1: %\documentclass[useAMS, usenatbib, referee]{mn2e}
2: \documentclass[useAMS, usenatbib, onecolumn]{mn2e}
3: \usepackage{epsfig}
4: \usepackage{amsmath}
5: \voffset -1.8cm
6:
7: \def\be{\begin{equation}}
8: \def\ee{\end{equation}}
9: \def\ba{\begin{eqnarray}}
10: \def\ea{\end{eqnarray}}
11: \def\go{\mathrel{\raise.3ex\hbox{$>$}\mkern-14mu
12: \lower0.6ex\hbox{$\sim$}}}
13: \def\lo{\mathrel{\raise.3ex\hbox{$<$}\mkern-14mu
14: \lower0.6ex\hbox{$\sim$}}}
15: \def\bB{{\bf B}}
16: \def\bw{{\bf w}}
17: \def\bv{{\bf v}}
18: \def\bu{{\bf u}}
19: \def\br{{\bf r}}
20: \def\hate{{\hat{\bf e}}}
21: \def\bfe{{\bf e}}
22: \def\cB{{\cal B}}
23: \def\cA{{\cal A}}
24: \def\d{\partial}
25:
26: \def\barR{{\bar R}}
27: \def\barz{{\bar z}}
28: \def\tomega{\tilde\omega}
29:
30: \begin{document}
31:
32: \title[Super-Reflection in Fluid Discs]
33: {Super-Reflection in Fluid Discs: Corotation Amplifier, Corotation
34: Resonance, Rossby Waves, and Overstable Modes}
35:
36: \author[D. Tsang and D. Lai]{David Tsang$^{1,2}$\thanks{Email:
37: dtsang@astro.cornell.edu; dong@astro.cornell.edu} and Dong
38: Lai$^{1}$\footnotemark[1] \\
39: $^1$Center for Radiophysics and Space Research, Department of Astronomy,
40: Cornell University, Ithaca, NY 14853, USA \\
41: $^2$Department of Physics,
42: Cornell University, Ithaca, NY 14853, USA \\}
43:
44:
45: \volume{387}\pagerange{446--462} \pubyear{2008}
46: \label{firstpage}
47:
48: \maketitle
49:
50: \begin{abstract}
51: In differentially rotating discs with no self-gravity,
52: density waves cannot propagate around
53: the corotation, where the wave pattern rotation speed equals the fluid
54: rotation rate. Waves incident upon the corotation barrier may be
55: super-reflected (commonly referred to as corotation amplifier), but
56: the reflection can be strongly affected by wave absorptions at the corotation
57: resonance/singularity. The sign of the absorption is related to the
58: Rossby wave zone very near the corotation radius.
59: We derive the explicit expressions for the
60: complex reflection and transmission coefficients, taking into account
61: wave absorption at the corotation resonance.
62: We show that for generic discs, this absorption plays a much more
63: important role than wave transmission across the corotation barrier.
64: Depending on the sign of
65: the gradient of the vortensity of the disc,
66: $\zeta=\kappa^2/(2\Omega\Sigma)$
67: (where $\Omega$ is the rotation rate, $\kappa$ is the epicyclic frequency,
68: and $\Sigma$ is the surface density), the corotation resonance can
69: either enhance or diminish the super-reflectivity, and this
70: can be understood in terms of the location of the Rossby wave zone
71: relative to the corotation radius.
72: Our results provide the explicit conditions
73: (in terms of disc thickness, rotation
74: profile and vortensity gradient) for which super-reflection can be
75: achieved. Global overstable disc modes may be possible for discs
76: with super-reflection at the corotation barrier.
77: \end{abstract}
78:
79: \begin{keywords}
80: accretion, accretion discs -- hydrodynamics -- waves -- instabilities
81: \end{keywords}
82:
83: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
84: \section{Introduction}
85:
86: Differentially rotating fluid discs, ubiquitous in astrophysics,
87: are known to exhibit rich dynamics and possible instabilities
88: (e.g. Papaloizou \& Lin 1995; Balbus \& Hawley 1998).
89: While local instabilities, such as Rayleigh's centrifugal instability
90: (for discs with specific angular momentum decreasing outwards),
91: gravitational instability (for self-gravitational discs with too
92: large a surface density, or more precisely, Toomre $Q\lo 1$),
93: and magnetorotational instability
94: (for discs with a sub-thermal magnetic field), are well
95: understood (at least in the linear regime), global effects and
96: instabilities are more subtle, since they involve couplings and feedbacks
97: of fluid at different locations (see Goldreich 1988 for an
98: introduction/review). A well-known example is the corotation
99: amplifier (e.g. Mark 1976; Narayan, Goldreich \& Goodman 1987), which arises
100: from the interaction across the corotation between waves carrying opposite
101: signs of angular momentum. Much stronger corotation amplifications
102: (WASER -- wave amplification by the stimulated emission of radiation,
103: and SWING amplifiers) can be achieved for self-gravitating
104: discs (e.g., Goldreich \& Lynden-Bell 1965; Julian \& Toomre 1966;
105: Lin \& Lau 1975; see Shu 1992 for a review).
106: Another well-known example is the Papaloizou-Pringle instability
107: in finite accretion tori (confined between two free surfaces),
108: in which coupling between waves inside the
109: corotation with those outside, combined with reflecting inner and outer
110: boundaries, leads to violent overstable modes (Papaloizou \& Pringle
111: 1984; Goldreich et al.~1986). Recent works on global
112: disc instabilities include the Rossby wave instability (for discs with a
113: strong enough density or vortensity bump; Lovelace et al.~1999;
114: Li et al.~2000) and the accretion-ejection instability (for magnetized
115: discs; Tagger \& Pellat 1999, Tagger \& Varniere 2006).
116:
117: In this paper we are interested in 2D fluid discs without self-gravity
118: and magnetic field. For disturbances of the form $e^{im\phi-i\omega t}$,
119: where $m>0$ and $\omega$ is the wave (angular) frequency
120: (and thus $\omega_p=\omega/m$ is the pattern frequency),
121: the well-known WKB dispersion relation for density waves
122: takes the form (e.g., Shu 1992)
123: \be
124: (\omega-m\Omega)^2=\tomega^2=\kappa^2+k_r^2c^2,
125: \label{eq:disp}
126: \ee
127: where $\Omega$ is the disc rotation frequency, $\tomega=\omega-m\Omega$
128: is the Doppler-shifted wave frequency, $\kappa$ is the radial
129: epicyclic frequency, $k_r$ is the radial wavenumber, and $c$ is
130: the sound speed. Thus waves can propagate either inside the
131: inner Lindblad resonance radius $r_{\rm IL}$ (defined by $\tomega=-\kappa$)
132: or outside the outer Lindblad resonance radius $r_{\rm OL}$
133: (defined by $\tomega=\kappa$),
134: while the region around the corotation radius $r_c$ (set by $\tomega=0$)
135: between $r_{\rm IL}$ and $r_{\rm OL}$
136: is evanescent. Since the wave inside $r_{\rm IL}$ has pattern speed
137: $\omega_p$ smaller than the fluid rotation rate $\Omega$, it carries
138: negative wave action (or angular momentum), while the wave outside
139: $r_{\rm OL}$ carries positive wave action. As a result, a wave incident
140: from small radii toward the corotation barrier will be super-reflected,
141: (with the reflected wave having a larger amplitude than the incident wave
142: amplitude) if it can excite a wave on the other side of the corotation barrier.
143: If there exists a reflecting boundary at the inner disc radius
144: $r_{\rm in}$, then a global overstable mode partially trapped between
145: $r_{\rm in}$ and $r_{\rm IL}$ will result (see, e.g. Narayan et al.~1987 for
146: specific examples in the shearing sheet model, and Goodman \& Evans,1999
147: and Shu et al.~ 2000 for global mode analysis of singular isothermal discs).
148:
149: The simple dispersion relation (\ref{eq:disp}), however, does not capture
150: an important effect in the disc, i.e., corotation resonance or
151: corotation singularity.
152: Near corotation $|\tomega|\ll\kappa$, the WKB dispersion relation for
153: the wave is [see equation (\ref{eq:disper}) below]
154: \be
155: \tomega={2\Omega k_\phi\over k_r^2+k_\phi^2
156: +\kappa^2/c^2}\left({d\over dr}\ln{\kappa^2
157: \over 2\Omega\Sigma}\right)_c,
158: \label{eq:disp2}\ee
159: where $k_\phi=m/r$ and $\Sigma$ is the surface density,
160: and the subscript ``c'' implies that the quantity is evaluated
161: at $r=r_c$. The quantity
162: \be
163: \zeta\equiv {\kappa^2\over 2\Omega\Sigma}=
164: {(\nabla\times {\bf u_0})\cdot {\hat z}\over\Sigma}
165: \label{eq:zeta}\ee
166: is the vortensity of
167: the (unperturbed) flow (where ${\bf u_0}$ is the flow velocity).
168: The dispersion relation (\ref{eq:disp2}) describes Rossby waves, analogous
169: to those studied in geophysics (e.g. Pedlosky 1987)
170: \footnote{A Rossby wave propagating in the Earth's atmosphere satisfies the
171: dispersion relation $\tomega=(2k_\phi/k^2R)(\partial\Omega_3/
172: \partial\theta)$,
173: where $k^2=k_\phi^2+k_\theta^2$,
174: $\Omega_3=\Omega\cos\theta$ is the projection of the rotation rate on
175: the local surface normal vector and $\theta$ is the polar angle
176: (co-latitude).}.
177: For $k_r^2\gg \kappa^2/c^2$ and $k_r^2\gg k_\phi^2$,
178: we see that Rossby waves can propagate either outside the
179: rotation radius $r_c$ (when $d\zeta/dr>0$) or inside $r_c$ (when
180: $d\zeta/dr<0$). In either case, we have $k_r\rightarrow\infty$ as $r\rightarrow
181: r_c$. This infinite wavenumber signifies wave absorption
182: (cf. Lynden-Bell \& Kalnajs 1972 in stellar dynamical context;
183: Goldreich \& Tremaine 1979 in the context of wave excitation in discs by
184: a external periodic force; see also Kato 2003, Li et. al. 2003, and Zhang \& Lai 2006 for wave
185: absorption at the corotation in 3D discs).
186: At corotation, the wave pattern angular speed $\omega/m$ matches $\Omega$,
187: and there can be efficient energy transfer between the wave and the background
188: flow, analogous to Landau damping in plasma physics.
189: Narayan et al.~(1987) treated this effect as
190: a perturbation of the shearing sheet model, and showed that the corotational
191: absorption can convert neutral modes in a finite shearing sheet
192: into growing or decaying modes. Papaloizou \& Pringle (1987) used a WKB
193: method to examine the effect of wave absorption at corotation on the
194: nonaxisymmetric modes in an unbound (with the outer boundary extending to
195: infinity) cylindrical torus.
196:
197: In this paper, we derive explicit expressions
198: for the complex reflection coefficient and transmission coefficient
199: for waves incident upon the corotation barrier. We pay particular
200: attention to the behavior of
201: perturbations near the corotation resonance/singularity.
202: Our general expressions include both the effects of corotation amplifier
203: and wave absorption at corotation (which depends on
204: $d\zeta/dr$). We show explicitly that depending on the sign of
205: $d\zeta/dr$, the corotation resonance/singularity can either enhance or
206: diminish the super-reflectivity, and this can be understood in terms of
207: the location of the Rossby wave zone relative to the corotation radius.
208:
209: Our paper is organized as follows. After presenting the general
210: perturbation equations (section 2), we discuss the the wave
211: dispersion relation and propagation diagram, and derive
212: the local solutions for the wave equation around the Lindblad resonances
213: and corotation resonance (section 3). We then construct global WKB
214: solution for the wave equation, and derive the wave reflection, transmission
215: and corotational damping coefficients in section 4. An alternative
216: derivation of the wave damping coefficient is presented in section 5.
217: Readers not interested in technical details can skip Sections 2-5 and
218: concentrate on Section 6, where we illustrate our results and discuss their
219: physical interpretations. Section 6.1 contains a numerical calculation of the
220: wave reflectivity across corotation and discusses the limitation of the WKB
221: analysis. We discuss how global overstable
222: modes may arise when super-reflection at the corotation is present in
223: section 7 and conclude in section 8.
224:
225:
226: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
227: \section{Perturbation Equations}
228:
229: We consider a geometrically thin gas disc and adopt cylindrical coordinate
230: system $(r,\phi,z)$. The unperturbed disc has an integrated
231: surface density $\Sigma(r)$ and velocity ${\bf u_0} = (0, r\Omega, 0)$.
232: The flow is assumed to be barotropic, so that the integrated pressure $P$
233: depends only on $\Sigma$. Self gravity of the disc is neglected.
234:
235: The linear perturbation equations for the flow read
236: \ba
237: &&{\partial \over\partial t}\delta\bu+(\bu_0\cdot\nabla)\delta\bu
238: +(\delta\bu\cdot\nabla)\bu_0
239: =-\nabla\delta h, \label{eq:u}\\
240: && {\partial\over\partial t}\delta\Sigma+\nabla\cdot(\Sigma\,\delta \bu
241: +\bu_0\,\delta\Sigma)=0,
242: \label{eq:rho}\ea
243: where $\delta\Sigma,~\delta \bu$ and $\delta h=\delta P/\Sigma$
244: are the (Eulerian) perturbations of surface density, velocity and enthalpy,
245: respectively. For barotropic flow, $\delta h$ and $\delta\Sigma$ are related by
246: \be
247: \delta h=c^2{\delta\Sigma\over\Sigma},
248: \ee
249: where $c$ is the sound speed, with $c^2=dP/d\Sigma$.
250:
251: We assume that the $\phi$ and $t$ dependence of the perturbation are of
252: the form
253: \be
254: \delta\bu,~\delta\Sigma,~\delta h \propto e^{ i m \phi-i\omega t},
255: \ee
256: where $m$ is a positive integer, and $\omega$ is the wave (angular) frequency.
257: %In general, $\omega=\omega_r+i\omega_i$ is complex, and
258: We presume $\omega>0$ so that the pattern (angular) speed of the perturbation
259: $\omega_p=\omega/m$ is positive (in the same direction as the flow rotation).
260: Note that we usually assume $\omega$ is real, except
261: in section 3.2 (dealing with the perturbation near corotation)
262: where we include a small imaginary part ($\omega=\omega_r+i\omega_i$, with
263: $\omega_i>0$) to represent slowly growing disturbances.
264: The perturbation equations (\ref{eq:u})-(\ref{eq:rho}) become
265: \ba
266: && -i\tomega {\Sigma\over c^2}\delta h + \frac{1}{r} \frac{\d}{\d r}
267: (\Sigma r \delta u_r) + \frac{im}{r}\Sigma \delta u_\phi = 0,\label{eq:sig}\\
268: && -i\tomega \delta u_r - 2\Omega \delta u_\phi = -{\partial\over\partial r}
269: \delta h,\label{eq:ur}\\
270: && -i\tomega \delta u_\phi + \frac{\kappa^2}{2\Omega}\delta u_r =
271: -\frac{im}{r} \delta h,\label{eq:uphi}
272: \ea
273: where the epicyclic
274: frequency $\kappa$ is given by
275: \be
276: \kappa^2={2\Omega\over r}{d\over dr}(r^2\Omega).
277: \ee
278:
279: Eliminating $\delta u_r$ and $\delta u_\phi$ from
280: equations (\ref{eq:sig})-(\ref{eq:uphi}), we obtain a standard second-order
281: differential equation governing $\delta h$ (e.g.,
282: Goldreich \& Tremaine 1979):
283: \be
284: \left[\frac{d^2}{dr^2} -
285: \frac{d}{dr}\left(\ln\frac{D}{r\Sigma}\right)\frac{d}{dr} -
286: \frac{2m\Omega}{r\tomega}\left(\frac{d}{dr}\ln\frac{\Omega\Sigma}{D}\right)
287: - \frac{m^2}{r^2} - \frac{D}{c^2}\right]\delta h = 0,\label{perturbeq1}
288: \ee
289: where
290: \be
291: D \equiv \kappa^2 - \tomega^2=\kappa^2-(\omega-m\Omega)^2.
292: \ee
293: Defining
294: \be
295: S = D/(r\Sigma),\quad
296: \eta =S^{-1/2}\delta h,
297: \ee
298: we can rewrite \eqref{perturbeq1} as a wave equation
299: \be
300: \left[\frac{d^2}{dr^2}-\frac{D}{c^2} - \frac{m^2}{r^2} -
301: \frac{2m\Omega}{r\tomega}
302: \left(\frac{d}{dr}\ln\frac{\Omega\Sigma}{D}\right) -
303: S^{1/2}\frac{d^2}{dr^2}S^{-1/2}\right]\eta =0~. \label{perturbeq2}
304: \ee
305: This is our basic working equation.
306:
307: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
308: \section{Propagation Diagram and Local Solutions Near Resonances}
309:
310: Consider local free wave solution of the form
311: \be
312: \delta h\propto \exp\left[i\int^r\!k_r(s) ds\right]~.
313: \ee
314: For $|k_rr|\gg m$ and away from the $D=0$ region,
315: we find from equation (\ref{perturbeq2})
316: \be
317: k_r^2+{D\over c^2}+
318: \frac{2m\Omega}{r\tomega}\left(\frac{d}{dr}\ln\frac{\Omega\Sigma}{D}\right)
319: \simeq 0.
320: \label{eq:disper}\ee
321: This is the general WKB dispersion relation.
322: Away from the region where $\tomega=0$, this reduces to the well-known
323: result $k_r^2\simeq -D/c^2$
324: [equation (\ref{eq:disp})]; in the vicinity of $\tomega=0$
325: this describes local Rossby waves, with
326: [see equations (\ref{eq:disp2})-(\ref{eq:zeta})]
327: \be
328: \tomega\simeq {2m\Omega\over r k_r^2}
329: \left({d\ln\zeta\over dr}\right)_c.
330: \ee
331:
332: Before studying global solutions to the wave equation (\ref{perturbeq1})
333: or (\ref{perturbeq2}), it is useful to consider the special resonant
334: locations in the disc. These can be
335: recognized by investigating the singular points and turning
336: points of the wave equation (\ref{perturbeq1}) or
337: (\ref{perturbeq2}), or by examining the characteristics of
338: the dispersion relation (\ref{eq:disper}).
339: The special radii are
340:
341: (i) Lindblad resonances (LRs), where $D=0$ or $\tomega ^2=\kappa^2$,
342: including the outer Lindblad resonance (OLR) at $\tomega=\kappa$ and the inner
343: Lindblad resonance (ILR) at $\tomega=-\kappa$. The LRs are
344: apparent singularities of equation (\ref{perturbeq1})
345: or (\ref{perturbeq2}) -- all physical quantities
346: are finite at $D=0$. The LRs are turning points
347: at which wave trains are reflected or transmitted. In the presence of an external force, waves are launched from LRs.
348:
349: (ii) Corotation resonance (CR), where $\tomega=0$. In general,
350: the CR is a singular point of the wave equation
351: except in the special case of $d\zeta/dr=0$ at corotation.
352: Some physical quantities (e.g., azimuthal velocity
353: perturbation) are divergent at corotation. Physically, this singularity
354: signifies that a steady emission or absorption of wave action
355: may occur there.
356:
357: From equation (\ref{perturbeq2}), we define the effective potential
358: for wave propagation by
359: \ba
360: V_{\rm eff}(r) &=&\frac{D}{c^2}+\frac{m^2}{r^2} +
361: \frac{2m\Omega}{r\tomega}
362: \left(\frac{d}{dr}\ln\frac{\Omega\Sigma}{D}\right)
363: +S^{1/2}\frac{d^2}{dr^2}S^{-1/2}\nonumber\\
364: &=&V_{\rm eff,0}(r)+\Delta V_{\rm eff}(r),
365: \ea
366: where
367: \ba
368: V_{\rm eff,0}(r) &=&\frac{D}{c^2}+\frac{m^2}{r^2} -
369: \frac{2m\Omega}{r\tomega} \left(\frac{d}{dr}\ln\zeta\right),\\
370: \Delta V_{\rm eff}(r) &=&
371: -\frac{2m\Omega}{r\tomega} \frac{d}{dr}\ln{D\over\kappa^2}
372: +S^{1/2}\frac{d^2}{dr^2}S^{-1/2}.
373: \ea
374: Clearly, wave propagation is possible only in the region where
375: $V_{\rm eff}(r)<0$. Figures 1-3 depict the wave propagation diagrams
376: for the cases of $(d\zeta/dr)_c=0,~<0$ and $>0$, respectively.
377: We are interested in the parameter regime $c/(r\Omega)\ll 1$ and $m$ is
378: of order unity. Note that the apparent singularity in $\Delta V_{\rm eff}(r)$
379: at $D=0$ can be eliminated by analysing the wave solution around the LRs
380: (see section 3.1 below). Thus we also show $V_{\rm eff,0}(r)$ in Figs.~1-3.
381:
382: We now consider the behaviors of the perturbations around the LRs and CR.
383:
384:
385: \begin{figure}
386: \centering
387: \includegraphics[width=13cm]{potential1}
388: {\vskip -3cm}
389: \caption{Wave propagation diagram in Keplerian discs:
390: A sketch of the effective potential $V_{\rm eff,0}(r)$
391: (solid line) and $V_{\rm eff}(r)$ (dashed line) as a function of $r$,
392: for the case of $(d\zeta/dr)_c=0$.
393: Waves can propagate only in the region where $V_{\rm eff}(r)< 0$, i.e.,
394: where the curves are below the dotted line.
395: The three special locations are denoted by
396: IL (Inner Lindblad Resonance), OL (Outer Lindblad Resonance) and CR
397: (Corotation Resonance).
398: The divergence in the $V_{\rm eff}(r)$ curve around IL and OL represents
399: an apparent singularity.}
400: \label{fig1}
401: \end{figure}
402:
403: \begin{figure}
404: \centering
405: \includegraphics[width=13cm]{potential2a}
406: {\vskip -3cm}
407: \caption{Same as Fig.~1, except for the case of
408: negative vortensity gradient, $(d\zeta/dr)_c<0$ (or
409: $\nu<0$). Note the CR represents a singularity, and the Rossby wave
410: zone lies inside the corotation radius.}
411: \label{fig2}
412: \end{figure}
413:
414: \begin{figure}
415: \centering
416: \includegraphics[width=13cm]{potential2b}
417: {\vskip -3cm}
418: \caption{Same as Fig.~1, except for the case of
419: positive vortensity gradient, $(d\zeta/dr)_c>0$ (or
420: $\nu>0$), for which the Rossby wave
421: zone lies outside the corotation radius.}
422: \label{fig3}
423: \end{figure}
424:
425:
426: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
427: \subsection{Solution Around Lindblad Resonances}
428:
429: Equation \eqref{perturbeq2} has an apparent singularity at the LRs,
430: where $D \rightarrow 0$. For concreteness we will
431: explicitly examine the outer Lindblad resonance (OLR);
432: a similar solution can be
433: found for the inner Lindblad resonance (ILR).
434:
435: In the vicinity of the OLR, equation \eqref{perturbeq2} becomes
436: \be
437: \frac{d^2}{dr^2} \eta + \left(k^2 - k \frac{d^2}{dr^2} \frac{1}{k}
438: +{4m\Omega\over r\tomega k}{dk\over dr}\right)\eta = 0, \label{LRdiffeq}
439: \ee
440: where $k^2\equiv -D/c^2$.
441: The last term inside $(\cdots)$ is smaller than the second term and will be
442: neglected. Changing the independent variable from $r$ to the dimensionless
443: integrated phase
444: \be
445: z = \int_{r_{\rm OL}}^r k\,dr,
446: \ee
447: we have
448: \be
449: V'' + \left[ \frac{k''}{2k} - \frac{3}{4}\left(\frac{k'}{k}\right)^2
450: + 1\right]V = 0,
451: \label{Vequation}\ee
452: where
453: \be
454: V= \sqrt{k}\,\eta=\sqrt{k\over S}\delta h,
455: \ee
456: and the prime denotes differentiation with respect to $z$.
457: Note that near OLR, $k^2 \simeq C(r-r_{\rm OL})$, with
458: $C=(-c^{-2}dD/dr)_{\rm OL}>0$ a constant, we have
459: \be
460: z =
461: \Biggl\{
462: \begin{array}{ll}
463: \frac{2}{3}C^{1/2}(r-r_{OL})^{3/2} &\qquad \qquad \textrm{for }
464: r > r_{\rm OL} \\
465: \frac{2}{3} C^{1/2}(r_{\rm OL}-r)^{3/2}e^{i3\pi/2}
466: %e^{\pm i3\pi/2}
467: &\qquad \qquad \textrm{for } r < r_{\rm OL}\end{array}
468: \ee
469: We can then express $k$ in terms of $z$ as
470: \be
471: k = \left(\frac{3}{2} C\right)^{1/3} z^{1/3}~.\label{kintermsofy}
472: \ee
473: Using equation \eqref{kintermsofy} in equation \eqref{Vequation} we have
474: \be
475: V'' + \left( 1 - \frac{7}{36z^2}\right) V = 0 \label{finalVeqn}~.
476: \ee
477:
478: Equation (\ref{finalVeqn}) has two independent solutions in terms of Bessel
479: function (Abramowitz \& Stegun 1964)
480: \be
481: V = \sqrt{z} J_{\pm 2/3} (z)
482: = \frac{1}{(12 z)^{1/6}}\left[\pm\sqrt{3}{\rm Ai}'(-Z)+{\rm Bi}'(-Z)\right]~,
483: \ee
484: where $Z=(3z/2)^{2/3}$, and ${\rm Ai}',~{\rm Bi}'$ are the derivatives of
485: the Airy functions. Instead of using $\sqrt{z}J_{\pm 2/3}(z)$, we
486: we can construct two linearly independent solutions for $\eta$ in a form
487: convenient for asymptotic matching:
488: \be
489: \eta_1 = -\left({\pi\over k}\right)^{1/2}\!\left({2\over 3z}\right)^{1/6}\!
490: {\rm Ai}'(-Z)\sim
491: \Biggl\{\begin{array}{ll}
492: \frac{1}{\sqrt{k}}\cos\left ( z + \pi/4\right) & \quad \textrm{for }
493: |z| \gg 1\textrm{ and }\arg(z) = 0\\
494: \frac{1}{2\sqrt{k}}\exp\left(-|z|\right) & \quad \textrm{for }|z| \gg 1
495: \textrm{ and }\arg(z) = 3\pi/2
496: \end{array}
497: \ee
498: \be
499: \eta_2 = \left({\pi\over k}\right)^{1/2}\!\left({2\over 3z}\right)^{1/6}\!
500: {\rm Bi}'(-Z)\sim
501: \Biggl\{\begin{array}{ll}
502: \frac{1}{\sqrt{k}}\sin\left( z + \pi/4\right) & \quad \textrm{for }
503: |z| \gg 1\textrm{ and }\arg(z) = 0\\
504: \frac{1}{\sqrt{k}}\exp\left(|z|\right) & \quad \textrm{for }|z|
505: \gg 1\textrm{ and }
506: \arg(z) = 3\pi/2\end{array}
507: \ee
508: where $\sim$ indicates asymptotic expansions.
509: This gives the connection formulae for the enthalpy perturbation
510: at the OLR\footnote{The usage here of $\gg$ is used as a shorthand for the range of validity for an asymptotic expansion of a local solution. The fitting formulae are to be used far enough away from the resonances so that the asymptotic expansion is valid, but close enough that the local approximation made in (22) holds.}:
511: \be
512: \delta h_1 \sim
513: \Biggl\{\begin{array}{ll}
514: \frac{1}{2}\sqrt{S/k}\, \exp\left(-\int_r^{r_{\rm OL}}\!|k| \,dr \right)
515: & \qquad \textrm{for } r \ll r_{\rm OL}\\
516: \sqrt{S/k}\, \cos\left(\int_{r_{\rm OL}}^r\! k\, dr + \pi/4\right)
517: &\qquad \textrm{for } r\gg r_{\rm OL}
518: \end{array}
519: \label{eq:con1}\ee
520: \be
521: \delta h_2 \sim
522: \Biggl\{\begin{array}{ll}
523: \sqrt{S/k} \exp\left(\int_r^{r_{\rm OL}}\! |k|\, dr \right)
524: & \qquad \textrm{for } r \ll r_{OL}\\
525: \sqrt{S/k}\, \sin\left(\int_{r_{\rm OL}}^r\! k\, dr + \pi/4\right)
526: &\qquad \textrm{for } r\gg r_{\rm OL}
527: \end{array}
528: \label{eq:con2}\ee
529:
530: The connection formulae for ILR can be similarly derived:
531: \footnote{Note that for the ILR, $k$ is real for $r < r_{\rm IL}$
532: and imaginary for $r > r_{\rm IL}$, while for
533: the OLR, $k$ is real for $r > r_{\rm OL}$ and imaginary for
534: $r < r_{\rm OL}$.}
535: \be
536: \delta h_1 \sim
537: \Biggl\{\begin{array}{ll}
538: \frac{1}{2}\sqrt{S/k}\, \exp\left(-\int^r_{r_{\rm IL}}\!|k| \,dr \right)
539: & \qquad \textrm{for } r \gg r_{\rm IL}\\
540: \sqrt{S/k}\, \cos\left(\int^{r_{\rm IL}}_r\! k\, dr + \pi/4\right)
541: &\qquad \textrm{for } r\ll r_{\rm IL}
542: \end{array}
543: \label{eq:conn1}\ee
544: \be
545: \delta h_2 \sim
546: \Biggl\{\begin{array}{ll}
547: \sqrt{S/k} \exp\left(\int^r_{r_{\rm IL}}\! |k|\, dr \right)
548: & \qquad \textrm{for } r \gg r_{IL}\\
549: \sqrt{S/k}\, \sin\left(\int^{r_{\rm IL}}_r\! k\, dr + \pi/4\right)
550: &\qquad \textrm{for } r\ll r_{\rm IL}
551: \end{array}
552: \label{eq:conn2}\ee
553:
554: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
555: \subsection{Solution Around Corotation Radius}
556:
557: In the vicinity of the corotation radius $r_c$, we can rewrite \eqref{perturbeq2} dropping the $m^2/r^2$ and $S$ terms compared to the singular term proportional to $1/\tomega$, giving
558: \be
559: \left[\frac{d^2}{dr^2} - \tilde{k}^2
560: +{2\over q}\left({d\over dr}\ln{\kappa^2\over\Omega\Sigma}\right)_c
561: {1\over r-R_c}\right]\eta = 0,
562: \label{pnearcr}\ee
563: where
564: \be
565: \tilde{k}^2 \equiv \frac{D}{c^2},\quad
566: q\equiv -\left({d\ln \Omega\over d\ln r}\right)_c,\quad
567: R_c\equiv r_c-i{r_c\omega_i\over q\omega_r}.
568: \ee
569: Here we have introduced a small imaginary part to the wave frequency, so that
570: $\omega=\omega_r+i\omega_i$. To study the response of the disc to a slowly
571: increasing perturbation, we require $\omega_i>0$.
572: Defining
573: \be
574: x = \int_{r_{c}}^r 2\tilde{k}\,dr, \quad \eta = \frac{1}{\sqrt{\tilde{k}}}\psi,
575: \ee
576: and recognizing that $\tilde k$ can be treated as a constant around corotation,
577: we have
578: \be
579: \frac{d^2}{dx^2} \psi + \left(
580: -\frac{1}{4} + \frac{\nu}{x+i\epsilon}\right)\psi =0
581: \label{Whittakerdiffeq} \ee
582: where $\epsilon=2{\tilde k}r_c\omega_i/(q\omega_r)$ and
583: \be
584: \nu={1\over q\tilde k}\left({d\over dr}\ln{\kappa^2\over\Omega\Sigma}\right)_c
585: =\left({c\over q\kappa}{d\over dr}\ln\zeta\right)_c. \label{eq:nu}
586: \ee
587: In equation
588: (\ref{Whittakerdiffeq}) $\epsilon > 0$, consistent with the initial value
589: problem in which the perturbation is gradually turned on starting
590: from $t=-\infty$. The parameter $\nu$ determines the width of the Rossby wave region,
591: $\Delta r_R = (2c/\kappa)|\nu|$. For $d\ln\zeta/dr\sim 1/r$, we have
592: $|\nu|\sim (H/r)_c$ ($H$ is the disc scale height). \footnote{Note that the WKB wavenumber $k_{r}$ in the Rossby region ranges from $\infty$ (at $r=r_c$) to $H^{-1}$ (at $r\sim r_c \pm \Delta r_R$) [see equation (18)] while the width of the region is of order $H^2/r$ for most Keplerian discs. Free Rossby waves tend to be sheared away by differential rotation (Tagger 2001).}
593:
594: Equation (\ref{Whittakerdiffeq}) is the differential equation for the
595: Whittaker function in complex variable $z = x +
596: i\epsilon$ with index $1/2$ (Abramowitz \& Stegun 1964).
597: The two linearly independent solutions convenient for the connection are
598: \be
599: \psi_- = \textrm{W}_{\nu,1/2} (z), \qquad
600: \psi_+ = e^{-i\pi\nu}\textrm{W}_{-\nu, 1/2} (ze^{-i\pi}) +
601: \frac{1}{2} T_0 \textrm{W}_{\nu,1/2} (z),
602: \label{eq:psi}\ee
603: where $T_0$ is the Stokes multiplier defined
604: below, and $z$
605: is defined in the complex plane so that $\arg(z)$ ranges from $0$ to $\pi$.
606: The particular linear combinations of Whittaker functions in (\ref{eq:psi})
607: are chosen so that appropriate asymptotic expansions can be obtained.
608: To obtain these asymptotic expansions and the
609: connection formulae around the corotation, one must carefully consider
610: the Stokes phenomenon, which alters the form of the asymptotic expansion
611: of a function depending on the position of $z$ in the complex
612: plane. Since the appropriate asymptotic expansions (which connect
613: the solution $W_{\pm\nu,1/2}$ analytically in different regions of the
614: complex plane) are not readily available, we relegate the discussion of the
615: Stokes phenomenon for the Whittaker function around the corotation
616: to Appendix A\footnote{The Stokes phenomenon is present
617: in our Lindblad resonance analysis as well, as we see, for example,
618: the asymptotic expansion for $Bi'(-(3z/2)^{2/3})$ takes the form
619: $e^{+|z|}$ for $\arg(z) = 3\pi/2$ but $\sin(z + \pi/4)$
620: (as opposed to $e^{iz}$) for $\arg(z) =0$.
621: The resulting connection formulae are well known for the Airy functions.}.
622: The resulting connection formulae are
623: \ba
624: %= \sqrt{\tilde k} \psi_-
625: \delta h_- &\sim&
626: \Biggl\{\begin{array}{ll}
627: \sqrt{S/k}\, \exp\left(-\int_{r_c}^{r}\! \tilde k \,dr \right)
628: & \qquad \qquad \qquad ~~\textrm{for } r \gg r_{c}\\
629: \sqrt{S/k}\, e^{i\pi\nu}\exp\left(+\int_r^{r_{c}}\! \tilde k \,dr \right)
630: + \sqrt{S/k}\, \frac{T_1}{2} e^{-i\pi\nu} \exp\left(-\int_r^{r_{c}}\! \tilde k \,dr \right)
631: &\qquad \qquad \qquad~~\textrm{for } r\ll r_{c}~.
632: \end{array}\label{whitaker1}\\
633: %= \sqrt{\tilde k}\psi_+
634: \delta h_+ &\sim&
635: \Biggl\{\begin{array}{ll}
636: \sqrt{S/k}\, \exp\left(+\int_{r_c}^{r}\! \tilde k \,dr \right)
637: & \qquad \textrm{for } r \gg r_{c}\\
638: \sqrt{S/k}\, \frac{T_0}{2} e^{i\pi\nu}\exp\left(+\int_r^{r_{c}}\! \tilde k \,dr \right)
639: + \sqrt{S/k}\, \left(1 + \frac{T_1T_0}{4}\right) e^{-i\pi\nu}\exp\left(-\int_r^{r_{c}}\! \tilde k \,dr \right)
640: &\qquad \textrm{for } r\ll r_{c}
641: \end{array}\label{whitaker2}
642: \ea
643: where $T_0$ and $T_1$ are the Stokes multipliers,
644: \be
645: T_0={2\pi i\over\Gamma(\nu)\Gamma(1+\nu)},\qquad
646: T_1={2\pi i\,e^{i2\pi\nu}\over\Gamma(-\nu)\Gamma(1-\nu)}.
647: \ee
648: Note that for $|\nu| \ll 1$,
649: $[\Gamma(\pm \nu)]^{-1}=\pm\nu +\gamma\nu^2+\cdots$,
650: where $\gamma=0.5772$ is the Euler constant.
651:
652: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
653: \section{Global WKB Solutions and Calculation of Reflectivity}
654:
655: In this section we consider a wave train which approaches corotation
656: from small radii ($r\ll r_{\rm IL}$). Its propagation is impeded
657: by the potential barrier between $r_{\rm IL}$ and $r_{\rm OL}$.
658: The incident wave is partially transmitted beyond the OLR and a reflected
659: wave propagates from the ILR toward small radii.
660: We will derive the explicit expressions for the (complex) reflection
661: coefficient ${\cal R}$ and transmission coefficient ${\cal T}$.
662:
663: From the dispersion relation [equation (\ref{eq:disp}), or equation
664: (\ref{eq:disper})
665: away from corotation], we find that the radial group velocity of the waves is
666: \be
667: c_g={d\omega\over dk_r}={k_rc^2\over\tomega (1-\kappa^2/\tomega^2)}.
668: \ee
669: Thus the sign of $c_g/c_p$ (where $c_p=\omega/k_r$ is the phase velocity)
670: is positive for $r>r_{\rm OL}$ and negative for $r<r_{\rm IL}$.
671: This implies that in the $r>r_{\rm OL}$ region, the outgoing (transmitted)
672: wave has the form $\exp(i\int^r k_r\,dr)$ (assuming $k_r>0$).
673: In the $r<r_{\rm IL}$ region, the incident wave (propagating from small
674: radii toward corotation) has the form $\exp(-i\int^r k_r\,dr)$, while
675: the reflected wave has the form $\exp(i\int^r k_r\,dr)$.
676:
677: A well-known property of density waves is that for $r<r_{\rm IL}$
678: the wave carries negative energy (or angular momentum),
679: while for $r>r_{\rm OL}$
680: the wave carries positive energy. An incident
681: wave $\exp(-i\int^r k_r\,dr)$, carrying energy of the amount $(-1)$,
682: will give rise to a reflected wave ${\cal R}\exp(i\int^r k_r\,dr)$
683: and a transmitted wave ${\cal T}\exp(i\int^r k_r\,dr)$. Let ${\cal D}_c$
684: be the energy dissipated at the corotation. Then energy conservation gives
685: $-1=(-1)|{\cal R}|^2+|{\cal T}|^2+{\cal D}_c$, or
686: \be
687: |{\cal R}|^2=1+|{\cal T}|^2+{\cal D}_c. \label{eq:Econs}
688: \ee
689: Because of the singularity at corotation and the associated
690: energy absorption, we first consider the simple case where
691: the corotation singularity is neglected (section 4.1)
692: before examining the general case (section 4.2).
693:
694:
695: \subsection{Neglecting Corotation Singularity: Corotation Amplifier}
696:
697: Here we consider the case where the vortensity has zero slope
698: at corotation, i.e., $(d\zeta/dr)_c=0$. This would occur for disc models
699: where the specific angular momentum $r^2\Omega$ is constant (as in the
700: original Papaloizou \& Pringle 1984 analysis), or for shearing sheet
701: approximation (as in Narayan et al.~1987). In this case, there is no
702: corotation singularity and no absorption of wave energy, and we always
703: obtain super-reflection
704: \be
705: |{\cal R}|^2=1+|{\cal T}|^2>1.
706: \ee
707: This is the essence of the corotation amplifier.
708:
709: To derive the complex ${\cal R}$ and ${\cal T}$, we assume
710: that the outgoing wave in the $r> r_{\rm OL}$ region
711: (region III in Fig.\ref{fig1}) is given by [see
712: eqs.~(\ref{eq:con1})-(\ref{eq:con2})]
713: \be
714: \delta h =\sqrt{S/k}\exp\left( i \int_{r_{\rm OL}}^r k\,dr
715: + \frac{\pi}{4}\right),
716: \label{eq:outside}\ee
717: where $k^2\equiv -D/c^2$.
718: The connection formulae (\ref{eq:con1})-(\ref{eq:con2})
719: then give for the evanescent zone (region II in Fig.~1):
720: \ba
721: \delta h &\simeq& \frac{\sqrt{S/k}}{2}\exp\left(-\int_r^{r_{\rm OL}}|k|\,dr
722: \right) + i\sqrt{S/k}\exp\left(\int_r^{r_{\rm OL}} |k| dr\right) \nonumber\\
723: &=& \frac{\sqrt{S/k}}{2}
724: \exp(-\Theta_{\rm{II}})
725: \exp\left(\int_{r_{\rm IL}}^{r}|k| dr \right) + i\sqrt{S/k}
726: \exp(+\Theta_{\rm{II}})\exp\left(-\int_{r_{\rm IL}}^r |k| dr\right)
727: \ea
728: where
729: \be
730: \Theta_{\rm{II}} = \int_{r_{\rm IL}}^{r_{\rm OL}} |k|\, dr
731: = \int_{r_{\rm IL}}^{r_{\rm OL}} {\sqrt{|D|}\over c}\, dr.
732: \ee
733: Using the connection formulae at ILR [eqs.~(\ref{eq:conn1})-(\ref{eq:conn2})],
734: we find that for $r < r_{\rm IL}$ (region I in Fig.~1)
735: \be
736: \delta h \simeq \frac{\sqrt{S/k}}{2}e^{-\Theta_{\rm{II}}}
737: \sin\left(\int_r^{r_{\rm IL}} k\,dr +\frac{\pi}{4}\right) + i2\sqrt{S/k}
738: e^{\Theta_{\rm{II}}}\cos\left(\int_r^{r_{\rm IL}} k\,dr
739: +\frac{\pi}{4}\right)~.
740: \ee
741: Expressing this in terms of traveling waves, and defining
742: $y = \int_{r_{\rm IL}}^r k dr - \pi/4$ we have
743: \be
744: \delta h \simeq i\sqrt{S/k} \left[ e^{-iy} \left(e^{+\Theta_{\rm{II}}}
745: - \frac{1}{4}e^{-\Theta_{\rm{II}}}\right) + e^{+iy} \left(
746: e^{\Theta_{\rm{II}}} + \frac{1}{4}e^{-\Theta_{\rm{II}}}\right)\right].
747: \label{eq:inside}\ee
748: Thus the reflection coefficient is
749: \be
750: {\cal R} = \frac{e^{\Theta_{\rm{II}}} + \frac{1}{4}e^{-\Theta_{\rm{II}}}}{e^{\Theta_{\rm{II}}} -
751: \frac{1}{4}e^{-\Theta_{\rm{II}}}}.\label{SimpleReflection}
752: \ee
753: Comparing equation (\ref{eq:inside}) with equation (\ref{eq:outside}),
754: we obtain the transmission coefficient
755: \be
756: {\cal T}={-i\over e^{+\Theta_{\rm II}}-{1\over 4}e^{-\Theta_{\rm II}}}.
757: \label{Simpletrans}\ee
758: As expected, $|{\cal R}|^2=1+|{\cal T}|^2>1$.
759:
760:
761: \subsection{Including Corotation Singularity}
762:
763: As noted before, for discs with nonzero vortensity gradient
764: ($d\zeta/dr\ne 0$), the singularity at corotation implies the absorption of
765: wave energy (or angular momentum). Similar situations
766: occur in geophysical wave systems (Dickenson 1968) and for waves in plasmas
767: (Landau damping).
768: In Appendix B, we discuss the toy problem of resonant tunneling which
769: shares the similar energy absorption feature as the corotation singularity.
770: Previous works on global modes in disc tori
771: (e.g. Papaloizou \& Pringle 1987; Goldreich et al.~1986; Narayan et al.~1987)
772: suggest that the sign of $d\zeta/dr$ determines whether the singularity
773: acts to stabilize or destabilize a global mode.
774: Here we derive the explicit expression for
775: the reflectivity and the related source term ${\cal D}_c$.
776:
777: As in section 4.1, we assume an outgoing wave in Region III,
778: and the connection formulae at the OLR then give for Region IIb of
779: Fig.~\ref{fig2} or Fig.~\ref{fig3}
780: \be
781: \delta h \simeq
782: \frac{\sqrt{S/k}}{2}\exp(-\Theta_{\rm{IIb}})\exp\left(\int_{r_{c}}^r
783: |k|\,dr\right) + i\sqrt{S/k}\exp(+\Theta_{\rm{IIb}})\exp
784: \left(-\int_{r_{c}}^r |k|\,dr\right),
785: \label{eq:dhc}\ee
786: where
787: \be
788: \Theta_{\rm{IIb}} = \int_{r_{c}}^{r_{\rm OL}} |k|\,dr.
789: \ee
790: Equation (\ref{eq:dhc}) is the asymptotic solution away from the $r_c$ in
791: region IIb. The corresponding general solution around $r_c$ is
792: \be
793: \delta h = \frac{\sqrt{S/k}}{2}\exp(-\Theta_{\rm IIb})\psi_+(r) + i\sqrt{S/k}\exp(\Theta_{\rm IIb}) \psi_-(r),
794: \label{eq:general}\ee
795: where $\psi_+$ and $\psi_-$ are given by (\ref{eq:psi}).
796: Using equations \eqref{whitaker1} and \eqref{whitaker2} to match asymptotes
797: over the corotation singularity, we obtain in Region $\rm{IIa}$
798: \ba
799: \delta h &\simeq& \left[\frac{1}{2}\exp(-\Theta_{\rm{IIb}})(1+\frac{1}{4} T_0 T_1)
800: +\frac{i}{2} T_1
801: \exp(+\Theta_{\rm{IIb}})
802: \right]\sqrt{S/k}~e^{-i\pi\nu}\exp\left(-\int_r^{r_{c}}\!|k|\,dr \right)\nonumber\\
803: &\qquad& +i\Bigl[
804: \exp(+\Theta_{\rm{IIb}})
805: -\frac{i}{4} T_0 \exp(-\Theta_{\rm IIb})\Bigr]
806: \sqrt{S/k} ~ e^{i\pi\nu}\exp\left(\int_r^{r_{c}}|k|\,dr\right).
807: \ea
808: Using the connection formulae at the ILR, we have for Region I:
809: \ba
810: \delta h &\simeq & \left[\frac{1}{2}\exp(-\Theta_{\rm{II}})
811: \left(1+\frac{1}{4} T_0 T_1\right)
812: + \frac{i}{2} T_1\exp(\Theta_{\rm{IIb}}-\Theta_{\rm{IIa}}) \right]
813: \sqrt{S/k}~e^{-i\pi\nu}\sin\left(\!\int_r^{r_{\rm IL}}\!\!k\,dr
814: +\frac{\pi}{4}\right)\nonumber\\
815: &\quad & + 2i \left[ \exp(+\Theta_{\rm{II}})
816: -\frac{i}{4} T_0 \exp(\Theta_{\rm{IIa}}
817: -\Theta_{\rm{IIb}}) \right]\sqrt{S/k}~e^{i\pi\nu}\cos\left(\!\int_r^{r_{\rm IL}}\!\!k\,dr
818: +\frac{\pi}{4}\right)\nonumber\\
819: &=& \left[1+{1\over 4}\,e^{-i2\pi\nu}e^{-2\Theta_{\rm II}}\left(1+\frac{1}{4} T_0 T_1\right)
820: +{i\over 4} T_1\,e^{-i2\pi\nu}e^{-2\Theta_{\rm IIa}}
821: -{i\over 4} T_0 e^{-2\Theta_{\rm IIb}}\right]i\sqrt{S/k}\,
822: e^{\Theta_{\rm II}}e^{i\pi\nu}e^{iy}\nonumber\\
823: & \quad & +\left[1-{1\over 4}\,e^{-i2\pi\nu}e^{-2\Theta_{\rm II}}\left(1+{1\over 4} T_0 T_1\right)
824: -{i\over 4} T_1 e^{-i2\pi\nu}e^{-2\Theta_{\rm IIa}}
825: -{i\over 4} T_0 e^{-2\Theta_{\rm IIb}}\right]i\sqrt{S/k}\,
826: e^{\Theta_{\rm II}} e^{i\pi\nu} e^{-iy},
827: \ea
828: where $y = \int_{r_{\rm IL}}^r k dr - \pi/4$ and
829: \be
830: \Theta_{\rm{IIa}} = \int_{r_{\rm IL}}^{r_{c}}\!\!|k|\,dr,\qquad
831: \Theta_{\rm II}=\Theta_{\rm{IIa}}+\Theta_{\rm{IIb}}
832: =\int_{r_{\rm IL}}^{r_{\rm OL}}\!\!|k|\,dr.
833: \ee
834: The reflection coefficient and transmission coefficient are then
835: \ba
836: &&{\cal R}={ 1+{1\over 4}\,e^{-i2\pi\nu}e^{-2\Theta_{\rm II}}\left(1+{1 \over 4} T_0 T_1\right)
837: +{i\over 4} T_1 e^{-i2\pi\nu}e^{-2\Theta_{\rm IIa}}
838: -{i\over 4} T_0e^{-2\Theta_{\rm IIb}}
839: \over
840: 1-{1\over 4}\,e^{-i2\pi\nu}e^{-2\Theta_{\rm II}}\left(1+{1 \over 4} T_0 T_1\right)
841: -{i\over 4} T_1 e^{-i2\pi\nu}e^{-2\Theta_{\rm IIa}}
842: -{i\over 4} T_0 e^{-2\Theta_{\rm IIb}}},\nonumber\\
843: && \quad = { 1 + {1 \over 4} \left(e^{-i 2\pi \nu} + \sin^2 \pi \nu \right)e^{-2\Theta_{\rm II}}
844: + {\pi \nu \over 2} {e^{-2\Theta_{\rm IIa}}\over (\Gamma(1-\nu))^2}
845: - {\pi \nu \over 2} {e^{-2\Theta_{\rm IIb}}\over (\Gamma(1 + \nu))^2}
846: \over
847: 1 - {1 \over 4} \left(e^{-i 2\pi \nu} + \sin^2 \pi \nu \right)e^{-2\Theta_{\rm II}}
848: - {\pi \nu \over 2} {e^{-2\Theta_{\rm IIa}}\over (\Gamma(1-\nu))^2}
849: -{\pi \nu \over 2} {e^{-2\Theta_{\rm IIb}}\over (\Gamma(1 + \nu))^2}
850: }
851: \label{eq:RwithGamma}\\
852: &&{\cal T}={-i\, e^{-\Theta_{\rm II}} e^{i\pi\nu}
853: \over
854: 1-{1\over 4}\,e^{-i2\pi\nu}e^{-2\Theta_{\rm II}}\left(1+{1 \over 4} T_0 T_1\right)
855: -{i\over 4} T_1 e^{-i2\pi\nu} e^{-2\Theta_{\rm IIa}}
856: -{i\over 4}T_0 e^{-2\Theta_{\rm IIb}}}\nonumber\\
857: &&\quad =
858: {- i e^{-\Theta_{\rm II}}e^{i\pi\nu}
859: \over
860: 1 - {1 \over 4} \left(e^{-i 2\pi \nu} + \sin^2 \pi \nu \right)e^{-2\Theta_{\rm II}}
861: - {\pi \nu \over 2} {e^{-2\Theta_{\rm IIa}}\over (\Gamma(1-\nu))^2}
862: -{\pi \nu \over 2} {e^{-2\Theta_{\rm IIb}}\over (\Gamma(1 + \nu))^2}
863: }
864: \label{eq:trans}\ea
865: The dissipation term due to the corotation singularity obtained from
866: ${\cal D}_c=|{\cal R}|^2 - 1- |{\cal T}|^2$ is
867: \be
868: {\cal D}_c = {
869: {\pi \nu \over 2} {\cos^2\pi\nu\over (\Gamma(1 + \nu))^2} e^{-2\Theta_{\rm II} - 2\Theta_{\rm IIb}}
870: + {2\pi \nu \over (\Gamma(1-\nu))^2} e^{-2\Theta_{\rm IIa}}
871: \over |1 - {1 \over 4} \left(e^{-i 2\pi \nu} + \sin^2 \pi \nu \right)e^{-2\Theta_{\rm II}}
872: - {\pi \nu \over 2} {e^{-2\Theta_{\rm IIa}}\over (\Gamma(1-\nu))^2}
873: -{\pi \nu \over 2} {e^{-2\Theta_{\rm IIb}}\over (\Gamma(1 + \nu))^2}
874: |^2} .
875: \ee
876:
877: For $|\nu| \ll 1$
878: equations (\ref{eq:RwithGamma}) and (\ref{eq:trans}) can be simplified,
879: and we have
880: \ba
881: &&{\cal R}\rightarrow { e^{\Theta_{\rm II}} + {1 \over 4} e^{-\Theta_{\rm II}}
882: \over { e^{\Theta_{\rm II}} - {1 \over 4} e^{-\Theta_{\rm II}}}} + { e^{+2\Theta_{\rm IIb}}-{1\over 4} e^{-2\Theta_{\rm IIb}} \over \left(e^{\Theta_{\rm II}} - {1 \over 4} e^{-\Theta_{\rm II}}\right)^2} \pi \nu + {\cal O}[\nu^2] \label{eq:R},\label{eq:eqr}\\
883: &&{\cal T}\rightarrow { -i
884: \over { e^{\Theta_{\rm II}} - {1 \over 4} e^{-\Theta_{\rm II}}}}
885: - {i \over 2} { e^{\Theta_{\rm IIb}-\Theta_{\rm IIa}}-e^{\Theta_{\rm IIa}-\Theta_{\rm IIb}} \over \left(e^{\Theta_{\rm II}} - {1 \over 4} e^{-\Theta_{\rm II}}\right)^2} \pi \nu + {\cal O}[\nu^2],\label{eq:eqt}\\
886: &&{\cal D}_c\rightarrow {2 \left( e^{\Theta_{\rm II}} + {1\over 4} e^{-\Theta_{\rm II}}\right)
887: \left( e^{2\Theta_{\rm IIb}}-{1 \over 4} e^{-2\Theta_{\rm IIb}}\right)
888: + \left( e^{\Theta_{\rm IIb}-\Theta_{\rm IIa}}-e^{\Theta_{\rm IIa}-\Theta_{\rm IIb}}\right)
889: \over \left(e^{\Theta_{\rm II}} - {1 \over 4} e^{-\Theta_{\rm II}}\right)^3}
890: \pi \nu + {\cal O}[\nu^2]. \label{eq:S}
891: \ea
892: Clearly, for $\nu=0$, equations (\ref{eq:eqr})-(\ref{eq:eqt}) reduce to
893: (\ref{SimpleReflection}) and (\ref{Simpletrans}).
894:
895: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
896: \section{Wave Damping at Corotation: Alternative Calculation}
897:
898: In the previous section we obtained the expression for the
899: dissipation term ${\cal D}_c$ at the corotation resonance using
900: the reflection and transmission coefficients. Here we provide
901: a more direct derivation of ${\cal D}_c$ using the change of angular momentum
902: flux across the corotation radius.
903:
904: In the absence of self-gravity, the angular momentum
905: flux carried by the the waves in the disc is entirely due to advection.
906: The time-averaged transfer rate of the $z$-component of angular momentum
907: across a cylinder of radius $r$ (in the outward direction) is given by
908: (e.g. Goldreich \& Tremaine 1979)
909: \be
910: F(r)=r^2\Sigma(r)\int_0^{2\pi}\!d\phi\,{\rm Re}[\delta u_r(r,\phi,t)]
911: {\rm Re}[\delta u_\phi(r,\phi,t)].
912: \ee
913: Using equations (\ref{eq:ur})-(\ref{eq:uphi}) to express $\delta u_r$
914: and $\delta u_\phi$ in terms of $\delta h$, this reduces to
915: (see Tanaka et al.~2002; Zhang \& Lai 2006)
916: \be
917: F(r) = \frac{\pi m r \Sigma}{D}{\rm Im}\left(\delta h\frac{d\delta
918: h^*}{dr} \right) .
919: \ee
920:
921: In Region III (see Figs.~1-3) the outgoing wave has the enthalpy
922: perturbation given by (up to a proportional constant)
923: \be
924: \delta h = \sqrt{S/k}\, {\cal T} \exp \left( i \int^r_{r_{\rm OL}} k dr
925: +{\pi\over 4} \right)
926: \label{eq:ddh}\ee
927: Calculating the angular momentum flux (setting the incoming wave flux to 1),
928: \be
929: F(r\gg r_{\rm OL})\simeq \pi m |{\cal T}|^2,
930: \ee
931: we see that angular momentum is transferred outwards (positive flux)
932: since waves in $r>r_{\rm OL}$ carries positive angular momentum.
933: For Region I we have [up to the same proportional constant as in
934: (\ref{eq:ddh})]
935: \be
936: \delta h = \sqrt{S/k}\left[\exp \left( - i \int^r_{r_{\rm IL}} k dr
937: +{\pi\over 4}\right) + {\cal R} \exp \left( i \int^r_{r_{\rm IL}} k dr
938: -{\pi\over 4}\right) \right],
939: \ee
940: which gives the angular momentum flux:
941: \be
942: F(r\ll r_{\rm IL}) \simeq \pi m (|{\cal R}|^2 - 1).
943: \ee
944: We see that the incident wave carries negative angular momentum outward,
945: and the reflected wave transfers positive angular momentum. The net angular
946: momentum transfer is positive (in the outward direction) for $|{\cal R}|>1$.
947:
948: Now consider the angular flux near the corotation radius, at $r=r_c^-$
949: (just inside corotation ) and at $r=r_c^+$ (just outside corotation).
950: Integrating equation~\eqref{pnearcr} across the singularity,
951: we find the discontinuity in the enthalpy perturbation derivatives:
952: \be
953: \frac{d\,\delta h}{dr}\bigg|_{r_c^+} - \frac{d\,\delta
954: h}{dr}\bigg|_{r_c^-} = {2\pi i\over q}\left({d\over dr}\ln\zeta\right)\delta h
955: \bigg|_{r_c}=2\pi\nu i {\kappa\over c}\delta h\bigg|_{r_c}.
956: \ee
957: Here we have chosen to integrate from $r_c^-$ to $r_c^+$ by going through the
958: upper complex plane, to be consistent with the physical requirement of
959: a gradually growing perturbation, turned on at $t = -\infty$. Thus the change in the angular momentum
960: flux across the corotation is
961: \be
962: \Delta F_c=F(r_c^+)-F(r_c^-)
963: = -\frac{2\pi^2 m r \Sigma\nu\kappa }{cD} |\delta h|^2\bigg|_{r_c}.
964: \ee
965: The wavefunction around $r_c$ is given by equation (\ref{eq:general})
966: multiplied by ${\cal T}$.
967: Noting that
968: \ba
969: \psi_-(r_c) &=& {\rm W}_{\nu, 1/2}(0) = \frac{1}{\Gamma(1-\nu)},\\
970: \psi_+(r_c) &=& e^{-i\pi\nu}\textrm{W}_{-\nu, 1/2} (0) + \frac{1}{2} T_0 \textrm{W}_{\nu,1/2} (0) = \frac{\cos \pi\nu}{\Gamma(1 + \nu)} ,
971: \ea
972: where we have used the convenient identity $\Gamma(\nu)\Gamma(1 - \nu) =
973: \pi/\sin\pi\nu$. We can evaluate the enthalpy perturbation at the
974: corotation, giving
975: \be
976: \delta h(r_c) = \sqrt{S/k}\,{\cal T}\left[ {1\over 2} e^{-\Theta_{\rm IIb}} {\cos \pi \nu \over \Gamma(1 + \nu)} + i e^{\Theta_{\rm IIb}}{1 \over \Gamma(1-\nu)}\right].
977: \ee
978: The change in angular momentum flux across the corotation is then
979: \be
980: \Delta F_c= -\pi m |{\cal T}|^2\nu \left[
981: {\pi \over 2} {\cos^2\pi\nu \over (\Gamma(1 + \nu))^2}e^{-2\Theta_{\rm IIb}}
982: + {2\pi \over (\Gamma(1-\nu))^2} e^{2\Theta_{\rm IIb}}
983: \right].
984: \ee
985: With $F(r\ll r_{\rm IL})=F(r_c^-)$, $F(r\gg r_{\rm OL})=F(r_c^+)$,
986: and thus $F(r\ll r_{\rm IL})=F(r\gg r_{\rm OL})-\Delta F_c$,
987: we find
988: \be
989: {\cal D}_c = |{\cal T}|^2 \nu \left[
990: {\pi \over 2} {\cos^2\pi\nu \over (\Gamma(1 + \nu))^2}e^{-2\Theta_{\rm IIb}}
991: + {2\pi \over (\Gamma(1-\nu))^2} e^{2\Theta_{\rm IIb}}
992: \right].
993: \label{eq:scc}\ee
994: This expression exactly agrees with ${\cal D}_c$ given in section 4.2.
995:
996: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
997: \section{Results and Discussion}
998:
999: The key new results of this paper consist of the analytical
1000: expressions for the reflection coefficient ${\cal R}$, transmission
1001: coefficient ${\cal T}$ and the dissipation term ${\cal D}_c$
1002: when a wave impinges upon the corotation barrier from small radii.
1003: These expressions, (\ref{eq:RwithGamma})-(\ref{eq:S}) and (\ref{eq:scc}),
1004: can be applied to discs with generic rotation and surface density profiles.
1005:
1006: \begin{figure}
1007: \centering
1008: \includegraphics[width=13cm]{Rvc1}
1009: {\vskip -1cm}
1010: \caption{
1011: The reflection coefficient as a function of $\beta=c/(r\Omega)$ for
1012: Keplerian discs with surface density profile $\Sigma\propto r^{-p}$.
1013: Note that for $p=1.5$, wave absorption at the corotation resonance is
1014: absent.
1015: }
1016: \label{fig4}
1017: \end{figure}
1018:
1019: \begin{figure}
1020: \centering
1021: \includegraphics[width=13cm]{RTS2}
1022: \caption{The reflection coefficient, wave damping coefficient and
1023: transmission coefficient as a function of $\nu$ for two different values
1024: of $\beta=c/(r\Omega)$. Note that for $\nu=0$ (no corotation resonance),
1025: wave damping is zero (${\cal D}_c=0$)
1026: and $|{\cal R}|^2-1$ is positive.
1027: }
1028: \label{fig5}
1029: \end{figure}
1030:
1031:
1032: For definiteness, here we illustrate our results using a (Newtonian)
1033: Keplerian disc model with
1034: \be
1035: \Omega=\kappa\propto r^{-3/2},\quad
1036: \Sigma\propto r^{-p},\quad
1037: {c\over r\Omega}=\beta,
1038: \ee
1039: where $p$ and $\beta$ are constants. The important parameter that
1040: determines the behavior of the corotation singularity is [see eq.~
1041: (\ref{eq:nu})]
1042: \be
1043: \nu=\left({2c\over 3\kappa}{d\over dr}\ln\zeta\right)_c=
1044: {2\over 3}\beta\left(p-{3\over 2}\right).
1045: \ee
1046: Clearly the models are scale-free, and ${\cal R}$, ${\cal T}$
1047: and ${\cal D}_c$ depend only on the two parameters $p$ and $\beta$
1048: (or $\nu$). Figure 4 depicts $|{\cal R}|$ as a function of $\beta$ for
1049: different values of $p$, while Figure 5 shows $|{\cal R}|,~{\cal D}_c$
1050: and $|{\cal T}|$ as a function of $\nu$ for $\beta=0.05$ and $\beta=0.1$.
1051:
1052: A key result of paper is that wave absorption at the corotation resonance
1053: plays an important role in determining the
1054: reflection and transmission of waves across the corotation barrier.
1055: Without corotation resonance (as for discs with zero
1056: vortensity gradient, or $\nu=0$), super-reflection is
1057: always achieved, but $|{\cal R}|^2-1\simeq \exp(-2\Theta_{\rm II})$
1058: (assuming $\Theta_{\rm II}\gg 1$, where
1059: $\Theta_{\rm II}=\Theta_{\rm IIa}+\Theta_{\rm IIb}=
1060: \int_{r_{\rm IL}}^{r_{\rm OL}}\,|k|\,dr$, with $|k|=|D|^{1/2}/c$)
1061: is rather small. In the presence of
1062: wave absorption at the corotation resonance (when $\nu\neq 0$),
1063: we find (assuming $\Theta_{\rm IIa}\gg 1$ and $|\nu|\ll 1$),
1064: \be
1065: |{\cal R}|^2-1\simeq \exp(-2\Theta_{\rm II})+2\pi\nu\exp(-2\Theta_{\rm IIa}),
1066: \ee
1067: and super-reflection can be much more prominent. From Fig.~5 we see that
1068: the transmission coefficient is generally much smaller than ${\cal D}_c$.
1069: Thus the reflectivity depends mainly on the wave damping at corotation.
1070:
1071: Equation (\ref{eq:scc}) or (\ref{eq:S}) clearly
1072: shows that for $\nu>0$, the wave damping term due to corotation resonance
1073: is positive, ${\cal D}_c > 0$.
1074: This can be understood from the fact that for $\nu>0$ the Rossby wave
1075: region lies outside $r_c$ (see Fig.~\ref{fig3}), and the dissipation
1076: at the corotation singularity carries away positive energy (just like the
1077: transmitted wave) so that energy conservation [see
1078: eq.~(\ref{eq:Econs})] requires $-1 = -|{\cal R}|^2 + |{\cal T}|^2 +
1079: |{\cal D}_c|$. In this case the corotation singularity enhances
1080: super-reflection, as seen in Figs.~4-5.
1081: On the other hand, for $\nu < 0$, the Rossby wave zone lies inside
1082: $r_c$ (Fig.~\ref{fig2}) and the dissipation carries away negative
1083: energy (like the reflected wave) so that $-1 = -|{\cal R}|^2 + |{\cal
1084: T}|^2 - |{\cal D}_c|$. In this case the corotation singularity tends
1085: to reduce super-reflection, and there is a competition between the
1086: effect of the corotation amplifier [the first term in eq.~(\ref{eq:R})]
1087: and the effect of the corotational absorption [the second term]. To
1088: obtain $|{\cal R}| > 1$ we require
1089: \be
1090: \nu > -{1 \over 2\pi} {1 - {1 \over 4}e^{-2\Theta_{\rm II}} \over
1091: e^{2\Theta_{\rm IIb}} - {1\over 4} e^{-2\Theta_{\rm IIb}}} \simeq -{1
1092: \over 2\pi} e^{-2\Theta_{\rm IIb}},
1093: \label{eq:ineq}\ee
1094: where the second inequality applies in the limit of $\Theta_{\rm IIb}
1095: \gg 1$. This puts a constraint on the disc thickness and the specific
1096: vorticity slope (note that the sound speed $c$ enters into both $\nu$ and
1097: $\Theta_{\rm IIb}$) in order to achieve super-reflection.
1098: For example, for a given $c$, the inequality (\ref{eq:ineq}) determines
1099: the critical value of $p$ for which $|{\cal R}|=1$.
1100:
1101: Figures 4-5 also reveal an intriguing oscillatory behavior of the reflection,
1102: transmission and damping coefficients.
1103: This non-monotonic behavior may be qualitatively
1104: understood from the Rossby wave zone
1105: around $r_c$ (see Figs.~2-3). The WKB wavenumber $k_r$ in the near
1106: vicinity of $r_c$ is given by
1107: \be
1108: k_r^2\simeq -{\kappa^2\over c^2}+{2m\Omega\over r\tomega}{d\over dr}\ln
1109: \zeta={\kappa^2\over c^2}\left(-1+{2\beta\nu r\over r-r_c}\right).
1110: \ee
1111: For $\nu>0$, the Rossby wave zone lies between $r_c$ and $r_c+\Delta r_R$,
1112: where $\Delta r_R=2\beta\nu r_c$.
1113: For quasi-normal modes to be ``trapped in" the Rossby wave zone they must obey the
1114: the Sommerfeld ``quantization'' condition\footnote{The phase factor $\pi/2$
1115: arises from a detailed analysis of the wave behavior at $r=r_c$
1116: and at $r=r_c+\Delta r_R$: The former gives a phase of $-3\pi/4$ and the latter
1117: gives $\pi/4$.}
1118: $\int_{r_c}^{r_c+\Delta r_R}\!\!dr\, k_r=\pi \nu\sim n\pi+{\pi/2}$,
1119: where $n=0,1,2,\cdots$. Thus when $\nu\simeq n+1/2$, the wave propagating in
1120: the Rossby zone (Fig.~3)
1121: is maximally reflected back to the singularity, leading to
1122: maximum negative damping and enhanced net reflection,
1123: as seen in Fig.~\ref{fig5}.
1124: For $\nu<0$ the Rossby wave zone lies between $r_c-|\Delta r_R|$
1125: and $r_c$ (Fig.~2), the wave in the Rossby zone is mostly
1126: absorbed at the corotation singularity. The asymmetry in the
1127: $\nu>0$ case and the $\nu<0$ case can also be seen in the toy problem of
1128: resonant tunneling (Appendix B). Note that for thin Keplerian disks, $|\nu|$ is
1129: much less than unity for reasonable density profiles ($|p|\sim 1$),
1130: so this non-monotonic behavior is of no practical interest.
1131:
1132:
1133: \subsection{Numerical Calculation of Reflectivity and Improved WKB Treatment}
1134:
1135: \begin{figure}
1136: \centering
1137: \includegraphics[width=13cm]{keprun}
1138: \caption{Numerical calculation of wave reflection and transmission across corotation for a Keplerian disc with $\nu = 0.033$ and $\beta = 0.1$. The top and middle panel show the enthalpy perturbation $\delta h$, with the solid lines depicting the real components and the dashed lines the imaginary components. Note that there is a discontinuity in the derivative of $\delta h$ at the corotation resonance, resulting in an absorption of flux at the corotation. The bottom panel shows the angular momentum flux carried by the wave [see eq. (67)]. Note that $F(r)$ is conserved away from corotation and the discontinuity in $F$ at $r=r_c$ results from wave absorption. The Lindblad and corotation resonances are indicated by three vertical dotted lines.}
1139: \label{fig6}
1140: \end{figure}
1141:
1142:
1143: Our analytical expressions derived in section 4-5
1144: are based on global WKB analysis and involves several approximations.
1145: In particular, in our treatment of the corotation resonance
1146: (section 3.2), we assumed $c\ll r\Omega$ --- if this is not satisfied,
1147: some of the neglected terms must be included in equation (\ref{pnearcr})
1148: and the numerical values of our solution may be modified. Thus our results
1149: for $\beta\go 0.1-0.2$ should be treated with caution.
1150:
1151: To assess the validity of our WKB analysis, we also carry out
1152: computation of the reflectivity by numerical integration of
1153: equations (8)-(10). The outgoing wave boundary condition,
1154: equation (\ref{eq:outside}),
1155: %equation (48),
1156: is imposed at some radius $r_{\rm out} \gg r_{\rm OL}$, such that
1157: \be
1158: \delta h' = \left(ik+{S'\over 2S}-{k'\over 2k}\right)\delta h,
1159: \ee
1160: where $'$ specifies derivatives with respect to $r$.
1161: At some inner radius $r_{\rm in}\ll r_{\rm IL}$, the solution takes the form of
1162: equation (71). The reflection coefficient can be obtained from
1163: \be
1164: |{\cal R}| = \left|\frac{\left[\left(\frac{S'}{2S} - \frac{k'}{2k} \right) - i k\right]\delta h - \delta h'}
1165: {\left[\left(\frac{S'}{2S} - \frac{k'}{2k} \right) + i k\right]\delta h - \delta h'}\right|_{r_{\rm in}}
1166: \ee
1167: while the transmission is
1168: \be
1169: |{\cal T}| = \left|\frac{2k}
1170: {\left[\left(\frac{S'}{2S} - \frac{k'}{2k} \right) + i k\right]\delta h - \delta h'}\right|_{r_{\rm in}} |\delta h|_{r_{\rm out}}~.
1171: \ee
1172: Because of the singularity at corotation, we include a small
1173: positive $\omega_i={\rm Im}(\omega)$ for the frequency so that singularity can be
1174: avoided. Figure \ref{fig6} depicts an example calculation for a Keplerian disc with $\beta = 0.1$ and $p = 2$ (so that $\nu = 0.033$).
1175:
1176: Figure \ref{fig7} shows our numerical result compared with the calculation
1177: from the WKB analysis. There is qualitative agreement between these
1178: results. In particular, both the numerical and WKB results show that wave
1179: absorption at corotation plays the dominant role in determining $|{\cal R}|$,
1180: and wave transmission is unimportant even for small (but nonzero) $\nu$.
1181:
1182: \begin{figure}
1183: \centering
1184: \includegraphics[width=13cm]{Rvnu3}
1185: \vskip -1cm
1186: \caption{The reflection coefficient as a function of $\nu$ calculated using the WKB, improved WKB, and numerical methods for a scale free Keplerian disk with sound speed $c = 0.05 \Omega r$.}
1187: \label{fig7}
1188: \end{figure}
1189:
1190: However, the WKB solution matches the numerical result closely only
1191: for $|\nu| \ll 1$, indicating that our WKB analysis can be improved.
1192: For the Keplerian disk considered here, the variation in $\nu$ was
1193: achieved by changing the background density index $p$. The WKB results
1194: shown in Figs.~\ref{fig5}-\ref{fig6} assume that the quantities
1195: $\Theta_{\rm IIa}$ and $\Theta_{\rm IIb}$ are not effected by changing
1196: $\nu$. However $p$ plays a non-negligible role in determining the
1197: effective wave number away from the corotation and Lindblad
1198: resonances. Indeed, in obtaining equation \eqref{LRdiffeq} or
1199: \eqref{pnearcr} from \eqref{perturbeq2}, we have neglected several
1200: terms that are negligible near $r_c$ or $r_{\rm IL/OL}$, but
1201: nevertheless important away from these resonances. Noting that our connection
1202: formulae [equations (32)-(35) and (42)-(43) involve the asymptotic expansions of
1203: local solutions around the resonances,
1204: we can improve our WKB results by adopting the following ansatz: we modify the
1205: integrands in $\Theta_{\rm IIa}$ and $\Theta_{\rm IIb}$ to include these dropped
1206: terms,
1207: \ba
1208: \Theta &=& \int \sqrt{-k_{\rm eff}^2} dr,\\ k_{\rm eff}^2
1209: &=& -\frac{D}{c^2} - \frac{m^2}{r^2} - \frac{p^2 - 1}{4r^2} -
1210: D^{1/2}\left(\frac{d}{dr}D^{-1/2}\right)\frac{1-p}{2}.
1211: \ea
1212: Note that
1213: in the expression above, we have left out the singular term ($\propto
1214: \tomega^{-1}$) at corotation and the dominant double singular term
1215: ($\propto D^{-2}$) at the Linblad resonances since they have already
1216: been accounted for by the connection formulae in sections 3. As shown
1217: in Fig. \ref{fig6}, the improved WKB result matches the numerical
1218: solution for a much larger range of $\nu$.
1219: The increased values of
1220: $\Theta_{\rm IIa}$ and $\Theta_{\rm IIb}$ for larger $|p|$ act to
1221: suppress super-reflection, and drive reflection coefficient
1222: toward $|{\cal R}| = 1$.
1223:
1224:
1225: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1226: \section{Global Overstable Modes}
1227:
1228: As mentioned in Section 1, global disc instabilities are often related to
1229: the super-reflection at the corotation barrier. To illustrate this,
1230: we consider a simple boundary condition
1231: \be
1232: \delta h(r_{{\rm in}}) = 0
1233: \ee
1234: at the inner radius of the disc, $r_{\rm in} \ll r_{\rm IL}$. The outgoing
1235: boundary condition at $r>r_{\rm OL}$ implies that the solution
1236: in the region $r<r_{\rm IL}$ is given by
1237: \be
1238: \delta h = \sqrt{S/k}\exp\left(-i\int_{r_{\rm IL}}^{r} k
1239: dr+\pi/4\right)+ {\cal R}\sqrt{S/k}
1240: \exp\left(i\int_{r_{\rm IL}}^{r}kdr-\pi/4\right).\label{eq:rinside}
1241: \ee
1242: Letting ${\cal R} = |{\cal R}| e^{i\varphi}$, and applying the boundary condition (\ref{eq:rinside}) yields the eigenvalue condition:
1243: \be
1244: \tan\left(\int_{r_{\rm in}}^{r_{\rm IL}} k dr
1245: - \pi/4-\varphi/2\right) = -i\left(\frac{|{\cal R}|-1}{|{\cal R}|+1}\right)~.\label{eq:eigcond}
1246: \ee
1247: Noting that for complex eigenvalue $\omega = \omega_r + i \omega_i$, the wavenumber $k$ is also complex
1248: \be
1249: k = k_r + ik_i = \frac{1}{c} \sqrt{(\tomega_r + i\tomega_i)^2 - \kappa^2} \simeq \frac{1}{c} \sqrt{\tomega_r^2 - \kappa^2} + i \frac{\omega_i\tomega_r}{c\sqrt{\tomega_r^2 -\kappa^2}},
1250: \ee
1251: where we have assumed $|\omega_i| \ll |\omega_r|$. $\tomega_r = \omega_r - m\Omega$ ($< 0$ for region I) and thus the real part of the eigenvalue condition gives
1252: \be
1253: \int_{r_{\rm in}}^{r_{\rm IL}} {1 \over c} \sqrt{(\omega_r - m\Omega)^2 - \kappa^2} dr - \pi/4 -\varphi/2= n\pi~,
1254: \ee
1255: where $n$ is an integer.
1256: The imaginary part of (\ref{eq:eigcond}) gives (assuming $|k_i| \ll |k_r|$)
1257: \be
1258: \int_{r_{\rm in}}^{r_{\rm IL}} k_i dr \simeq -\left(\frac{|{\cal R}|-1}{|{\cal R}|+1}\right)~,
1259: \ee
1260: from which we find the growth rate
1261: \be
1262: \omega_i = \left(\frac{|{\cal R}| -1}{|{\cal R}|+1}\right)\left[\int_{r_{\rm in}}^{r_{\rm IL}} |\tomega_r|/c\sqrt{\tomega_r^2 - \kappa^2} dr\right]^{-1}.
1263: \ee
1264: Thus the modes are overstable ($\omega_i > 0$) for $|{\cal R}| > 1$, and stable ($\omega_i < 0$) for $|{\cal R}| < 1$.
1265:
1266: %%%%%%%%%%%%%%%%%%%%%%
1267: \section{Conclusion}
1268:
1269: In this paper we have derived
1270: explicit expressions for the reflection coefficient, transmission
1271: coefficient and wave absorption coefficient
1272: when a wave is scattered by the corotation barrier in a disc.
1273: These expressions include both the effects of corotation amplifier
1274: (which exists regardless of the gradient of the vortensity
1275: $\zeta=\kappa^2/\Omega\Sigma$ of the
1276: background flow) and wave absorption at the corotation (which depends on
1277: $d\zeta/dr$). They demonstrate clearly that the corotation wave absorption plays a
1278: dominant role in determining the reflectivity and that the sign of
1279: $d\zeta/dr$ determines whether the corotation singularity enhances or
1280: diminishes the super-reflectivity.
1281: Our result can be understood in terms of the location of the Rossby wave zone
1282: relative to the corotation radius. We also carried out numerical calculations
1283: of the reflectivity. Our result provides the conditions
1284: (in terms of disc thickness, rotation
1285: profile and surface density profile) for which super-reflection is
1286: achieved and global overstable modes in discs are possible.
1287:
1288: In future works we will explore global oscillation modes and their
1289: stabilities in a variety of astrophysical contexts, ranging from
1290: accreting white dwarfs to accreting black hole systems. The possible
1291: overstabilities of these modes are directly
1292: linked to the effects studied in this paper and may provide
1293: explanations for some of the
1294: quasi-periodic variabilities observed in these systems.
1295:
1296:
1297: \section*{Acknowledgments}
1298:
1299: DL thanks Peter Goldreich and Ramesh Narayan for useful conversations.
1300: This work has been supported in part by NASA Grant NNX07AG81G, NSF
1301: grants AST 0707628, and by {\it Chandra} grant TM6-7004X
1302: (Smithsonian Astrophysical Observatory).
1303:
1304: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%5
1305: \appendix
1306:
1307: \renewcommand{\theequation}{A-\arabic{equation}}
1308: % redefine the command that creates the equation no.
1309: \setcounter{equation}{0} % reset counter
1310: % use *-form to suppress numbering
1311:
1312:
1313: \section{Stokes Phenomenon and the Matching Condition Across the Corotation
1314: Singularity}
1315:
1316: As we saw in section 3.1 the perturbation equation near the
1317: corotation resonance can be solved in terms of the Whittaker
1318: function. However in order to provide the matching conditions we must
1319: carefully consider the effect of the Stokes Phenomenon on the
1320: asymptotic expansions.
1321:
1322: Stokes phenomenon causes the functional form of the asymptotic expansion of an entire
1323: function to be different at different points in the complex plane.
1324: The general asymptotic solution of the Whittaker equation
1325: for $|z|\gg 1$ can be written as a linear combination of two functions
1326: \be
1327: P(z)=e^{z/2}z^{-\nu},\qquad
1328: Q(z)=e^{-z/2}z^\nu,
1329: \ee
1330: i.e., $AP(z)+BQ(z)$.
1331: However, because of the Stokes phenomenon, the coefficients $A$ and $B$
1332: can change when crossing the Stokes lines. For Whittaker functions,
1333: the Stokes lines are the positive real axis (where $P$ is dominant and
1334: $Q$ is sub-dominant) and negative real axis (where $Q$ is dominant and
1335: $P$ is sub-dominant).
1336:
1337: Consider specific solution to the Whittaker equation, with the asymptotic
1338: expansion ($|z|\gg 1$) given by
1339: \be
1340: F(z)\rightarrow AP(z)+BQ(z),\qquad ({\rm for}~~\arg(z)=0)
1341: \label{eq:f}\ee
1342: on the real axis. Our goal is to derive the expansion coefficients
1343: of $F(z)$ on the negative real axis ($\arg(z)=\pi$).
1344: To achieve this, we use the general results obtained by Heading (1962):
1345:
1346: (1) ``The coefficient of the subdominant term after crossing the Stokes
1347: line $=$ the coefficient of the subdominant term before crossing
1348: the Stokes line + $ T_n \times$ the coefficient of the dominant
1349: term on the Stokes line.''
1350:
1351: (2) ``The coefficient of the subdominant term on the Stokes line $=$ the
1352: coefficient of the subdominant term before the Stokes line $+
1353: {1\over 2} T_n \times$ the coefficient of the dominant term on the
1354: Stokes line.''
1355:
1356: Here $T_n$ is the Stokes multiplier for crossing the Stokes line $\arg(z) = n\pi$ in the direction of increasing $\arg(z)$ given by
1357: \footnote{Note that a typo in equation (18) of Heading (1962) has
1358: been corrected here.}
1359: \be
1360: T_n ={2\pi i e^{-2\pi ins\nu}\over \Gamma(s\nu)\Gamma(1 + s\nu)}
1361: \ee
1362: where $s = (-1)^n$. Note that since the pole of equation (\ref{Whittakerdiffeq}) lies below the real axis, $\arg(z)$ increases as we we move along the contour from $z$ positive and real ($\arg(z) = 0$) to $z$ negative and real ($\arg(z) = \pi$).
1363:
1364: Since equation (\ref{eq:f}) is given on one of the Stokes line,
1365: we first determine $F(z)$ in the region $-\pi<\arg(z)<0$:
1366: \be
1367: F(z)\sim AP(z)+\left(B-{1\over 2} T_0 A\right)Q(z),
1368: \qquad ({\rm for}~ -\pi<\arg(z)<0).
1369: \ee
1370: Then in region $0<\arg(z)<\pi$ we have
1371: \be
1372: F(z)\sim AP(z)+\left(B+{1\over 2} T_0 A\right)Q(z),
1373: \qquad ({\rm for}~ 0<\arg(z)<\pi).
1374: \ee
1375: Thus on the negative real axis, we obtain
1376: \be
1377: F(z)\sim
1378: \left[A+{1\over 2} T_1
1379: \left(B+{1\over 2} T_0 A\right)\right]P(z)+\left(B+{1\over 2} {\rm T_0}A\right)Q(z),
1380: \qquad ({\rm for}~ \arg(z)=\pi).
1381: \ee
1382:
1383: The connection formulae for the two independent Whittaker functions for $|z| \gg 1$ are
1384: \be
1385: \psi_- = W_{\nu,1/2}(z)\sim
1386: \Biggl\{
1387: \begin{array}{ll}
1388: Q(z) & ~~~\arg(z)=0\\
1389: Q(z)+{1\over 2}{\rm T_1} P(z) & ~~~\arg(z)=\pi
1390: \end{array}
1391: \ee
1392: \be
1393: \psi_+ = e^{-i\pi\nu}\textrm{W}_{-\nu, 1/2} (ze^{-i\pi}) + \frac{1}{2} T_0 \textrm{W}_{\nu,1/2} (z).
1394: \sim
1395: \Biggl\{
1396: \begin{array}{ll}
1397: P(z) & ~~~\arg(z)=0\\
1398: {1\over 2}T_0 Q(z)+
1399: \left(1+{1\over 4} T_0 T_1\right)P(z) & ~~~\arg(z)=\pi,
1400: \end{array}
1401: \ee
1402: where $P(z)$ and $Q(z)$ have the asymptotic behavior
1403: \ba
1404: P(z) = e^{+z/2 -\nu \log(z)} &\sim&
1405: \Biggl\{\begin{array}{ll}
1406: \exp\left(\int_{r_c}^{r}\! \tilde k \,dr \right)
1407: & \qquad \qquad \qquad ~~\textrm{for } r \gg r_{c}\\
1408: e^{-i\pi\nu}\exp\left(-\int_r^{r_{c}}\! \tilde k \,dr \right)
1409: &\qquad \qquad \qquad~~\textrm{for } r\ll r_{c}~.
1410: \end{array}\\
1411: %\rightarrow e^{\left( \int_{r_c}^r \tilde k dr \right)}\\
1412: Q(z) = e^{-z/2 + \nu \log(z)} %\rightarrow e^{\left( -\int_{r_c}^r \tilde k dr \right)},
1413: &\sim&
1414: \Biggl\{\begin{array}{ll}
1415: \exp\left(-\int_{r_c}^{r}\! \tilde k \,dr \right)
1416: & \qquad \qquad \qquad \qquad~\textrm{for } r \gg r_{c}\\
1417: e^{i\pi\nu}\exp\left(\int_r^{r_{c}}\! \tilde k \,dr \right)
1418: &\qquad \qquad \qquad\qquad~\textrm{for } r\ll r_{c}~.
1419: \end{array}
1420: \ea
1421: for $|z| \gg 1$, since in this limit $|z| \gg \log |z|$.
1422:
1423:
1424:
1425: %%%%%%%%%%%%%%%%%%%%%%
1426: \section{Resonance Tunneling}
1427:
1428: \renewcommand{\theequation}{B-\arabic{equation}}
1429: % redefine the command that creates the equation no.
1430: \setcounter{equation}{0} % reset counter
1431: % use *-form to suppress numbering
1432:
1433: To help understand the effect of wave absorption at the corotation
1434: singularity we consider the following toy problem. For a 1-D
1435: quantum mechanical potential of the form
1436: \be
1437: V(x) = {C\over x},
1438: \ee
1439: we have the time-independent wave equation
1440: \be
1441: \left(\frac{1}{2}\frac{d^2}{dx^2} + E - \frac{C}{x + i\epsilon}\right)
1442: \psi = 0,
1443: \label{eq:wave}\ee
1444: where we have set the Planck constant and particle mass to unity.
1445: In equation (\ref{eq:wave}), a small imaginary part $i\epsilon$ is added
1446: to x, to account for the physical requirement of a growing incoming
1447: perturbation. The sign of $\epsilon$ must be the same as the sign of $C$,
1448: so as to give us the correct physical behavior.
1449: For concreteness we will assume $C>0$ (and thus $\epsilon > 0$),
1450: though the same calculation can be performed for $C < 0$ ($\epsilon < 0$)
1451: with similar results.
1452:
1453: Let $y = 2ik(x + i\epsilon)$, where $k = \sqrt{2E}$, we have
1454: \be
1455: \frac{d^2\psi}{dy^2} + \left(-\frac{1}{4} + \frac{\nu}{y} \right)\psi = 0,
1456: \qquad {\rm with}~~\nu=iC/k,
1457: \label{qmwhittaker},
1458: \ee
1459: which we recognize as the Whittaker differential equation.
1460: With $C > 0$, $y$ lies in the domain $\pi/2 \leq \arg(y) \leq 3\pi/2$.
1461: Using the same procedure as discussed in Appendix A, we obtain the
1462: connection formulae for the general solution of
1463: equation \eqref{qmwhittaker} as
1464: \be
1465: \psi \rightarrow \Biggl\{\begin{array}{ll}
1466: A P(z) + B Q(z)
1467: &\qquad \textrm{for } \arg(z) = \pi/2 \\
1468: (A + T_1 B) P(z) + B Q(z)
1469: &\qquad \textrm{for } \arg(z) = 3\pi/2.
1470: \end{array}
1471: \ee
1472:
1473: Consider a wave propagating from $x=-\infty$ and impinging on the
1474: potential. The transmitted wave at $x>0$ has the form
1475: \be
1476: \psi_+ \rightarrow e^{y/2} y^{-\nu} = e^{ikx -i(C/k)\ln|2kx| + \pi C/2 k}
1477: \qquad \textrm{for } x \gg 1/k\\.
1478: \ee
1479: The corresponding wave solution in the $x<0$ region has the asymptotic form
1480: \be
1481: \psi_+ \rightarrow e^{y/2} y^{-\nu} = e^{ikx -i(C/k)\ln|2kx| + 3\pi C/2k}
1482: \qquad \textrm{for } x \ll -1/k.
1483: \ee
1484: This gives
1485: \be
1486: |{\cal R}| = 0,\qquad |{\cal T}| = e^{-\pi C/ k}
1487: \ee
1488: for a wave incident from $x < 0$.
1489:
1490: Now consider a wave incident from the $x > 0$ region towards small $x$.
1491: The transmitted wave is given by
1492: \be
1493: \psi_- \rightarrow e^{-y/2} y^{+\nu} = e^{-ikx +i(C/k)\ln|2kx| - 3\pi
1494: C/2k}\qquad \textrm{for } x \ll -1/k.
1495: \ee
1496: Connecting to the $x>0$ region, we have
1497: \be
1498: \psi_- \rightarrow e^{-y/2} y^{+\nu} - T_1 e^{y/2} y^{-\nu}= e^{-ikx
1499: +i(C/k)\ln|2kx| - \pi C/2k} - T_1 e^{ikx -i(C/k)\ln|2kx| + \pi C/2k}
1500: \qquad \textrm{for } x \gg 1/k.
1501: \ee
1502: This gives the reflection and transmission coefficients
1503: \ba
1504: && |{\cal R}| =e^{\pi C/k}|T_1|=
1505: \frac{2\pi e^{-\pi C/ k}}{|\Gamma(-iC/k) \Gamma(1 -
1506: iC/k)|}
1507: = 2 e^{-\pi C/k} \sinh(\pi C/k) = 1- e^{-2\pi C/k}, \\
1508: &&|{\cal T}| = e^{-\pi C/k},
1509: \ea
1510: for a wave incident from the positive side of the singularity
1511: [a similar result was obtained by Budden (1979) in the context of
1512: wave propagation in cold plasma].
1513:
1514: We can define the wave absorption coefficient at the singularity as
1515: \be
1516: {\cal D} = 1- |{\cal T}|^2 - |{\cal R}|^2
1517: \ee
1518: For the forward moving incident wave ($+$) and the backwards moving incident
1519: wave ($-$), we have
1520: \be
1521: {\cal D}_+ = 1 - e^{-2\pi C/k}, \qquad {\cal D}_- = e^{-2\pi C/k} \left(1 - e^{-2\pi C/k} \right).
1522: \ee
1523:
1524:
1525:
1526:
1527:
1528:
1529: %%%%%%%%%%%%%%%%%%%%%%
1530:
1531: \begin{thebibliography}{99}
1532:
1533: \bibitem[]{} Abramowitz, M, Stegun, I.A. 1964, Handbook of
1534: Mathematical Functions (Dover: New York)
1535:
1536: \bibitem[]{}
1537: Balbus, S.A., Hawley, J.F. 1998, Rev. Mod. Phys., 70, 1.
1538:
1539: \bibitem[]{}
1540: Budden, K.G., Phil. Trans. R. Soc. Lond. A, 1979, 290, 405
1541:
1542: \bibitem[]{}
1543: Dickenson, R.E., 1968, J. Atmos. Sci., 25, 984
1544:
1545: \bibitem[]{}
1546: Goldreich, P. 1988, in ``Origin, structure and evolution of galaxies''
1547: (Proceedings of the Guo Shoujing Summer School of Astrophysics,
1548: Tunxi, China), ed. L.Z. Fang (World Scientific: Singapore), p.~127
1549:
1550: \bibitem[]{}
1551: Goldreich, P., Lynden-Bell, D. 1965, MNRAS, 130, 125
1552:
1553: \bibitem[]{}
1554: Goldreich, P., Goodman, J., Narayan, R. 1986, MNRAS, 221, 339
1555:
1556: \bibitem[]{}
1557: Goldreich, P., Tremaine, S. 1979, ApJ, 233, 857
1558:
1559: \bibitem[]{}
1560: Goodman, J., Evans, N. 1999, MNRAS, 309, 599
1561:
1562: \bibitem[]{}
1563: Heading, J. 1962, J. Lond. Math. Soc., 37, 195
1564:
1565: \bibitem[]{}
1566: Julian, W.H., Toomre, A. 1966, ApJ, 146, 810
1567:
1568: \bibitem[]{}
1569: Kato, S., 2003, PASJ, 55, 257
1570:
1571: \bibitem[]{}
1572: Li, H., Finn, J.M., Lovelace, R.V.E., Colgate, S.A. 1999, ApJ, 533, 1023
1573:
1574: \bibitem[]{}
1575: Li, L., Goodman, J., Narayan, R., 2003, ApJ, 593, 980
1576:
1577: \bibitem[]{}
1578: Lin, C.C., Lau, Y.Y. 1975, SIAM J. Appl. Math., 29, 352
1579:
1580: \bibitem[]{}
1581: Lovelace, R.V.E., Li, H., Colgate, S.A., Nelson, A.F. 1999, ApJ, 513, 805
1582:
1583: \bibitem[]{}
1584: Lynden-Bell, D., Kalnajs, A.J. 1972, MNRAS, 157, 1
1585:
1586: \bibitem[]{}
1587: Mark, J.W.K. 1976, ApJ, 205, 363
1588:
1589: \bibitem[]{}
1590: Narayan, R., Goldreich, P., Goodman, J. 1987, MNRAS, 228, 1
1591:
1592: \bibitem[]{}
1593: Papaloizou, J.C.B., Lin, D.N.C. 1995, ARAA, 33 505
1594:
1595: \bibitem[]{}
1596: Papaloizou, J.C.B., Pringle, J.E. 1984, MNRAS, 208, 721
1597:
1598: \bibitem[]{}
1599: Papaloizou, J.C.B., Pringle, J.E. 1987, MNRAS, 225, 267
1600:
1601: \bibitem[]{}
1602: Pedlosky, J. 1987, Geophysical Fluid Dynamics. Springer-Verlag, Berlin
1603:
1604: \bibitem[]{}
1605: Shu, F.H. 1992, The Physics of Astrophysics II: Gas Dynamics. University
1606: Science Books, Mill Valley, CA, chapter 12
1607:
1608: \bibitem[]{}
1609: Shu, F.H., Laughlin, G., Lizano, S., Galli, D., 2000, ApJ, 535, 190
1610:
1611: \bibitem[]{}
1612: Tagger, M., Pellat, R. 1999, A\&A, 349, 1003
1613:
1614: \bibitem[]{}
1615: Tagger, M., 2001 A\&A, 380, 750
1616:
1617: \bibitem[]{}
1618: Tagger, M., Varniere, P. 2006, ApJ, 652, 1457
1619:
1620: \bibitem[]{}
1621: Tanaka, H., Takeuchi, T., Ward, W.R. 2002, ApJ, 565, 1257
1622:
1623: \bibitem[]{}
1624: Zhang, H., Lai, D., MNRAS, 2006, 368, 917
1625:
1626: \end{thebibliography}
1627:
1628: \end{document}
1629: