betaFeedback

WorksPhysics and Astronomy

Exact analytical solutions for the kinks, the solitons and the shocks in discrete nonlinear transmission line with nonlinear capacitance

Abstract

We studied discrete transmission lines constructed from ideal linear inductors and nonlinear capacitors (and possibly resistors). The localised travelling waves in the lossless transmission lines are the kinks and the solitons, which speeds and profiles were calculated. The localised travelling waves in the lossy transmission lines are the dissipative kinks and the shocks, which speeds and profiles were also calculated.

Introduction

The nonlinear electrical transmission lines are of much interest both due to their applications, and as the laboratories to study nonlinear waves[1]. Among the latter – the travelling waves, especially the kinks and the solitons, attract a lot of attention[2]. The nonlinear waves become especially interesting (and challenging) when one takes into account the discrete nature of the transmission line. The kinks and the solitons in discrete transmission lines were studied in quite a few publications[3, 4, 5].

The field of nonlinear waves in discrete transmission lines is closely connected with Fermi-Pasta-Ulam-Tsingou (FPUT) problem[6]. In particular, dynamics of lattice kinks and solitons were studied in a large number of publications[7, 8, 9, 10]

We studied previously the kinks and the solitons in discrete Josephson transmission line (JTL) JTL[11, 12]. There we formulated the quasi-static approximation, which allows to reduce the discrete equations to the continuous ones. In the present short note we would like to apply this approximation to different kind of discrete transmission lines – with nonlinear capacitors. Such transmission lines in the continuum approximation were studied in the classical papers by R. Landauer[13, 14, 15].

The kinks and the solitons

The travelling waves

The discrete transmission line is constructed from the identical inductors and nonlinear capacitors is shown on Figure 1.

The lossless discrete transmission line

Figure 1. The lossless discrete transmission line.

We take the voltages VnV_n and the currents which pass though the inductors In−1,nI_{n-1,n} as the dynamical variables. The circuit equations (Kirchhoff laws) are

dQndt=Inc(1)\frac{dQ_n}{dt}=I_n^c \tag*{(1)}
LdIn,n+1dt=Vn−Vn+1(2)L\frac{dI_{n,n+1}}{dt}=V_n-V_{n+1} \tag*{(2)}
Inc=In−1,n−In,n+1.(3)I_n^c=I_{n-1,n}-I_{n,n+1}. \tag*{(3)}

where Qn=Q(Vn)Q_n=Q(V_n) and the nonlinear nature of the transmission line is determined by the (monotonically increasing) function Q=Q(V)Q=Q(V). We can exclude In,n+1I_{n,n+1} from (1b) and (1c) and present these two equations as a single one

LdIncdt=Vn+1−2Vn+Vn−1.(4)L\frac{dI_n^c}{dt}=V_{n+1}-2V_n+V_{n-1}. \tag*{(4)}

From (1a) and (2) we can obtain closed wave equation for the voltage VV

Ld2Qndt2=Vn+1−2Vn+Vn−1.(5)L\frac{d^2Q_n}{dt^2}=V_{n+1}-2V_n+V_{n-1}. \tag*{(5)}

Let us write down Eq. (3) in the quasi-continuum approximation, which was proposed in our previous publications[11, 12]. The approximation consists in treating nn as the continuous variable ZZ, expanding the r.h.s. of the equation (3) in Taylor series with respect to ZZ and keeping only the first two terms of the expansion, thus obtaining the equation in the form

L∂2Q(V)∂t2=∂2V∂Z2+112∂4V∂Z4.(6)L\frac{\partial^2Q(V)}{\partial t^2}=\frac{\partial^2V}{\partial Z^2}+\frac{1}{12}\frac{\partial^4V}{\partial Z^4}. \tag*{(6)}

In this paper we are interested in the travelling wave solutions, for which the dependence of all the variables upon ZZ and tt is of the form

X(Z,t)=X(Z+Ut),(7)X(Z,t)=X(Z+Ut), \tag*{(7)}

where UU is the speed of the travelling wave (the distance is measured in the units of the transmission line spacial period AA). This from (4) we obtain an ordinary differential equation

U2Ld2Q(V)dZ2=d2VdZ2+112d4VdZ4.(8)U^2L\frac{d^2Q(V)}{dZ^2}=\frac{d^2V}{dZ^2}+\frac{1}{12}\frac{d^4V}{dZ^4}. \tag*{(8)}

After finding V(Z)V(Z) we can from (1b) obtain the current

I(Z)=−V(Z)UL+const.(9)I(Z)=-\frac{V(Z)}{UL}+\mathrm{const}. \tag*{(9)}

Performing an integration in (6) we can get

112d3VdZ3=U2LdQdZ−dVdZ.(10)\frac{1}{12}\frac{d^{3}V}{dZ^{3}}=U^{2}L\frac{dQ}{dZ}-\frac{dV}{dZ}. \tag*{(10)}

Notice that equation with the same form as (10) one would obtain after making the travelling wave ansatz in the generalised Korteweg-de Vries-modified Korteweg-de Vries (KDV-mKDV) equation.

We may perform the second integration to obtain

112d2VdZ2=U2LQ(V)−V+F,(11)\frac{1}{12}\frac{d^{2}V}{dZ^{2}}=U^{2}LQ(V)-V+F, \tag*{(11)}

where FF is the constant of integration. Introducing the new dependent variable

E=dVdZ,(12)E=\frac{dV}{dZ}, \tag*{(12)}

considering VV as the independent variable and taking into account that

d2VdZ2=12dE2dV,(13)\frac{d^{2}V}{dZ^{2}}=\frac{1}{2}\frac{dE^{2}}{dV}, \tag*{(13)}

we may present (9) as

124dE2dV=U2LQ(V)−V+F.(14)\frac{1}{24}\frac{dE^{2}}{dV}=U^{2}LQ(V)-V+F. \tag*{(14)}

However, for technical reasons, it will be more convenient to return to (8) and avoid the integration of the equation. After presenting dQ/dZdQ/dZ as (dQ/dV)(dV/dZ)(dQ/dV)(dV/dZ), taking into account that

d3VdZ3=12E2d2E2dV2,(15)\frac{d^{3}V}{dZ^{3}}=\frac{1}{2}E^{2}\frac{d^{2}E^{2}}{dV^{2}}, \tag*{(15)}

and dividing both parts of (8) by the common multiplier EE in both parts, we may reduce the order of the equation and write it down as

124d2E2dV2=U2LdQ(V)dV−1.(16)\frac{1}{24}\frac{d^{2}E^{2}}{dV^{2}}=U^{2}L\frac{dQ(V)}{dV}-1. \tag*{(16)}

Consider the ”sound” waves that is small amplitude, smooth waves propagating on the homogeneous backgrounds VV. These waves are described by Eq. (16), with the l.h.s. being put equal to zero. The speed of the sound is given by the equation

1u2(V)=LdQdV.(17)\frac{1}{u^{2}(V)}=L\frac{dQ}{dV}. \tag*{(17)}

Thus Eq. (16) can be presented as

124d2E2dV2=U2u2(V)−1.(18)\frac{1}{24}\frac{d^{2}E^{2}}{dV^{2}}=\frac{U^{2}}{u^{2}(V)}-1. \tag*{(18)}

Note that in the framework of the quasi-continuum approximation EE is just the voltage on the inductor. Thus the physical meaning of Eq. (18) is simple. Its solution gives the voltage on the inductor as the function of the voltage on the capacitor.

A simple mechanical interpretation of Eq. (18) may be helpful. If we call VV the ”time”, the equation can be considered as describing motion of the fictitious Newtonian particle, E2E^{2} being its coordinate and 1/241/24 it’s mass, under the influence of the time-dependent force, given by the r.h.s. of the equation.

We’ll consider localised travelling waves, that is impose upon V(Z)V(Z) the boundary conditions

lim⁡Z→−∞V(Z)=V1,lim⁡Z→+∞V(Z)=V2.(19)\lim_{Z\to-\infty}V(Z)=V_{1},\qquad\lim_{Z\to+\infty}V(Z)=V_{2}. \tag*{(19)}

The value of V2V_{2} may be different from V1V_{1}, in which case we’ll talk about the kink, or equal to V1V_{1}, in which case we’ll talk about the soliton. The quantity E2(V)E^{2}(V) obviously has minimum (equal to zero) at V=V1,2V=V_{1,2}. Hence from Eq. (18) follows

U2>u2(V1,2),(20)U^{2}>u^{2}(V_{1,2}), \tag*{(20)}

that is the travelling waves are supersonic[11, 12].

The boundary conditions (19) obviously give

E(V1,2)=0.(21)E(V_{1,2})=0. \tag*{(21)}

Also, from (13) and (19) follows that in the vicinity of V1,2V_{1,2}

E2∼(V−V1,2)2.(22)E^{2}\sim(V-V_{1,2})^{2}. \tag*{(22)}

The conditions (21) and (22) mean that the particle is at rest at the origin at ”time” equal to V1V_{1} and then again at ”time” equal to V2V_{2}.

The main result of the Section II A is Eq. (16). The natural question arises: why were we not happy with Eq. (9) (which can be easily integrated in quadratures) and introduced instead of it the former equation. The detailed answer to this question will be given in Section III, but we can formulate it announce it right now. The main aim of the present paper is obtaining exact analytical solutions for the kinks and the shocks in the lossy transmission line. When the loss terms are included into the equations, (14) can be integrated analytically much easier than (9).

The kinks

With four boundary conditions specifying both the position and the velocity of the particle at V=V1V=V_{1} and V=V2V=V_{2}, of the three parameters V1V_{1}, V2V_{2}, U2U^{2} only one (say V1V_{1}) can be considered as a free parameter, the other two are connected with it. One such constraint we could obtain at much earlier stage. By integrating (14) with respect to VV from V1V_{1} to V2V_{2} and taking into account the boundary conditions we obtain the speed of the kink

1U2=LQ2−Q1V2−V1.(23)\frac{1}{U^2} = L\frac{Q_2-Q_1}{V_2-V_1}. \tag*{(23)}

where Qi=Q(Vi)Q_i=Q(V_i).

If Q(V)Q(V) is a polynomial function of VV, from (22) follows that the kink can be presented as

E2(V)=P1(V)(V−V1)2(V−V2)2,(24)E^2(V)=P_1(V)(V-V_1)^2(V-V_2)^2, \tag*{(24)}

where P1(V)P_1(V) is some polynomial function.

If we approximate Q(V)Q(V) between V1V_1 and V2V_2 by the third degree polynomial

Q(V)=const+C0(αV+βV2+γV3)(25)Q(V)=\mathrm{const}+C_0\left(\alpha V+\beta V^2+\gamma V^3\right) \tag*{(25)}

(for brevity we’ll call Eq. (25) the cubic case), then

1u2(V)=LC0(α+2βV+3γV2),(26)\frac{1}{u^2(V)}=LC_0\left(\alpha+2\beta V+3\gamma V^2\right), \tag*{(26)}
1U2=LC0[α+β(V1+V2)+γ(V12+V1V2+V22)].(27)\frac{1}{U^2}=LC_0\left[\alpha+\beta(V_1+V_2)+\gamma\left(V_1^2+V_1V_2+V_2^2\right)\right]. \tag*{(27)}

We also can specify (24) to

E(V)=ψ(V−V1)(V−V2),(28)E(V)=\psi(V-V_1)(V-V_2), \tag*{(28)}

where ψ\psi is a constant (positive, if V1>V2V_1>V_2, and negative in the opposite case). Substituting (28) into (18) and using Vieta’s formula we obtain after simple algebra

ψ2=6U2LC0γ,(29)\psi^2=6U^2LC_0\gamma, \tag*{(29)}
V1+V2=−2β3γ.(30)V_1+V_2=-\frac{2\beta}{3\gamma}. \tag*{(30)}

Thus the necessary condition for the existence of the kink is the inequality γ>0\gamma>0, that is the speed of sound u2(V)u^2(V) should be a convex function of VV.

To find V(Z)V(Z) we have to solve Eq. (12). The solution is

V(Z)=V1+V22−V1−V22tanh⁡[V1−V22ψZ].(31)V(Z)=\frac{V_1+V_2}{2}-\frac{V_1-V_2}{2}\tanh\left[\frac{V_1-V_2}{2}\psi Z\right]. \tag*{(31)}

The solitons

To specify the soliton we need to introduce an additional parameter, and as such we’ll take the extremal value of the voltage V3V_3. That is VV changes between V1V_1 and V3V_3, and ∣V3−V1∣\lvert V_3-V_1\rvert we’ll call the amplitude of the soliton. Because V3V_3 is the extremal value of VV we have

(dVdZ)V=V3=0,(d2VdZ2)V=V3≠0.(32)\left(\frac{dV}{dZ}\right)_{V=V_3}=0,\qquad\left(\frac{d^2V}{dZ^2}\right)_{V=V_3}\ne0. \tag*{(32)}

Hence from (12) and (13) follows that in the vicinity of V3V_3

E2∼(V−V3).(33)E^2\sim(V-V_3). \tag*{(33)}

Recalling our mechanical analogy we see that the soliton corresponds to the motion of the fictitious particle with zero initial coordinate and velocity at V=V1V=V_1 and zero coordinate (but not velocity) at V=V3V=V_3. Mechanical interpretation of (16) in the case of the soliton leads to the situation which doesn’t normally occur in mechanics. The boundary condition V2=V1V_2=V_1 means that the ”time” VV flows along the contour from V1V_1 to V3V_3 and then back from V3V_3 to V1V_1. This is why E(V)E(V) for the soliton will later turn out to be a double-valued function.

With the position and the velocity of the particle at V=V1V=V_1 and its position at V=V3V=V_3 being specified, of the three parameters V1V_1, V3V_3, U2U^2 only two (say V1V_1 and V3V_3) can be considered as free parameters. Multiplying both sides of (10) by V3−VV_3-V, integrating the equation with respect to VV from V1V_1 to V3V_3 and taking into account the boundary conditions we obtain the speed of the soliton

1U2=2(V3−V1)2∫V1V3V3−Vu2(V) dV.(34)\frac{1}{U^2}=\frac{2}{(V_3-V_1)^2}\int_{V_1}^{V_3}\frac{V_3-V}{u^2(V)}\,dV. \tag*{(34)}

If Q(V)Q(V) is a polynomial function of VV, from (22) and (33) follows that the soliton can be presented as

E2(V)=P2(V)(V−V1)2(V−V3),(35)E^2(V)=P_2(V)(V-V_1)^2(V-V_3), \tag*{(35)}

where P2(V)P_2(V) is some polynomial function.

In the cubic case, calculating the integral in (34) we obtain

1U2=LC0[α+2β3(2V1+V3)+γ2(3V12+2V1V3+V32)].(36)\frac{1}{U^2}=LC_0\left[\alpha+\frac{2\beta}{3}(2V_1+V_3)+\frac{\gamma}{2}(3V_1^2+2V_1V_3+V_3^2)\right]. \tag*{(36)}

We also can specify (35) to

E2(V)=ψ(V−V1)2(V−V3)(V−V4),(37)E^2(V)=\psi(V-V_1)^2(V-V_3)(V-V_4), \tag*{(37)}

where ψ\psi is a constant. Substituting (37) into (18) and using Vieta’s formula we obtain after simple algebra

ψ=6U2LC0γ,(38)\psi=6U^2LC_0\gamma, \tag*{(38)}
2V1+V3+V4=−4β3γ.(39)2V_1+V_3+V_4=-\frac{4\beta}{3\gamma}. \tag*{(39)}

To make life easier to ourselves, let us restrict the consideration by the case ∣β/γ∣≫∣V1∣,∣V3∣\lvert\beta/\gamma\rvert\gg\lvert V_1\rvert,\lvert V_3\rvert. Physically the approximation means that V1V_1 and V3V_3 are close enough to each other, and we can ignore the cubic term in (25). In this case, excluding V4V_4 and introducing the amplitude ϕ\phi instead of ψ\psi we obtain that (37) and (34) are modified to

E2(V)=ϕ(V−V1)2(V−V3),(40)E^2(V)=\phi(V-V_1)^2(V-V_3), \tag*{(40)}
ϕ4=2U2LC0β.(41)\frac{\phi}{4}=2U^2LC_0\beta. \tag*{(41)}

From (35) we see that the soliton exists both for positive and negative β\beta. For positive β\beta we have dark soliton (V3<V1V_3<V_1), and for negative β\beta — bright soliton (V3>V1V_3>V_1). To find V(Z)V(Z) we have to solve Eq. (12). The solution is

V(Z)=V1+(V3−V1)sech⁡2(V3−V12ϕZ).(42)V(Z)=V_1+(V_3-V_1)\operatorname{sech}^2\left(\frac{V_3-V_1}{2}\phi Z\right). \tag*{(42)}

How good is the quasi-continuum approximation?

We would like to discuss an opportunity (and motivation) to go beyond the limits of the quasi-continuum approximation used in the paper. To do it, one may keep the third term in the Taylor expansion of the discrete second order derivative in the r.h.s. of (2)[11, 12]

Vn+1−2Vn+Vn−1=∂2V∂Z2+112∂4V∂Z4+1360∂6V∂Z6.(43)V_{n+1}-2V_n+V_{n-1}=\frac{\partial^2 V}{\partial Z^2}+\frac{1}{12}\frac{\partial^4 V}{\partial Z^4}+\frac{1}{360}\frac{\partial^6 V}{\partial Z^6}. \tag*{(43)}

Thus we obtain instead of (8) the equation

U2LdQdZ=dVdZ+112d3VdZ3+1360d5VdZ5.(44)U^2L\frac{dQ}{dZ}=\frac{dV}{dZ}+\frac{1}{12}\frac{d^3V}{dZ^3}+\frac{1}{360}\frac{d^5V}{dZ^5}. \tag*{(44)}

Looking at the results of the previous Sections we understand that the fifth order derivative in (44) can be discarded if γ(V1−V2)2/α≪1\gamma(V_1-V_2)^2/\alpha\ll1 and β(V1−V3)/α≪1\beta(V_1-V_3)/\alpha\ll1. The small numerical coefficient before the sixth order derivative in the r.h.s. of (38) means that the small parameters shouldn’t be actually very small.

With all that we have to admit that the transition from the discrete to the continuous system is an uncontrollable approximation. More specifically, the mere fact of existence of the travelling wave of the form (5) in the discrete system should be proved analytically or, at least, numerically. We hope to return to this issue in the future.

The dissipative kinks and shocks

Let us ask ourselves what happens with the kinks when the transmission line becomes lossy? To answer this question let us include in the system the resistors RLR_L shunting the inductors and the resistors RCR_C in series with the nonlinear capacitors (see Figure 2). In this case the circuit equations are modified to

The lossy discrete transmission line

Figure 2. The lossy discrete transmission line.

LdIn,n+1dt=(1+RCCddt)(Vn−Vn+1),(45)L\frac{dI_{n,n+1}}{dt}=\left(1+R_CC\frac{d}{dt}\right)(V_n-V_{n+1}), \tag*{(45)}
Inc=(1+LRLddt)(In−1,n−In,n+1)(46)I_n^c=\left(1+\frac{L}{R_L}\frac{d}{dt}\right)(I_{n-1,n}-I_{n,n+1}) \tag*{(46)}

((1) remains as it was.) Hence (3) is modified to

Ld2Qndt2=(1+LRLddt)(1+RCCddt)(Vn+1−2Vn+Vn−1),(47)L\frac{d^2Q_n}{dt^2}=\left(1+\frac{L}{R_L}\frac{d}{dt}\right)\left(1+R_CC\frac{d}{dt}\right)(V_{n+1}-2V_n+V_{n-1}), \tag*{(47)}

and (4) is modified due to the presence of ohmic resistors to

L∂2Q(V)∂t2=∂2V∂Z2+τR∂3V∂t∂Z2+112∂4V∂Z4,(48)L\frac{\partial^2Q(V)}{\partial t^2}=\frac{\partial^2V}{\partial Z^2}+\tau_R\frac{\partial^3V}{\partial t\partial Z^2}+\frac{1}{12}\frac{\partial^4V}{\partial Z^4}, \tag*{(48)}

where

τR≡LRL+RCC.(49)\tau_R\equiv\frac{L}{R_L}+R_CC. \tag*{(49)}

In deriving (41) we assumed that the dissipation is small, so we ignored the terms which correspond to the product of the dissipation coefficients and the higher order derivatives with respect to ZZ. For the travelling wave we obtain from (41) after integration the generalization of (8)

112d3VdZ3+UτRd2VdZ2=U2LdQdZ−dVdZ.(50)\frac{1}{12}\frac{d^3V}{dZ^3}+U\tau_R\frac{d^2V}{dZ^2}=U^2L\frac{dQ}{dZ}-\frac{dV}{dZ}. \tag*{(50)}

Similar to what we said in Section II A, notice that equation with the same form as (43) one would obtain after making the travelling wave ansatz in the generalised Korteweg-de Vries-modified Korteweg-de Vries (KDV-mKDV) equation with dissipation.

Performing the second integration, we obtain the generalization of (9)

112d2VdZ2+UτRdVdZ=U2LQ(V)−V+F.(51)\frac{1}{12}\frac{d^2V}{dZ^2}+U\tau_R\frac{dV}{dZ}=U^2LQ(V)-V+F. \tag*{(51)}

From (39) we obtain that in the present case the sum of the currents flowing through the nonlinear inductance and the ohmic resistance RLR_L is given by the equation

I(Z)=−V(Z)UL−τRLdV(Z)dZ+const.(52)I(Z)=-\frac{V(Z)}{UL}-\frac{\tau_R}{L}\frac{dV(Z)}{dZ}+\mathrm{const.} \tag*{(52)}

(51) is Newton equation, describing motion with friction of the fictitious particle (VV being the coordinate of the particle and ZZ - the ”time”) in the force field given by the r.h.s. of the equation. We again impose on V(Z)V(Z) the boundary conditions (17). Hence the motion starts at Z=−∞Z=-\infty at V=V1V=V_1 and ends at Z=+∞Z=+\infty at V=V2V=V_2 (obviously that both V1V_1 and V2V_2 are the equilibrium positions). Equating the r.h.s. of (44) to zero at V=V1V=V_1 and V=V2V=V_2 we recover (23) for the speed of the wave.

It will be convenient to present (51) as

112d2VdZ2+UτRdVdZ=−dΠ(V)dV,(53)\frac{1}{12}\frac{d^2V}{dZ^2}+U\tau_R\frac{dV}{dZ}=-\frac{d\Pi(V)}{dV}, \tag*{(53)}

where

dΠ(V)dV=V2−V1Q2−Q1Q(V)+V+V2Q1−V1Q2Q2−Q1.(54)\frac{d\Pi(V)}{dV}=\frac{V_2-V_1}{Q_2-Q_1}Q(V)+V+\frac{V_2Q_1-V_1Q_2}{Q_2-Q_1}. \tag*{(54)}

The constraint for the travelling wave asymptotic boundary conditions is the inequality

Π(V1)>Π(V2).(55)\Pi(V_1) > \Pi(V_2). \tag*{(55)}

Inequality (55) can be explicitly presented as

∫V2V1Q(V) dV<12(Q1+Q2)(V1−V2).(56)\int_{V_2}^{V_1} Q(V)\,\mathrm{d}V < \frac{1}{2}(Q_1+Q_2)(V_1-V_2). \tag*{(56)}

For γ>0\gamma> 0, β≥0\beta\ge0, the condition (56) means just ∣V1∣>∣V2∣|V_1| > |V_2|.

The point V=V1V=V_1 should be the position of unstable equilibrium. The point V=V2V=V_2 can be the point of either stable, or unstable equilibrium. (We met with similar dichotomy while studying localised travelling waves in Josephson transmission line[16]). The waves corresponding to the stable equilibrium at the point V=V2V=V_2 we called the shocks, to the unstable equilibrium – the kinks.

Differentiating (47) we obtain the r.h.s. of (52) (see below)

−d2Π(V)dV2=U2u2(V)−1;(57)-\frac{\mathrm{d}^2\Pi(V)}{\mathrm{d}V^2}=\frac{U^2}{u^2(V)}-1; \tag*{(57)}

the point V=ViV=V_i being the point of the stable/unstable equilibrium corresponds to the travelling wave being subsonic/supersonic with respect to the background. Stable equilibrium at the point V=V2V=V_2 (together with unstable equilibrium at the point V=V1V=V_1) reflect the well-known in the nonlinear waves theory fact: the shock speed is larger than the sound speed in the region before the shock but smaller than the sound speed in the region behind the shock[17].

Let us return to (24) giving the sound and the kink/shock speeds for the cubic case. The kinks/shocks phase diagram in the space V1,V2V_1,V_2 can be obtained by comparing U2U^2 and u2(V2)u^2(V_2). Consider the case γ>0,β≥0\gamma> 0,\beta\ge0. If V1V_1 and V2V_2 have the same sign, the travelling wave is always the shock. In the opposite case, the wave can be either the kink or the shock, depending upon V1V_1 and V2V_2. The boundary in the phase space between the kinks and the shocks is especially simple for β=0\beta=0

V1+2V2=0(58)V_1+2V_2=0 \tag*{(58)}

(very much similar to what we had for the Josephson transmission line[16]).

The most important distinction of the dissipative kinks from the shocks: since the particle stops at the unstable equilibrium point, for the kink to exist, the fine tuning is necessary - the parameters V1V_1, V2V_2 and RR should satisfy definite relation (see below). The shocks don’t need such find tuning – they are generic.

To study quantitatively dissipative kinks and shocks we will use the approach of the previous Section. After repeating exactly what was done there starting from (50), we obtain modified (16)

124d2E2dV2+UτRdEdV=U2u2(V)−1.(59)\frac{1}{24}\frac{\mathrm{d}^2E^2}{\mathrm{d}V^2}+U\tau_R\frac{\mathrm{d}E}{\mathrm{d}V}=\frac{U^2}{u^2(V)}-1. \tag*{(59)}

Considering again the cubic case we can easily check up that (25) is still the solution of (59), with ψ\psi being still given by (29), but (30) is generalized to

V1+V2=−2β3γ+4UτRψ.(60)V_1+V_2=-\frac{2\beta}{3\gamma}+\frac{4U\tau_R}{\psi}. \tag*{(60)}

Hence the dissipative kinks and the shocks are of the same shape (given by (27)) as the kinks in the lossless line.

Now we understand what all the trouble of introducing the new dependent and independent variables was for. We were able to integrate analytically (59) (while we didn’t see any way to integrate analytically (51)).

We need to emphasise the different meaning of the condition (53) for the kinks and the shocks. For the kinks, (53) is the condition of the solution existence (and the solution happens to be described by the elementary function). For the shock, (53) is just the condition for the solution to be described by an elementary function. But, of course, the shock with any other values of V1,V2,τRV_1,V_2,\tau_R also exists (provided that V1,V2V_1,V_2 satisfy the constraint (48)). So the obtained shock can be called the particular shock of the first kind. Note also that by integrating (52) we can obtain Abel equation of the second kind, and (53) is the condition of the solvability of the equation[18].

Let us assume that V1V_1 and V2V_2 are so close to each other, that we can keep in the r.h.s. of (23) only the first two terms, thus simplifying (52) to

124d2E2dV2+UτRdEdV=U2LC0(α+2βV)−1(61)\frac{1}{24}\frac{\mathrm{d}^2E^2}{\mathrm{d}V^2}+U\tau_R\frac{\mathrm{d}E}{\mathrm{d}V}=U^2LC_0(\alpha+2\beta V)-1 \tag*{(61)}

(we assume for the sake of definiteness that β>0\beta>0 and V1,V2>0V_1,V_2>0). The elementary solution of (54) satisfying the boundary conditions (19) is of the form[16]

E=χ(V−V2)[(V−V2)1/2−(V1−V2)1/2].(62)E=\chi(V-V_2)\left[(V-V_2)^{1/2}-(V_1-V_2)^{1/2}\right]. \tag*{(62)}

Substituting (55) into (54) we obtain

χ2=8U2LC0β(63)\chi^2=8U^2LC_0\beta \tag*{(63)}
UτR=5χ24(V1−V2)1/2.(64)U\tau_R=\frac{5\chi}{24}(V_1-V_2)^{1/2}. \tag*{(64)}

Similar to (60) in the case of shocks, (64) is just the condition for the shock to be expressed by an elementary function. The shock exists for any other value of V1−V2V_1-V_2. So the obtained shock can be called the particular shock of the second kind.

The function E(V)E(V) being found, we can find the function V(Z)V(Z) by solving (12). The solution is

V(Z)=V2+V1−V2[exp⁡(χZ/2)+1]2.(65)V(Z)=V_2+\frac{V_1-V_2}{[\exp(\chi Z/2)+1]^2}. \tag*{(65)}

Conclusions

We considered localised travelling waves of the voltage and the current in the discrete lossless transmission line with nonlinear capacitors. We have shown that these waves are of two kinds: the kinks, when the voltage has different asymptotic values at both sides of the transmission line, and the solitons (solitary waves), when the voltage has the same asymptotic value at both sides of the transmission line. We also studied lossy transmission lines. The localised travelling waves in such systems are the dissipative kinks and the shocks.

The velocity of the kink, both in the lossless and in the lossy line, and of the shock in the lossy line is given by the same equations

1U2=LQ2−Q1V2−V1,(66)\frac{1}{U^2}=L\frac{Q_2-Q_1}{V_2-V_1}, \tag*{(66)}

where Q1Q_1, V1V_1 and Q2Q_2, V2V_2 are the asymptotic values of the capacitor charge and the voltage on both ends of the transmission line.

To our opinion the most important equation of the paper is Eq. (59) describing the lossy transmission line

124d2E2dV2+UτRdEdV=U2u2(V)−1.(67)\frac{1}{24}\frac{d^2E^2}{dV^2}+U\tau_R\frac{dE}{dV}=\frac{U^2}{u^2(V)}-1. \tag*{(67)}

When the charge of the non-linear capacitor and the voltage on it are connected by the equation

Q(V)=const+C0(αV+βV2+γV3),(68)Q(V)=\mathrm{const}+C_0\left(\alpha V+\beta V^2+\gamma V^3\right), \tag*{(68)}

the profiles of the kink, both in the lossless and in the lossy line, and of the particular shock of the first kind in the lossy line, can be easily calculated on the basis of the solution of Eq. (67)

E(V)=ψ(V−V1)(V−V2),(69)E(V)=\psi(V-V_1)(V-V_2), \tag*{(69)}

and are given by the same equation

V(Z)=V1+V22−V1−V22tanh⁡[V1−V22ψZ],(70)V(Z)=\frac{V_1+V_2}{2}-\frac{V_1-V_2}{2}\tanh\left[\frac{V_1-V_2}{2}\psi Z\right], \tag*{(70)}

where ψ2=6U2LC0γ\psi^2=6U^2LC_0\gamma, and the condition on the asymptotic values of the voltage on both ends of the transmission line V1V_1 and V2V_2 are given by Eq. (60). The profile of the particular shock of the second kind is given by Eq. (65).

Acknowledgments

I am grateful to A. M. Kamchatnov, B. Malomed and A. Vainchtein for the discussion. The paper was finalized during the Author’ participation in Workshop on Twistronics and Moire Materials: Bridging Theory and Experiments, ICTP, Trieste. The Author is grateful to the Center for the hospitality.1

References

References

  1. [1]E. Kengne, W. M. Liu, L. Q. English, and B. A. Malomed, Phys. Rep. 982, 1 (2022).
  2. [2]M. Remoissenet, Waves Called Solitons: Concepts and Experiments, Springer-Verlag Berlin Heidelberg GmbH (1996).DOI
  3. [3]E. Kengne, R. Vaillancour, And B. A. Malomed, Int. J. Mod. Phys. B 23, 133 (2009).
  4. [4]S. Abdoulkary, A. Mohamadou, and T. Beda, Comm. Nonl. Sci Numer Sim. 16, 3525 (2011).
  5. [5]M.M. Bogdan and D. V. Laptev, J. Phys. Soc. Japan 83, 064007 (2014).
  6. [6]The Fermi-Pasta-Ulam Problem: A Status Report (Ed: G. Gallavotti), Lecture Notes in Physics, Vol. 728, Springer, New York, NY 2007.
  7. [7]P. G. Kevrekidis and M. I. Weinstein, Physica D: Nonlinear Phenomena 142, 113 (2000).
  8. [8]A. Vainchtein, Physica D: Nonlinear Phenomena, 434, 133252 (2022).
  9. [9]A. Vainchtein and Lev Truskinovsky, arXive:5376227; Physica D: Nonlinear Phenomena 134187 (2024).
  10. [10]L. F. Calazans de Brito and A. M. Kamchatnov, Phys. Rev. E 109, 015102 (2024).
  11. [11]E. Kogan, Phys. Stat. Sol. (b) 259, 2200160 (2022). https://doi: 10.1002/pssb.20220160
  12. [12]E. Kogan, Phys. Stat. Sol. (b) 260, 2200475 (2023). https://doi: 10.1002/pssb.202200475.DOI
  13. [13]R. Landauer, IBM J. Res. Dev. 4, 391 (1960).
  14. [14]S. T. Peng, R. Landauer, IBM J. Res. Dev. 17, 299 (1973).
  15. [15]R. Landauer, J. Appl. Phys. 51 5594 (1980).
  16. [16]E. Kogan, Phys. Stat. Sol. (b), 2300336 (2024); https://doi.org/10.1002/pssb.202300336.
  17. [17]G. B. Whitham, Linear and Nonlinear Waves, John Wiley & Sons Inc., New York (1999).
  18. [18]A. D. Polyanin and V. F. Zaitsev, Handbook of exact solutions for ordinary differential equations, Second edition, (Chapman & Hall/CRC, Boca Raton London New York Washington, D.C. 2003).

Paper details

vlnc:W_59rquuuz
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents