1: \documentclass[apjl]{emulateapj}

2: %\documentclass[12pt,preprint]{aastex}

3: 

4: %%%%% AUTHORS - PLACE YOUR OWN MACROS HERE %%%%%

5: \newcommand{\Rs}{R_s}

6: \newcommand{\Rt}{R_t}

7: 

8: \begin{document}

9: 

10: \title{Dynamical effects of self-generated magnetic fields in cosmic

11:   ray modified shocks}

12: 

13: \author{D. Caprioli\altaffilmark{1}, P. Blasi\altaffilmark{2,3},

14:   E. Amato\altaffilmark{2}, M. Vietri\altaffilmark{1}}

15:   

16: \altaffiltext{1}{Scuola Normale Superiore, Pisa, Italy}

17: \altaffiltext{2}{INAF/Osservatorio Astrofisico di Arcetri, Firenze,

18:   Italy}

19: \altaffiltext{3}{Fermilab, Center for Particle Astrophysics, USA }

20: 

21: 

22: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%{}

23: 

24: \begin{abstract}

25: Recent observations of greatly amplified magnetic fields ($\delta B/B

26: \sim 100$) around supernova shocks are consistent with the

27: predictions of the non-linear theory of particle acceleration (NLT),

28: if the field is generated upstream of the shock by cosmic ray induced

29: streaming instability. The high acceleration efficiencies and large

30: shock modifications predicted by NLT need however to be mitigated to

31: confront observations, and this is usually assumed to be accomplished

32: by some form of turbulent heating. We show here that magnetic fields

33: with the strength inferred from observations have an important

34: dynamical role on the 

35: shock, and imply a shock modification substantially reduced with respect

36: to the naive unmagnetized case. The effect appears as soon as the

37: pressure in the turbulent magnetic field becomes comparable with the

38: pressure of the thermal gas. The

39: relative importance of this unavoidable effect and of the poorly known

40: turbulent heating is assessed. More specifically we conclude that even

41: in the cases in which turbulent heating may be of some importance, the

42: dynamical reaction of the field cannot be neglected, as instead is

43: usually done in most current calculations. 

44: \end{abstract}

45: 

46: \keywords{acceleration of particles --- shock waves --- magnetic field}

47: 

48: \maketitle

49: 

50: \section{Introduction}

51: 

52: The supernova remnant (SNR) paradigm for the origin of galactic cosmic

53: rays is based on the assumption that at least $\sim 10-20\%$ of the

54: kinetic energy of the expanding shell is converted into cosmic rays

55: (CRs). Moreover, as recent observations have proved, the magnetic

56: field (MF) in the shock vicinity is amplified by a large amount

57: (e.g. \cite{ballet}) as would be expected if cosmic rays induce

58: streaming instability (SI) upstream of the shock. We stress that such

59: MF amplification is required to accelerate protons up to

60: $\sim 10^6$ GeV. The need for a satisfactory and self-consistent

61: description of these points is sufficient to justify the

62: development of a NLT of particle acceleration. 

63: 

64: The developments of the theory are summarized in (\cite{druryrev},

65: \cite{je91}, \cite{maldrury}). The kinetic theory for arbitrary

66: diffusion coefficients (\cite{amato1}), and even in the case of

67: self-generated MFs (\cite{amato2} and \cite{elli06}) has 

68: been recently developed.

69: 

70: All approaches to NL shock acceleration find that the large

71: pressure of the accelerated particles decelerates the incoming gas, and

72: leads to total compression factors that scale with the Mach number of

73: the shock as $\Rt\sim M_0^{3/4} \sim 20-50$. Such large shock

74: modifications however are at odds with observations which are better 

75: fit by $\Rt\sim 7-10$. The problem with larger values of $\Rt$

76: resides in both the estimated distance between the forward and reverse

77: shocks (\cite{warren}) and in the fit to multifreqeuncy

78: observations with concave spectra (VBK05 and references therein). The

79: reduction in the compression factor is almost invariably attributed to

80: turbulent heating (TH) in the precursor (\cite{bv97} and

81: later) as due to damping of waves on the background plasma

82: (\cite{mck-v82}, hereafter MKV82). 

83: In fact this mechanism was originally proposed in

84: order to keep the MF amplification in the linear regime

85: (e.g. $\delta B/B\ll 1$), but is now commonly applied to cases in

86: which $\delta B/B\gg 1$.  

87: Unfortunately, the heating process is quite model dependent and

88: even its applicability to situations of interest for SNRs can and

89: should be seriously questioned. The effectiveness of the heating

90: process can easily be reduced to negligible levels or artificially

91: amplified to unphysical levels.   

92: 

93: As mentioned above, a breakthrough in the field has recently been

94: provided by X-ray observations: the detection of X-ray bright

95: filaments in the outskirts of some SNRs allows one to infer the

96: strength of the

97: MF in these filaments, found to be $B\sim 100-500\mu

98: G$. Such strong fields are generally attributed to the SI induced by

99: CRs efficiently accelerated at the shock front, although alternative

100: explanations have been proposed (\cite{gj07}). In Tab. \ref{tab} we

101: list some SNRs with estimated MFs: $u_0$ is the shock

102: velocity, $B_2$ is the MF downstream of the shock as

103: inferred from the X-ray brightness profile and

104: $P_{w2}=B_2^2/(8\pi\rho_0 u_0^2)$ is the downstream magnetic pressure

105: normalized to the bulk one. The values of the parameters are from

106: \cite{P+06} and from VBK05 (in parenthesis in Tab. \ref{tab}).

107: 

108: \begin{table}

109: \begin{center}

110: \caption{\label{tab}Parameters for 5 well known SNRs.}

111: \begin{tabular}{cccc}

112: \tableline

113: SNR & $u_0 (km/s)$ & $B_2(\mu G)$ & $P_{w2}\times 10^3$\\

114: \tableline

115: Cas A& 5200 (2500)& 250--390 & 32 (36) \\

116: Kepler & 5400 (4500)& 210--340 &  23 (25)\\

117: Tycho & 4600 (3100)& 300--530 &  27 (31)\\

118: SN 1006 & 2900 (3200)& 91--110 & 40 (42) \\

119: RCW 86 & (800) & 75--145 & 14-35 (16-42)\\

120: \tableline

121: \end{tabular}

122: \tablecomments{For SNRs discussed by \cite{P+06} we used 

123: $\rho_0=0.1\,m_p/cm^3$ in the case of SN 1006  and $\rho_0=0.5\,m_p/cm^3$ 

124: in the other cases, while VBK05 provide directly $P_{w2}$.}

125: \end{center}

126: \end{table}

127: We show below that for the field stregths inferred for SNRs, the

128: magnetic pressure is comparable or even in excess of the thermal

129: pressure of the background plasma and that whenever this happens the

130: dynamical reaction of the field on the fluid is such that the

131: compression factors are substantially reduced and fall in the range

132: suggested by observations. It is crucial to keep in mind that, 

133: contrary to the TH, which can be either suppressed or amplified 

134: by changing parameters on which there is little or no control,

135: the feedback of the self-generated turbulent MF on the plasma is not

136: model dependent and must be included. 

137: 

138: \section{Dynamics of a magnetized CR modified shock} 

139: The reaction of accelerated particles upstream of the shock leads to

140: the formation of a precursor, in which the fluid speed 

141: decreases while approaching the shock. One can describe this effect by

142: introducing two compression factors $\Rt=u_0/u_2$ and

143: $\Rs=u_1/u_2$, where $u$ is the fluid speed and the indexes '$0$',

144: '$1$' and '$2$' refer to quantities at upstream infinity, upstream and

145: downstream of the subshock respectively. 

146: 

147: The most general equations of conservation of mass, momentum and

148: energy in the stationary case for a parallel shock are: 

149: %\begin{equation}

150: %\frac{\partial}{\partial x}\left( \rho u \right) = 0,

151: %\label{eq:mass}

152: %\end{equation}

153: \begin{equation}

154: \frac{\partial}{\partial x}\left( \rho u \right) = 0,\,\,\,\,\,\,\,

155: \frac{\partial}{\partial x} \left( \rho u^2 + p + p_c + p_w \right)=0, 

156: \label{eq:massmomentum}

157: \end{equation}

158: \begin{equation}

159: \frac{\partial}{\partial x} \left( \frac{1}{2}\rho u^3 + 

160: \frac{\gamma p u}{\gamma-1} + F_w \right) = -u \frac{\partial

161:   p_c}{\partial x}.  

162: \label{eq:energy}

163: \end{equation}

164: 

165: As usual, $\rho$, $u$, $p$ and $\gamma$ stand for density, velocity,

166: pressure and the ratio of specific heats of the gas, while $p_w$ and

167: $F_w$ represent the pressure and energy flux in the form of Alfv\'en

168: waves. $p_c$ is the CR pressure. The continuity of the distribution

169: function of the accelerated particles through the subshock implies

170: that the CR pressure is also continuous ($p_{c1}=p_{c2}$), and that

171: the term $\partial p_c/\partial x$ gives null contribution when

172: Eq.~\ref{eq:energy} is integrated from $x=0^-$ to $x=0^+$. All

173: previous equations at the subshock read as the usual Rankine-Hugoniot

174: relations at a magnetized gaseous shock. 

175:   

176: In order to treat the presence of Alfv\'en waves correctly, we use the

177: approach of \cite{vs99} (hereafter VS99), considering two upstream

178: wave trains with helicities $H_c=\pm 1$, and their respective

179: downstream counterparts. If $\delta \vec B_{i}$ is a mode of the

180: MF perturbation, we write the velocity perturbation as

181: $\delta \vec u=-H_c\frac{\delta\vec B}{\sqrt{4\pi\rho}}$. Neglecting

182: the electric field contribution, which is of order $(u/c)^2$, the

183: magnetic pressure and the energy flux, which for Alfv\'en waves is the

184: sum of the normal component of Poynting vector $\vec u\times\delta\vec

185: B\times\delta\vec B/4\pi$ and the transverse kinetic energy flux

186: $\rho\delta\vec u^2/2$, are

187: \begin{equation}\label{Fw}

188: p_w=\frac{1}{8\pi}\left(\sum_i\delta\vec B_i\right)^2;\quad

189: F_w=\sum_i\frac{\delta \vec B_i^2}{4\pi}\tilde{u_i}+

190: p_w u\,,

191: \end{equation}

192: having posed $\tilde{u_i}=u+H_{c,i} v_A$. 

193: The upstream magnetic turbulence typically shows two opposite

194: helicities (\cite{bl2001}), each of which splits into a reflected and a

195: transmitted wave crossing the subshock. 

196: According to VS99, the transmission and reflection

197: coefficients, in the limit $M_A^2\gg \Rs$ (large Alfv\'enic Mach

198: number), do not depend on $H_c$ and read 

199: \begin{equation}

200: \label{TR}

201: T \simeq(\Rs+\sqrt{\Rs})/2\,; \qquad 

202: R \simeq(\Rs-\sqrt{\Rs})/2\,. 

203: \end{equation}

204: For a typical supernova shock, the Alfv\'enic Mach number is

205: $M_{A,1}=u_1/v_A \geq 100$, hence in the following we adopt these

206: coefficients and neglect $v_A$ with respect to the fluid velocity in

207: Eq.~\ref{Fw}. For each $H_c$ we therefore have $\delta B_2/\delta

208: B_1=T+R=\Rs$ and thus $p_{w2}=p_{w1} \Rs^2$.  

209: 

210: As pointed out above, the subshock can be viewed as a simple shock in

211: a magnetized gas, therefore the pressure jump is (VS99) 

212: \begin{equation}\label{p2p1}

213: \frac{p_2}{p_1}=\frac{(\gamma+1)\Rs-(\gamma-1)+(\gamma-1)(\Rs-1)\Delta}

214: {\gamma+1-(\gamma-1)\Rs}\;, 

215: \end{equation}

216: with $\Delta$ defined as:  

217: \begin{equation}

218: \Delta=\frac{\Rs+1}{\Rs-1}\frac{\left[p_{w}\right]}{p_1}-

219: \frac{2\Rs}{\Rs-1}\frac{\left[F_w\right]}{p_1 u_1}\;. 

220: \end{equation}

221: Using the expressions for $T$ and $R$ (Eq.~\ref{TR}) we get

222: \begin{equation}\label{delta}

223: \Delta=(\Rs-1)^2\frac{p_{w1}}{p_1}+\Rs~\frac{\vec B_-\cdot\vec

224: B_+}{2\pi p_1}\,. 

225: \end{equation}

226: Following VS99, we assume that the two opposite--propagating waves

227: carry MFs $\vec B_\pm$ displaced in such a way that $\vec

228: B_-\cdot\vec B_+=0$. This is not the most general configuration, but

229: it is nevertheless rather common since it occurs when: 1) there is

230: only one wave train, 2) when the two fields are orthogonal, and 3) on

231: average, when the relative phase between the wave trains is arbitrary.  

232: 

233: At this point we normalize all quantities to the ones at upstream

234: infinity: $U(x)= u(x)/u_0$, $P_w(x) = p_w/\rho_0u_0^2$ and

235: $P(x)=p(x)/\rho_0u_0^2=\frac{U(x)^{-\gamma}}{\gamma M_0^2}$. In the

236: latter, we used the assumption of adiabatic heating in the precursor

237: and the conservation of mass. 

238: 

239: Substituting Eq.~\ref{p2p1}, Eq.~\ref{delta} and the above expression

240: for $P(x)$ in the equation for momentum conservation, the compression

241: factors $\Rs$ and $\Rt$ are related through the equation 

242: \begin{equation}\label{rsrt}

243: \Rt^{\gamma+1}=\frac{M_0^2\Rs^\gamma}{2}\left[\frac{\gamma+1-\Rs(\gamma-1)}  

244: {1+\Lambda_B}\right],

245: \end{equation}

246: which is the same as the standard relation apart from the factor

247: $(1+\Lambda_B)$, with  

248: \begin{equation}

249: \Lambda_B=W\left[1+\Rs\left(2/\gamma-1\right)\right]\;,

250: \end{equation}

251: and $W=P_{w1}/P_1$. It is clear that the net effect of the magnetic

252: turbulence is to make the fluid less compressible, noticeably reducing

253: $\Rt$ if $W=P_{w1}/P_1$ is of order 1. 

254: Moreover, the pressure and temperature jumps at the subshock are

255: enhanced (Eq.~\ref{p2p1}). 

256: 

257: We should notice that if one naively assumed that $F_w=3up_w$

258: everywhere, neglecting the $T$ and $R$ coefficients needed to satisfy

259: Maxwell equations at the subshock, one would obtain

260: $\Delta'=[(\Rs-1)^2-2\Rs] W < \Delta\,,$ 

261: leading to an incorrect pressure jump. This approach, adopted by

262: \cite{elli06}, also leads to a less marked decrease of $\Rt$, since

263: $\Lambda_B'=W\left[1+\Rs\left(3/\gamma-2\right)\right]$.

264: 

265: \section{Confronting observations}

266: 

267: Here we show that the magnetization levels estimated in SNRs as

268: reported in Tab. \ref{tab} imply that $W\geq 1$, so that the dynamical

269: feedback of the amplified MF needs to be taken into

270: account. 

271: 

272: In Fig. \ref{fig:M100} we plot $\Rt$ versus $\Rs$ for $M_0=100$: the

273: three shadowed regions  represent the relation between $\Rs$ and

274: $\Rt$ for fixed $P_{w2}\in [0.02,0.04]$ and $W=1,~3,~10$:

275: $P_{w2}/\Rs^2 = P_{w1} = W P_{g1}

276: =W\left(\Rt/\Rs\right)^\gamma/(\gamma M_0^2)\,$.  

277: 

278: The three solid lines represent the relation $\Rt-\Rs$ for the three

279: given values of $W$ as given by Eq. \ref{rsrt}; the dashed line refers

280: to $W=0$, when $p_w$ is not included.    

281: 

282: \begin{figure}

283: \includegraphics[width=.45\textwidth]{f1.eps}

284: \caption{$\Rs$-$\Rt$ for $W=1,3,10$ (intersections of

285:   solid lines with corresponding shadowed regions) and for 

286:   $\chi=200,500$ (dash-dotted and dotted lines). The dashed line

287:   represents the case $W=0$ (see text for details).} 

288: \label{fig:M100}

289: \end{figure}

290: 

291: The compression factor lies at the intersection between the curve and

292: the shadowed region for a given value of $W$. If $W<0.7$ there are no

293: intersections. This implies that the values of the magnetic

294: pressure inferred from observations require substantial MF

295: amplification upstream, 

296: and that the conservation equations are affected by the dynamical

297: reaction of the field. Only values $W\geq 3$ are compatible with the

298: whole range $0.02\leq P_{w2}\leq 0.04$ inferred from observations. 

299: This means that in order to account for the inferred values of

300: $B_2$, the magnetic pressure must be \emph{at least} comparable to the

301: gas pressure, and thus its dynamical role cannot be neglected. From

302: Fig. \ref{fig:M100} one also sees that the magnetic reaction leads to

303: values of $\Rt$ lower by roughly a factor $\sim 2$ compared with the

304: case $W=0$. We will comment further on this point below. 

305: 

306: Up until this point we never used the physically crucial point that

307: the observed fields may be generated through a cosmic ray

308: induced SI upstream of the shock. The instability may operate in the

309: resonant (\cite{bell78}) and in the non-resonant (\cite{bell2004})

310: regime.  

311: 

312: The growth rates of these different modes can be easily estimated

313: only in the context of quasi-linear theory. Given the difficulty in

314: deriving this information in the general non linear case, here we

315: assume the following general relation between the pressures of CRs and

316: wave pressure upstream of the subshock: 

317: \begin{equation}\label{defchi}

318: \xi_1 = \chi P_{w1},

319: \end{equation}

320: where $\xi(x) = p_{cr}(x)/\rho_0u_0^2$ is the normalized CR

321: pressure. For resonant SI, one has $\chi\simeq M_A=u_0/v_A$, while for

322: non-resonant modes, $\chi\sim 4c/u_0$. In both cases, for typical

323: values of the parameters, one obtains $200\leq\chi\leq 500$.  

324: 

325: From Eq. \ref{eq:massmomentum} applied to the

326: precursor, namely between upstream of the subshock ($x=0^-$) and

327: upstream infinity, we can write:  

328: \begin{equation}

329: \frac{\Rs}{\Rt} + \frac{1}{\gamma M_0^2} \left[ \left(

330:   \frac{\Rt}{\Rs}\right)^\gamma -1 \right] + P_{w1}(1+\chi) = 1.  

331: \label{eq:prec}

332: \end{equation}

333: The physical values of $\Rs$ and $\Rt$ for a given $\chi$ are obtained

334: by determining the intersection of the corresponding curve with that

335: obtained for a given value of $W$ at the subshock. Whether the

336: solution reproduces the estimated value of $P_{w2}$ depends on whether

337: the intersection falls within or outside the shadowed region for

338: the same $W$ in Fig. \ref{fig:M100}. The dash-dotted and the dotted

339: line show the results for $\chi=200$ and $500$ respectively: it is

340: evident that the chosen values of $\chi$ allow for a consistent

341: explanation of the downstream magnetic pressures as inferred from

342: observations, and, 

343: equally important, lead to compression factors which are much lower

344: than those predicted by the standard NLT (\cite{bv97} and papers that

345: followed).   

346: 

347: \section{Heating in the precursor}

348: 

349: The strong shock modification predicted by NLT when the magnetic

350: pressure is ignored is usually assumed to be somewhat mitigated by

351: heating of the precursor as a result of damping of Alfven waves

352: (\cite{v-mck81},hereafter VMK81, and MKV82) on the background gas. 

353: Other phenomena (for instance acoustic instability) may also lead to heating of the

354: precursor. In the original description, that remained basically

355: unchanged to the present time, VMK81 assumed that the rate of

356: damping ($\Gamma$) equals the rate of growth ($\sigma$) of Alfven

357: waves. The main implication of this assumption is that the growth of

358: the waves never reaches the non-linear regime, which is in fact the very

359: reason why the mechanism was invoked in the 80's. The recent observations 

360: prove that waves can grow to $\delta B/B\gg 1$. It is therefore at

361: least not self-consistent to apply the standard treatment for TH to

362: situations in which MF 

363: amplification to the non-linear regime takes place. In a minimal attempt

364: to include faster growth one may assume that $\Gamma=\alpha \sigma$,

365: with $\alpha<1$. Following MKV82 and \cite{be99} one can then 

366: obtain a generalized relation between $\Rt$ and $\Rs$ in the form

367: \begin{equation}\label{rsrtTH}

368: \Rt^{\gamma+1}=\frac{M_0^2\Rs^\gamma}{2}\left[\frac{\gamma+1-\Rs(\gamma-1)}  

369: {(1+\Lambda_B)(1+\Lambda_{TH})}\right],

370: \end{equation}

371: where

372: \begin{equation}	

373: \Lambda_{TH}=\alpha (\gamma-1)\frac{M_0^2}{M_A}

374: \left[1-\left(\frac{\Rs}{\Rt}\right)^\gamma\right]\,,

375: \end{equation}

376: which becomes equivalent to the standard Eq. 50 of \cite{be99} only for

377: $\alpha=1$. Now it is easy to check that for typical values of $\Rs$

378: and $\Rt$ $\Lambda_{TH}>\Lambda_B$ if $\alpha\gtrsim

379: 3W\frac{M_A}{M_0^2}$. For instance for $M_A\sim 10^3$ and $M_0\sim

380: 100$ one requires $\alpha$ to be of order unity. In this case

381: however it is not easy to amplify the MF to $\delta B\gg

382: B_0$. If $\alpha$ is appreciably smaller than unity, the main process

383: for the smoothening of the precursor is the dynamical reaction of the

384: self-generated MF. In both cases the role of TH can be

385: seriously questioned. 

386: 

387: A deeper look at the physical processes that may result in the heating

388: of the precursor make the role of TH even more uncertain: in the

389: original papers of VMK81 and MKV82 the Alfven heating was

390: considered as a result of non-linear Landau damping in a gas in the

391: hot coronal phase. The authors reached the conclusion that the damping

392: is important if $u_0\ll 4000 km/s (T_0/5\times 10^5 K)^{1/2}$, where

393: $u_0$ is the shock velocity and $T_0$ is the temperature of the

394: unshocked gas. It is all but clear whether for the velocities and

395: temperatures that apply to the SNRs in Tab. \ref{tab}, non-linear

396: Landau damping is such to lead to $\alpha\sim 1$. We stress that at

397: the same time, $\alpha$ cannot be too close to unity, otherwise TH

398: inhibits the growth of $\delta B$ to the observed levels. 

399: 

400: Other types of turbulent heating may be at work but a quantitative

401: analysis of these phenomena is lacking at the present time. The

402: expression for $\Lambda_{TH}$ is however rather general, in that we

403: did not specify the mathematical form of the growth and damping

404: rates. Therefore we expect to draw similar conclusions in terms of the

405: parameter $\alpha$. 

406: 

407: This section strongly suggests that, contrary to the common wisdom,

408: the most likely reason for the smoothening of the precursor is the

409: dynamical reaction of the generated MFs rather than some

410: form of non adiabatic heating in the precursor.

411: 

412: \section{Conclusions}

413: It is well known that the effect of a MF is in general

414: that of reducing the plasma compressibility. We showed here that when

415: applied to a cosmic ray modified shock, 1) this finding implies that 

416: CR induced SI is adequate to explain the magnetization inferred from 

417: X-ray observations; 2) the downstream

418: MFs imply that $W\sim 1-10$, so that the field becomes  

419: dynamically important, since this happens whenever the magnetic

420: pressure upstream becomes comparable with the gas pressure, namely

421: when $W>0.7-1$; 3) the dynamical reaction of the

422: MF reduces the compression in the precursor, leading to

423: smaller (larger) values of $\Rt$ ($\Rs$) in agreement with the values

424: required to explain the distance between forward and reverse shock and

425: the multifrequency observations of several SNRs; 4) this effect comes

426: from first principles, though in our calculations we restricted our

427: attention to the case of Alfven waves, and is not affected by the huge

428: uncertainties typical of TH; 5) an efficient TH may smoothen the

429: precursor if $\alpha$ is close to unity, but in this case it is likely

430: to inhibit the growth of the field to $\delta B\gg B_0$. 

431: 

432: Although the underlying physics is well known, the dynamical effect of

433: the magnetic pressure has not been included in the calculations of 

434: multifrequency emission from SNRs (\cite{bv97} and successive papers),

435: so that the strong modifications predicted by NLT had to be

436: compensated by assuming TH. The only exception that we are aware of is

437: the recent work by \cite{elli06}, in which the authors perform Monte Carlo

438: simulations of the particle acceleration process including the

439: pressure of self-generated MFs. In such simulations, which

440: represent the state of the art in the field, however, thermal

441: and accelerated particles are treated in the same way, therefore the

442: condition $W\sim 1$ could not be tested. We suspect that for this

443: reason the smoothening of the precursor was attributed mostly to the

444: backreaction of the accelerated particles on the field through

445: injection. This effect is certainly present but as we showed here by

446: using only a hydrodynamical approach, the smoothening is in fact

447: mainly due to the reaction of the magnetic pressure on the background

448: plasma.

449: 

450: The smoothening of the precursor also results in two important

451: effects: 1) spectra of accelerated particles closer to power laws,

452: though the concavity which is peculiar of NL DSA remains evident. 2)

453: The maximum momentum of accelerated particles for given Mach number is

454: predicted to be somewhat larger (see \cite{bac07} for a detailed

455: discussion). Both these effects will be discussed in detail in a

456: forthcoming paper.  

457: 

458: \acknowledgments{The authors are grateful to the referee, Don Ellison, 

459: for his precious comments. This work was partially supported by 

460: PRIN-2006, by ASI through contract ASI-INAF I/088/06/0 and (for PB) by

461: the US DOE and by NASA grant NAG5-10842. Fermilab is operated by Fermi

462: Research Alliance, LLC under Contract No. DE-AC02-07CH11359 with the

463: United States DOE.}

464: 

465: \begin{thebibliography}{99}

466: 

467: \bibitem[\protect\citeauthoryear{Amato \& Blasi}{2005}]{amato1}

468: Amato, E. and Blasi, P., 2005, MNRAS Lett., 364, 76

469: 

470: \bibitem[\protect\citeauthoryear{Amato \& Blasi}{2006}]{amato2}

471: Amato, E. and Blasi, P., 2006, MNRAS, 371, 1251

472: 

473: \bibitem[\protect\citeauthoryear{Ballet}{2006}]{ballet}

474: Ballet, J., 2006, Adv. Sp. Res., 37, 1902

475: 

476: \bibitem[\protect\citeauthoryear{Bell}{1978a}]{bell78}

477: Bell, A.R., 1978a, MNRAS, 182, 147

478: 

479: \bibitem[\protect\citeauthoryear{Bell}{2004}]{bell2004}

480: Bell, A.R., 2004, MNRAS 353, 550

481: 

482: \bibitem[\protect\citeauthoryear{Bell \& Lucek}{2001}]{bl2001}

483: Bell, A.R. and Lucek S.G., 2001, MNRAS 321, 438

484: 

485: \bibitem[\protect\citeauthoryear{Berezhko \& Ellison}{1999}]{be99}

486: Berezhko E.G. and Ellison D.C., 1999, ApJ, 526, 385

487: 

488: \bibitem[\protect\citeauthoryear{Berezhko \& V\"{o}lk}{1997}]{bv97}

489: Berezhko E.G. and V\"{o}lk, H.J., 1997, APh, 7, 183

490: 

491: \bibitem[\protect\citeauthoryear{Blasi, Amato \& Caprioli}{2007}]{bac07}

492: Blasi, P., Amato, E. and Caprioli, D., 2007, MNRAS, 375, 1471B

493: 

494: \bibitem[\protect\citeauthoryear{Drury}{1983}]{druryrev}

495: Drury, L.O'C, 1983, Rep. Prog. Phys., 46, 973

496: 

497: \bibitem[\protect\citeauthoryear{Drury \& Falle}{1986}]{df86}

498: Drury, L.O'C and Falle, S.A.E.G., 1986, MNRAS, 223, 353D

499: 

500: \bibitem[\protect\citeauthoryear{Drury \& V\"{o}lk}{1981}]{dr_v81}

501: Drury, L.O'C and V\"{o}lk, H.J., 1981, ApJ, 248, 344

502: 

503: \bibitem[\protect\citeauthoryear{Giacalone \& Jokipii}{2007}]{gj07}

504: Giacalone J. and Jokipii J.R., 2007, ApJ, 663, L41

505: 

506: \bibitem[\protect\citeauthoryear{Jones \& Ellison}{1991}]{je91}

507: Jones, F.C., and Ellison, D.C., 1991, Sp. Sci. Rev., 58, 259

508: 

509: \bibitem[\protect\citeauthoryear{Malkov \& Drury} {2001}]{maldrury}

510: Malkov, M.A., Drury, L.O'C, 2001, Rep. Progr. Phys., 64, 429 

511: 

512: \bibitem[\protect\citeauthoryear{McKenzie \& V\"{o}lk}{1982}]{mck-v82}

513: McKenzie, J.F., and V\"{o}lk, H.J., A\&A, 116, 191

514: 

515: \bibitem[\protect\citeauthoryear{Parizot et al.}{2006}]{P+06} 

516: Parizot, E. et al., A\&A, 453, 387{}

517: 

518: \bibitem[\protect\citeauthoryear{Vainio \& Schlickeiser}{1999}]{vs99}

519: Vainio C., and Schlickeiser R., 1999, A\&A, 343, 303

520: 

521: \bibitem[\protect\citeauthoryear{Vladimirov, Ellison \&

522:     Bykov}{2006}]{elli06} 

523: Vladimirov, A., Ellison, D.C. and Bykov, A., 2006, ApJ, 652, 1246

524: 

525: \bibitem[\protect\citeauthoryear{V\"{o}lk, Berezhko \&

526: Ksenofontov}{2005}]{V+05}   V\"{o}lk H. J. , E. G. Berezhko and

527: L. T. Ksenofontov, A\&A, 433, 229 

528:     

529: \bibitem[\protect\citeauthoryear{V\"{o}lk \& McKenzie}{1981}]{v-mck81}

530: V\"{o}lk, H.J., and McKenzie, J.F., 1981, ICRC, 9, 246V

531: 

532: \bibitem[\protect\citeauthoryear{Warren et al}{2005}]{warren} 

533: Warren, J.S. et al., 2005, ApJ, 634, 376

534: 

535: \end{thebibliography}

536: 

537: \end{document}

538: 

539: 

540: 

541: 

542: 

543: 

544: 

545: 

546: 

547: 

548: 

549: 

550: 

551: 

552: 

553: 

554: 

555: 

556: 

557: 

558: 

559: