This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Modeling of three-dimensional betatron oscillation and radiation reaction in plasma accelerators

Yulong Liu    Ming Zeng [email protected] Institute of High Energy Physics, Chinese Academy of Sciences, Beijing 100049, China University of Chinese Academy of Sciences, Beijing 100049, China
Abstract

Betatron oscillation is a commonly known phenomenon in laser or beam driven plasma wakefield accelerators. In the conventional model, the plasma wake provides a linear focusing force to a relativistic electron, and the electron oscillates in one transverse direction with the betatron frequency proportional to 1/γ1/\sqrt{\gamma}, where γ\gamma is the Lorentz factor of the electron. In this work, we extend this model to three-dimensional by considering the oscillation in two transverse and one longitudinal directions. The long-term equations, with motion in the betatron time scale averaged out, are obtained and compared with the original equations by numerical methods. In addition to the longitudinal and transverse damping due to radiation reaction which has been found before, we show phenomena including the longitudinal phase drift, betatron phase shift and betatron polarization change based on our long-term equations. This work can be highly valuable for future plasma based high-energy accelerators and colliders.

I Introduction

The new generation of accelerators, using plasma as the acceleration media, offer high acceleration gradient in the order of 10–100 GV/m\rm GV/m and strong transverse focusing field [1, 2, 3]. Depends on the driver type, the plasma accelerators are named laser wakefield accelerators (LWFAs), which are driven by laser pulses, and plasma wakefield accelerators (PWFAs), which are driven by charged particle beams. When a high intensity laser pulse (1018W/cm2\gtrsim 10^{18}\ \rm W/cm^{2}) or a high current particle beam (1kA\gtrsim 1\ \rm kA) propagates through an underdense plasma, the radiation pressure of the laser or the space charge of the beam expels all plasma electrons away from axis radially, leaves behind a nearly uniform ion channel. This high-intensity three-dimensional (3D) regime has been referred to as the blowout regime [4, 5]. In this regime, the expelled electrons are pulled back by the ion channel and thereby bubble-like plasma wake wave is created. The wake consists of a longitudinal electric field that is a function of distance behind the driver, and transversal electromagnetic fields that are proportional to the off-axis distance. Consequently, in addition to the longitudinal acceleration / deceleration, the electrons reside in the wake also perform radial oscillation, called betatron oscillation (BO), under the action of transverse focusing field, with the frequency ωβ=ωpκ/γ\omega_{\beta}=\omega_{p}\kappa/\sqrt{\gamma}, where ωp\omega_{p} is the plasma frequency, γ\gamma is the relativistic factor of the electron, and κ\kappa is the focusing constant which takes 1/21/\sqrt{2} for the blow-out regime [6, 7].

Electrons emit synchrotron radiation when performing BO [8, 9, 10], which affects the electrons in return. Such effect is called the radiation reaction (RR) and its classical expression is the Lorentz–Abraham–Dirac (LAD) equation or the Landau-Lifshitz equation [11, 12]. Because the RR force is proportional to the classical electron radius re2.81×1015mr_{e}\approx 2.81\times 10^{-15}\ \rm m, it is generally negligible unless under extreme conditions [13, 14] or for sufficiently long interaction time [15]. The BO in a plasma accelerator is another good case for such long interaction time. The radiation leads to the energy loss of electrons and in return affects the energy-dependent betatron frequency, as well as the other beam properties, such as the energy spread and emittance [16, 17, 18, 19, 20, 21, 22, 23].

Although there are many established theories on the long-term RR damping effect of BO, their models assert the electron only moves in one plane, spanned by the longitudinal direction and one transverse direction, thus only linear polarization is considered. Moreover, these models usually neglect the longitudinal and energy oscillations during one betatron period. In this paper, we establish a 3D BO model with RR effect, which generalizes the betatron polarization from linear to elliptical, and considers both the longitudinal and energy oscillations. Long-term equations (LTEs), without resolving the betatron period, are derived and verified by numerical methods. The LTEs reproduce the previous results, such as longitudinal and transverse damping due to RR, as the special cases, and meanwhile reveal new phenomena such as betatron phase shift and polarization change.

The rest of this paper is organized as follows. Sec. II gives the original form of the force, and shows the equations of motion expressed by the transverse motion only. Sec. III derives the LTEs by averaging the equations of motion through one betatron period. Sec. IV discusses the different phenomena in the RR dominant regime and the betatron phase shift dominant regime. Sec. V numerically verifies the LTEs by comparing with the code PTracker which solves the equations with the original form of force. Before start, it is worth noting that we use plasma normalization units described in Appx. A, and some symbols and calculation rules used often during the derivation are described in Appx. B.

II The electromagnetic field and the equations of motion

Consider an electron with γ1\gamma\gg 1 is trapped in a plasma wakefield with the longitudinal co-moving coordinate ζ=zβwt\zeta=z-\beta_{w}t, where the wake is moving in the z+z+ direction with the phase velocity βw\beta_{w}. Neglect the interaction between the beam particles, the electromagnetic field provided by the wake can be modeled as [7]

Ez\displaystyle E_{z} =\displaystyle= Ez0+λζ1,\displaystyle E_{z0}+\lambda\zeta_{1}, (1)
E\displaystyle\vec{E}_{\perp} =\displaystyle= κ2(1λ)r,\displaystyle\kappa^{2}\left(1-\lambda\right)\vec{r}, (2)
Bθ\displaystyle B_{\theta} =\displaystyle= κ2λr,\displaystyle-\kappa^{2}\lambda r, (3)

where Ez0=Ez|ζ=ζE_{z0}=\left.E_{z}\right|_{\zeta=\left<\zeta\right>}, λ=dEz/dζ|ζ=ζ\lambda=\left.dE_{z}/d\zeta\right|_{\zeta=\left<\zeta\right>}, and r=(x,y)\vec{r}=\left(x,y\right) is the transverse offset. The force can be expressed as

fz\displaystyle f_{z} =\displaystyle= Ez0λζ1+κ2λ(xβx+yβy)+fzrad,\displaystyle-E_{z0}-\lambda\zeta_{1}+\kappa^{2}\lambda(x\beta_{x}+y\beta_{y})+f_{z}^{\rm rad}, (4)
fx\displaystyle f_{x} =\displaystyle= κ2(1λ+λβz)x+fxrad,\displaystyle-\kappa^{2}\left(1-\lambda+\lambda\beta_{z}\right)x+f_{x}^{\rm rad}, (5)
fy\displaystyle f_{y} =\displaystyle= κ2(1λ+λβz)y+fyrad,\displaystyle-\kappa^{2}\left(1-\lambda+\lambda\beta_{z}\right)y+f_{y}^{\rm rad}, (6)

where βx=x˙\beta_{x}=\dot{x}, βy=y˙\beta_{y}=\dot{y}, βz=z˙=βz0+ζ1˙\beta_{z}=\dot{z}=\beta_{z0}+\dot{\zeta_{1}}, βz0=βw+ζ˙=βz\beta_{z0}=\beta_{w}+\dot{\left<\zeta\right>}=\left<\beta_{z}\right>, and frad\vec{f}^{\rm rad} is the RR force. The formulas of 3D BO with RR can be written in the form of transverse terms only (see Appx. C)

γ˙\displaystyle\dot{\gamma} =\displaystyle= Ez0βz0+(λβz04+κ2λκ2)(xβx+yβy)23reγ2κ4(x2+y2),\displaystyle-E_{z0}\beta_{z0}+\left(\frac{\lambda\beta_{z0}}{4}+\kappa^{2}\lambda-\kappa^{2}\right)\left(x\beta_{x}+y\beta_{y}\right)-\frac{2}{3}r_{e}\gamma^{2}\kappa^{4}\left(x^{2}+y^{2}\right), (7)
pz˙\displaystyle\dot{p_{z}} =\displaystyle= Ez0+λ(14+κ2)(xβx+yβy)23reγ2κ4(x2+y2),\displaystyle-E_{z0}+\lambda\left(\frac{1}{4}+\kappa^{2}\right)\left(x\beta_{x}+y\beta_{y}\right)-\frac{2}{3}r_{e}\gamma^{2}\kappa^{4}\left(x^{2}+y^{2}\right), (8)
px˙\displaystyle\dot{p_{x}} =\displaystyle= κ2x+κ2λ2(γ2+βx2+βy2)x23reγ2κ4(x2+y2)βx,\displaystyle-\kappa^{2}x+\frac{\kappa^{2}\lambda}{2}\left(\left<\gamma\right>^{-2}+\beta_{x}^{2}+\beta_{y}^{2}\right)x-\frac{2}{3}r_{e}\gamma^{2}\kappa^{4}\left(x^{2}+y^{2}\right)\beta_{x}, (9)
py˙\displaystyle\dot{p_{y}} =\displaystyle= κ2y+κ2λ2(γ2+βx2+βy2)y23reγ2κ4(x2+y2)βy,\displaystyle-\kappa^{2}y+\frac{\kappa^{2}\lambda}{2}\left(\left<\gamma\right>^{-2}+\beta_{x}^{2}+\beta_{y}^{2}\right)y-\frac{2}{3}r_{e}\gamma^{2}\kappa^{4}\left(x^{2}+y^{2}\right)\beta_{y}, (10)

where

p\displaystyle\vec{p} =\displaystyle= γβ,\displaystyle\gamma\vec{\beta}, (11)
βz0\displaystyle\beta_{z0} =\displaystyle= 112(γ2+βx2+βy2),\displaystyle 1-\frac{1}{2}\left(\left<\gamma\right>^{-2}+\left<\beta_{x}^{2}\right>+\left<\beta_{y}^{2}\right>\right), (12)

and rer_{e} is also normalized to kp1k_{p}^{-1}. One may note the second terms in Eqs. (7) - (10), which come from the oscillation of βz\beta_{z} and the modulation of γ\gamma due to transverse oscillation, were neglected in previous works. In the following sections we show these terms lead to new phenomena.

III The long-term equations of 3D betatron oscillation

In this section we use the same averaging method as Ref. [20]. We firstly introduce two complex variables

U\displaystyle U =\displaystyle= (xiκ1γ12βx)eiφ,\displaystyle\left(x-i\kappa^{-1}\gamma^{\frac{1}{2}}\beta_{x}\right)e^{-i\varphi}, (13)
V\displaystyle V =\displaystyle= (yiκ1γ12βy)eiφ,\displaystyle\left(y-i\kappa^{-1}\gamma^{\frac{1}{2}}\beta_{y}\right)e^{-i\varphi}, (14)

where

φ=ωβ𝑑t=κγ12𝑑t\varphi=\int\omega_{\beta}dt=\kappa\int\gamma^{-\frac{1}{2}}dt (15)

is the betatron phase. Obviously |U1||U|\left|U_{1}\right|\ll\left|\left<U\right>\right| and |V1||V|\left|V_{1}\right|\ll\left|\left<V\right>\right| are satisfied, and we apply the rules in Appx. B often in the following. Because the equations for xx and yy directions are symmetric, we may derive for xx direction only, then exchange xx and yy, UU and VV for the yy direction. With the help of Eqs. (7), (9) and (11), we may write the time derivative of Eq. (13) as

U˙=\displaystyle\dot{U}= i12κ1γ12Ez0βz0βxeiφ+i13reκ3γ32(x2+y2)βxeiφ\displaystyle-i\frac{1}{2}\kappa^{-1}\gamma^{-\frac{1}{2}}E_{z0}\beta_{z0}\beta_{x}e^{-i\varphi}+i\frac{1}{3}r_{e}\kappa^{3}\gamma^{\frac{3}{2}}\left(x^{2}+y^{2}\right)\beta_{x}e^{-i\varphi} (16)
+i12κ1γ12[λβz04+κ2(λ1)](xβx+yβy)βxeiφi12κλγ12(γ2+βx2+βy2)xeiφ.\displaystyle+i\frac{1}{2}\kappa^{-1}\gamma^{-\frac{1}{2}}\left[\frac{\lambda\beta_{z0}}{4}+\kappa^{2}\left(\lambda-1\right)\right]\left(x\beta_{x}+y\beta_{y}\right)\beta_{x}e^{-i\varphi}-i\frac{1}{2}\kappa\lambda\gamma^{-\frac{1}{2}}\left(\left<\gamma\right>^{-2}+\beta_{x}^{2}+\beta_{y}^{2}\right)xe^{-i\varphi}.

In the following, we omit \left<\right> on UU and VV for convenience, so that all UU and VV actually mean U\left<U\right> and V\left<V\right>. We perform average on Eq. (16) to obtain (note only the terms with ei0φe^{i0\varphi} survive after averaging)

U˙=\displaystyle\dot{U}= 14Ez0βz0γ1U124reκ4γ(|U|2U+2|V|2UV2U)+i164κλβz0γ32(|U|2U+V2U)\displaystyle\frac{1}{4}E_{z0}\beta_{z0}\left<\gamma\right>^{-1}U-\frac{1}{24}r_{e}\kappa^{4}\left<\gamma\right>\left(\left|U\right|^{2}U+2\left|V\right|^{2}U-V^{2}U^{*}\right)+i\frac{1}{64}\kappa\lambda\beta_{z0}\left<\gamma\right>^{-\frac{3}{2}}\left(\left|U\right|^{2}U+V^{2}U^{*}\right) (17)
i116κ3γ32[(|U|2+2λ|V|2)U(2λ1)V2U]i14κλγ52U.\displaystyle-i\frac{1}{16}\kappa^{3}\left<\gamma\right>^{-\frac{3}{2}}\left[\left(\left|U\right|^{2}+2\lambda\left|V\right|^{2}\right)U-\left(2\lambda-1\right)V^{2}U^{*}\right]-i\frac{1}{4}\kappa\lambda\left<\gamma\right>^{-\frac{5}{2}}U.

By asserting V=0V=0 and omitting the last three terms in Eq. (17), which comes from the second terms in Eqs. (7) - (10), we can reproduce Eq. (19) in Ref. [20].

The average of Eq. (7) leads to

γ˙=Ez0βz013reκ4γ2(|U|2+|V|2),\left<\dot{\gamma}\right>=-E_{z0}\beta_{z0}-\frac{1}{3}r_{e}\kappa^{4}\left<\gamma\right>^{2}\left(\left|U\right|^{2}+\left|V\right|^{2}\right), (18)

with the second term reproduces Eq. (B2) in Ref. [18]. Ez0E_{z0} is a function of ζ\left<\zeta\right>, which obeys

ζ˙=12γw212γ214κ2γ1(|U|2+|V|2),\dot{\left<\zeta\right>}=\frac{1}{2}\gamma_{w}^{-2}-\frac{1}{2}\left<\gamma\right>^{-2}-\frac{1}{4}\kappa^{2}\left<\gamma\right>^{-1}\left(\left|U\right|^{2}+\left|V\right|^{2}\right), (19)

where γw=(1βw2)1/2\gamma_{w}=\left(1-\beta_{w}^{2}\right)^{-1/2} and we have used Eq. (46).

The above averaged equations Eqs. (17), (18) and (19) are already enough to predict the long-term behavior of BO. However, the equations for the complex variables are not explicit. To make them more physically meaningful, we introduce

U=|U|eiΦx,\displaystyle U=\left|U\right|e^{i\Phi_{x}}, (20)
V=|V|eiΦy,\displaystyle V=\left|V\right|e^{i\Phi_{y}}, (21)
ΔΦ=ΦyΦx.\displaystyle\Delta\Phi=\Phi_{y}-\Phi_{x}. (22)

|U|\left|U\right| has the meaning of the BO amplitude in the xx direction, and Φx\Phi_{x} the phase shift. For the yy direction they are similar. Thus ΔΦ\Delta\Phi is the phase difference of the two directions. By Applying d|U|/dt=(U˙U+UU˙)/2|U|d\left|U\right|/dt=\left(\dot{U}U^{*}+U\dot{U}^{*}\right)/2\left|U\right| and Φ˙x=(U˙UU˙U)/2i|U|2\dot{\Phi}_{x}=\left(\dot{U}U^{*}-\dot{U}^{*}U\right)/2i\left|U\right|^{2} we get

d|U|dt=\displaystyle\frac{d{\left|U\right|}}{dt}= 14Ez0βz0γ1|U|124reκ4γ[|U|3+|V|2|U|(2cos2ΔΦ)]\displaystyle\frac{1}{4}E_{z0}\beta_{z0}\left<\gamma\right>^{-1}\left|U\right|-\frac{1}{24}r_{e}\kappa^{4}\left<\gamma\right>\left[\left|U\right|^{3}+\left|V\right|^{2}\left|U\right|\left(2-\cos 2\Delta\Phi\right)\right] (23)
116κ[14λβz0κ2(12λ)]γ32|V|2|U|sin2ΔΦ,\displaystyle-\frac{1}{16}\kappa\left[\frac{1}{4}\lambda\beta_{z0}-\kappa^{2}\left(1-2\lambda\right)\right]\left<\gamma\right>^{-\frac{3}{2}}\left|V\right|^{2}\left|U\right|\sin 2\Delta\Phi,
Φ˙x=\displaystyle\dot{\Phi}_{x}= 124reκ4γ|V|2sin2ΔΦ+164κλβz0γ32[|U|2+|V|2cos2ΔΦ]\displaystyle\frac{1}{24}r_{e}\kappa^{4}\left<\gamma\right>\left|V\right|^{2}\sin 2\Delta\Phi+\frac{1}{64}\kappa\lambda\beta_{z0}\left<\gamma\right>^{-\frac{3}{2}}\left[\left|U\right|^{2}+\left|V\right|^{2}\cos 2\Delta\Phi\right] (24)
116κ3γ32[|U|2+2λ|V|2+(12λ)|V|2cos2ΔΦ]14κλγ52,\displaystyle-\frac{1}{16}\kappa^{3}\left<\gamma\right>^{-\frac{3}{2}}\left[\left|U\right|^{2}+2\lambda\left|V\right|^{2}+\left(1-2\lambda\right)\left|V\right|^{2}\cos 2\Delta\Phi\right]-\frac{1}{4}\kappa\lambda\left<\gamma\right>^{-\frac{5}{2}},
dΔΦdt=124reκ4γ(|V|2+|U|2)sin2ΔΦ+18κ[14λβz0κ2(12λ)]γ32(|V|2|U|2)sin2ΔΦ.\frac{d\Delta\Phi}{dt}=-\frac{1}{24}r_{e}\kappa^{4}\left<\gamma\right>\left(\left|V\right|^{2}+\left|U\right|^{2}\right)\sin 2\Delta\Phi+\frac{1}{8}\kappa\left[\frac{1}{4}\lambda\beta_{z0}-\kappa^{2}\left(1-2\lambda\right)\right]\left<\gamma\right>^{-\frac{3}{2}}\left(\left|V\right|^{2}-\left|U\right|^{2}\right)\sin^{2}\Delta\Phi. (25)

Note when doing exchange of UU and VV for the yy direction, one also has to change the ±\pm sign of ΔΦ\Delta\Phi.

To further simplify we notice Eq. (23) can be rewritten with the help of Eq. (18)

d|U|dt=14γ˙γ|U|18reκ4γ[|U|3+4cos2ΔΦ3|V|2|U|]116κ[14λβz0κ2(12λ)]γ32|V|2|U|sin2ΔΦ,\frac{d{\left|U\right|}}{dt}=-\frac{1}{4}\frac{\left<\dot{\gamma}\right>}{\left<\gamma\right>}\left|U\right|-\frac{1}{8}r_{e}\kappa^{4}\left<\gamma\right>\left[\left|U\right|^{3}+\frac{4-\cos 2\Delta\Phi}{3}\left|V\right|^{2}\left|U\right|\right]-\frac{1}{16}\kappa\left[\frac{1}{4}\lambda\beta_{z0}-\kappa^{2}\left(1-2\lambda\right)\right]\left<\gamma\right>^{-\frac{3}{2}}\left|V\right|^{2}\left|U\right|\sin 2\Delta\Phi, (26)

which reproduces Eq. (66) in Ref. [22] if V=0V=0. Introduce

Sx\displaystyle S_{x} =\displaystyle= κγ12|U|2,\displaystyle\kappa\left<\gamma\right>^{\frac{1}{2}}\left|U\right|^{2}, (27)
Sy\displaystyle S_{y} =\displaystyle= κγ12|V|2,\displaystyle\kappa\left<\gamma\right>^{\frac{1}{2}}\left|V\right|^{2}, (28)

which have the physical meaning of the areas (divided by 2π2\pi) of the ellipses encircled by the particle trajectory in xx-pxp_{x} and yy-pyp_{y} phase spaces. Then Eqs. (18), (19), (23), (24) and (25) can be rewritten as

γ˙=Ez0βz013reκ3γ32(Sx+Sy),\left<\dot{\gamma}\right>=-E_{z0}\beta_{z0}-\frac{1}{3}r_{e}\kappa^{3}\left<\gamma\right>^{\frac{3}{2}}\left(S_{x}+S_{y}\right), (29)
ζ˙=12γw212γ214κγ32(Sx+Sy),\dot{\left<\zeta\right>}=\frac{1}{2}\gamma_{w}^{-2}-\frac{1}{2}\left<\gamma\right>^{-2}-\frac{1}{4}\kappa\left<\gamma\right>^{-\frac{3}{2}}\left(S_{x}+S_{y}\right), (30)
S˙x=14reκ3γ12(Sx2+4cos2ΔΦ3SxSy)18[14λβz0κ2(12λ)]γ2SxSysin2ΔΦ,\dot{S}_{x}=-\frac{1}{4}r_{e}\kappa^{3}\left<\gamma\right>^{\frac{1}{2}}\left(S_{x}^{2}+\frac{4-\cos 2\Delta\Phi}{3}S_{x}S_{y}\right)-\frac{1}{8}\left[\frac{1}{4}\lambda\beta_{z0}-\kappa^{2}\left(1-2\lambda\right)\right]\left<\gamma\right>^{-2}S_{x}S_{y}\sin 2\Delta\Phi, (31)
Φ˙x=\displaystyle\dot{\Phi}_{x}= 124reκ3γ12Sysin2ΔΦ+164λβz0γ2(Sx+Sycos2ΔΦ)\displaystyle\frac{1}{24}r_{e}\kappa^{3}\left<\gamma\right>^{\frac{1}{2}}S_{y}\sin 2\Delta\Phi+\frac{1}{64}\lambda\beta_{z0}\left<\gamma\right>^{-2}\left(S_{x}+S_{y}\cos 2\Delta\Phi\right) (32)
116κ2γ2[Sx+2λSy+(12λ)Sycos2ΔΦ]14κλγ52,\displaystyle-\frac{1}{16}\kappa^{2}\left<\gamma\right>^{-2}\left[S_{x}+2\lambda S_{y}+\left(1-2\lambda\right)S_{y}\cos 2\Delta\Phi\right]-\frac{1}{4}\kappa\lambda\left<\gamma\right>^{-\frac{5}{2}},
dΔΦdt=124reκ3γ12(Sy+Sx)sin2ΔΦ+18[14λβz0κ2(12λ)]γ2(SySx)sin2ΔΦ.\frac{d\Delta\Phi}{dt}=-\frac{1}{24}r_{e}\kappa^{3}\left<\gamma\right>^{\frac{1}{2}}\left(S_{y}+S_{x}\right)\sin 2\Delta\Phi+\frac{1}{8}\left[\frac{1}{4}\lambda\beta_{z0}-\kappa^{2}\left(1-2\lambda\right)\right]\left<\gamma\right>^{-2}\left(S_{y}-S_{x}\right)\sin^{2}\Delta\Phi. (33)

It is generally safe to take βz0=1\beta_{z0}=1 here. But Eq. (12), or βz0=112[γ2+12κγ3/2(Sx+Sy)]\beta_{z0}=1-\frac{1}{2}\left[\left<\gamma\right>^{-2}+\frac{1}{2}\kappa\left<\gamma\right>^{-3/2}\left(S_{x}+S_{y}\right)\right], gives a better accuracy. The above long-term equations, Eqs. (29) - (33), show that the BO experiences acceleration (for Ez0<0E_{z0}<0) or deceleration (for Ez0>0E_{z0}>0), radiation damping, longitudinal phase drift, and betatron phase shift. These equations may be used for the long-term behavior of BO without resolving the betatron period.

IV Discussion on two regimes

From Eqs. (31) - (33) we note two regimes. One is the RR dominant regime, where reγ5/21r_{e}\left<\gamma\right>^{5/2}\gg 1, so that the first terms in Eqs. (31) - (33) dominate. This regime has been discussed before [20], although only for the linearly polarized case ΔΦ=0\Delta\Phi=0 (so that the ratio between SxS_{x} and SyS_{y} is a constant). The other is the betatron phase shift dominant regime, where reγ5/21r_{e}\left<\gamma\right>^{5/2}\ll 1, so that the remaining terms in Eqs. (31) - (33) dominate. These terms were previously proposed [22], but the betatron phase shift is found for the first time in the present work.

In the RR dominant regime, an interesting phenomenon is that an elliptical polarization (in the xx-yy plane) always approaches linear polarization, because ΔΦ\Delta\Phi always approaches the nearest integer multiple of π\pi according to Eq. (33). This phenomenon can also be viewed by rotating the xx axis to the major axis of the ellipse, so that Sx>SyS_{x}>S_{y} and ΔΦ=π/2\Delta\Phi=\pi/2. Define R=Sy/SxR=S_{y}/S_{x} and perform time derivative with the help of Eq. (31)

R˙=16reκ3γ12R(SxSy)<0,\dot{R}=-\frac{1}{6}r_{e}\kappa^{3}\left<\gamma\right>^{\frac{1}{2}}R\left(S_{x}-S_{y}\right)<0, (34)

which suggests that the ellipse monotonically becomes thinner.

The betatron phase shift dominant regime requires a moderate γ\gamma, or straightforwardly re0r_{e}\rightarrow 0, which corresponds to very dilute plasma case, leads to a constant SSx+SyS\equiv S_{x}+S_{y}. It can be proved that the time integral of Eq. (30) reproduces Eq. (6) in Ref. [21], which is the ζ\left<\zeta\right> drift, in the case that γ\left<\gamma\right> linearly depends on tt. In another case that ζ\left<\zeta\right> drifts around the zero point of Ez0E_{z0}, the drift frequency can be obtained by using Eqs. (29) and (30), and asserting Ez0=λζE_{z0}=\lambda\left<\zeta\right>

ωζ=λβz0(1+38κγ12S)γ3.\omega_{\left<\zeta\right>}=\sqrt{\lambda\beta_{z0}\left(1+\frac{3}{8}\kappa\left<\gamma\right>^{\frac{1}{2}}S\right)\left<\gamma\right>^{-3}}. (35)

We also note the angular momentum Lz=γxβyγyβxL_{z}=\gamma x\beta_{y}-\gamma y\beta_{x} and its changing rate

Lz\displaystyle\left<L_{z}\right> =\displaystyle= Sx12Sy12sinΔΦ,\displaystyle-S_{x}^{\frac{1}{2}}S_{y}^{\frac{1}{2}}\sin\Delta\Phi, (36)
Lz˙\displaystyle\dot{\left<L_{z}\right>} =\displaystyle= 13reκ3γ12(Sx+Sy)Lz,\displaystyle-\frac{1}{3}r_{e}\kappa^{3}\left<\gamma\right>^{\frac{1}{2}}\left(S_{x}+S_{y}\right)\left<L_{z}\right>, (37)

which obeys the law of conservation of angular momentum if Lz=0\left<L_{z}\right>=0 initially, or re0r_{e}\rightarrow 0. Especially for re0r_{e}\rightarrow 0, the particle trajectory in the xx-yy plane generally encircles an ellipse with constant area and shape, which also has precession leading to the rotation of the major and minor axes of the ellipse.

V Numerical comparison of long-term equations and the original ones

To verify the LTEs, we solve them numerically using the backward-differentiation formulas (BDF) in the SciPy integration package [24]. Meanwhile, the original equations of motion with the force expressions Eqs. (4) - (6) are solved by Runge-Kutta 4th order method using the code PTracker (PT) [25]. We choose four cases with their parameters and initial values listed in Tab. 1, and the comparison results are plotted from Fig. 1 to 4. Note that Φx\Phi_{x} cannot be obtained directly from PT. Thus we perform the following treatment to the PT results

xcosφ=|U|2[cosΦx+cos(2φ+Φx)],x\cdot\cos\varphi=\frac{\left|U\right|}{2}\left[\cos\Phi_{x}+\cos\left(2\varphi+\Phi_{x}\right)\right], (38)

because x=|U|cos(φ+Φx)x=\left|U\right|\cos\left(\varphi+\Phi_{x}\right), where φ\varphi is obtained by numerical integral based on Eq. (15). Then cosΦx\cos\Phi_{x} can be obtained by a low-pass filter. Similar treatment is performed to obtain cosΔΦ\cos\Delta\Phi, according to

xy=|U||V|2[cosΔΦ+cos(2φ+Φx+Φy)].x\cdot y=\frac{\left|U\right|\left|V\right|}{2}\left[\cos\Delta\Phi+\cos\left(2\varphi+\Phi_{x}+\Phi_{y}\right)\right]. (39)

A case in the betatron phase shift dominant regime is shown in Fig. 1. We see Sx+SyS_{x}+S_{y} is a constant, although SxS_{x}, SyS_{y} and ΔΦ\Delta\Phi change gradually. The approximate “phase-locking” is chosen, i.e. γwγz0\gamma_{w}\approx\gamma_{z0}, thus ζ\left<\zeta\right> oscillates near the zero point of EzE_{z} with the drift frequency ωζ\omega_{\left<\zeta\right>} according to Eq. (35). γ\left<\gamma\right> oscillates with the same drift frequency as shown in Fig. 1 (b).

A second case during the transition of the two regimes is shown in Fig. 2. Sx+SyS_{x}+S_{y} is approximately a constant initially, and starts to decrease near the regime transition γ=re2/5\gamma=r_{e}^{-2/5}.

A third case in the RR dominant regime is shown in Fig. 3. The initial values are chosen so that the particle trajectory in the xx-yy plane is a ellipse with its major axis laying on the xx axis. As shown in Fig. 3 (a), R=Sy/SxR=S_{y}/S_{x} decreases monotonically, as predicted by Eq. (34).

The last case shown in Fig. 4 is also in the RR dominant regime, but the particle trajectory in the xx-yy plane is a oblique ellipse. As shown in Fig. 4 (c), ΔΦ\Delta\Phi gradually approaches π\pi, which is in accordance with the discussion in Sec. IV.

In all these plots, the results from PT and LTEs show agreement with high accuracy, demonstrating the correctness of LTEs. Because the BO frequency is the highest frequency in our physical process, the LTEs largely reduce the numerical complexity and meanwhile keep the long-term accuracy.

\begin{overpic}[width=208.13574pt]{S_gamma_Phi_DPhi_change_zeta0.pdf} \put(16.0,54.0){(a)} \put(65.0,54.0){(b)} \put(15.0,12.0){(c)} \put(65.0,35.0){(d)} \end{overpic}
Figure 1: The numerical comparison of the LTEs and the original equations solved by PTracker in the betatron phase shift dominant regime. (a) SxS_{x} and SyS_{y} change with time, but Sx+SyS_{x}+S_{y} is a constant. (b) γ\gamma has oscillation frequencies of 2ωβ0.142\omega_{\beta}\approx 0.14 due to the BO and ωζ3.11×103\omega_{\left<\zeta\right>}\approx 3.11\times 10^{-3} due to the drift oscillation of ζ\left<\zeta\right>. (c) The gray curve shows xyx\cdot y obtained from PT, and the black curve shows its low-pass filtered result, which is compared with the LTE solution according to Eq. (39). (d) The gray curve shows xcosφx\cdot\cos\varphi obtained from PT, and the black curve shows its low-pass filtered result, which is compared with the LTE solution according to Eq. (38).
\begin{overpic}[width=208.13574pt]{S_gamma_Phi_DPhi_change_fz0_plusRRforce.pdf} \put(12.0,54.0){(a)} \put(64.0,74.0){(b)} \put(10.0,35.0){(c)} \put(63.0,35.0){(d)} \end{overpic}
Figure 2: The numerical comparison of the LTEs and the original equations solved by PTracker in the transition between the betatron phase shift dominant and the RR dominant regimes. (a) Sx+SyS_{x}+S_{y} is initially approximately a constant, but decreases later. (b) γ\gamma increases due to the acceleration field, and passes the regime transition at γ=re2/5=104\gamma=r_{e}^{-2/5}=10^{4}. (c) and (d) show the same treatments as in Fig. 1 (c) and (d).
\begin{overpic}[width=208.13574pt]{S_gamma_Phi_DPhi_change_R_plusRRforce.pdf} \put(18.0,53.0){(a)} \put(65.0,74.0){(b)} \put(15.0,34.0){(c)} \put(65.0,34.0){(d)} \end{overpic}
Figure 3: The numerical comparison of the LTEs and the original equations solved by PTracker in the RR dominant regime. ΔΦ=π/2\Delta\Phi=\pi/2 and Sx>SyS_{x}>S_{y}, thus the major axis of the particle trajectory ellipse lays on the xx axis. (a) R=Sy/SxR=S_{y}/S_{x} decreases with time due to Eq. (34), thus the ellipse is getting thinner. (b) γ\gamma increases due to the acceleration field. (c) and (d) show the same treatments as in Fig. 1 (c) and (d).
\begin{overpic}[width=208.13574pt]{S_gamma_Phi_DPhi_change_DeltPhi_plusRR.pdf} \put(15.0,73.0){(a)} \put(65.0,73.0){(b)} \put(15.0,12.0){(c)} \put(65.0,12.0){(d)} \end{overpic}
Figure 4: The numerical comparison of the LTEs and the original equations solved by PTracker in the RR dominant regime. Sx=SyS_{x}=S_{y}, thus the particle trajectory in the xx-yy plane is an oblique ellipse. (a) SxS_{x} and SyS_{y} decrease with the same rate. (b) γ\gamma initially decreases with time because the longitudinal RR damping is stronger than the acceleration. Later the RR damping becomes weaker due to the decrease of SxS_{x} and SyS_{y}, and γ\gamma increases. (c) and (d) show the same treatments as in Fig. 1 (c) and (d), but the oscillation amplitudes are divided so that the changes of ΔΦ\Delta\Phi and Φx\Phi_{x} are clearer. ΔΦ\Delta\Phi is between π/2\pi/2 and π\pi, thus ΔΦ\Delta\Phi gradually approaches π\pi.
Table 1: The cases for comparing PT and LTEs
Case Parameters Initial Values
EzE_{z} λ\lambda κ\kappa rer_{e} γw\gamma_{w} |U|\left|U\right| |V|\left|V\right| Φx\Phi_{x} Φy\Phi_{y} γ\left<\gamma\right> ζ\left<\zeta\right>
Fig. 1 λζ\lambda\zeta 14\dfrac{1}{4} 12\dfrac{1}{\sqrt{2}} 0 14 1.12 0.87 0 π6\dfrac{\pi}{6} 10210^{2} -0.05
Fig. 2 -0.001 0 12\dfrac{1}{\sqrt{2}} 101010^{-10} 10410^{4} 1.12 0.87 0 π6\dfrac{\pi}{6} 10310^{3} 0
Fig. 3 λζ\lambda\zeta 12\dfrac{1}{2} 12\dfrac{1}{\sqrt{2}} 101010^{-10} 10410^{4} 0.2 0.18 0 π2\dfrac{\pi}{2} 10310^{3} -0.1
Fig. 4 -0.1 0 12\dfrac{1}{\sqrt{2}} 101010^{-10} 10410^{4} 0.2 0.2 π4\dfrac{\pi}{4} π\pi 10610^{6} 0

VI Conclusions

We have established a three-dimensional betatron oscillation model including radiation reaction to study the long-term behavior of an electron in laser or beam driven plasma wakefield. The original equations of motion have been expressed by the transverse oscillation terms as Eqs. (7) - (10), and then averaged in one betatron period to obtain the long-term equations Eqs. (29) - (33). The conditions of our model are r2γr^{2}\ll\gamma, r2γ1r^{2}\gamma\gg 1 and rγre/2α1r\gamma r_{e}/2\alpha\ll 1, as discussed in Appx. C. Our model, on one hand, reproduces previous results such as longitudinal deceleration and transverse damping, and on the other hand reveals new phenomena such as longitudinal phase drift oscillation, betatron phase shift and betatron polarization change. Two regimes with distinct behaviors, determined by reγ5/2r_{e}\gamma^{5/2}, are discussed in Sec. IV, and are demonstrated by numerical methods in Sec. V. The numerical comparisons of the long-term equations and the original equations of motion show the high accuracy of our model. This model can be fundamental for future plasma based high-energy accelerators and colliders [26].

Acknowledgements.
MZ greatly appreciates the fruitful discussion on the averaging method with Igor Kostyukov and Anton Golovanov from Institute of Applied Physics RAS, Russia. This work is supported by Research Foundation of Institute of High Energy Physics, Chinese Academy of Sciences (Grant Nos. E05153U1, E15453U2).

Appendix A Plasma normalization units

The plasma normalization units are used throughout the paper, as listed in Tab. 2, where cc is the speed of light in vacuum, ωp\omega_{p} is the plasma frequency, ee is the elementary charge, and mem_{e} is the electron mass. For example, the time is normalized to ωp1\omega_{p}^{-1}, means any time related quantity such as tt in this paper actually means ωpt\omega_{p}t in the unnormalized form.

Table 2: The plasma normalization units
Physical quantities Variables Normalization units
time tt ωp1\omega_{p}^{-1}
frequency ω\omega ωp\omega_{p}
length x,y,z,rex,y,z,r_{e} c/ωpc/\omega_{p}
velocity vv cc
momentum pp mecm_{e}c
angular momentum LL mec2/ωpm_{e}c^{2}/\omega_{p}
electric field EE mecωp/em_{e}c\omega_{p}/e
magnetic field BB meωp/em_{e}\omega_{p}/e (in SI)
force ff mecωpm_{e}c\omega_{p}

Appendix B Symbols and rules

If any variable XX, either real or complex, can be expressed as X=X+X1X=\left<X\right>+X_{1}, where \left<\right> means taking average in the betatron period time scale, and X1X_{1} is the BO term, taking average and derivative can permute

ddtX=ddtX.\left<\frac{d}{dt}X\right>=\frac{d}{dt}\left<X\right>. (40)

We use a dot on the top to express the time derivative if there is no ambiguity. We have the order-of-magnitude estimation

X1˙ωβX1γ12X1.\dot{X_{1}}\sim\omega_{\beta}X_{1}\sim\gamma^{-\frac{1}{2}}X_{1}. (41)

If |X1||X|\left|X_{1}\right|\ll\left|\left<X\right>\right|, for any power α\alpha we have

Xα=Xα[1+𝒪(X12X2)].\left<X^{\alpha}\right>=\left<X\right>^{\alpha}\left[1+\mathcal{O}\left(\frac{X_{1}^{2}}{\left<X\right>^{2}}\right)\right]. (42)

And if another variable Y=Y+Y1Y=\left<Y\right>+Y_{1} also has |Y1||Y|\left|Y_{1}\right|\ll\left|\left<Y\right>\right|,

XY=XY[1+𝒪(X1Y1XY)].\left<XY\right>=\left<X\right>\left<Y\right>\left[1+\mathcal{O}\left(\frac{X_{1}Y_{1}}{\left<X\right>\left<Y\right>}\right)\right]. (43)

If XX is a complex, taking average and modulus can permute

|X|=|X|[1+𝒪(|X1|2|X|2)].\left<\left|X\right|\right>=\left|\left<X\right>\right|\left[1+\mathcal{O}\left(\frac{\left|X_{1}\right|^{2}}{\left|\left<X\right>\right|^{2}}\right)\right]. (44)

However, taking modulus and derivative cannot permute.

Appendix C Equations of motion expressed by transverse oscillations

In Eqs. (4) - (6), the longitudinal and transverse oscillations are coupled. As shown in the following, the longitudinal variables ζ1\zeta_{1} and βz\beta_{z} are dependent variables which can be expressed by the transverse ones.

We treat frad\vec{f}^{\rm rad} as a perturbation and omit it first. On one hand we have

γ2\displaystyle\gamma^{-2} =1βz2βx2βy2\displaystyle=1-\beta_{z}^{2}-\beta_{x}^{2}-\beta_{y}^{2} (45)
=γz022βz0ζ1˙βx2βy2+𝒪(ζ1˙2),\displaystyle=\gamma_{z0}^{-2}-2\beta_{z0}\dot{\zeta_{1}}-\beta_{x}^{2}-\beta_{y}^{2}+\mathcal{O}\left(\dot{\zeta_{1}}^{2}\right),

where γz0=(1βz02)1/2\gamma_{z0}=\left(1-\beta_{z0}^{2}\right)^{-1/2}. By taking average we get

γz02γ2+βx2+βy2.\gamma_{z0}^{-2}\approx\left<\gamma\right>^{-2}+\left<\beta_{x}^{2}\right>+\left<\beta_{y}^{2}\right>. (46)

Write γ=γ+γ1\gamma=\left<\gamma\right>+\gamma_{1} in the form

γ2=γ2[12γ1γ+𝒪(γ12γ2)],\gamma^{-2}=\left<\gamma\right>^{-2}\left[1-2\frac{\gamma_{1}}{\left<\gamma\right>}+\mathcal{O}\left(\frac{\gamma_{1}^{2}}{\left<\gamma\right>^{2}}\right)\right], (47)

we have

γ1[βx2βx22+βy2βy22+βz0ζ1˙]γ3.\gamma_{1}\approx\left[\frac{\beta_{x}^{2}-\left<\beta_{x}^{2}\right>}{2}+\frac{\beta_{y}^{2}-\left<\beta_{y}^{2}\right>}{2}+\beta_{z0}\dot{\zeta_{1}}\right]\left<\gamma\right>^{3}. (48)

On the other hand,

γ˙=Ez0βz0λβz0ζ1κ2(1λ)(xβx+yβy)\dot{\gamma}=-E_{z0}\beta_{z0}-\lambda\beta_{z0}\zeta_{1}-\kappa^{2}\left(1-\lambda\right)\left(x\beta_{x}+y\beta_{y}\right) (49)

by applying γ˙=βE\dot{\gamma}=-\vec{\beta}\cdot\vec{E} and Eqs. (1) and (2), or

γ1˙=λβz0ζ1κ2(1λ)(xβx+yβy).\dot{\gamma_{1}}=-\lambda\beta_{z0}\zeta_{1}-\kappa^{2}\left(1-\lambda\right)\left(x\beta_{x}+y\beta_{y}\right). (50)

Note Eq. (41), Eq. (48) seams incompatible with Eq. (50), unless

ζ1˙=βx2βx22βy2βy22,\dot{\zeta_{1}}=-\frac{\beta_{x}^{2}-\left<\beta_{x}^{2}\right>}{2}-\frac{\beta_{y}^{2}-\left<\beta_{y}^{2}\right>}{2}, (51)

which leads to

ζ1=xβx+yβy4,\zeta_{1}=-\frac{x\beta_{x}+y\beta_{y}}{4}, (52)

which is a general form of Eq. (18) in Ref. [22]. Then

1βz=1βz0ζ1˙=12(γ2+βx2+βy2),1-\beta_{z}=1-\beta_{z0}-\dot{\zeta_{1}}=\frac{1}{2}\left(\left<\gamma\right>^{-2}+\beta_{x}^{2}+\beta_{y}^{2}\right), (53)

and the formulas of 3D BO with negligible RR are

γ˙\displaystyle\dot{\gamma} =Ez0βz0+(λβz04+κ2λκ2)(xβx+yβy),\displaystyle=-E_{z0}\beta_{z0}+\left(\frac{\lambda\beta_{z0}}{4}+\kappa^{2}\lambda-\kappa^{2}\right)\left(x\beta_{x}+y\beta_{y}\right), (54)
pz˙\displaystyle\dot{p_{z}} =Ez0+λ(14+κ2)(xβx+yβy),\displaystyle=-E_{z0}+\lambda\left(\frac{1}{4}+\kappa^{2}\right)\left(x\beta_{x}+y\beta_{y}\right), (55)
px˙\displaystyle\dot{p_{x}} =κ2x+κ2λ2(γ2+βx2+βy2)x,\displaystyle=-\kappa^{2}x+\frac{\kappa^{2}\lambda}{2}\left(\left<\gamma\right>^{-2}+\beta_{x}^{2}+\beta_{y}^{2}\right)x, (56)
py˙\displaystyle\dot{p_{y}} =κ2y+κ2λ2(γ2+βx2+βy2)y.\displaystyle=-\kappa^{2}y+\frac{\kappa^{2}\lambda}{2}\left(\left<\gamma\right>^{-2}+\beta_{x}^{2}+\beta_{y}^{2}\right)y. (57)

From Eq. (54) we may write

γ=γ+(λβz04+κ2λκ2)x2x2+y2y22,\gamma=\left<\gamma\right>+\left(\frac{\lambda\beta_{z0}}{4}+\kappa^{2}\lambda-\kappa^{2}\right)\frac{x^{2}-\left<x^{2}\right>+y^{2}-\left<y^{2}\right>}{2}, (58)

indicating the prerequisite of the above derivation, which has used Eq. (42), is r2γr^{2}\ll\left<\gamma\right>.

Now we consider RR as a perturbation. The LAD equation for the RR four-force is [11]

Fμrad=23re[d2Pμdτ2+(dPνdτdPνdτ)Pμ].F_{\mu}^{\rm rad}=\frac{2}{3}r_{e}\left[\frac{d^{2}P_{\mu}}{d\tau^{2}}+\left(\frac{dP_{\nu}}{d\tau}\frac{dP^{\nu}}{d\tau}\right)P_{\mu}\right]. (59)

with the metric (1,1,1,1)\left(1,-1,-1,-1\right), where rer_{e} is the classical electron radius (also normalized to kp1k_{p}^{-1}), PμP_{\mu} is the four-momentum, and τ\tau is the proper time (dτ=dt/γd\tau=dt/\gamma). Use Eqs. (54) - (57) we can verify

dPνdτdPνdτ=γ2(γ˙2|p˙|2)γ2(px˙2+py˙2)\frac{dP_{\nu}}{d\tau}\frac{dP^{\nu}}{d\tau}=\gamma^{2}\left(\dot{\gamma}^{2}-\left|\dot{\vec{p}}\right|^{2}\right)\approx-\gamma^{2}\left(\dot{p_{x}}^{2}+\dot{p_{y}}^{2}\right) (60)

as long as r2γ21r^{2}\gamma^{2}\gg 1. We can also prove that the first term in Eq. (59) is negligible compared with the second term as long as r2γ1r^{2}\gamma\gg 1. Finally the equations of motion expressed by the transverse oscillations are obtained as Eqs. (7) - (10). As it has been discussed in Ref. [22] and [23], this classical RR model is valid as long as rγre/2α1r\gamma r_{e}/2\alpha\ll 1, where α\alpha is the fine structure constant.

References

  • Tajima and Dawson [1979] T. Tajima and J. M. Dawson, Laser electron accelerator, Phys. Rev. Lett. 43, 267 (1979).
  • Chen et al. [1985] P. Chen, J. M. Dawson, R. W. Huff, and T. Katsouleas, Acceleration of electrons by the interaction of a bunched electron beam with a plasma, Phys. Rev. Lett. 54, 693 (1985).
  • Blumenfeld et al. [2007] I. Blumenfeld, C. E. Clayton, F.-J. Decker, M. J. Hogan, C. Huang, R. Ischebeck, R. Iverson, C. Joshi, T. Katsouleas, N. Kirby, W. Lu, K. A. Marsh, W. B. Mori, P. Muggli, E. Oz, R. H. Siemann, D. Walz, and M. Zhou, Energy doubling of 42 gev electrons in a metre-scale plasma wakefield accelerator, Nature 44510.1038/nature05538 (2007).
  • Rosenzweig et al. [1991] J. B. Rosenzweig, B. Breizman, T. Katsouleas, and J. J. Su, Acceleration and focusing of electrons in two-dimensional nonlinear plasma wake fields, Phys. Rev. A 44, R6189 (1991).
  • Pukhov and Meyer-ter Vehn [2002] A. Pukhov and J. Meyer-ter Vehn, Laser wake field acceleration: the highly non-linear broken-wave regime, Applied Physics B 7410.1007/s003400200795 (2002).
  • Kostyukov et al. [2004] I. Y. Kostyukov, A. Pukhov, and S. Kiselev, Phenomenological theory of laser-plasma interaction in “bubble” regime, Physics of Plasmas 11, 5256 (2004).
  • Lu et al. [2006] W. Lu, C. Huang, M. Zhou, M. Tzoufras, F. S. Tsung, W. B. Mori, and T. Katsouleas, A nonlinear theory for multidimensional relativistic plasma wave wakefields, Physics of Plasmas 13, 056709 (2006)https://doi.org/10.1063/1.2203364 .
  • Esarey et al. [2002] E. Esarey, B. A. Shadwick, P. Catravas, and W. P. Leemans, Synchrotron radiation from electron beams in plasma-focusing channels, Phys. Rev. E 65, 056505 (2002).
  • Rousse et al. [2004] A. Rousse, K. T. Phuoc, R. Shah, A. Pukhov, E. Lefebvre, V. Malka, S. Kiselev, F. Burgy, J.-P. Rousseau, D. Umstadter, and D. Hulin, Production of a kev x-ray beam from synchrotron radiation in relativistic laser-plasma interaction, Phys. Rev. Lett. 93, 135005 (2004).
  • Corde et al. [2013] S. Corde, K. Ta Phuoc, G. Lambert, R. Fitour, V. Malka, A. Rousse, A. Beck, and E. Lefebvre, Femtosecond x rays from laser-plasma accelerators, Rev. Mod. Phys. 85, 1 (2013).
  • Dirac [1938] P. A. M. Dirac, Classical theory of radiating electrons, Proc. R. Soc. Lond. A 167, 148 (1938).
  • Landau and Lifshitz [1975] L. D. Landau and E. M. Lifshitz, Chapter 9 - radiation of electromagnetic waves, in The Classical Theory of Fields (Fourth Edition), Course of Theoretical Physics, Vol. 2, edited by L. D. Landau and E. M. Lifshitz (Pergamon, Amsterdam, 1975) fourth edition ed., pp. 171–225.
  • Cole et al. [2018] J. M. Cole, K. T. Behm, E. Gerstmayr, T. G. Blackburn, J. C. Wood, C. D. Baird, M. J. Duff, C. Harvey, A. Ilderton, A. S. Joglekar, K. Krushelnick, S. Kuschel, M. Marklund, P. McKenna, C. D. Murphy, K. Poder, C. P. Ridgers, G. M. Samarin, G. Sarri, D. R. Symes, A. G. R. Thomas, J. Warwick, M. Zepf, Z. Najmudin, and S. P. D. Mangles, Experimental evidence of radiation reaction in the collision of a high-intensity laser pulse with a laser-wakefield accelerated electron beam, Phys. Rev. X 8, 011020 (2018).
  • Poder et al. [2018] K. Poder, M. Tamburini, G. Sarri, A. Di Piazza, S. Kuschel, C. D. Baird, K. Behm, S. Bohlen, J. M. Cole, D. J. Corvan, M. Duff, E. Gerstmayr, C. H. Keitel, K. Krushelnick, S. P. D. Mangles, P. McKenna, C. D. Murphy, Z. Najmudin, C. P. Ridgers, G. M. Samarin, D. R. Symes, A. G. R. Thomas, J. Warwick, and M. Zepf, Experimental signatures of the quantum nature of radiation reaction in the field of an ultraintense laser, Phys. Rev. X 8, 031004 (2018).
  • Nielsen et al. [2021] C. F. Nielsen, J. B. Justesen, A. H. Sørensen, U. I. Uggerhøj, R. Holtzapple, and C. NA63, Experimental verification of the landau–lifshitz equation, New Journal of Physics 23, 085001 (2021).
  • Michel et al. [2006] P. Michel, C. B. Schroeder, B. A. Shadwick, E. Esarey, and W. P. Leemans, Radiative damping and electron beam dynamics in plasma-based accelerators, Phys. Rev. E 74, 026501 (2006).
  • Kostyukov et al. [2006] I. Y. Kostyukov, E. N. Nerush, and A. M. Pukhov, Radiative losses in plasma accelerators, Journal of Experimental and Theoretical Physics 10310.1134/S1063776106110173 (2006).
  • Schroeder et al. [2010] C. B. Schroeder, E. Esarey, C. G. R. Geddes, C. Benedetti, and W. P. Leemans, Physics considerations for laser-plasma linear colliders, Phys. Rev. ST Accel. Beams 13, 101301 (2010).
  • Deng et al. [2012] A. Deng, K. Nakajima, J. Liu, B. Shen, X. Zhang, Y. Yu, W. Li, R. Li, and Z. Xu, Electron beam dynamics and self-cooling up to pev level due to betatron radiation in plasma-based accelerators, Phys. Rev. ST Accel. Beams 15, 081303 (2012).
  • Kostyukov et al. [2012] I. Y. Kostyukov, E. N. Nerush, and A. G. Litvak, Radiative damping in plasma-based accelerators, Phys. Rev. ST Accel. Beams 15, 111001 (2012).
  • Ferran Pousa et al. [2019] A. Ferran Pousa, A. Martinez de la Ossa, and R. W. Assmann, Intrinsic energy spread and bunch length growth in plasma-based accelerators due to betatron motion, Scientific Reports 910.1038/s41598-019-53887-8 (2019).
  • Zeng and Seto [2021] M. Zeng and K. Seto, Radiation reaction of betatron oscillation in plasma wakefield accelerators, New J. Phys. 23, 075008 (2021).
  • Golovanov et al. [2022] A. A. Golovanov, E. N. Nerush, and I. Y. Kostyukov, Radiation reaction-dominated regime of wakefield acceleration, New Journal of Physics 24, 033011 (2022).
  • BDF [2022] https://docs.scipy.org/doc/scipy/reference/generated/scipy.integrate.BDF.html (2022), the documentation of BDF integration in SciPy.
  • Zeng [2022] M. Zeng, https://github.com/mingzeng7/PTracker (2022), the git repository for the code PTracker.
  • Leemans and Esarey [2009] W. Leemans and E. Esarey, Laser-driven plasma-wave electron accelerators, Physics Today 62, 44 (2009).