1: %% Electronic version

2: 

3: \documentclass[apjl]{emulateapj}

4: 

5: \usepackage{epsfig}

6: 

7: \citestyle{apj}

8: 

9: \hyphenation{brems-strah-lung}

10: \hyphenation{an-i-so-tro-pies}

11: 

12: \usepackage{ulem} % DCE

13: \usepackage{color} % DCE

14: \newcommand{\red}{\textcolor{red}}

15: \newcommand{\blue}{\textcolor{blue}}

16: \newcommand{\green}{\textcolor{green}}

17: 

18: \newcount\listnorom

19: \listnorom=0

20: \newcommand\listromanDE{\global\advance \listnorom by 1

21: {\lowercase\expandafter{(\romannumeral\listnorom)}\ }}

22: \newcommand\newlistroman{\listnorom=0}

23: %

24: \newcommand{\Iratio}{I_\mathrm{MC}/I_\mathrm{SNR}}

25: \newcommand{\CD}{contact discontinuity}

26: \newcommand{\fpp}{f_p(p)}

27: \newcommand{\fep}{f_e(p)}

28: \newcommand{\DKol}{D_\mathrm{Kol}}

29: \newcommand{\Dkol}{D_\mathrm{Kol}}

30: \newcommand{\Egam}{E_\gamma}

31: \newcommand{\McloudPP}{`molecular cloud'}

32: \newcommand{\Mcloud}{molecular cloud}

33: \newcommand{\SC}{self-consistent}

34: \newcommand{\SCly}{self-consistently}

35: \newcommand{\RadFS}{R_\mathrm{FS}}

36: \newcommand{\gamray}{$\gamma$-ray}

37: \newcommand{\gamrays}{$\gamma$-rays}

38: \newcommand{\muG}{$\mu$G}

39: \newcommand{\pcc}{cm$^{-3}$}

40: \newcommand{\nH}{n_p}

41: \newcommand{\nMC}{n_\mathrm{MC}}

42: \newcommand{\rel}{relativistic}

43: \newcommand{\Rel}{Relativistic}

44: \newcommand{\nonrel}{non-relativistic}

45: \newcommand{\rg}{r_g}

46: \newcommand{\rgz}{r_{g0}}

47: \newcommand{\MS}{M_S}

48: \newcommand{\MFA}{magnetic field amplification}

49: \newcommand{\Rtot}{r_\mathrm{tot}}

50: \newcommand{\Rsub}{r_\mathrm{sub}}

51: \newcommand{\rTP}{r_\mathrm{TP}}

52: \newcommand{\fsk}{f_\mathrm{sk}}

53: \newcommand{\etamfp}{\eta_\mathrm{mfp}}

54: \newcommand{\EffRel}{\epsilon_\mathrm{acc}}

55: \newcommand{\EffCRs}{E_\mathrm{CR}/E_\mathrm{SN}}

56: \newcommand{\Eesc}{\epsilon_\mathrm{esc}}

57: \newcommand{\tstep}{t_\mathrm{step}}

58: \newcommand{\coul}{Coulomb}

59: \newcommand{\Syn}{Synchrotron}

60: \newcommand{\syn}{synchrotron}

61: \newcommand{\synch}{synchrotron}

62: \newcommand{\brem}{bremsstrahlung}

63: \newcommand{\brems}{bremsstrahlung}

64: \newcommand{\IC}{inverse-Compton}

65: \newcommand{\pion}{pion-decay}

66: \newcommand{\Nshell}{N_\mathrm{shell}}

67: \newcommand{\Acut}{\alpha_\mathrm{cut}}

68: \newcommand{\alfcut}{\alpha_\mathrm{cut}}

69: \newcommand{\epRatio}{(e/p)_\mathrm{rel}}

70: \newcommand{\Tep}{(T_e/T_p)}

71: \newcommand{\Msun}{\mathrm{M}_{\odot}}

72: \newcommand{\Bcloud}{B_\mathrm{MC}}

73: \newcommand{\Bism}{B_\mathrm{ISM}}

74: \newcommand{\BISM}{B_\mathrm{ISM}}

75: \newcommand{\Emax}{E_\mathrm{max}}

76: \newcommand{\pmax}{p_\mathrm{max}}

77: \newcommand{\tSNR}{t_\mathrm{SNR}}

78: \newcommand{\dSNR}{D_\mathrm{SNR}}

79: \newcommand{\EnSN}{E_\mathrm{SN}}

80: \newcommand{\Mej}{M_\mathrm{ej}}

81: \newcommand{\SA}{semi-analytic}

82: \newcommand{\Inj}{\chi_\mathrm{inj}}

83: \newcommand{\NT}{non-thermal}

84: \newcommand{\NL}{nonlinear}

85: \newcommand{\TP}{test-particle}

86: \newcommand{\Pcr}{P_\mathrm{cr}}

87: \newcommand{\xx}[1]{\times 10^{#1}}

88: \newcommand{\EffAcc}{\varepsilon_\mathrm{acc}}

89: \newcommand{\EffEsc}{\varepsilon_\mathrm{esc}}

90: 

91: %

92: %DCE

93: % BELOW IS OUTLINE FORMAT

94: \newcount\Inum

95: \newcount\IInum

96: \newcount\IIInum

97: \newcount\IVnum

98: \Inum=0

99: \def\I{\global\multiply\IInum by 0 \global\multiply\IIInum by 0

100:             \global\multiply\IVnum by 0 \global\advance \Inum by 1

101:             {\the\Inum. }}

102: \IInum=0

103: \def\II{\global\multiply\IIInum by 0\global\multiply\IVnum by 0

104:        \global\advance \IInum by 1 {\the\Inum.\the\IInum. }}

105: \IIInum=0

106: \def\III{\global\multiply\IVnum by 0\global\advance \IIInum by 1

107:             {\the\Inum.\the\IInum.\the\IIInum. }}

108: \IVnum=0

109: %\def\IV{\global\advance \IVnum by 1

110: %            {\the\Inum.\the\IInum.\the\IIInum.\the\IVnum.}}

111: \def\IV{\global\advance \IVnum by 1

112:             {\the\IVnum. }}

113: %

114: \def\newoutline{\Inum=0}

115: %

116: \parindent=18truept

117: \def\reff{\noindent \hangafter=1 \hangindent=1truecm}

118: \def\izz{\vskip6pt\leftskip0pt\reff{\I}}

119: \def\ii{\vskip0pt\leftskip20pt\reff{\II}}

120: \def\iii{\vskip0pt\leftskip35truept\reff{\III}}

121: \def\iv{\vskip0pt\leftskip45truept\reff{\IV}}

122: %

123: \def\back{\leftskip0pt}

124: %

125: %DCE

126: 

127: \newcommand{\vdag}{\ensuremath{(v)^\dagger}}

128: \newcommand{\myemail}{kamae@slac.stanford.edu}

129: 

130: \shorttitle{Broadband Spectrum of SNR}

131: \shortauthors{Lee et al.}

132: 

133: \newcommand{\ka}{\ensuremath{\rm{K}_\alpha\ }}

134: \newcommand{\kb}{\ensuremath{\rm{K}_\beta\ }}

135: 

136: \newcommand{\imp}[1]{\textbf{#1}}

137: 

138: \begin{document}

139: 

140: \submitted{Accepted by ApJ, June 24, 2008}

141: %top matter

142: %\usefont{T1}{ptm}{m}{n}

143: \title{

144: %A Broadband Emission Spectrum Modeling 

145: %       of Shell-Type Supernova Remnants on the Non-Linear Diffusive Shock 

146: %       Acceleration Theory

147: 3-D Model of Broadband Emission from

148: Supernova Remnants Undergoing Non-linear Diffusive Shock

149: Acceleration}

150: \author{Shiu-Hang Lee, Tuneyoshi Kamae}

151: \affil{Stanford Linear Acceleration Center and Kavli Institute for 

152:        Particle Astrophysics and Cosmology, Stanford University, 

153:        Menlo Park, CA 94025}

154: \email{kamae@slac.stanford.edu,shia520@slac.stanford.edu}

155: \and

156: \author{Donald C. Ellison}

157: \affil{Department of Physics, North Carolina State University, 

158: Box 8202, Raleigh, NC 27695}

159: \email{don\_ellison@ncsu.edu}

160: 

161: %%\date{November 7, 2007}

162: 

163: \begin{abstract}

164: %

165: We present a 3-dimensional model of supernova remnants

166:   (SNRs) where the hydrodynamical evolution of the remnant is modeled

167:   consistently with nonlinear diffusive shock acceleration

168:   occuring at the outer blast wave. The model

169:   includes particle escape and diffusion outside of the forward shock,

170:   and particle interactions with arbitrary distributions of external

171:   ambient material, such as molecular clouds. We

172:   include synchrotron emission and cooling, bremsstrahlung radiation,

173:   neutral pion production, inverse-Compton (IC), and Coulomb

174:   energy-loss. Boardband spectra have been calculated for typical

175:   parameters including dense regions of gas external to a $1000$ year

176:   old SNR.

177: %

178: In this paper, we describe the details of our model but do not

179: attempt a detailed fit to any specific remnant. We also do not include

180: magnetic field amplification (MFA), even though this effect may be

181: important in some young remnants. In this first presentation of

182: the model we don't attempt a detailed fit to any specific remnant.

183: %

184: Our aim is to develop a flexible platform, which  can be

185: generalized to include effects such as MFA, and which can be easily

186: adapted to various SNR environments, including Type Ia SNRs,

187: which explode in a constant density medium, and Type II SNRs, which

188: explode in a pre-supernova wind.

189: %

190: When applied to a specific SNR, our

191: model will predict cosmic-ray spectra and multi-wavelength morphology

192: in projected images for instruments with varying spatial and spectral

193: resolutions.

194: %

195: We show examples of these spectra and images and emphasize 

196: the importance of measurements in the hard

197: X-ray, GeV, and TeV gamma-ray bands for investigating key ingredients in

198: the acceleration mechanism, and for deducing whether or not TeV emission

199: is produced by IC from electrons or \pion\ from protons.  

200: %

201: \end{abstract}

202: 

203: \keywords{acceleration of particles --- supernova remnants --- 

204: cosmic rays --- X-rays: general, gamma-ray}

205: 

206: \maketitle

207: %end of top matter

208: 

209: \section{Introduction}

210: 

211: Supernovae (SNe) are the only known sources capable of providing the

212: energy needed to power the bulk of the galactic cosmic rays (CRs) with

213: energies below the spectral feature called the ``knee'' around $3 \times

214: 10^{15}$\,eV  \citep[e.g.,][]{Drury83}.

215: %

216: If SNe are the main sources of Galactic CRs, the acceleration mechanism

217:   must be efficient so that $\gtrsim 10$\% of the total SN explosion

218:   energy in our Galaxy ends up in cosmic rays

219:   \citep[e.g.,][]{Hillas2005}.

220: %

221: Observational evidence that the outer blast wave shock accelerates

222: electrons to ultra-relativistic energies

223: in some young SNRs \citep[e.g.,][]{KoyamaSN1006_95},

224: %

225: and the existence of a well-developed model of particle acceleration at

226: shocks, i.e., diffusive shock acceleration (DSA)

227: \cite[e.g.,][]{Drury83,BE87,JE91} support the above contention.

228: 

229: When confronting observations with theoretical models, however, there

230: remain a number of important ambiguities and uncertainties from both the

231: observational and theoretical perspectives.  

232: Resolution of these ambiguities and uncertainties by new telescopes will

233: be essential to claim evidence for the pion-decay feature in the

234: GeV-TeV emission from SNRs.  The Gamma-ray Large Area Space

235: Telescope (GLAST), to be launched in 2008, will probe this crucial

236: energy range with unprecedented sensitivity and resolution.

237: 

238: Fundamental questions for CR origin also concern the spectral shape and

239: maximum ion energy a given SNR can produce. Electron energy spectra

240: inferred from young SNRs vary and can be 

241: substantially harder than CR electron spectra observed at Earth, even

242: after correction for propagation in the galaxy \citep[e.g.,][]{BV2006}.

243: The maximum CR ion energy SNRs actually produce will remain uncertain

244: until a firm identification of \pion\ emission is obtained and gamma-ray

245: emission is detected past a few 100~TeV, the maximum possible electron

246: energy in SNRs.

247: 

248: There remain other basic questions concerning the DSA mechanism. For

249: instance, is DSA efficient enough for nonlinear 

250: effects, such as shock smoothing and magnetic field amplification, 

251: to become important in young SNRs? 

252: %

253: How does particle injection occur and how does injection and

254: acceleration vary between electrons and protons? While the

255: galactic CR electron-to-proton ratio, $\epRatio$, of $0.01$--$0.0025$

256: observed at Earth at \rel\ energies is often used to constrain the

257: ratio in SNRs, this ratio has not been observed outside of

258: the heliosphere.\footnote{We note that while energetic electrons and

259: protons are observed from solar flares and at low Mach number

260: heliospheric shocks, these observations provide limited help for

261: understanding the high Mach number shocks expected in young SNRs and

262: other astrophysical sourses where a large fraction of the shock energy

263: is put into \rel\ particles.}

264: %

265: The $\epRatio$ ratio is crucial in deciding whether the $\gamma$-ray

266: emission from different SNRs, or observed in

267: different parts of an individual SNR, is of hadronic or leptonic origin.

268: 

269: The recent discovery of spatially thin, hard X-ray filaments in some

270: young SNRs \citep[e.g.,][]{BambaEtal2003,Uchiyama_J1713_2007} supports previous suggestions

271: \citep[e.g.,][]{Cowsik80,BL2001,RE92} that the particle acceleration

272: process can amplify the ambient magnetic field by large factors. If

273: magnetic field amplification in DSA is as large as now appears to be the

274: case \citep[e.g.,][]{BKV2003}, it will have far-reaching consequences

275: not only for understanding the origin of Galactic CRs, but for

276: interpreting \syn\ emission from shocks throughout the universe.  Since

277: shocks and related superthermal particle populations exist in diverse

278: environments, the knowledge gained from studying SNRs will have wide

279: applicability.

280: 

281: The advent of new space- and ground-based telescopes will result in

282: observations of SNRs at many different wavelengths with greatly improved

283: sensitivity and resolution. 

284: It is even conceivable that features in the CR

285: spectrum observed at Earth might be associated with nearby SNRs with

286: future observations \citep[e.g.,][]{EW1999,KobayashiEtal2001}.

287: 

288: In order to take full advantage of current and future observations, and

289: to improve our understanding of the DSA mechanism, the data must be

290: analyzed with consistent, broadband photon emission models including

291: nonlinear effects.

292: %

293: This has prompted us to develop a three-dimensional model of young SNRs

294: where the evolution of the remnant is coupled to nonlinear diffusive

295: shock acceleration (NL-DSA) \citep[e.g.,][]{EDB2004,EC2005}, in an

296: environment with an arbitrary mass distribution. We focus on radiation

297: from CR electrons and protons and leave the modeling of heavier ions for

298: future work. 

299: %

300: In this preliminary study, we also ignore other possible

301: acceleration processes, most notably second-order stochastic

302: acceleration, and do not include magnetic field amplification.

303: 

304: We believe our work is a significant advance over previous work for

305: several reasons.  Of particular importance is that we include

306: ``escaping'' particles \SCly. In NL-DSA, a sizeable fraction of the SN

307: explosion energy can be put into very energetic CRs that escape the

308: forward shock and stream into the surrounding ISM. These particles will

309: produce detectable radiation if they interact with dense, external

310: material.  Another advantage is that we have a ``coherent'' model,

311: easily expandable to include more complex effects, where the various

312: environmental and theoretical parameters can be straightforwardly varied

313: and the resulting radiation can be compared directly with observations.

314: %

315: This is important because all SNe and SNRs are different and complex

316: with many poorly constrained parameters. It is essential that the

317: underlying theory consistently model broad-band emission from radio

318: to TeV $\gamma$-rays taking into account individual characteristics of

319: the remnants and their environments.

320: 

321: In Sections~\ref{section:DSA} and \ref{section:Diff_Interaction}

322: we give a brief general description of nonlinear diffusive

323: shock acceleration and describe the environmental and model

324: parameters required for a hydrodynamical solution.

325: We place a time-dependent, spherically symmetric, hydrodynamic

326: calculation of a SNR, including NL-DSA, in a three-dimensional box

327: consisting of $51 \times 51 \times 51$ cells.\footnote{The resolution of

328: the 3-D box is, of course, adjustable and limited only by computational

329: considerations.}

330: %

331: The energetic particles produced by the outer blast

332: wave shock propagate through the simulation box where

333: they interact with an arbitrary distribution of matter placed 

334: external to the outer shock.

335: %

336: The energetic particles in the box, including those within the SNR,

337: suffer energy losses and produce broad-band continuum emission spectra

338: by interacting with the magnetic field, photon field, and matter density

339: of each cell. In Section~\ref{section:results} we show some

340: examples including line-of-sight projections of the emitted radiation

341: which are suitable for comparison with observations.

342: 

343: There are a number of young SNRs under active investigation,

344: including SNR RX~J1713

345: \citep[e.g.,][]{AharonianJ1713_2006,Uchiyama_J1713_2007}, Vela

346: Jr. \citep[e.g.,][]{AharonianVela2005}, RCW~86

347: \citep[][]{Hoppe_etal_RCW86_2007,Ueno_etal_RCW_2007,RhoEtal2002},

348: IC~443 \citep[][]{Albert_etal_IC443,Humensky_etalIC443} and

349: W~28 \citep[][]{AharonianW28}.

350: %

351: However, here we concentrate on a general study using various

352: parameters typical of young, shell Type Ia SNRs and leave detailed

353: modeling of individual remnants for future work.

354: 

355: \section{Diffusive Shock Acceleration in SNRs}

356: \label{section:DSA}

357: %

358: \subsection{The Diffusive Shock Acceleration Theory}

359: %

360: In the \TP\ approximation, diffusive shock acceleration

361: produces superthermal particles with a power law distribution where the

362: power-law index depends only on the shock compression ratio, i.e., $f(p)

363: \propto p^{-\sigma}$, where $\sigma= 3\rTP/(\rTP -1)$, $\rTP$ is the

364: \TP\ shock compression ratio, $p$ is the particle momentum, and $f(p)$

365: is the phase-space distribution function \citep[see][ and references

366: therein]{Drury83,BE87}.

367: This \TP\ result holds as long as the pressure exerted by the

368: accelerated particles (i.e., cosmic rays), $\Pcr$, is small compared to

369: the far upstream momentum flux, $\rho_0 u_0^2$ ($\rho_0$ is the

370: unshocked density and $u_0$ is the unmodified shock speed). There is

371: considerable observational evidence, however, that DSA is intrinsically

372: efficient and shocks with high sonic Mach numbers $\MS \gtrsim 10$ are

373: expected to accelerate particles efficiently enough that $\Pcr \sim

374: \rho_0 u_0^2$. In this case, the pressure in accelerated particles feeds

375: back on the shock structure in a strongly nonlinear fashion

376: \citep[e.g.,][]{JE91}.

377: 

378: \newlistroman

379: In NL-DSA, the following effects become important:

380: %

381: \listromanDE a precursor is formed upstream of the viscous subshock with

382: a length scale comparable to the diffusion length

383: %

384: of the highest momentum particles the shock produces. In the shock

385: reference frame, the incoming plasma is decelerated and heated in the

386: precursor before it reaches the subshock;

387: %

388: \listromanDE the production of relativistic particles, and the escape of

389: some fraction of the highest energy particles from the precursor, soften

390: the equation of state of the plasma, making the plasma

391: more compressible and allowing the overall shock

392: compression ratio to increase, i.e., $\Rtot > \rTP$;

393: \listromanDE the simple power law of the \TP\ approximation is replaced

394: by a concave spectrum at superthermal energies. The spectrum is softer than

395: the \TP\ power law for low momentum particles and harder for high

396: momentum particles; and

397: %

398: \listromanDE the weak subshock has a compression ratio $\Rsub< \rTP$ so

399: that the shocked plasma has a lower temperature than would be the case

400: in the \TP\ approximation \citep[see][ and references therein for

401: detailed discussions of these effects]{BE99}.  

402: 

403: The modification of the equation-of-state by the production of \rel\

404: particles and the escaping energy flux in NL-DSA, influences the

405: evolution of the SNR and numerical approaches have been developed to

406: describe this process \citep*[e.g.,][]{BEK96a,EDB2004}.

407: %

408: Here we generalize the basic NL-SNR model by including CR propagation

409: within the remnant and, most importantly, in surrounding material using

410: a three-dimensional simulation. The escaping particle flux is expected

411: to dominate interactions outside of the SNR blast wave.

412: 

413: \subsection{CR-Hydro Simulation}

414: %

415: We calculate the hydrodynamic evolution of a SNR with a spherically

416: symmetric model described in detail in \citet{EPSBG2007} and references

417: therein (see Fig.~\ref{Boxel_fig}).

418: %

419: The model couples efficient DSA to the hydrodynamics using the \SA\ model

420: of \citet*{BGV2005} \citep[see also][]{AB2005,AB2006}.

421: %

422: Given an injection parameter, $\Inj$ \citep[this is $\xi$ in

423: equation~(25) in][]{BGV2005}, the \SA\ model 

424: %

425:   calculates the full proton distribution function $\fpp$ at each

426:   time-step of the hydro simulation, along with the overall shock

427:   compression ratio, $\Rtot$, and the subshock compression ratio,

428:   $\Rsub$. The hydro provides the required input for the \SA\

429:   calculation, i.e., the shock speed, shock radius, ambient density and

430:   temperature, and the ambient magnetic field, and $\fpp$ reflects the

431:   \NL\ effects from efficient acceleration.

432: %

433: The coupling between the hydro and NL-DSA is accomplished by using $\fpp$,

434:   and the escaping particle flux, to calculate an effective ratio of

435:   specific heats which is then used in the hydrodynamic equations. The

436:   electron spectrum, $\fep$, is determined from $\fpp$ with two

437:   additional parameters, the electron-to-proton ratio at \rel\ energies,

438:   $\epRatio$, and the temperature ratio immediately behind the shock, $\Tep$

439:   \citep[see][ for a full discussion]{EDB2004}.

440: %

441: 

442: In this paper, we only consider Type Ia supernovae with no pre-SN wind.

443: We also ignore any CR production that might occur at the reverse

444: shock. Both of these restrictions are for clarity and our model can

445: be applied to Type II SNe with winds and can calculate particle heating

446: and acceleration at reverse shocks.

447: The parameters controlling our results fall into two catagories:

448: Environmental parameters and model parameters. These are listed in the

449: following sections with either default values or the

450: range of values used for our examples.

451: 

452: \newlistroman

453: \subsubsection{Environment Parameters}

454: \label{section:param_envir}

455: %

456: The environmental parameters include:

457: \listromanDE

458: the SN explosion energy, $\EnSN = 10^{51}$\,erg,

459: %

460: \listromanDE

461: the ejecta mass, $\Mej = 1.4\,\Msun$,

462: %

463: \listromanDE

464: the distance to the SNR, $\dSNR = 1$\,kpc,

465: %

466: \listromanDE

467: the age of the SNR, $\tSNR = 1000$\,yr,

468: %

469: \listromanDE the ISM proton number density,

470: $n_p= 0.1, 1,$ or $10$~cm$^{-3}$,

471: %

472: \listromanDE the proton number density in the \Mcloud\ if present

473: $\nMC=10^3$\,\pcc,

474: %

475: \listromanDE

476: the ambient, i.e., unshocked, magnetic field, $\Bism = 3$\,\muG, and

477: %

478: \listromanDE

479: the ambient proton temperature, $T_p = 10^4$~K.

480: %

481: The quantities $n_p$, $\Bism$, and $T_p$ are assumed to be constant in the

482: region outside of the forward shock.

483: 

484: \newlistroman

485: \subsubsection{Model Parameters}

486: \label{section:param_model}

487: %

488: The model parameters used in this simulation are:

489: \listromanDE

490: an exponential ejecta density profile applicable to Type Ia SNe,

491: %

492: \listromanDE the acceleration efficiency for DSA, $\EffRel=\EffCRs$,

493: where we consider two possibilities: the test-particle case where 1\% of

494: the total SN explosion energy is put into CR energy,

495: $E_\mathrm{CR}$, during the 1000\,yr evolution of the SNR,

496: and \NL\ DSA, where 75\% of the SN explosion energy is put into CRs

497: during 1000\,yr,\footnote{These percentages include CRs that escape

498: upstream from the forward shock during the SNR evolution.}

499: %

500: \listromanDE

501: the electron to proton ratio at relativistic energies, 

502:       $\epRatio = 0.01$,

503: %

504: \listromanDE the electron to proton temperature ratio immediately behind

505: the forward shock, $\Tep = 1$,

506: %

507: \listromanDE

508: the cutoff index for the shape of particle spectra near $\Emax$, 

509:       $\alfcut = 1$,

510: %

511: \listromanDE

512: %

513: the number of gyroradii in a mean free path, $\etamfp=1$,\footnote{This

514:   parameter is discussed more fully in Section~\ref{section:diff} below.}

515: % 

516: \listromanDE

517: %

518: the fraction of the forward shock radius, $\fsk=0.05$, used to truncate

519: DSA,\footnote{The maximum proton energy produced by the shock,

520: $\Emax$, is determined by either the finite shock age, $\tSNR$, or

521: the finite size of the shock, whichever occurs first. Our choice of

522: $\fsk=0.05$ is arbitrary but is consistent with previous work

523: \citep[e.g.,][]{EC2005}. For this particular $\fsk$, $\Emax$ is

524: determined by the finite shock size in all of our

525: examples.\label{FN:maxsize}}

526: %

527: \listromanDE the number of shells between the forward shock

528: and the contact discontinuity at the end of the simulation,

529: $\Nshell = 20$, and

530: %

531: \listromanDE

532: the diffusive time step interval, $\tstep = 10$\,yr.

533: %

534: All of these parameters, except $\nMC$ and $\tstep$, are described in

535: detail in \citet{EC2005} and \citet{EPSBG2007}.

536: 

537: The geometry of the magnetic field that is used as input to the DSA

538: calculation and to calculate the \syn\ emission is not described

539: explicitly in the CR-hydro simulation. Instead, it is assumed that the

540: field immediately upstream from the FS, $B_0=\Bism$, is turbulent and,

541: as in \citet{VBKR2002}, we set the immediate downstream compressed field

542: to $B_2 = B_0\sqrt{1/3 + 2\Rtot^2/3}$, where $\Rtot$ is the overall

543: shock compression ratio.

544: %

545: The magnitude of the shocked field evolves as the density of the plasma

546: changes, as described in \citet{EC2005}, and the magnetic pressure is

547: included in the hydrodynamics, although it is insignificant for the

548: results we show here.

549: %

550: An important limitation of our current model is that we do not include

551: self-generated magnetic turbulence or \MFA. Magnetic field amplification

552: is only now being studied in nonlinear calculations

553: \citep[e.g.,][]{AB2006,VEB2006} and we leave implementation of this

554: important aspect of DSA for future work.  We also neglect other

555: wave-particle effects, such as wave-damping \citep[e.g.,][]{Pohl2005},

556: and simply assume that the shocked field is turbulent enough for Bohm

557: diffusion to occur with a background field that is compressed at the

558: shock and evolves adiabatically behind the shock.

559: 

560: \subsection{Model Geometry and Simulation Method}

561: %

562: We treat the SNR hydrodynamics in 1-D by assuming a spherically symmetric

563: structure for the region of the remnant between the forward shock (FS)

564: and the contact discontinuity (CD).

565: %

566: The main generalization we have made to the CR-hydro model of

567: \citet{EPSBG2007} is to imbed the SNR in a fully 3-D astrophysical

568: environment where CRs accelerated by the remnant propagate and interact

569: with ambient material. A cross-section of the 3-D simulation box is

570: shown in Fig.~\ref{Boxel_fig}.

571: %

572: Spatially dependent environmental aspects, like matter density in a

573: molecular cloud, magnetic field strengths, and the

574: magnetic turbulence spectrum are all defined and stored in 3-D

575: simulation cells.

576: 

577: \begin{figure}        % Fig 1

578: \epsscale{1.0} \plotone{f1.eps}

579: \caption{Cross-section of the 3-D boxel simulation (not to scale). For

580:   the results shown in this paper, the box is divided into $51 \times 51

581:   \times 51$ cells. Only a few representative SNR shells are shown in

582:   this sketch. The actual number of spherically symmetric shells between

583:   the forward shock (FS) and \CD (CD) increases with time

584:   and equals 20 at the end of the simulation. We show the `molecular

585:   cloud' discussed in Section~\ref{section:profile}. The `molecular

586:   cloud' shell discussed in Section~\ref{section:MC_shell} is not shown

587:   for clarity.

588: \label{Boxel_fig}}

589: \end{figure}

590: 

591: The various interactions and photon emission processes

592: are computed throughout the simulation box so that mult-wavelength

593: spectra and projected morphology are obtained.

594: %

595: 

596: The temporal sequence of the evolving SNR is as follows:

597: For a SNR of a given age, $\tSNR$, we divide its life span into

598:   $\Nshell$ epochs. During each epoch, the forward shock propagates into

599:   the ambient medium and new CRs (and shock-heated ISM plasma) are

600:   produced. A new spherical shell containing the shocked thermal

601:   plasma and new CRs is created.  In every subsequent epoch, this

602:   spherical shell of material evolves while another new shell is

603:   produced. In this way, an ``onion skin'' structure is formed of shells

604:   containing CRs of various ages (see Fig.~\ref{Boxel_fig}). The

605:   evolution of the shells includes the hydrodynamics (i.e., adiabatic

606:   effects), changes in the assumed frozen-in magnetic field, and losses

607:   from radiation and \coul\ processes for electrons. 

608: %

609: Spatial diffusion of CRs,

610:   magnetic field evolution in the shells, and fast synchrotron

611:   losses for electrons  

612:   are treated using a finer timescale through further dividing each epoch into 

613:   a number of time steps $\tstep = 10$\,yr. 

614: %

615: As the local

616:   magnetic field evolves in the shocked material, the local diffusion

617:   coefficient is modified accordingly.

618: 

619: As mentioned in footnote~\ref{FN:maxsize}, the maximum CR energy in

620: our examples is determined by the finite shock size. Particles that

621: reach this energy escape and, for efficient DSA, carry away a

622: sizable fraction of the total energy flux.

623: %

624:  

625: %

626: For each epoch, the CR-hydro simulation determines the escaping flux and

627: maximum CR energy, $\Emax$, for electrons and protons in the outermost

628: shell immediately behind the FS where CR acceleration is taking

629: place. These particles are added to the simulation box in a spherical

630: shell immediately in front of the FS.  While the precise energy

631: distribution of the escaped particles is still largely unknown (the

632: shape is not determined by the CR-hydro model), we assume the escaped

633: CRs have a Gaussian distribution in momentum-space centered at $\Emax$

634: and normalized to the total escaped flux ($\Emax$ and the total escaped

635: flux are determined by the CR-hydro code) \citep{ZP2008}. 

636: The width of the Gaussian is

637: determined by fitting the high-energy spectral cut-off around $\Emax$ of

638: the newly accelerated CRs. The width of this cut-off depends on our

639: model parameter, $\alfcut$.

640: %

641: 

642: As time progresses, the energetic electrons and protons diffuse in both

643: the SNR shells and in the external material with momentum-dependent

644: diffusion coefficients described in the next section. As the CRs

645: diffuse, they interact with the astrophysical environment, such as the

646: cosmic microwave background (CMB) radiation or a

647: molecular cloud, and the photon emissivity is recorded as a 3-D map for

648: later analysis.

649: %

650: 

651: \section{Diffusion and Interaction Processes}

652: \label{section:Diff_Interaction}

653: 

654: \subsection{Diffusion}  % Fig 2

655: \label{section:diff}

656: 

657: In the simulation, particles spatially diffuse in two distinct regions;

658: the volume inside the shocked SNR shells and the ambient ISM

659: outside of the FS. For the shocked material, we assume Bohm diffusion

660: while for the unshocked ISM we assume much weaker diffusion. For

661:   this study, we assume Kolmogorov turbulence dominates outside of the

662:   SNR but other forms could be used instead. 

663: %

664: Inside and outside of the SNR, we assume the

665: turbulence is strong enough to ensure isotropic diffusion over length

666: scales $<1$\,pc.

667: %

668: 

669: If $\lambda$ is the scattering mean free path, the diffusion

670:   coefficient can be written as:

671: %

672: %

673: \begin{equation} \label{eq:Diff_general}

674: D =\frac{1}{3} \lambda v   

675: \ ,

676: \end{equation}

677: %

678: or, if we assume $\lambda$ is proportional to some power of the

679: gyroradius,

680: %

681: \begin{equation}

682: D = D_0 \beta \left ( \rg / \rgz \right )^s

683: \ ,

684: \end{equation}

685: %

686: where $v$ is the particle speed, $\beta=v/c$, $\rg = pc/(eB)$ is the

687: gyroradius in cgs units, 

688: $\rgz$ is some constant reference length, 

689: $D_0 = \etamfp \rgz c/3$ is a normalization

690: constant, and $s$ depends on the magnetic turbulence spectrum.

691: 

692: For Bohm diffusion, $\etamfp=s=1$ and

693: %

694: \begin{equation} \label{eq:Diff_Bohm}

695: D_{B}=\frac{v}{3} \left ( \frac{pc}{e B(\mathbf{r},t)} \right )

696: \ .

697: \end{equation}

698: Bohm diffusion is assumed throughout the shocked gas and the magnetic

699:   field in a particular shell, $B(\mathbf{r},t)$, depends on the

700:   location of the shell and its age.

701: %

702: We assume $B(\mathbf{r},t)$ is `frozen-in' and the details of the field

703:   evolution are given in \citet{EC2005}.

704: %

705: 

706: For the volume outside of the FS, Kolmogorov turbulence

707: is assumed ($s=1/3$) and the normalization of the diffusion

708: coefficient $D_\mathrm{Kol}$ is taken from \citet{PMJSZ2006}, 

709: a value determined to reproduce the observed CR spectra at Earth, i.e.,

710: %

711: \begin{equation} \label{eq:Diff_Kol}

712: D_\mathrm{Kol} =

713: 0.25 \beta \left ( \frac{R}{10\mathrm{GV}} \right )^{1/3}

714: \frac{\mathrm{kpc}^{2}}{\mathrm{Myr}}

715: \ ,

716: \end{equation}

717: %

718: where $R=pc/(Ze)$ is the magnetic rigidity.

719: The diffusion coefficients are shown in Fig.~\ref{fig:Diff_Kol_Bohm}

720:   and, as expected, $D_B \ll D_\mathrm{Kol}$ since the self-generated

721:   turbulence in the shocked material is far stronger than turbulence in

722:   the relatively undisturbed ISM.\footnote{We make no attempt to \SCly\

723:   calculate the turbulence generated by CRs as they escape from the SNR

724:   and stream through the ISM.}\footnote{We note that 

725:   Eq.\ref{eq:Diff_Kol} is significantly different from one assumed 

726:   in a recent paper \citet{Gabici_Aharonian}

727:   where a strategy to search for ``PeV accelerators" in SNRs is discussed. 

728:   The difference is due to their assumption that 

729:   generation of plasma waves can suppress the diffusion coefficient 

730:   by an order of magnitude relative to that for Galactic cosmic rays.}

731: %

732: 

733: \begin{figure}        % Fig 2

734: \begin{center}

735: \epsscale{1.2} \plotone{f2.eps}

736: \caption{ \sloppy Momentum-dependence of spatial diffusion coefficients

737: for CR protons. Bohm diffusion (dashed curve)  is

738: implemented for the space inside the SNR shells ($D_{B}$);

739: The Kolmogorov spectrum (solid curve) is

740: employed for the space outside the shells for CR diffusion in the

741: ISM ($\DKol$). A field $B=128$\,\muG\ is

742: used for the $D_{B}$ plot. The

743: normalization of $\DKol$ is taken from the calculation for the Galactic ridge by

744: \citet{PMJSZ2006}.  }

745: \label{fig:Diff_Kol_Bohm}  

746: \end{center}

747: \end{figure}

748: 

749: Simple diffusion of CR particles is incorporated in the simulation in a

750: discretized manner.  In each time step and each spatial grid in the 3-D

751: simulation box, particles are exchanged between the adjacent boxels

752: according to the particle momentum, location, and density

753: gradient. 

754: %

755: The particle's location determines which diffusion coefficient is used,

756:   and the simulation resolution is mainly determined by the boxel size

757:   and time step, $\tstep$, which are user-tunable.

758: 

759: \subsection{Interaction Processes}

760: %

761: The CR interaction processes considered include

762: synchrotron radiation, bremsstrahlung, inverse-Compton scattering, and

763: neutral pion decay. Energy changes from adiabatic effects and radiation,

764: as well as \coul\ energy losses, are also included.  All of these

765: processes are treated in a fully space- and time-dependent fashion where

766: the evolution of relevant parameters, such as the magnetic field and

767: shell densities, are taken into account in each time step and boxel.

768: 

769: The details of the radiation processes can be found in \citet{Sturner97}

770: and \citet{BaringEtal99} but we note that, for IC emission, we only

771: consider CR electrons colliding with a monoenergetic and isotropic

772: photon field with an average energy density equal to that of

773: the CMB field.

774: 

775: For hadronic interactions we employ the latest parametric proton-proton

776: (p-p) model developed by \citet{Kamae06}. In this model, the total

777: inclusive inelastic p-p cross-section includes the non-diffractive (with

778: Feymann scaling violation) and diffractive components, plus the

779: $\Delta(1232)$ and $Res(1600)$ resonance-excitation contributions

780: important in the 10\,MeV to 1\,GeV range. This model alone can account

781: for $\sim 20\%$ of the GeV $\gamma$ ray excess between the EGRET

782: Galactic diffuse spectrum and previous model prediction using proton

783: data in the solar system \citep{Hunter97}.

784: %

785: 

786: \subsubsection{Coulomb Losses}

787: %

788: Coulomb losses for superthermal electrons are included in our

789:   model using equation~(10) from \citet{Sturner97}, i.e.,

790: %

791: \begin{equation}

792: \label{coul}

793: \dot{E}_\mathrm{coul} = -\left ( \frac{4 \pi e^4}{m_e c}\right ) 

794: %

795: \left [

796: \frac{\lambda(t) n_\mathrm{SNR} \eta^e_\mathrm{He}}{\beta_e} \right ]

797: [\psi(t) - \psi'(t)]

798: \ ,

799: \end{equation}

800: % 

801: where $n_\mathrm{SNR}$ is the proton number density in a shocked shell,

802:   $\beta_e = v_e/c$ is the electron $\beta$, and $t$ is the time. The

803:   definitions of the other terms are in \citet{Sturner97} but are not

804:   important for our discussion here. Equation~(\ref{coul}) shows that

805:   Coulomb losses increase for large ambient densities and low electron

806:   speeds. As a shell of shocked material ages, Coulomb losses cause the

807:   low energy part of the superthermal electron distribution to become

808:   depleted, as indicated in Fig.~\ref{fig:total_all_stack}. In all

809:   cases, Coulomb losses are insignifcant for protons.

810:   

811: \begin{figure}       % Fig 3

812: \begin{center}

813: \epsscale{1.1} \plotone{f3.eps}

814: \caption

815: { \sloppy CR spectra integrated over the whole SNR FS-CD region at

816:     1000\,yr, plotted as $p^4f(p)$, where f(p) is the phase space

817:     distribution function. In all panels, solid curves are protons and

818:     dashed curves are electrons. Also in all panels, the heavy curves

819:     are from Model A and these are compared to thin curves for Model B

820:     (top panel), Model C (middle panel), and Model D (bottom

821:     panel). Parameters for the various models are given in

822:     Table~\ref{table:models}}

823: \label{fig:total_all_stack}

824: \end{center}

825: \end{figure}

826:    

827: \section{Results}

828: \label{section:results}

829: %

830: In this initial presentation of our \hbox{3-D} simulation, we show

831: results for a set of generic Type Ia SNR models where we vary the

832: acceleration efficiency for DSA, $\EffRel=\EffCRs$, and the

833: ambient proton number density, $\nH$ (we assume the ISM is made of

834: hydrogen).  All of the other environmental and model parameters are kept

835: constant with the values given in Sections~\ref{section:param_envir} and

836: \ref{section:param_model}.

837: 

838: \input{tab1}

839: 

840: The values of $\EffRel$ and $\nH$ for the four

841: models we compute are given in 

842: Table \ref{table:models}. We have included the injection efficiency,

843: $\Inj$, in Table~\ref{table:models} where $\Inj$ determines $\EffCRs$

844: and in practice we vary $\Inj$ until we obtain the desired acceleration

845: efficiency.  We also show the fraction of $\EnSN$ 

846: that is in escaping particles, $\Eesc$.

847: %

848: Model A is used as a reference for the other three and, for all of the

849: models, the duration of each epoch is 50\,yr so we have 20 shells in the

850: SNR when it is 1000\,yr old.

851: 

852: \subsection{Electron and Proton Spectra}

853: %

854: In Fig.~\ref{fig:total_all_stack} we show electron and proton

855: phase-space distributions for the four models listed in

856: Table~\ref{table:models}.

857: %

858: We plot $p^4f(p)$ to emphasize the spectral

859: curvature at \rel\ energies and the spectra are integrated over the

860: entire shocked region between the FS and CD at the end of the

861: simulation. 

862: 

863: In the top panel, we compare Models A (efficient NL-DSA; bold lines) and

864: B (inefficient DSA or TP acceleration; thin lines).  The \TP\ model

865: shows flat electron and proton spectra at \rel\ energies [$f(p)\propto

866: p^{-4}$] with considerably lower fluxes at \rel\ energies than Model

867: A. The `thermal' portions of the spectra show that the TP shock produces

868: higher temperatures than in Model A, a characteristic feature of NL-DSA.

869: %

870: The structure seen in the `thermal' portions of the spectra comes

871: about because these spectra are summed over the various shells and

872: the ones produced early on have less efficient DSA and have a

873: higher temperature than later shells.

874: 

875: \newlistroman

876: In the middle and lower panels of Fig.~\ref{fig:total_all_stack} we keep

877: $\EffRel=0.75$ but vary $\nH$; $\nH=0.1$\,\pcc\ in the middle panel and

878: $\nH=10$\,\pcc\ in the lower panel. The important points for this

879: comparison are: 

880: %

881: \listromanDE the CR flux at \rel\ energies scales approximately as

882: $\nH$, as expected,

883: %

884: \listromanDE the maximum proton momentum scales inversely as $\nH$

885: \citep[see, for example,][]{BaringEtal99},

886: %

887: \listromanDE the electron cutoff energy also scales inversely as $\nH$

888: but is influenced by radiation losses and the dependence is

889: weaker than for protons, and

890: %

891: \listromanDE the shocked temperature scales inversely as $\nH$,

892: although this is not immediately clear from the figures since the

893: `thermal' portions of the distributions are made up of contributions

894: from a range of temperatures and densities.

895: %

896: 

897: Of course, other aspects of the hydrodynamics depend strongly on

898: $\nH$. The radius of the SNR at $\tSNR=1000$\,yr is considerably greater

899: for $\nH=0.1$\,\pcc\ ($\RadFS=6.7$\,pc) than for

900: $\nH=10$\,\pcc\ ($\RadFS=3.1$\,pc).

901: %

902: It is also expected that the FS will weaken faster with time 

903: for a denser upstream medium. However the strength, in terms of the

904: efficiency of NL-DSA, also depends on the magnetic field and 

905: for the parameters used here, Model D has a larger compression ratio at

906: $\tSNR=1000$\,yr ($\Rtot=11.9$) than Model A ($\Rtot = 10.7$).  

907: 

908: Coulomb losses also increase as $\nH$ increases and the dip which

909: appears just above the thermal peak in the Model D electron spectrum

910: (light dashed curve in lower panel) reflects Coulomb losses experienced

911: by the superthermal electrons as they collide with the shocked thermal

912: gas. Coulomb losses can be expected to be more pronounced in NL models

913: because the larger compression ratio results in a larger post-shock

914: density. 

915: 

916: \subsection{Spatial Variation and Escaping Flux}

917: 

918: At any given time, the spatial variation of the CR spectrum can be

919: calculated. Fig.~\ref{fig:spatial_A} shows CR spectra of Model A at

920: three different locations: (i) just behind the forward shock (solid

921: lines); (ii) mid-way between the forward shock and the contact

922: discontinuity (dashed lines); and (iii) at a distance of $d=9$\,pc from

923: the center of the SNR which is approximately $2 - 6$\,pc

924: beyond the FS, depending on the model (dotted lines).

925: The heavy-weight curves are protons and the light-weight curves are

926: electrons.

927: 

928: \begin{figure}       % Fig 4

929: \begin{center}

930: \epsscale{1.2} \plotone{f4.eps}

931: \caption

932: {\sloppy 

933: %

934: Spatial variation of CR momentum distribution for Model A at t =

935: 1000yr. The distribution is plotted at three locations: (i) just

936: behind the FS (solid); (ii) mid-way between the FS and the CD (dashed);

937: and (iii) at a distance of 9pc away from the SNR center

938: (dash-dot). Thick lines are for protons and thin lines are for

939: electrons.  }

940: \label{fig:spatial_A}

941: \end{center}

942: \end{figure}   

943: 

944: Compared with the freshly accelerated electrons at location (i), many of

945: the highest energy electrons are lost in the mid-point location (ii)

946: mainly due to synchrotron losses with a small contribution from

947: adiabatic losses.  The protons show a smaller change which is due to

948: adiabatic losses only.

949: 

950: At location (iii), only those CRs that escaped from the shock are

951: present and their spectra lack a low energy component since low

952: energy CRs remain trapped in the remnant. The hardness of the spectra at

953: 9\,pc from the center of the SNR reflects the strong momentum dependence

954: of the escape probability and the spatial diffusion coefficients. The

955: escape probability from the SNR increases with energy and high-energy

956: CRs diffuse faster in the ISM.

957: 

958: \subsection{Boardband Photon Spectrum}

959: %

960: Once the particle spectra are determined, the photon emission can be

961: calculated throughout the simulation box for arbitrary 3-dimensional

962: distributions of matter and ambient photon fields.

963: 

964: We consider two simple matter distributions (i.e., `molecular clouds')

965:  outside of the FS.  The first is a spherical shell,

966:  concentric with the SNR where the inner and outer radii are equal

967:  to 9 and 10\,pc, respectively. The second is a hemisphere centered at one side

968:  of the simulation box with radius = 3.2\,pc (see Fig.~\ref{Boxel_fig}).

969: %

970: In both cases, the proton number density in the `molecular cloud' is

971: $\nMC=10^3$\,\pcc\ and the magnetic field is $\Bism=3$\,\muG, the

972: same field as in the ISM.

973: %

974: The entire simulation box is 20\,pc on a side and is divided into

975: $51\times51\times51$ boxels.

976: %

977: The density in the ISM between the molecular clouds and the FS is $\nH$

978: and the photon field throughout

979: the simulation box is the uniform CMB field for all models.

980: 

981: \subsubsection{Emission from the SNR Shells}

982: \label{section:shell}

983: 

984: Fig.~\ref{fig:shell_total_comp} shows the broadband

985: photon emission for Models A to D integrated over the shocked SNR shells

986: between the FS and CD. The bottom panel shows the total

987: spectra while the upper three panels show the

988: individual components from $\pi^0$-decay (solid), IC (dashed),

989: synchrotron (dash-dotted), and bremsstrahlung (dotted) compared with

990: Model A.  Emission from CRs outside of the SNR is not shown.

991: %

992: 

993: \begin{figure}       % Fig 5

994: \begin{center}

995: \epsscale{1.1}

996: \plotone{f5.eps}

997: \caption{%\sloppy

998: Photon spectra of all four models integrated over the

999: region from the CD to the FS. Upper three panels:

1000: Models B to D are compared to Model A and

1001: are split into individual components for different emission

1002: mechanisms: $\pi^0$-decay (red), IC (blue),

1003: bremsstrahlung (green) and synchrotron radiation (black). Solid lines

1004: represent spectra for Model B, C and D in each panel while Model A is

1005: shown as dashed lines.  Bottom panel: The contributions from all

1006: mechanisms are summed for each

1007: model: Model A (thick solid), Model B (thin solid), Model C (dash) and

1008: Model D (dash-dot).  

1009: %This array shows how the relative dominance of

1010: %emission mechanisms varies with the model parameters, namely

1011: %acceleration efficiency (Model A vs B) and ambient particle densities

1012: %(Model A vs C and D), in different energy bands.

1013: }

1014: \label{fig:shell_total_comp}

1015: \end{center}

1016: \end{figure}

1017: 

1018: In Model A, the photon flux in the radio to X-ray energy range is

1019: dominated by synchrotron emission up to $\sim 100$\,keV.  The second

1020: largest contribution is from thermal bremsstrahlung which dominates 

1021: between $\sim 100$\,keV and $\sim 50$\,MeV. 

1022: Between $\sim 50$\,MeV and $\sim 10$\,GeV, \pion\ and IC compete.

1023: Beyond $\sim 10$\,GeV, the emission is dominated by IC.

1024: 

1025: As seen in the A vs. B comparison panel, thermal bremsstrahlung

1026: plays an important role in the TP Model B and dominates \syn\ emission

1027: in the entire X-ray energy band.  Thermal bremsstrahlung is strong in

1028: the TP model because the shocked temperatures are considerably higher

1029: than those in efficient DSA.  The emission from \syn, IC, and

1030: \pion\ are all weak in the TP case as expected.

1031: %

1032: 

1033: In the three NL Models A, C, and D, the acceleration efficiency is set

1034: at $\EffRel=0.75$, but the ambient density

1035: is varied with $\nH =1$, $0.1$ and $10$\,\pcc\ respectively.

1036: In the X-ray band, the thermal bremsstrahlung scales approximately as

1037: $\nH^2$ and dominates \syn\ in Model D, where

1038: $\nH=10$\,\pcc.

1039: 

1040: Above $\sim 100$\,MeV, the competition is mainly between IC and \pion\

1041: but \brem\ is also important for $\nH=10$\,\pcc. For Model C ($\nH=0.1$\,\pcc), 

1042: both \pion\ and \brems\ are suppressed relative to

1043: IC. For Model D, \pion\ dominates until near the maximum energies where

1044: \brem\ becomes comparable.  

1045: 

1046: \subsubsection{Emission from a Shell 

1047: `Molecular Cloud'}

1048: \label{section:MC_shell}

1049: %

1050: We first consider the shell of external material centered with the SNR.

1051: %

1052: Protons and electrons which have sufficiently high energy and,

1053: therefore, long diffusion lengths can escape from the FS and enter the ISM.

1054: %

1055: These CRs also interact with the ambient ISM material of density $\nH$

1056: and the CMB radiation.

1057: 

1058: In Fig.~\ref{fig:MC_total_comp} we show the photon spectra from the

1059: molecular cloud shell for Models A--D in the same representation as

1060: Fig.~\ref{fig:shell_total_comp} but now integrated over the

1061: \Mcloud volume, all calculated at $\tSNR=1000$\,yr.

1062: %

1063: As expected, these spectra are considerably harder than their

1064: counterparts inside the remnant. 

1065: %

1066: The escape of CRs from the forward shock during acceleration depends on

1067:  the diffusion coefficient and the strong momentum dependence of the

1068:  Bohm diffusion coefficient, $D_B$, favors the escape of the highest

1069:  energy particles. Once in the ISM, the \rel\ CRs diffuse with a

1070:  diffusion coefficient $D_\mathrm{Kol} \propto p^{1/3}$, hardening the

1071:  spectrum even more, as shown in Fig.~\ref{fig:spatial_A}.

1072: %

1073: The photon spectra reflect the hard particle spectra.

1074: 

1075: \begin{figure}         % Fig 6

1076: \begin{center}

1077: \epsscale{1.1}

1078: \plotone{f6.eps} 

1079: \caption

1080: {

1081: \sloppy

1082: Same as Fig.~\ref{fig:shell_total_comp} but the photon spectra

1083: are now integrated over the shell \Mcloud\ volume.

1084: The strong momentum dependent diffusion results

1085: in hard spectra and a change in the relative dominance between the

1086: emission mechanisms, compared to spectra integrated over the SNR.

1087: }

1088: \label{fig:MC_total_comp}

1089: \end{center}

1090: \end{figure}

1091: 

1092: \begin{figure*}        % Fig 7 

1093: \epsscale{1.0} \plotone{f7.eps}

1094: \caption{ 

1095: Photon flux maps projected along the

1096: line-of-sight for Model A (left panels) and Model B

1097: (right panels), with a hemisphere \Mcloud\

1098: centered at pixel coordinate (50,25,25) with a radius of 3.2\,pc.

1099: The horizontal and virtical scales are in pixels

1100: where the pixel size is 0.38\,pc $\times$ 0.38\,pc.

1101: %

1102: The upper panels

1103: are integrated over the energy range $1-300$\,GeV, while

1104: the bottom panels are integrated over energies

1105: $E_{\gamma} \ge 1$\,TeV. The color scale is logarithmic in

1106: $\log(N_{\gamma}/\mathrm{cm}^2/\mathrm{s})$.

1107: The dashed circle in each panel indicates the position of the contact discontinuity. 

1108: %

1109: \label{fig:maps}}

1110: \end{figure*}

1111: 

1112: \begin{figure}        % Fig 8

1113: \epsscale{1.1} \plotone{f8.eps}

1114: \caption{

1115: %

1116: Line-of-sight emission profiles in the radial

1117: direction from the SNR center at $R=0$ to the hemispheric

1118: \McloudPP\ centered at $R=10$\,pc. 

1119: %

1120: Results for Model A (solid curves) and Model B (dashed curves)

1121: are displayed. Markers represent the centers of the cubic spatial bins which are

1122: $0.385$\,pc on a side. Four wavebands

1123: are considered - (i) Soft X-rays from 1 to 5\,keV (circle); (ii) Hard

1124: X-rays from 5 to 10\,keV (triangle); (iii) 1 to 300\,GeV $\gamma$-rays

1125: (square) and (iv) $\gamma$-rays with energy above 1\,TeV

1126: (cross). The vertical axis shows the photon flux

1127: in log($N_{\gamma}$/cm$^2$/s) for each pixel bin.

1128: }

1129: \label{fig:R_profile_model_A_vs_B}

1130: \end{figure}

1131: 

1132: \begin{figure}         % Fig 9

1133: \begin{center}

1134: \epsscale{1.1} \plotone{f9.eps} 

1135: \caption{\sloppy Same as Fig.~\ref{fig:R_profile_model_A_vs_B}

1136:     with individual emission processes shown, i.e., \pion\

1137:     (circles), IC (triangles), and bremsstrahlung (squares).

1138:     Synchrotron emission has a negligible flux in the $\gamma$-ray energy band.}

1139: \label{fig:R_profile_comp}

1140: \end{center}

1141: \end{figure}

1142: 

1143: With a number density of $\nMC=10^3$\,\pcc, and a column density of

1144: $n_\mathrm{Col} \sim 10^{21-22}$\, cm$^{-2}$ in the `molecular cloud,'

1145: $\pi^0$-decay is the main $\gamma$-ray source for all models, followed

1146: by relativistic bremsstrahlung and then IC emission, as shown in

1147: Fig.~\ref{fig:MC_total_comp}.

1148: %

1149: For Models A, B, C and D, the separation between the

1150: FS and the inner edge of the MC is found to be around $4.3$, $3.8$,

1151: $2.3$ and $5.9$\,pc respectively at $t=1000$\,yr.

1152: 

1153: %

1154: For the environmental parameters studied here, the emission at all

1155: wavebands from the \Mcloud\ is weaker than that from the SNR

1156: shell, but the difference depends on the photon energy.  With the

1157: assumption $B_\mathrm{MC} = 3$\,\muG, the X-ray \syn\ flux from the

1158: \Mcloud\ stays at the ISM level and will be difficult to

1159: detect. The GeV $\gamma$-ray flux is more model-dependent and the

1160: flux stays around a factor of 10--100

1161: smaller than the flux from the SNR.

1162: %

1163: For the TeV flux, which is detected by atmospheric Cherenkov

1164: telescopes, the \Mcloud\ emission is about $1-10\%$ of that from the

1165: SNR.

1166: % 

1167: 

1168: 

1169: 

1170: \subsection{Boardband Images and Projected Emission Profiles}

1171: \label{section:profile}

1172: %

1173: Multi-wavelength projection maps are useful for studying the

1174: energy-dependent morphology of SNRs. 

1175: %

1176: We use our hemispherical \McloudPP\ example (Fig.~\ref{Boxel_fig}) to

1177: calculate 2-D projection maps in various energy bands at

1178: $\tSNR=1000$\,yr for Models A and B.  After the photon emissivity is

1179: calculated in each boxel in the 3-D simulation box, we perform a

1180: line-of-sight projection through the box.

1181: %

1182: \newlistroman

1183: We choose four energy bands:

1184: %

1185: \listromanDE soft X-rays with $\Egam = 1-5$\,keV;

1186: %

1187: \listromanDE hard X-rays with $\Egam = 5-10$\,keV;

1188: %

1189: \listromanDE $\Egam = 1-300$\,GeV; and

1190: %

1191: \listromanDE $\Egam >1$\,TeV.  

1192: %

1193: The parameters we use result in a column density of $\sim

1194: 10^{21}$\,cm$^{-2}$ for the cloud, which is small enough to ignore in

1195: the present context.

1196: 

1197: Fig.~\ref{fig:maps} shows the

1198: $\gamma$-ray projected flux maps in

1199: $\log(N_{\gamma}/\mathrm{cm}^2/\mathrm{s})$ at a source distance of

1200: $\dSNR=1$\,kpc for Models A (NL) and B (TP) in

1201: the GeV and TeV bands [i.e., bands (iii) and (iv)].

1202: %

1203: The color scales are different for the GeV and TeV images,

1204: but the spatial resolution is the same.

1205: %

1206: The difference between the example with efficient DSA (left panels)

1207:   and the \TP\ case (right panels) is mainly one of intensity if only

1208:   GeV-TeV emission is concerned. In both cases, the brightest regions of

1209:   the SNR are considerably brighter than the cloud and for the \TP\ case

1210:   (Model B), the cloud is almost invisible on these scales. For the SNR

1211:   in both the NL and TP cases, the region between the CD and FS clearly

1212:   shows up in the maps even with the projection through the

1213:   remnant.\footnote{The dashed circle in each panel shows the

1214:   position of the CD at 1000\,yr.} 

1215: %

1216: There is also a clear limb darkening effect from projection

1217:   seen at the edge of both remnants and at the edge of the cloud in the

1218:   NL case. At the \Mcloud, however, there is no noticable drop in

1219:   intensity towards the center of the cloud as occurs for the SNR.

1220: 

1221: 

1222: 

1223: 

1224: These details, of course, depend on the particular parameters we

1225:   have chosen but some general statements can be made. 

1226:   Unless there is a source of soft photons associated with the

1227:   external material, the brightness of the external material (MC)

1228:   compared to the SNR, the $\Iratio$ ratio, will be independent of the

1229:   density ratio $\nMC/\nH$ if IC dominates the GeV-TeV emission. If

1230:   \pion\ or \brem\ dominate, the $\Iratio$ ratio will scale

1231:   approximately as the first power of the density ratio, $\nMC/\nH$.

1232:   In all cases, $\Iratio$ will decrease with the distance the external material is

1233:   from the FS. Another important result, which is implicit in

1234:   Fig.~\ref{fig:maps} and important for comparing \pion\ and IC

1235:   emission, is that emission from the SNR and the external material must

1236:   be considered together. To first order, an increase in acceleration

1237:   efficiency or ambient matter density not solely associated with the

1238:   cloud, $\nH$, will leave $\Iratio$ unchanged.

1239: %The brightness of

1240: %  the external material (MC) compared to the SNR, the $\Iratio$ ratio,

1241: %  will scale approximately as the first power of the density ratio,

1242: %  $\nMC/\nH$, if IC dominates the GeV-TeV emission, and approximately as

1243: %  the second power of $\nMC/\nH$ if \pion\ dominates. 

1244: %

1245: 

1246: In Fig.~\ref{fig:R_profile_model_A_vs_B} we show emission calculated

1247: along a horizontal line from the center of the remnant at $R=0$

1248: across the \Mcloud\ for all four energy bands. These fluxes are

1249: determined, as are the 2-D maps, by summing the emission from each boxel

1250: along a line-of-sight.

1251: The plateaus on the left hand side of the plot within $R \lesssim 5$\,pc

1252:  show emission from the SNR.  The subtle increase of the projected flux

1253:  with $R$ in this region is the result of projection through the

1254:  shell of material between the CD and the FS. 

1255: %

1256: Beyond $R\sim 5$\,pc, the fluxes drop abruptly to the ISM level. Here,

1257: escaping CRs stream through the ISM with a large diffusion

1258: coefficient $\DKol$.

1259: At $R \sim 6.8$\,pc, the CRs impact the hemisphere \McloudPP\ with

1260: $\nMC=10^3$\,\pcc\ and the fluxes for energy bins (iii) and (iv)

1261: increase by almost 2 orders of magnitude from the ISM level. 

1262: %

1263: These photons are mainly from \pion. There is no increase at the

1264:   edge of the cloud for energy bins (i) and (ii) since this emission is

1265:   totally from \syn\ and we have assumed the field in the \Mcloud\

1266:   equals the ISM field, $\Bcloud=\Bism$.

1267:   

1268: 

1269: 

1270: Fig.~\ref{fig:R_profile_comp} shows the emission profiles in bands

1271: (iii) and (iv) for Models A and B separated into individual

1272: emission mechanisms. While the total fluxes at these energies

1273: depend strongly on acceleration efficiency, the $\Iratio$ ratio is much

1274: less sensitive to $\EffRel$,

1275: as mentioned above. There is no increase in IC emission at the edge of

1276: the cloud near $R\sim 6.9$\,pc since we only consider electron

1277: scattering off the CMB. Unless there is an additional source of photons

1278: associated with the external material, IC will be strongly suppressed

1279: relative to \pion\ in external material.

1280: 

1281:    

1282: \section{Summary and Conclusions}

1283: \label{section:summary}

1284: %

1285: We have introduced a 3-D simulation of an evolving SNR where the \NL\

1286: acceleration of CRs is coupled to the SNR evolution.  The model follows

1287: the diffusion and interaction of CRs within the spherically

1288: symmetric remnant, as well as high-energy CRs that escape from the

1289: forward shock and diffuse into the surrounding medium.

1290: %

1291: For any set of model and environmental parameters, and for arbitrary

1292: distributions of matter surrounding the remnant, we can calculate

1293: broadband photon spectra and obtain line-of-sight projections and

1294: morphologies that will allow for efficient comparisons with observations

1295: in various energy bands.

1296: 

1297: We have illustrated the capabilities of this simulation with several

1298: models that differ from each other in the CR acceleration efficiency,

1299: $\EffRel$, the ambient ISM proton density, $\nH$, and the matter

1300: distribution of a \McloudPP\ external to the SNR.

1301: %

1302: Of course, all of the results discussed here assume particular

1303:   values for parameters, such as a shocked

1304:   electron to proton temperature ratio $T_e/T_p = 1$ and an

1305:   electron/proton ratio at \rel\ energies $\epRatio=0.01$.  These

1306:   parameters are critical for understanding DSA and applying the

1307:   mechanism to astrophysical sources yet they are poorly constrained by

1308:   both observations and theory.  For instance, the value of $T_e/T_p$

1309:   determines the importance of \brems\ compared to \syn\ in the X-ray

1310:   range and also strongly influences the thermal X-ray line spectra

1311:   \citep[][]{EPSBG2007}.

1312: %

1313: The $\epRatio$ ratio is the most important factor after the ambient

1314: density determining the relative intensity of IC and \pion\ emission at

1315: GeV-TeV energies. The confirmation of CR {\it ion} production in SNRs

1316: depends on this parameter.

1317: 

1318: Other important parameters of DSA that remain uncertain are the

1319: injection and acceleration efficiencies, the amount of magnetic

1320: compression and amplification that occurs, and the diffusion coefficient

1321: of escaping particles as they leave the shock, which must differ

1322: substantially from Bohm diffusion \citep[e.g.,][]{BAC2007,EV2008}.

1323: %

1324: Due to the still limited dynamic range of particle-in-cell simulations,

1325: and the lack of strong, \NL\ shocks producing \rel\ particles in the

1326: heliosphere, we believe young SNRs are the best `laboratory' for

1327: studying NL-DSA. Broadband observations matched against \SC\ \NL\ models

1328: currently provide the best constraints on these important

1329: parameters.

1330: %

1331: 

1332: There are three important aspects of our 3-D

1333: simulation that are new and extend the large body of existing work on

1334: DSA in SNRs.

1335: %

1336: One is that the simulation consistently models high-energy CRs that

1337: escape from the forward shock of the SNR with the evolution of the SNR

1338: itself. In NL-DSA, the fraction of total explosion energy that ends up

1339: in escaping particles can be large (see Table~\ref{table:models}) and we

1340: believe this is the first work to include these particles in a coherent

1341: emission model.

1342: %

1343: Second, the 3-D simulation box

1344: allows for the modeling of CR interactions in arbitrary mass

1345: distributions outside of the SNR. This feature is essential for

1346: producing 2-D projection maps that can be compared with current and

1347: future observations. 

1348: %

1349: These maps, tuned to match the instrument response of telescopes, will

1350:  serve to help determine the importance of pre-SN shells and/or nearby

1351:  molecular clouds in producing \gamray\ emission.

1352: %

1353: Third, the simulation platform is extremely flexible making it

1354: straightforward to add important effects not present in this preliminary

1355: model.  These generalizations include shock acceleration and heating at

1356: the reverse shock as well as the forward shock, pre-SN winds for Type II

1357: SNe, various forms for particle diffusion in the ISM, 

1358: %

1359: %

1360: production and interaction of heavy CR ions, and a parameterized representation of

1361: magnetic field amplification.

1362: 

1363: Another physical effect that may importantly influence the photon

1364: spectrum is anisotropy from angular-dependent interactions. These

1365: include an angular-dependent neutral pion production cross-section

1366: \citep{Karlsson07} and anisotropic IC scattering with photon fields

1367: other than the CMB \citep{Moskalenko00}. Preliminary results show that

1368: anisotropies can change the spectral shape and flux of the observed

1369: photons drastically. 

1370: %

1371: When anisotropic interactions are implemented, the projection maps we

1372: calculate will show how the observed flux depends on the orientation of

1373: the FS and \Mcloud\ with respect to the line-of-sight. We leave this

1374: issue to future studies.

1375: 

1376: 

1377: Finally, in addition to modeling the photon emission from SNRs, our

1378: model can also determine the total contribution of CR ions and electrons

1379: injected into the Galaxy from an individual SNe over its lifetime.  This

1380: can serve as input to Galaxy-scale propagation models \citep[for

1381: example, GALPROP,][]{SMP2007} and also add to our knowledge on the

1382: Galactic \gamray\ diffuse emission.

1383: 

1384: \acknowledgments

1385: The authors wish to thank Roger Blandford,

1386: Steven Kahn, Igor Moskalenko, Niklas Karlsson, Stefan Funk, Takaaki Tanaka, Johan-Cohen 

1387: Tanugi and Masaru Ueno for helpful discussions. 

1388: They are grateful to the anonymous referee for bringing new

1389: publications to their attention.

1390: D.C.E. is grateful for the hospitality 

1391: of KIPAC where part of this work

1392: was done, as well as for support from a NASA ATP grant (06-ATP06-21) and

1393: a NASA LTSA grant (NNH04Zss001N-LTSA). This work was supported in part

1394: by the U.S. Department of Energy under Grant DE-AC02-76SF00515.

1395: 

1396: \begin{thebibliography}{41}

1397: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi

1398: \expandafter\ifx\csname url\endcsname\relax

1399:   \def\url#1{{\tt #1}}\fi

1400: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi

1401: 

1402: \bibitem[{{Aharonian} et~al.(2005){Aharonian}, {Akhperjanian}, {Bazer-Bachi}

1403:   et~al.}]{AharonianVela2005}

1404: {Aharonian} F., {Akhperjanian} A.G., {Bazer-Bachi} A.R., et~al., Jul. 2005,

1405:   \aap, 437, L7

1406: 

1407: \bibitem[{{Aharonian} et~al.(2006){Aharonian}, {Akhperjanian}, {Bazer-Bachi}

1408:   et~al.}]{AharonianJ1713_2006}

1409: {Aharonian} F., {Akhperjanian} A.G., {Bazer-Bachi} A.R., et~al., Apr. 2006,

1410:   \aap, 449, 223

1411:   

1412: \bibitem[{{Aharonian} et~al.(2008){Aharonian}, {Akhperjanian}, {Bazer-Bachi}

1413:   et~al.}]{AharonianW28}

1414: {Aharonian} F., {Akhperjanian} A.G., {Bazer-Bachi} A.R., et~al., Apr. 2008,

1415:   \aap, 481, 401

1416: 

1417: \bibitem[{{Albert} et~al.(2007){Albert}, {Aliu}, {Anderhub}

1418:   et~al.}]{Albert_etal_IC443}

1419: {Albert} J., {Aliu} E., {Anderhub} H., et~al., Aug. 2007, \apjl, 664, L87

1420: 

1421: \bibitem[{{Amato} \& {Blasi}(2005)}]{AB2005}

1422: {Amato} E., {Blasi} P., Nov. 2005, \mnras, 364, L76

1423: 

1424: \bibitem[{{Amato} \& {Blasi}(2006)}]{AB2006}

1425: {Amato} E., {Blasi} P., Sep. 2006, \mnras, 371, 1251

1426: 

1427: \bibitem[{{Bamba} et~al.(2003){Bamba}, {Yamazaki}, {Ueno}, \&

1428:   {Koyama}}]{BambaEtal2003}

1429: {Bamba} A., {Yamazaki} R., {Ueno} M., {Koyama} K., Jun. 2003, \apj, 589, 827

1430: 

1431: \bibitem[{Baring et~al.(1999)Baring, Ellison, Reynolds, Grenier, \&

1432:   Goret}]{BaringEtal99}

1433: Baring M.G., Ellison D.C., Reynolds S.P., Grenier I.A., Goret P., 1999, ApJ,

1434:   513, 311

1435: 

1436: \bibitem[{Bell \& Lucek(2001)}]{BL2001}

1437: Bell A.R., Lucek S.G., 2001, MNRAS, 321, 433

1438: 

1439: \bibitem[{Berezhko \& Ellison(1999)}]{BE99}

1440: Berezhko E.G., Ellison D.C., 1999, ApJ, 526, 385

1441: 

1442: \bibitem[{{Berezhko} \& {V{\"o}lk}(2006)}]{BV2006}

1443: {Berezhko} E.G., {V{\"o}lk} H.J., Jun. 2006, \aap, 451, 981

1444: 

1445: \bibitem[{{Berezhko} et~al.(1996){Berezhko}, {Elshin}, \&

1446:   {Ksenofontov}}]{BEK96a}

1447: {Berezhko} E.G., {Elshin} V.K., {Ksenofontov} L.T., Jan. 1996, Journal of

1448:   Experimental and Theoretical Physics, 82, 1

1449: 

1450: \bibitem[{{Berezhko} et~al.(2003){Berezhko}, {Ksenofontov}, \& {V{\"

1451:   o}lk}}]{BKV2003}

1452: {Berezhko} E.G., {Ksenofontov} L.T., {V{\" o}lk} H.J., Dec. 2003, \aap, 412,

1453:   L11

1454: 

1455: \bibitem[{{Blandford} \& {Eichler}(1987)}]{BE87}

1456: {Blandford} R., {Eichler} D., Oct. 1987, Physics Reports, 154, 1

1457: 

1458: \bibitem[{{Blasi} et~al.(2005){Blasi}, {Gabici}, \& {Vannoni}}]{BGV2005}

1459: {Blasi} P., {Gabici} S., {Vannoni} G., Aug. 2005, \mnras, 361, 907

1460: 

1461: \bibitem[{{Blasi} et~al.(2007){Blasi}, {Amato}, \& {Caprioli}}]{BAC2007}

1462: {Blasi} P., {Amato} E., {Caprioli} D., Mar. 2007, \mnras, 375, 1471

1463: 

1464: \bibitem[{Chevalier et~al.(1978)}]{Chevalier78}

1465: {{Chevalier}, R.~A., {Oegerle}, W.~R. \& {Scott}, J.~S.},

1466: 1978, \apj, 222, 527-536

1467: 

1468: \bibitem[{{Cowsik} \& {Sarkar}(1980)}]{Cowsik80}

1469: {Cowsik} R., {Sarkar} S., Jun. 1980, \mnras, 191, 855

1470: 

1471: \bibitem[{{Drury}(1983)}]{Drury83}

1472: {Drury} L.O., Aug. 1983, Reports of Progress in Physics, 46, 973

1473: 

1474: \bibitem[{{Ellison} \& {Cassam-Chena{\"{\i}}}(2005)}]{EC2005}

1475: {Ellison} D.C., {Cassam-Chena{\"{\i}}} G., Oct. 2005, \apj, 632, 920

1476: 

1477: \bibitem[{{Ellison} \& {Vladimirov}(2008)}]{EV2008}

1478: {Ellison} D.C., {Vladimirov} A., Jan. 2008, \apjl, 673, L47

1479: 

1480: \bibitem[{Ellison et~al.(2004)Ellison, Decourchelle, \& Ballet}]{EDB2004}

1481: Ellison D.C., Decourchelle A., Ballet J., 2004, A\&A, 413, 189

1482: 

1483: \bibitem[{{Ellison} et~al.(2007){Ellison}, {Patnaude}, {Slane}, {Blasi}, \&

1484:   {Gabici}}]{EPSBG2007}

1485: {Ellison} D.C., {Patnaude} D.J., {Slane} P., {Blasi} P., {Gabici} S., Jun.

1486:   2007, \apj, 661, 879

1487: 

1488: \bibitem[{{Erlykin} \& {Wolfendale}(1999)}]{EW1999}

1489: {Erlykin} A.D., {Wolfendale} A.W., Oct. 1999, \aap, 350, L1

1490: 

1491: \bibitem[{{Gabici} \& {Aharonian}(2007)}]{Gabici_Aharonian}

1492: {Gabici} S., {Aharonian} F.A., Aug. 2007, \apjl, 665, L131

1493: 

1494: \bibitem[{{Hillas}(2005)}]{Hillas2005}

1495: {Hillas} A.M., May 2005, Journal of Physics G Nuclear Physics, 31, 95

1496: 

1497: \bibitem[{{Hoppe} et~al.(2007){Hoppe}, {Lemoine-Goumard}, \& {for the

1498:   H.~E.~S.~S.~Collaboration}}]{Hoppe_etal_RCW86_2007}

1499: {Hoppe} S., {Lemoine-Goumard} M., {for the H.~E.~S.~S.~Collaboration}, Sep.

1500:   2007, ArXiv e-prints, 709

1501: 

1502: \bibitem[{{Hunter} et~al.(1997){Hunter}, {Bertsch}, {Catelli}

1503:   et~al.}]{Hunter97}

1504: {Hunter} S.D., {Bertsch} D.L., {Catelli} J.R., et~al., May 1997, \apj, 481, 205

1505: 

1506: \bibitem[{{Jones} \& {Ellison}(1991)}]{JE91}

1507: {Jones} F.C., {Ellison} D.C., 1991, Space Science Reviews, 58, 259

1508: 

1509: \bibitem[{{Kamae} et~al.(2006){Kamae}, {Karlsson}, {Mizuno}, {Abe}, \&

1510:   {Koi}}]{Kamae06}

1511: {Kamae} T., {Karlsson} N., {Mizuno} T., {Abe} T., {Koi} T., Aug. 2006, \apj,

1512:   647, 692

1513: 

1514: \bibitem[{{Karlsson} \& {Kamae}(2007)}]{Karlsson07}

1515: {Karlsson} N., {Kamae} T., Sep. 2007, ArXiv e-prints, 709

1516: 

1517: \bibitem[{{Kobayashi} et~al.(2001){Kobayashi}, {Nishimura}, {Komori}, \&

1518:   {Yoshida}}]{KobayashiEtal2001}

1519: {Kobayashi} T., {Nishimura} J., {Komori} Y., {Yoshida} K., 2001, Advances in

1520:   Space Research, 27, 653

1521: 

1522: \bibitem[{{Koyama} et~al.(1995){Koyama}, {Petre}, {Gotthelf}

1523:   et~al.}]{KoyamaSN1006_95}

1524: {Koyama} K., {Petre} R., {Gotthelf} E.V., et~al., Nov. 1995, \nat, 378, 255

1525: 

1526: \bibitem[{{Moskalenko} \& {Strong}(2000)}]{Moskalenko00}

1527: {Moskalenko} I.V., {Strong} A.W., Jan. 2000, \apj, 528, 357

1528: 

1529: \bibitem[{{Pohl} et~al.(2005){Pohl}, {Yan}, \& {Lazarian}}]{Pohl2005}

1530: {Pohl} M., {Yan} H., {Lazarian} A., Jun. 2005, \apjl, 626, L101

1531: 

1532: \bibitem[{{Ptuskin} et~al.(2006){Ptuskin}, {Moskalenko}, {Jones}, {Strong}, \&

1533:   {Zirakashvili}}]{PMJSZ2006}

1534: {Ptuskin} V.S., {Moskalenko} I.V., {Jones} F.C., {Strong} A.W., {Zirakashvili}

1535:   V.N., May 2006, \apj, 642, 902

1536: 

1537: \bibitem[{{Reynolds} \& {Ellison}(1992)}]{RE92}

1538: {Reynolds} S.P., {Ellison} D.C., Nov. 1992, \apjl, 399, L75

1539: 

1540: \bibitem[{{Rho} et~al.(2002){Rho}, {Dyer}, {Borkowski}, \&

1541:   {Reynolds}}]{RhoEtal2002}

1542: {Rho} J., {Dyer} K.K., {Borkowski} K.J., {Reynolds} S.P., Dec. 2002, \apj, 581,

1543:   1116

1544: 

1545: \bibitem[{{Strong} et~al.(2007){Strong}, {Moskalenko}, \& {Ptuskin}}]{SMP2007}

1546: {Strong} A.W., {Moskalenko} I.V., {Ptuskin} V.S., Nov. 2007, Annual Review of

1547:   Nuclear and Particle Science, 57, 285

1548: 

1549: \bibitem[{{Sturner} et~al.(1997){Sturner}, {Skibo}, {Dermer}, \&

1550:   {Mattox}}]{Sturner97}

1551: {Sturner} S.J., {Skibo} J.G., {Dermer} C.D., {Mattox} J.R., Dec. 1997, \apj,

1552:   490, 619

1553: 

1554: \bibitem[{{Torres} et~al.(2008)}]{Torres2008}

1555: {Torres}, D.~F., {Rodriguez Marrero}, A.~Y., \& {de Cea del Pozo}, E.,

1556: 2008, \mnras, preprint

1557:   

1558: \bibitem[{{Uchiyama} et~al.(2007){Uchiyama}, {Aharonian}, {Tanaka},

1559:   {Takahashi}, \& {Maeda}}]{Uchiyama_J1713_2007}

1560: {Uchiyama} Y., {Aharonian} F.A., {Tanaka} T., {Takahashi} T., {Maeda} Y., Oct.

1561:   2007, \nat, 449, 576

1562: 

1563: \bibitem[{{Ueno} et~al.(2007){Ueno}, {Sato}, {Kataoka}

1564:   et~al.}]{Ueno_etal_RCW_2007}

1565: {Ueno} M., {Sato} R., {Kataoka} J., et~al., Jan. 2007, \pasj, 59, 171

1566: 

1567: \bibitem[{{V{\" o}lk} et~al.(2002){V{\" o}lk}, {Berezhko}, {Ksenofontov}, \&

1568:   {Rowell}}]{VBKR2002}

1569: {V{\" o}lk} H.J., {Berezhko} E.G., {Ksenofontov} L.T., {Rowell} G.P., Dec.

1570:   2002, \aap, 396, 649

1571: 

1572: \bibitem[{{VERITAS Collaboration: T.~B.~Humensky}(2007)}]{Humensky_etalIC443}

1573: {VERITAS Collaboration: T.~B.~Humensky}, Sep. 2007, ArXiv e-prints, 709

1574: 

1575: \bibitem[{{Vladimirov} et~al.(2006){Vladimirov}, {Ellison}, \&

1576:   {Bykov}}]{VEB2006}

1577: {Vladimirov} A., {Ellison} D.C., {Bykov} A., Dec. 2006, \apj, 652, 1246

1578: 

1579: \bibitem[{{Zirakashvili} \& {Ptuskin}(2008)}]{ZP2008}

1580: {Zirakashvili} V.N., {Ptuskin} V.S., Jan. 2008, ArXiv e-prints, 801

1581: 

1582: \end{thebibliography}

1583: 

1584: \end{document}

1585: 

1586: 

1587: 

1588: 

1589: 

1590: 

1591:  

1592: 

1593: 

1594: 

1595: 

1596: 

1597: 

1598: 

1599: 

1600: 

1601: