3) and (1.6) define the same function: m(ξ) = m(ξ). Lemma 3.1 implies that the period (2.8) corresponds in the standard way (cf. [2]) to the elliptic modulus m, and the following formula is valid: dn2(s,m) = d2 du2 log Θ(u | τ) + E(m) K(m) , where Θ(u | τ) = θ3 ( u 2K(m) + 1 2 | τ ) = θ3 ( u 2K(m) + 1 2 ) , and θ3(s) is as in (1.5). For the remainder, we will fix ξ and omit it from our notation. Fixing ξ, we also fix m = a c−1 and, consequently, we will also omit m in the complete elliptic integrals: K := K(m), E := E(m). Therefore, qGP (x, t) = − c2 2K2 d2 dv2 log θ3 ( tc(6ξ − c2(1 +m2)) K + 1 2 + v ) + c2(1−m2)− 2E K . (3.2) Thus, to convince ourselves that this formula coincides with (1.5) (possibly, up to a phase shift), it is sufficient to show: Lemma 3.2. The following equalities hold: − c2 2K2 = Γ 6π dB dξ ; (3.3) −6cξ − c3(1 +m2) K = B 2π ; (3.4) c2 − E K = −h, (3.5) where B, Γ, and h are defined by (2.9), (2.10), and (2.4), respectively. Journal of Mathematical Physics, Analysis, Geometry, 2016, vol. 12, No. 1 13 I. Egorova, Z. Gladka, and G. Teschl P r o o f. Formulas (2.21), (2.10), and 781.01 of [9] imply Γ 6π dB dξ = −8Γ2 = −1 2  a∫ 0 ds√ (c2 − s2)(a2 − s2) −2 = − c2 2K2 . (3.6) Next, by (2.4) and formula 781.11 from [9], we have −h = a∫ 0 s2ds√ (c2 − s2)(a2 − s2)  a∫ 0 ds√ (c2 − s2)(a2 − s2) −1 = c(K − E) 1 cK . This gives (3.5). To prove (3.4), we use (2.20) and (2.21). Namely, we have tB 2π = −2Γx− 4Wt = −2(x− 2(c2 + a2)t)Γ. Here we used equation (4.4.13) of [22] which implies the equality W = (c2 +a2)Γ. By (3.6), we have Γ = c 4K . Thus, B 2π = − tc K ( 6ξ − (c2 + a2) ) , which proves (3.4). Note that the opposite sign with respect to (3.2) of the first summand in θ3 is not essential as θ3 is even. We proved that formulas (1.5) and (1.2) represent the same function up to the phase shift. Namely, instead of the summand (2π)−1∆ in the argument of the theta function for qRH(x, t), in the same formula for qGP(x, t) we have the summand 1 2 . Recall now that the transmission coefficients for the pure step potential have the representation (cf. [7, 11]) T (k) = 2ik w(k) , T1(k) = 2i √ k2 + c2 w(k) , w(k) = ik + i √ k2 + c2. Since in (2.12) k = is, s ∈ [0, c], then |w(k)|2 = | − s+ i √ c2 − s2|2 = c2, and |T (is)T1(is)| = 4sc−2 √ c2 − s2, s ∈ [0, c]. Therefore, the value of the phase shift for the pure step case is given by ∆ps 2π = ∫ c a log ( 4sc−2 √ c2 − s2 ) ( (c2 − s2)(s2 − a2) )−1/2 ds 2π ∫ a 0 ((c2 − s2)(s2 − a2))−1/2 ds = 1 2πK(m) 1∫ m log ( 4s √ 1− s2 ) √ (1− s2)(s2 −m2) ds. 14 Journal of Mathematical Physics, Analysis, Geometry, 2016, vol. 12, No. 1 On the Form of Dispersive Shock Waves of the Korteweg–de Vries Equation Acknowledgments. We thank Alexei Rybkin and Johanna Michor for valu- able discussions on this topic. I.E. is indebted to the Department of Mathematics at the University of Vienna for its hospitality and support during the fall of 2015, where part of this work was done. References [1] M.J. Ablowitz and D.E. Baldwin, Interactions and Asymptotics of Dispersive Shock Waves — Korteweg–de Vries Equation. — Phys. Lett. A 377 (2013), 555–559. [2] N.I. Akhiezer, Elements of the Theory of Elliptic Functions. Translation of Mathe- matical Monographs 79, Amer. Math. Soc., Providence, 1990. [3] R.F. Bikbaev, Structure of a Shock Waves in the Theory of the Korteweg–de Vries Equation. — Phys. Lett. A 141 (1989), 289–293. [4] R.F. Bikbaev and V.Yu. Novokshenov, Self-Similar Solutions of the Whitham Equa- tions and KdV Equation with Finite-Gap Boundary Conditions. — Proc. of the III Intern. Workshop. Kiev 1 (1988), 32–35. [5] R.F. Bikbaev and V.Yu. Novokshenov, Existence and Uniqueness of the Solution of the Whitham Equation. Asymptotic methods for solving problems in mathematical physics, 81–95, Akad. Nauk SSSR Ural. Otdel., Bashkir. Nauchn. Tsentr, Ufa, 1989. (Russian) [6] R.F. Bikbaev and R.A. Sharipov, The Asymptotic Behavior as t→∞ of the Solution of the Cauchy Problem for the Korteweg–de Vries Equation in a Class of Potentials with Finite-Gap Behavior as x→ ±∞. — Theor. Math. Phys. 78 (1989), 244–252. [7] V.S. Buslaev and V.N. Fomin, An Inverse Scattering Problem for the One- Dimensional Schrödinger Equation on the Entire Axis. Vestnik Leningrad. Univ. 17 (1962), 56–64. (Russian) [8] B.A. Dubrovin, Theta Functions and Nonlinear Equations. — Russian Math. Sur- veys 36 (1981), 11–92. [9] H.B. Dwight, Tables of Integrals and Other Mathematical Data. 4th ed. The Macmil- lan Company, New York, 1961. [10] I. Egorova, Z. Gladka, V. Kotlyarov, and G. Teschl, Long-Time Asymptotics for the Korteweg–de Vries Equation with Steplike Initial Data. — Nonlinearity 26 (2013), 1839–1864. [11] I. Egorova, Z. Gladka, T.L. Lange, and G. Teschl, Inverse Scattering Theory for Schrödinger Operators with Steplike Potentials. — J. Math. Phys., Anal., Geom. 11 (2015), 123–158. [12] H.M. Farkas and I. Kra, Riemann Surfaces. Graduate Texts in Mathematics 71, Springer-Verlag, New York, 1980. [13] B. Fornberg and G. B. Whitham, A numerical and Theoretical Study of Certain Nonlinear Wave Phenomena. — Phil. Trans. R. Soc. Lond 289 (1978), 373–404. Journal of Mathematical Physics, Analysis, Geometry, 2016, vol. 12, No. 1 15 I. Egorova, Z. Gladka, and G. Teschl [14] F. Gesztesy and H. Holden, Soliton Equations and Their Algebro-Geometric So- lutions; Volume I: (1+1)-Dimensional Continuous Models. Cambridge Studies in Advanced Mathematics 79, Cambridge Univ. Press, Cambridge, 2003. [15] A.V. Gurevich and L.P. Pitaevskii, Decay of Initial Discontinuity in the Korteweg– de Vries Equation. — JETP Letters 17 (1973), 193–195. [16] A.V. Gurevich and L.P. Pitaevskii, Nonstationary Structure of a Collisionless Shock Wave. — Soviet Phys. JETP 38 (1974), 291–297. [17] A.R. Its and V.B. Matveev, Schrödinger Operators with the Finite-Band Spectrum and the N -soliton Solutions of the Korteweg–de Vries Equation. — Teoret. Mat. Fiz. 23 (1975), No. 1, 51–68. (Russian) [18] V.P. Kotlyarov and A.M. Minakov, Riemann–Hilbert Problem to the Modified Korteweg–de Vries Equation: Long-Time Dynamics of the Step-like Initial Data. — J. Math. Phys. 51 (2010), 093506. [19] V.P. Kotlyarov and A.M. Minakov, Step-Initial Function to the MKdV Equation: Hyper-Elliptic Long-Time Asymptotics of the Solution. — J. Math. Phys., Anal. Geom. 8 (2012), 38–62. [20] S.B. Kuksin, Analysis of Hamiltonian PDEs. Oxford Lecture Series in Mathematics and its Applications 19, Oxford University Press, Oxford, 2000. [21] J.A. Leach and D.J. Needham, The Large-Time Development of the Solution to an Initial-Value Problem for the Korteweg–de Vries Equation. II. Initial Data has a Discontinuous Compressive Step. — Mathematika 60 (2014), 391–414. [22] V.A. Marchenko, Sturm–Liouville Operators and Applications, rev. ed. Amer. Math. Soc., Providence, 2011. [23] V.Yu. Novokshenov, Time Asymptotics for Soliton Equations in Problems with Step Initial Conditions. — J. Math. Sci. 125 (2005), 717–749. 16 Journal of Mathematical Physics, Analysis, Geometry, 2016, vol. 12, No. 1