1: 

2: \def\etal{{\frenchspacing\it et al.}}

3: \def\ie{{\frenchspacing\it i.e.}}

4: \def\eg{{\frenchspacing\it e.g.}}

5: \def\etc{{\frenchspacing\it etc.}}

6: \def\rms{{\frenchspacing r.m.s.}}

7: \def\ith{i^{\rm th}}

8: 

9: %%%% EQUATION STUFF: %%%%

10: \def\beq#1{\begin{equation}\label{#1}}

11: \def\eeq{\end{equation}}

12: \def\beqa#1{\begin{eqnarray}\label{#1}}

13: \def\eeqa{\end{eqnarray}}

14: \def\eq#1{equation~(\ref{#1})}

15: \def\Eq#1{Equation~(\ref{#1})}

16: \def\eqn#1{~(\ref{#1})}

17: 

18: %%%% FIGURE STUFF: %%%%

19: \def\fig#1{Figure~\ref{#1}}

20: \def\Fig#1{Figure~\ref{#1}}

21: 

22: \def\bfk{\mbox{\bf k}}

23: \def\bfr{\mbox{\bf r}}

24: \def\bfe{\mbox{\bf e}}

25: \def\bfd{\mbox{\bf d}}

26: \def\bfx{\mbox{\bf x}}

27: \def\bfp{\mbox{\bf p}}

28: \def\bfq{\mbox{\bf q}}

29: \def\ztwodf{z_{\rm 2df}}

30: \def\etal{{\frenchspacing\it et al.}}

31: 

32: \def\expec#1{\langle#1\rangle}

33: 

34: %\def\la{\mathrel{\mathpalette\fun <}}

35: %\def\ga{\mathrel{\mathpalette\fun >}}

36: \def\fun#1#2{\lower3.6pt\vbox{\baselineskip0pt\lineskip.9pt

37:         \ialign{$\mathsurround=0pt#1\hfill##\hfil$\crcr#2\crcr\sim\crcr}}}

38:         

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

40: 

41: 

42: %\documentclass{article}

43: %\usepackage{emulateapj5,pstricks,apjfonts}

44: %\usepackage{emulateapj5}

45: 

46: %\usepackage{emulateapj5,pstricks}

47: \usepackage{graphicx}

48: \usepackage{epsfig}

49: 

50: \newcommand{\be}{\begin{equation}}

51: \newcommand{\ee}{\end{equation}}

52: \newcommand{\ba}{\begin{eqnarray}}

53: \newcommand{\ea}{\end{eqnarray}}

54: \newcommand{\Om}{\Omega_m}

55: 

56: \shorttitle{Photometric Redshift Estimator for SNe~Ia}

57: \shortauthors{Wang et al. }

58: 

59: %\def\baselinestretch{2}

60: \begin{document}

61: 

62: 

63: \title{Survey Requirements for Accurate and Precise Photometric 

64: Redshifts for Type~Ia Supernovae}

65: \author{Yun~Wang\altaffilmark{1},

66: Gautham Narayan\altaffilmark{2}, and

67: Michael Wood-Vasey\altaffilmark{2}}

68: \altaffiltext{1}{Homer L. Dodge Department of Physics \& Astronomy, 

69: 		 Univ. of Oklahoma,

70:                  440 W Brooks St., Norman, OK 73019;

71:                  email: wang@nhn.ou.edu}

72: \altaffiltext{2}{Center for Astrophysics, 

73:                  Harvard University, 

74:                  60 Garden Street, Cambridge, MA 02138; 

75:                  email: gnarayan@cfa.harvard.edu, wmwood-vasey@cfa.harvard.edu}

76: \begin{abstract}

77: 

78: In this paper we advance the simple analytic photometric redshift

79: estimator for Type~Ia supernovae (SNe~Ia) proposed by \citet{Wang07},

80: and use it to study simulated SN~Ia data.

81: We find that better than 0.5\% accuracy in $z_{\rm phot}$ 

82: (with $\sigma[(z_{\rm phot}-z_{\rm spec})/(1+z_{\rm spec})]<0.005$) is possible for 

83: SNe~Ia with well sampled lightcurves in three observed passbands ($riz$)

84: with a signal-to-noise ratio of 25 at peak brightness, if the extinction

85: by dust is negligible. The corresponding bias in $z_{\rm phot}$ 

86: (the mean of $z_{\rm phot}-z_{\rm spec}$) is $5.4\times 10^{-4}$.

87: If dust extinction is taken into consideration in the $riz$

88: observer-frame lightcurves, the accuracy in $z_{\rm phot}$ deteriorates to 4.4\%,

89: with a bias in $z_{\rm phot}$ of $8.0\times 10^{-3}$.

90: Adding the $g$ band lightcurve improves the

91: accuracy in $z_{\rm phot}$ to 2.5\%,

92: and reduces the bias in $z_{\rm phot}$ to $-1.5\times 10^{-3}$.

93: Our results have significant implications for the design

94: of future photometric surveys of SNe~Ia from both ground and space

95: telescopes. Accurate and precise photometric redshifts boost

96: the cosmological utility of such surveys.

97: 

98: \end{abstract}

99: 

100: %\end{document}

101: 

102: \keywords{distance scale -- methods: data analysis -- supernovae: general}

103: 

104: 

105: \section{Introduction}

106: 

107: The discovery of cosmic acceleration \citep{Riess98,Perl99} was made using 

108: Type~Ia supernovae (SNe~Ia) as cosmological standard candles

109: \citep{Phillips93,Riess95}.

110: The unknown reason for the observed cosmic acceleration has

111: been dubbed ``dark energy''. 

112: Obtaining the spectroscopic redshifts of SNe~Ia is the most costly 

113: aspect of supernova surveys.

114: Large photometric surveys of SNe~Ia

115: have been planned to help illuminate the nature of dark energy.

116: Obtaining accurate and precise photometric redshifts (photo-$z$'s) 

117: will be key in maximizing the cosmological utility of such surveys \citep{Huterer04}.

118: 

119: In this paper, we advance the simple analytic photometric redshift

120: estimator for Type~Ia supernovae (SNe~Ia) proposed by \citet{Wang07}\footnote{Based

121: on the analytic photometric redshift estimator for galaxies proposed

122: by \cite{Wang98a}.},

123: and use it to study simulated SN~Ia data.

124: In order to derive unbiased and accurate photo-$z$'s,

125: it will be important to develop a variety of different techniques 

126: to cross-check sensitivities to various systematic uncertainties

127: and correctly calculating the covariance between distance modulus

128: and photo-$z$ when fitting for cosmological parameters.

129: 

130: We present our method in Sec.~\ref{sec:method} and the results in Sec.~\ref{sec:results}.

131: We discuss and summarize in Sec.~\ref{sec:discussion}.

132: 

133: \section{The Method}

134: \label{sec:method}

135: 

136: \subsection{The analytic photo-$z$ estimator}

137: 

138: The analytic photo-$z$ estimator for SNe~Ia proposed by \citet{Wang07}

139: is empirical, model independent (no templates used), 

140: and uses observables that reflect the properties

141: of SNe~Ia as calibrated standard candles.

142: It was developed using the SN~Ia data released by the Supernova

143: Legacy Survey \citep{Astier06}.

144: 

145: This estimator uses the fluxes in $griz$ (or $riz$) at the epoch 

146: of $i$ maximum flux to make an effective K-correction

147: to the $i$ flux. The first estimate of redshift is given by

148: \be

149: z_{\rm phot}^{0}=c_1 + c_2 g_f +c_3 r_f + c_4 i_f + c_5 z_f +c_6 i_f^2

150: + c_7 i_f^3

151: \label{eq:z0}

152: \ee

153: where $g_f=2.5\log(f_g)$, $r_f=2.5\log(f_r)$, $i_f=2.5\log(f_i)$,

154: and $z_f=2.5\log(f_z)$, and $f_g$, $f_r$, $f_i$, $f_z$ are fluxes

155: in counts, normalized to some fiducial zeropoint, in $griz$ at the epoch of $i$ maximum flux.

156: 

157: Next, it calibrates each SN~Ia in its estimated rest-frame using

158: \be

159: \Delta i_{15}= 2.5 \log(f_i^{15d}/f_i),

160: \label{eq:del_i15}

161: \ee

162: where $f_i^{15d}$ is the $i$ band flux at 15 days after

163: the $i$ flux maximum in the estimated rest-frame, corresponding

164: to the epoch of $\Delta t^{15d}=15 (1+z_{\rm phot}^{0})$ days after the

165: epoch of $i$ flux maximum.	    

166: 

167: The final estimate for the photometric redshift is given by

168: \be

169: z_{\rm phot} = \sum_{i=1}^8 c_i \,p_i,

170: %c_1 + c_2 g_f +c_3 r_f + c_4 i_f + c_5 z_f +c_6 i_f^2 + c_7 i_f^3

171: %		+c_8 \Delta i_{15}

172: \label{eq:z_a}

173: \ee

174: where the data vector ${\mbox {\bf p}}=\{1, g_f, r_f, i_f, z_f, i_f^2,

175: i_f^3, \Delta i_{15}\}$.

176: The coefficients $c_i$ (i=1,2,...,8) are found by using

177: a training set of SNe~Ia with $griz$ (or $riz$, for which $c_2=0$) 

178: lightcurves and

179: measured spectroscopic redshifts. 

180: We use the jackknife technique \citep{Lupton93} to estimate the 

181: mean and the covariance matrix of $c_i$ (see Sec.~\ref{sec:results}).

182: 

183: 

184: 

185: \subsection{Simulation of data}

186: 

187: SN~Ia lightcurves for training and testing were simulated using 

188: the templates and various routines from MLCS2k2 \citep{Jha07}.  

189: We created two sets of $griz$ lightcurves using the MLCS2k2 ``final'' 

190: templates at random redshifts in the range $0.001 \leq z \leq 1$ 

191: with S/N=25 at peak and $\Delta$, 

192: the parameter that MLCS2k2 uses to describe the variation in the 

193: SN~Ia lightcurves, fixed to $0$. The luminosity distance $\mu$, 

194: is calculated from the redshift assuming

195: a flat cosmology with $\Omega_M=0.3$, $h=0.65$ and $w=-1$. The particular choice 

196: of cosmology will not affect the photo-$z$ estimator as long as the training and testing 

197: lightcurves are consistent with each other. 

198: 

199: The first training set consists of 100 lightcurves and was 

200: simulated with no host extinction. 

201: The corresponding blind test set contains 1000 lightcurves 

202: similar to those in the training set.

203: 

204: The second training set consists of 200 lightcurves, and has $A_V$ 

205: drawn at random from the ``default'' distribution, an exponential 

206: decay with argument $0.46$~mag and zero for $A_V < 0$ mag, following \citet{Jha07}. 

207: The particular form of this exponential tail is derived 

208: from the observation \citep{Lira95} that SNe~Ia have a common 

209: color at $\sim40$ days past peak. We assume the extinction 

210: law to be $R_V = 3.1$.  We used a larger training 

211: set of 200 simulated lightcurves for training to compensate 

212: for the larger variation in the data 

213: caused by including extinction. 

214: The corresponding blind test set contains 1000 lightcurves 

215: similar to those in the training set.

216: 

217: The actual training sets and blind test sets used are

218: slightly smaller than described above, because

219: only SNe~Ia with $riz$ lightcurves, i.e. $z\le0.95$, are used in this paper.

220: We use the training sets to derive the photo-$z$ coefficients

221: and their covariance matrix (see Eq.~\ref{eq:z_a} and Sec.~\ref{sec:results}), 

222: and the blind test sets to evaluate the performance of the 

223: photo-$z$ estimator.   

224: 

225: \section{Results}

226: \label{sec:results}

227: 

228: We use a slightly modified version of the jackknife technique 

229: \citep{Lupton93} to estimate the 

230: mean and the covariance matrix of $c_i$. 

231: From the training set containing $N$ SNe~Ia, 

232: we extract $N$ subsamples each containing $N-1$ SNe~Ia

233: by omitting one SN~Ia. The coefficients $c_i^{(s)}$ (i=1,2,...,8)

234: for the $s$-th subsample are found by a maximum likelihood

235: analysis matching the predictions of

236: Eq.~\ref{eq:z_a} with the spectroscopic redshifts.

237: 

238: The mean of the coefficients $c_i$ (i=1,2,...,8) are given by

239: \be

240: \langle c_i \rangle = \frac{1}{N} \sum_{s=1}^{N} c_i^{(s)}.

241: \label{eq:mean}

242: \ee

243: Note that this is related to the usual ``bias-corrected jackknife

244: estimate'' for $c_i$, $c_i^J$, as follows:

245: \be

246: c_i^J \equiv c_i^N +(N-1) \left(c_i^N-\langle c_i \rangle\right),

247: \ee

248: where $c_i^N$ are estimated from the entire training set (with $N$ SNe~Ia).

249: We found that for small training sets (with $N<20$) that include

250: SNe~Ia at $z \sim 0$, $c_i^J$ give

251: biased estimates of $c_i$ by giving too much weight to the

252: SN~Ia with the smallest redshift. For training sets 

253: not including nearby SNe~Ia, $\langle c_i \rangle$ and $c_i^J$ 

254: are approximately equal.

255: We have chosen to use $\langle c_i \rangle$ from Eq.~\ref{eq:mean}

256: as the mean estimates for $c_i$ to avoid biased $z_{\rm phot}$ 

257: for SNe~Ia at $z$ close to zero.

258: 

259: The covariance matrix of $c_i$ (i=1,2,...,8) are given by

260: \be

261: {\rm Cov}(c_i,c_j)= \frac{N-1}{N} \sum_{s=1}^{N} 

262: \left(c_i^{(s)} - \langle c_i \rangle \right)\, 

263: \left(c_j^{(s)} - \langle c_j \rangle \right)

264: \ee

265: Since $z_{\rm phot}= \sum c_i p_i$, $\Delta z_{\rm phot}= \sum p_i\Delta c_i$,

266: since the uncertainty in $z_{\rm phot}$ is dominated by the

267: uncertainty in $c_i$.

268: Therefore estimated error in $z_{\rm phot}$ is

269: \be

270: {\rm d}z_{\rm phot}= \left\{ \sum_{i=1}^8 \sum_{j=1}^8 

271:  [p_i] \, {\rm Cov}(c_i,c_j) [p_j] \right\}^{1/2},

272: \ee

273: where ${p_i}$ and ${p_j}$ (i,j=1,2,...8) are

274: the $i$-th and $j$-th components of the data vector

275: ${\mbox {\bf p}}=\{1, g_f, r_f, i_f, z_f, i_f^2,

276: i_f^3, \Delta i_{15}\}$. 

277: 

278: In applying the photo-$z$ estimator to simulated data, we find

279: that applying it to the entire data set leads to substructure

280: in the estimated $z_{\rm phot}$. Therefore, we divide the

281: training set into three $i$ maximum flux ranges (corresponding 

282: to three roughly equal redshift intervals), 

283: and applying the photo-$z$ estimator to each flux range.

284: 

285: Fig.~\ref{fig1} shows the resultant performance of

286: the photo-$z$ estimator for simulated SN~Ia lightcurves

287: in $riz$, with S/N=25 at peak brightness, and zero

288: extinction due to dust. The top panel shows the

289: training set of 97 SNe~Ia, with different point types and

290: color denoting different flux ranges.

291: The $c_i$'s derived from the training set are used

292: to predict $z_{\rm phot}$ for a blind-test set of 940 SNe~Ia

293: (divided into the same flux ranges as the training set),

294: with the results shown in the bottom panel of Fig.~\ref{fig1}.

295: 

296: The results for the blind-test set closely mimic that

297: of the training set.

298: Since the data are very sparse at $z<0.02$ due to

299: the small size of the training set, 

300: the estimated uncertainty on $z_{\rm phot}$, d$z_{\rm phot}$,

301: becomes large for nearby SNe~Ia.

302: Since d$z_{\rm phot}$ is a reliable

303: indicator of the accuracy of $z_{\rm phot}$,

304: we can use d$z_{\rm phot}$ to exclude the SNe~Ia

305: with estimated $z_{\rm phot}$ not suitable for inclusion

306: in the cosmological data set.

307: 

308: We find that d$z_{\rm phot}<0.1$ only excludes 9 out of the

309: 940 SNe~Ia in the blind-test set, all at $z<0.02$.

310: For the culled blind-test set of 931 SNe~Ia (d$z_{\rm phot}<0.1$),

311: $\sigma[(z_{\rm phot}-z_{\rm spec})/(1+z_{\rm spec})]=0.0043$.

312: The corresponding bias in $z_{\rm phot}$ 

313: (the mean of $z_{\rm phot}-z_{\rm spec}$) is $5.4\times 10^{-4}$ overall

314: and is  $[0.48,  0.79,  1.5, -0.81]\times 10^{-3}$ in 

315: the 4 redshift bins bounded by $[0,0.25,0.5,0.75,1]$.

316: 

317: Fig.~\ref{fig2} shows the performance of

318: the photo-$z$ estimator for simulated SN~Ia lightcurves

319: in $riz$, with S/N=25 at peak brightness, and 

320: extinction due to dust parameterized by $A_V$.

321: Again, we cull the blind-test set by requiring

322: that d$z_{\rm phot}<0.1$; this excludes 16 out of the

323: 956 SNe~Ia, most of which are highly extinguished or

324: have very low $z$.

325: The inclusion of dust extinction leads to a deteriorated

326: accuracy in $z_{\rm phot}$ of 4.4\%.

327: The corresponding bias in $z_{\rm phot}$ 

328: is $8.0\times 10^{-3}$ overall

329: and is  $[-9.3, -22.,  25.,  53.]\times 10^{-3}$ in the 

330: 4 redshift bins bounded by $[0,0.25,0.5,0.75,1]$.

331: 

332: Fig.~\ref{fig3} is similar to Fig.~\ref{fig2}, but only the subsets of data

333: (both the training set and the blind-test set) with 

334: $g$ band lightcurves are included.

335: Note that the $i$ maximum flux ranges in Fig.~\ref{fig3} are similar

336: to Fig.~\ref{fig2}, but slightly adjusted to make sure that

337: the sub sample with the smallest fluxes contains

338: enough SNe~Ia. The SNe~Ia at $z>0.54$ in

339: our simulated data set do not have $g$ band lightcurves,

340: as these require rest-frame lightcurve templates bluer than $u$ band,

341: which are not yet readily available.

342: 

343: Imposing d$z_{\rm phot}<0.1$ excludes 12 out of the

344: 537 SNe~Ia, most of which are highly extinguished or

345: have very low $z$.

346: Adding the $g$ band lightcurves improve the

347: accuracy in $z_{\rm phot}$ to 2.5\%.

348: The corresponding bias in $z_{\rm phot}$ 

349: is reduced to $-1.5\times 10^{-3}$ overall

350: and is $[-6.0, 3.2]\times 10^{-3}$ in the 

351: 2 redshift bins bounded by $[0,0.266,0.532]$.

352: 

353: 

354: \section{Discussion and Summary}

355: \label{sec:discussion}

356: 

357: Accurate and precise photo-$z$'s will be critical for 

358: enabling cosmology with future large SNe~Ia photometric surveys

359:  \citep{Huterer04}.

360: In this paper, we have advanced the empirical and analytic photo-$z$

361: estimator proposed by \citet{Wang07} and used it

362: to study simulated data in order to derive

363: the survey requirements for accurate and precise photo-$z$'s

364: for SNe~Ia from large photometric surveys.

365: Future work will explore the critically-important question

366: of the degree of covariance between the redshift determined 

367: using this method and the distance modulus

368: determined by using a luminosity distance fitter

369: since it is these two parameters that form the basis

370: for the cosmological measurements with SNe~Ia.

371: 

372: We apply the photo-$z$ estimator to $riz$ data divided into

373: $i$ maximum flux ranges that roughly correspond to

374: equal redshift intervals (see Fig.~\ref{fig1} and Fig.~\ref{fig2}). 

375: We use d$z_{\rm phot}<0.1$ as a quality measurement to cull the blind-test set.

376: This cut only excludes $<2$\% of the SNe~Ia,

377: most of which are highly extinguished or are

378: at very low $z$. 

379: To derive accurate cosmological constraints,

380: it is optimal to combine the very large set

381: of SNe~Ia with well estimated photo-$z$'s with

382: a small spectroscopic sample of SNe~Ia at very low $z$ 

383: expected from ongoing nearby surveys

384: \citep{Huterer04}. 

385: SNe~Ia at $z<0.025$ are currently excluded

386: from SN~Ia cosmological analyses even for spectroscopic

387: data, because peculiar velocities due to cosmic large

388: scale structure modifies the Hubble diagram of

389: nearby standard candles \citep{Wang98b}, and

390: have the largest impact on the nearest SNe~Ia (see, e.g., 

391: \cite{Zehavi98,Cooray06,Hui06,Conley07}).

392: 

393: 

394: We find that determining $z_{\rm phot}$'s with better than 0.5\% accuracy 

395: (RMS[$\Delta z_{\rm phot}/(1+z_{\rm spec})]$ = 0.0043\%)

396: and negligible bias ($\langle z_{\rm phot}-z_{\rm spec}\rangle=5.4\times 10^{-4}$)

397: is possible if dust extinction can be neglected, for

398: SNe~Ia with well sampled lightcurves in $riz$ and S/N=25

399: at peak brightness.

400: The inclusion of dust extinction leads to a drastic deterioration

401: of the accuracy of $z_{\rm phot}$

402: (RMS[$\Delta z_{\rm phot}/(1+z_{\rm spec})]$ = 4.4\%,

403: and $\langle z_{\rm phot}-z_{\rm spec}\rangle =8.0\times 10^{-3}$).

404: Adding the $g$ band lightcurve significantly improves

405: the accuracy and precision of $z_{\rm phot}$

406: with RMS[$\Delta z_{\rm phot}/(1+z_{\rm spec})$] = 2.5\%,

407: and $\langle z_{\rm phot}-z_{\rm spec}\rangle = -1.5\times 10^{-3}$.

408: 

409: Future supernova surveys can easily obtain 

410: the multi-band photometry of a huge number of supernovae

411: \citep{Wang00,JEDI,Phillips06}.

412: We only considered $griz$ photometry in this paper,

413: due to the limitation of currently available SN Ia lightcurve

414: templates. The addition of photometry at longer wavelengths

415: should provide more dramatic improvement on the

416: accuracy and precision of $z_{\rm phot}$ since extinction

417: decreases with wavelength, and SNe~Ia are better standard

418: candles in the near IR \citep{KK04}.

419: In future work, we will study simulated SN~Ia data

420: over the entire wavelength range of relevance to future

421: observations, and examine the impact of realistic survey 

422: and observational parameters on the precision and

423: accuracy of our photo-$z$ estimator.

424: 

425: Our results are encouraging for the prospect of using photometric

426: surveys to constrain cosmology, such as those planned for 

427: the Advanced Liquid-mirror Probe for Astrophysics, Cosmology and 

428: Asteroids (ALPACA)\footnote{http://www.astro.ubc.ca/LMT/alpaca/};

429: Pan-STARRS\footnote{http://pan-starrs.ifa.hawaii.edu/}; 

430: the Dark Energy Survey (DES)\footnote{http://www.darkenergysurvey.org/};

431: and the Large Synoptic Survey Telescope (LSST)\footnote{http://www.lsst.org/}.

432: 

433: 

434: \bigskip

435: 

436: {\bf Acknowledgments}

437: We thank the Aspen Center for Physics for hospitality 

438: where part of the work was done.

439: We are grateful to Alex Kim for useful discussions.

440: YW  is supported in part by NSF CAREER grant AST-0094335. 

441: MWV is supported in part by NSF AST-057475.

442: 

443: 

444: \begin{thebibliography}{}

445: 

446: \bibitem[Astier et al.(2006)]{Astier06}

447: Astier,  P., et al. 2006, A \& A 447, 31-48 (2006) 

448: 

449: \bibitem[Conley et al. (2007)]{Conley07}

450: Conley, A., et al. 2007,

451: arXiv:0705.0367v2, ApJL, in press

452: 

453: \bibitem[Cooray \& Caldwell(2006)]{Cooray06}

454: Cooray, A.; Caldwell, R.R. 2006,

455: Phys.Rev.D73, 103002

456: 

457: \bibitem[Guy et al.(2007)]{Guy07}

458: Guy, J., et. al. 2007, A \& A, in press,

459: arXiv:astro-ph/0701828v1

460: 

461: \bibitem[Hui \& Greene(2006)]{Hui06}

462: Hui, L.;  \& Greene, P.B. 2006, 

463: Phys. Rev. D73, 123526

464: 

465: \bibitem[Huterer et al.(2004)]{Huterer04}

466: Huterer, D.; Kim, A.; Krauss, L.M.; Broderick, T. 2004, 

467: ApJ, 615, 595

468: 

469: \bibitem[Jha et al.(2007)]{Jha07}

470: Jha, S.; Reiss, A. G.; Kirshner, R. P. 2007,

471: ApJ, 659, 122

472: arXiv:astro-ph/0612666

473: 

474: \bibitem[Krisciunas, Phillips, \& Suntzeff(2004)]{KK04}

475: Krisciunas, K.; Phillips, M.M.; Suntzeff, N.B. 2004,

476: ApJ, 602, L81

477: 

478: \bibitem[Lira(1995)]{Lira95}

479: Lira, P. 1995,

480: Master's thesis, University of Chile

481: 

482: \bibitem[Lupton(1993)]{Lupton93}

483: Lupton, R., 1993, ``Statistics in Theory and Practice'',

484: Princeton University Press

485: 

486: \bibitem[Perlmutter et al.~(1999)]{Perl99} 

487: Perlmutter, S. {\etal}, 1999, ApJ, 517, 565

488: 

489: \bibitem[Phillips(1993)]{Phillips93}

490: Phillips, M.M. 1993, ApJ, 413, L105 

491: 

492: \bibitem[Phillips et al.(2006)]{Phillips06}

493: Phillips, M. M.; Garnavich, P.; Wang, Y. et al. 2006,

494: Proc. of SPIE, Volume 6265, 626529,

495: arXiv:astro-ph/0606691v1

496: 

497: \bibitem[Riess, Press, \& Kirshner(1995)]{Riess95}

498: Riess, A.G., Press, W.H., and Kirshner, R.P. 1995, ApJ, 438, L17 

499: 

500: \bibitem[Riess et al.~(1998)]{Riess98}

501: Riess, A. G, {\etal}, 1998, Astron. J., 116, 1009

502: 

503: \bibitem[Wang(2000)]{Wang00}

504: Wang, Y. 2000, ApJ 531, 676 

505: %arXiv:astro-ph/9806185

506: 

507: \bibitem[Wang, Bahcall, \& Turner(1998)]{Wang98a}

508: Wang, Y.; Bahcall, N.; Turner, E. L., 1998, AJ, 116, 2081

509: %arXiv:astro-ph/9804195

510: 

511: \bibitem[Wang, Spergel, \& Turner(1998)]{Wang98b}

512: Wang, Y.;  Spergel, D.N.; \& Turner, E.L. 1998,

513: ApJ, 498,  1

514: %arXiv:astro-ph/9708014v2

515: 

516: \bibitem[Wang et al.(2004)]{JEDI}

517: Wang, Y., et al. 2004 (the JEDI Collaboration), 

518: BAAS, v36, n5, 1560. See also http://jedi.nhn.ou.edu/.

519: 

520: %{2007ApJ...654L.123W}

521: \bibitem[Wang(2007)]{Wang07}

522: Wang, Y.\ 2007, \apjl, 654, L123 

523: %arXiv:astro-ph/0609639

524: 

525: \bibitem[Zehavi et al.(1998)]{Zehavi98}

526: Zehavi, I.; Riess, A.G.; Kirshner, R.P.; \& Dekel, A. 1998,

527: ApJ, 503, 483 

528: %arXiv:astro-ph/9802252

529: 

530: \end{thebibliography}

531: 

532: 

533: %\end{document}

534: \clearpage

535: \setcounter{figure}{0}

536: 

537: 

538: \begin{figure}

539: \includegraphics{za_sn25_zeroAV_riz.eps}

540: \caption{The analytic photo-$z$ estimator 

541: applied to simulated SN~Ia lightcurves

542: in $riz$, with S/N=25 at peak brightness, and zero

543: extinction due to dust. 

544: }

545: \label{fig1}

546: \end{figure}

547: 

548: \begin{figure}

549: 

550: \includegraphics{za_sn25_AV_riz.eps}

551: \caption{The analytic photo-$z$ estimator 

552: applied to simulated SN~Ia lightcurves

553: in $riz$, with S/N=25 at peak brightness, and 

554: dust extinction parameterized by $A_V$. 

555: Dust extinction leads to deterioration in

556: the accuracy of $z_{\rm phot}$.

557: The y-axis range is

558: the same as in Fig.~\ref{fig3}.

559: }

560: \label{fig2}

561: \end{figure}

562: 

563: \begin{figure}

564: \includegraphics{za_sn25_AV_griz.eps}

565: \caption{The analytic photo-$z$ estimator 

566: applied to simulated SN~Ia lightcurves

567: in $griz$, with S/N=25 at peak brightness, and 

568: dust extinction parameterized by $A_V$. 

569: Adding the $g$ band lightcurves improves the

570: accuracy of $z_{\rm phot}$.

571: The y-axis range is

572: the same as in Fig.~\ref{fig2}.

573: }

574: \label{fig3}

575: \end{figure}

576: 

577:  

578: 

579: \end{document}

580: 

581: % LocalWords:  al Yun Gautham Narayan Univ Huterer MWV phot Jha Lira Eq Cov IR

582: % LocalWords:  JEDI ALPACA DES AST ApJL Cooray Caldwell Hui Greene Phys Krauss

583: % LocalWords:  Broderick ApJ Reiss Suntzeff KK Proc SPIE Astron astro ph AJ

584: % LocalWords:  BAAS Zehavi Dekel

585: