1: %-
2: %- File : ms.tex
3: %- ----------------
4: %- Purpose : Manuscript for Atmos2002 conference proceedings.
5: %- Created : Fri 07 Jun 2002 08:18:08 AM CDT
6: %- Author : Rollin C. Thomas (RCT) - thomas@mail.nhn.ou.edu
7: %-
8:
9: \documentstyle[11pt,newpasp,twoside,epsf]{article}
10:
11: \markboth{Thomas}{Synthetic Spectra from 3D Models of Supernovae}
12: \pagestyle{myheadings}
13:
14: \nofiles
15: \marginparwidth 1.25in
16: \marginparsep .125in
17: \marginparpush .25in
18: \reversemarginpar
19:
20: \begin{document}
21: %%
22: \title
23: {
24: Synthetic Spectra from 3D Models of Supernovae
25: }
26: %%
27: \author
28: {
29: R. C. Thomas, E. Baron, David Branch
30: }
31: %%
32: \affil
33: {
34: University of Oklahoma
35: Nielsen Hall RM 131
36: Norman, Oklahoma 73019
37: }
38: %%
39: \begin{abstract}
40: We describe \texttt{Brute}, a code for generating synthetic spectra
41: from 3D supernova models. It is based on a Monte Carlo implementation
42: of the Sobolev approximation. The current version is highly
43: parameterized, but in the future it will be extended for full NLTE
44: spectrum modelling of supernovae.
45: \end{abstract}
46: %%
47:
48: \section{Introduction}
49:
50: The amount of evidence that the envelopes of supernovae (SNe) can
51: deviate from spherical symmetry is increasing. Net intrinsic polarization
52: measurements from some SNe are consistent with ellipsoidal envelopes
53: (Howell et al.\ 2001; Leonard et al.\ 2001; Leonard et al.\ 2000; Wang
54: et al.\ 2001; Wang et al.\ 1997; see Wheeler 2000 for a list of
55: measurements). In some flux spectra, a clumpy or macroscopically
56: mixed ejecta distribution (or at least a nonspherical excitation
57: structure) has been invoked to explain phenomena such as the ``Bochum
58: event'' in SN 1987A (Phillips \& Heathcote 1989; Utrobin, Chugai \&
59: Andronova 1995). Most conspicuously, the interesting morphologies of
60: SN remnants may indicate macroscopic mixing in the initial SN
61: envelopes (Decourchelle et al.\ 2001; Fesen \& Gunderson 1996; Hwang,
62: Holt \& Petre 2000; Tsunemi, Miyama \& Aschenbach 1995). These
63: findings have inspired new 3D explosion models (Khokhlov 2000;
64: Reinecke, Hillebrandt \& Niemeyer 2002; Kifonidis et al.\ 2000).
65: Testing new models as they arise will require detailed NLTE radiative
66: transfer calculations in 3D.
67:
68: Nonetheless, considering and deducing general constraints on SN
69: geometry without such detailed calculations can be fruitful (Thomas et
70: al.\ 2002). We have developed a parameterized synthetic spectrum code
71: for 3D models of SNe called \texttt{Brute} which we describe here. In
72: \S 2 we describe the assumptions and implementation of \texttt{Brute},
73: and in \S 3 we
74: outline future work to bring the code toward a detailed analysis
75: code.
76:
77: \section{Model}
78:
79: \texttt{Brute} was designed under some of the same basic assumptions
80: as the highly successful direct analysis code \texttt{Synow} (Fisher
81: 2000; Deng et al.\ 2000; Millard et al.\ 1999). Both codes use a
82: simplified model of a SN atmosphere which has been called the
83: ``elementary supernova'' (ES; Jeffery \& Branch 1990). The major
84: difference is that \texttt{Synow} is a strictly 1D code, but
85: \texttt{Brute} is for arbitrary geometry.
86:
87: \subsection{The Elementary Supernova Model}
88: The ES model is a simplified picture of a SN at phases between a few
89: days to a few months after explosion. It consists of a central,
90: optically thick continuum-emitting core surrounded by an extended
91: envelope within which line formation occurs. The SN expands
92: homologously so that $v \propto r$ for all matter elements.
93:
94: The ES core (or photosphere) is approximated by a surface which emits
95: blackbody radiation at some specified temperature. The photosphere is
96: sharp, meaning that all radiation regardless of wavelength originates
97: at the same physical surface.
98:
99: Radiation transfer in the envelope is accomplished by means of the
100: Sobolev approximation (Sobolev 1947; Rybicki \& Hummer 1978),
101: justified in detail for the SN case by Jeffery \& Branch (1990). The
102: assumption of homologous expansion provides a great simplification in
103: the calculations, because then (1) Sobolev optical depth is angle
104: independent, (2) common point velocity surfaces for intensity
105: integrals are spheres and (3) propagating photons only redshift with
106: respect to the matter and never blueshift. For each ion used in the
107: envelope, the optical depth of a ``reference line'' is parameterized
108: spatially using a contrast function. All other line optical depths of
109: the same ion are scaled assuming LTE. The source function is taken to
110: be that of pure resonance scattering ($S = J$).
111:
112: It should be noted that the parameterized prescription here can be
113: replaced by one with more detail. The resonance scattering assumption
114: can be generalized as well. The effects of special relativity in the
115: transfer can be included (to minor effect; Jeffery 1993). Electron
116: scattering also may be included. To speed up transfer, lines close
117: together in wavelength can be grouped into bands of small width
118: to prevent needless extra calculations. For the purpose of direct
119: analysis of SN spectra however, the above parameterized
120: prescription is sufficient.
121:
122: Once the source function has been determined for each line at all
123: points in the envelope, a formal solution along impact parameter beams
124: toward an observer is calculated. Again, the Sobolev approximation
125: simplifies this process, since emission in a given wavelength from the
126: SN arises only from common direction velocity planes perpendicular
127: to the line of sight.
128:
129: \subsection{3D Implementation}
130:
131: The key difference between the 1D and 3D implementation is the method
132: by which the mean intensity is calculated. In 1D, we can take
133: advantage of an azimuthal symmetry in the intensity arriving at a
134: point to simplify the mean intensity integral. But in 3D, such
135: symmetry is missing and the integral becomes more problematic. To
136: avoid this problem, we adopt a Monte Carlo approach, based largely on
137: 1D codes (Abbott \& Lucy 1985; Mazzali \& Lucy 1993; Mazzali 2000).
138:
139: Packets of equal energy with wavelengths drawn from the Planck
140: distribution are emitted from the surface of the core such that the
141: core is a Lambert radiator. The packet random walks from line
142: scattering to line scattering as it redshifts into resonance with
143: lines of increasing wavelength. The probability that a packet
144: undergoes a scattering (coherent in the comoving frame) is determined
145: by the optical depth at the resonance targets.
146:
147: As each packet enters a new cell of the envelope or immediately after a
148: scattering, an ``event'' optical depth is chosen to ``schedule'' the
149: next possible scattering : $\tau_{ev} = - \ln( R )$ where $R$ is a
150: random number between 0 and 1. As the packet propagates through the
151: cell, it accumulates optical depth from the lines it encounters until
152: the total exceeds $\tau_{ev}$. The packet then scatters coherently.
153:
154: From the perspective of a single cell in the envelope during a
155: simulation, packets of equal energy intersect the cell from many
156: directions. The locus of points defined by the packets' previous
157: resonance targets form the common point velocity surfaces. The Monte
158: Carlo technique assembles these surfaces implicitly. The radiation
159: field is built up by counting packets that come into resonance with
160: lines everywhere in the envelope. At the end of the simulation, the
161: total luminosity is used to calibrate the weight of a single packet,
162: converting the packet-in-resonance tallies into estimates of $J$ in
163: each line.
164:
165: \begin{figure}
166: \plotfiddle{fig01.eps}{2.0in}{0.0}{45}{45}{-124}{0}
167: \caption{The effect of increasing packet number on noise. The Ca infrared
168: triplet is plotted for (from bottom to top) $10^4$, $10^5$, $10^6$
169: and $10^7$ packets used.}
170: \end{figure}
171:
172: Of course in 3D, the appearance of a spectrum may depend on the line
173: of sight to the envelope. Rather than wait for enough packets to
174: emerge along all lines of sight (a serious problem), we employ the
175: same formal integral technique used in the 1D calculation. Once the
176: source function is determined everywhere by Monte Carlo, common
177: direction plane positions are computed and the flux is integrated
178: across impact parameter beams. This technique greatly reduces the
179: number of packets to follow in a simulation.
180:
181: \section{Future work}
182:
183: In order to resolve variations in optical depth that are small in
184: size, the resolution of the envelope must be increased. The
185: simplest solution is to use a big computer with a lot of memory,
186: or split the envelope up amongst multiple processors. The fact
187: that we need only sweep the line list from blue to red may help,
188: and we can load segments of the line list as the calculation
189: progresses.
190:
191: Including electron scattering in the calculations, and replacing
192: the assumption of pure resonance scattering with branching will
193: help to make the code more consistent overall. The eventual
194: goal is a NLTE code for use with new 3D explosion models. Until
195: the reliability and quantity of those models increases, a great
196: deal about deviations from spherical symmetry in SNe can be
197: learned from a parameterized code.
198:
199: \acknowledgements
200:
201: This work was supported in part by NASA grants NAG5-3505 and
202: NAG5-12127, and an IBM SUR grant to the University of Oklahoma.
203:
204: \begin{references}
205:
206: \reference Abbott, D., Lucy, L. 1985 ApJ 288, 679
207:
208: \reference Decourchelle, A., et al.\ 2001 A\&A 365, L218
209:
210: \reference Deng, J., Qiu, Y., Hu, J., Hatano, K., Branch, D. 2000 ApJ
211: 540, 452
212:
213: \reference Fesen, R., Gunderson, K. 1996 ApJ 470, 967
214:
215: \reference Fisher, A. 2000 PhD Thesis, U of Oklahoma
216:
217: \reference Reinecke, M., Hillebrandt, W., Niemeyer, J. 2002 A\&A
218: in press, also astro-ph/0206459
219:
220: \reference Howell, D., Hoflich, P., Wang, L., Wheeler, J. 2001 ApJ
221: 556, 302
222:
223: \reference Hwang, U., Holt, S., Petre, R. 2000 ApJ 537, L119
224:
225: \reference Jeffery, D. 1993 ApJ 415, 734
226:
227: \reference Jeffery, D., Branch, D. 1990, in Supernovae, Jerusalem
228: Winter School for Theoretical Physics, ed. J. C. Wheeler,
229: T. Piran \& S. Weinberg (Singapore : World Scientific),
230: 149
231:
232: \reference Khokhlov, A. 2001 ApJ in press, astro-ph/0008463
233:
234: \reference Kifonidis, K., Plewa, T., Janka, H.-Th., Muller, E. 2000
235: ApJ 531, L123
236:
237: \reference Leonard, D., Filippenko, A., Ardila, D., Brotherton, M. 2001
238: ApJ 553, 861
239:
240: \reference Leonard, D., Filippenko, A., Barth, A., Matheson, T. 2000
241: ApJ 536, 239
242:
243: \reference Mazzali, P. 2000 A\&A 363, 705
244:
245: \reference Mazzali, P., Lucy, L. 1993 A\&A 279, 447
246:
247: \reference Millard, J., et al.\ 1999 ApJ 527, 746
248:
249: \reference Phillips, M., Heathcote, S. 1989 PASP 101, 137
250:
251: \reference Rybicki, G., Hummer, D. 1978 ApJ 219, 654
252:
253: \reference Sobolev, V. 1947 Moving Envelopes of Stars (Leningrad:
254: Leningrad State UP) (English transl. S. Gaposchkin
255: [Cambridge: Harvard UP, 1960])
256:
257: \reference Thomas, R., Kasen, D., Branch, D., Baron, E., 2002 ApJ 567,
258: 1037
259:
260: \reference Tsunemi, H., Miyata, E., Aschenbach, B. 1999 PASJ 51, 711
261:
262: \reference Utrobin, V., Chugai, N., Andronova, A. 1995, A\&A 295, 129
263:
264: \reference Wang, L., Howell, D., Hoflich, P., Wheeler, J. 2001 ApJ
265: 550, 1030
266:
267: \reference Wang, L., Wheeler, J., Hoflich, P. 1997 ApJ 476, L27
268:
269: \reference Wheeler, H., 2000 in Cosmic Explosions, ed. S. S. Holt \&
270: W. W. Zhang (New York : AIP), 445; also astro-ph/9912403
271:
272: \end{references}
273:
274: \end{document}
275:
276: