1: % -*- Mode:LaTeX -*-
2: % For AASTeX version 5.02
3: \documentclass[12pt,preprint]{aastex}
4: %\documentclass[11pt,preprint]{aastex}
5: %\documentclass[preprint2,flushrt]{aastex}
6:
7: %\received{}
8: %\accepted{}
9:
10: \usepackage{psfig}
11: \usepackage{epsfig}
12:
13: %\doublespace
14:
15: %\received{2004 May 12}
16: \begin{document}
17:
18: \title{Redshift-Independent Distances to Type Ia Supernovae}
19:
20: \shorttitle{Z-free SNe Ia Distances}
21:
22: \author{Brian J. Barris\altaffilmark{1} and John L. Tonry\altaffilmark{1}}
23:
24: \altaffiltext{1}{Institute for Astronomy (IfA), University of Hawaii,
25: 2680 Woodlawn Drive, Honolulu, HI 96822; barris@ifa.hawaii.edu,
26: jt@ifa.hawaii.edu}
27:
28: \begin{abstract}
29: \label{batmloop-abstract}
30:
31: We describe a procedure for accurately determining luminosity
32: distances to Type Ia supernovae (SNe Ia) without knowledge of
33: redshift. This procedure, which may be used as an extension of any of
34: the various distance determination methods currently in use, is based
35: on marginalizing over redshift, removing the requirement of knowing
36: $z$ a priori. We demonstrate that the Hubble diagram scatter of
37: distances measured with this technique is approximately equal to that
38: of distances derived from conventional redshift-specific methods for a
39: set of 60 nearby SNe Ia. This indicates that accurate distances for
40: cosmological SNe Ia may be determined without the requirement of
41: spectroscopic redshifts, which are typically the limiting factor for
42: the number of SNe that modern surveys can collect. Removing this
43: limitation would greatly increase the number of SNe for which current
44: and future SN surveys will be able to accurately measure distance.
45: The method may also be able to be used for high-$z$ SNe Ia to
46: determine cosmological density parameters without redshift
47: information.
48:
49: \end{abstract}
50:
51: \keywords{distance scale --- methods: data analysis --- supernovae: general}
52:
53: \section{Introduction}
54: \label{batmloop-intro}
55:
56: The use of Type Ia supernovae (SNe Ia) as a tool for cosmological
57: studies became feasible upon the realization that there is a
58: relationship between luminosity and light-curve shape (Phillips 1993).
59: With this knowledge and the advent of wide-field CCD arrays which
60: allow for the efficient discovery of large numbers of SNe Ia,
61: cosmological investigations into the fundamental composition of the
62: universe using SNe Ia as probes have revealed the acceleration of the
63: expansion of the universe (Riess et al. 1998, Perlmutter et al. 1999).
64:
65: In order to derive distance from observed photometry, numerous factors
66: which alter the light-curve must be taken into account. Any distance
67: measurement technique must calculate the probability of a fit for a SN
68: light-curve in the parameter space of ($t_0$, $A_V$, $R_D$, $z$, $d$),
69: where $t_0$=time-of-maximum, $A_V$=extinction, $R_D$=decline rate
70: (using any of the several current parameterizations of SNe Ia
71: light-curve shapes---see below), $z$=redshift, and $d$=distance. We
72: then either marginalize over these factors or take cuts through
73: parameter space at specific values in order to obtain a measurement
74: for $d$, the primary quantity of interest.
75:
76: The initial discovery of the relationship between light-curve shape
77: and brightness led to parameterization by the rate of decline in
78: $B$-band brightness over the 15 days after maximum light ($\Delta
79: m_{15}$). A second method, the Multi-wavelength Light-Curve Shape
80: (MLCS) method (see Riess et al. 1996, 1998) is based on $\Delta$, the
81: difference in peak brightness between an observed SN and a fiducial
82: light-curve template. The ``stretch'' method (Perlmutter et al. 1997)
83: parameterizes the light-curve by a factor $s$ which broadens or
84: narrows a template light-curve in order to modify the light-curve
85: shape. The Bayesian Adapted Template Match (BATM) method introduced
86: by Tonry et al. (2003) measures distances through comparison with a
87: large set of well-observed nearby supernovae rather than a
88: parametrized template. These methods all seem to be fundamentally
89: equivalent for the purposes of measuring accurate distances and
90: resolving uncertainties between the effects mentioned above. Another
91: recently introduced distance measurement technique is CMAGIC (Wang et
92: al. 2003), which uses the relationship between light-curve shape and
93: brightness in a more indirect way than the above methods. It utilizes
94: an observed linear relationship in color-magnitude space for extended
95: periods of time after maximum light to estimate distance. As noted by
96: Wang et al. (2003), CMAGIC remains under development and a more
97: thorough demonstration of its utility as a distance estimator is
98: necessary.
99:
100: Redshift is an especially problematic parameter. It broadens the
101: light-curve by (1+$z$), which affects what we can deduce about
102: luminosity, and also modifies how the spectral energy distribution
103: (SED) maps into observed bandpasses, thereby changing the color of the
104: SN Ia. It thus exhibits considerable covariance with other fit
105: parameters, and needs to be known accurately. Redshift is normally
106: measured spectroscopically, which confers the added benefit of
107: confirming that an object is indeed a SN Ia. However, for large
108: surveys (see Barris et al. 2004), far more SNe are discovered than can
109: possibly be observed spectroscopically. The situation will get worse
110: with larger surveys such as ESSENCE (Smith et al. 2002), the
111: Canada-France-Hawaii Legacy Survey
112: (http://www.cfht.hawaii.edu/Science/CFHLS) and Pan-STARRS (Kaiser \&
113: Pan-STARRS Team 2002).
114:
115: There have been several attempts to determine redshift and type for SN
116: through photometric means. Barris et al. (2002) concluded that over a
117: wide range of redshift, SNe Ia lie in a narrow region of
118: color-magnitude space, indicating it is likely that tests
119: discriminating between SNe Ia and II may be reliable. Riess et
120: al. (2004) constructed cuts based on SEDs of SNe to optimize selection
121: of high-$z$ candidates for spectroscopy, distinguishing between the
122: various SN types at high redshifts based on the UV deficit of SNe Ia
123: relative to core collapse SNe. In combination with the photometric
124: colors of the host galaxy, they were also able to make a rough
125: estimate of the redshift of the SNe. Barris et al. (2004) used a
126: similar, although simplified, comparison to argue for the
127: identification of several SNe as Type Ia, despite lacking spectral
128: confirmation.
129:
130: Currently it is not apparent that photometric observations are
131: sufficient to accurately determine the redshift and/or distances of
132: SNe Ia. Comparisons with SEDs of SNe Ia may allow constraints on
133: (1+$z$) on the order of $\approx$0.1, but this is a poor constraint on
134: $z$ for $z<0.5$. Without knowledge of redshift, the standard methods
135: for determining distance cannot be meaningfully applied, since
136: observations must be corrected for cosmological effects such as
137: time-dilation and K-corrections before allowing accurate measurements
138: of other parameters. However, often the quantity of interest is
139: distance. If $d$ may be accurately measured despite a poor constraint
140: on $z$ or other parameters, it would be an important breakthrough for
141: cosmological studies using SNe.
142:
143: We describe in this paper a method for determining accurate distances
144: to SNe Ia without knowledge of redshift, based on marginalizing over
145: $z$. We have used the BATM technique introduced by Tonry et
146: al. (2003), but the method is not specific to BATM and can be employed
147: with any distance estimator. We describe the necessary steps for
148: implementing the method, and demonstrate that it produces accurate
149: distances for a large sample of Hubble flow SNe Ia. The potential to
150: remove the need for spectroscopic redshifts promises to greatly
151: increase the number of SNe that may be used for cosmological analysis,
152: which is of particular importance due to the number of extremely
153: wide-field surveys planned for the near future.
154:
155: \section{Marginalization over Redshift}
156: \label{batmloop-method}
157:
158: As described above, all distance measurement techniques for SNe Ia
159: calculate the probability of a fit for a SN as a function of ($t_0$,
160: $A_V$, $R_D$, $z$, $d$). The parameters are either fit as a group or
161: marginalized over, in order to obtain a measurement of $d$. After
162: marginalizing out parameters ($t_0$, $A_V$, $R_D$), one is left with
163: probability as a function of ($z$, $d$). For some questions knowledge
164: of ($z$, $d$) probability contours is very interesting. The most
165: immediately obvious is for probing cosmological density parameters,
166: which led to the accelerating universe result. For all conventional
167: distance determination methods, $z$ is required to be an input
168: parameter for the reasons described above, providing a final estimate
169: of $d$ by evaluating at the known redshift.
170:
171: One could alternatively treat $z$ as a free parameter to be
172: marginalized out. For many questions there is no added benefit to
173: knowing $z$ in addition to distance. For example, inquiries into
174: rates as a function of distance, galaxy-type, location within the
175: host, or host galaxy properties can be answered based solely upon $d$,
176: rather than $z$. Even with a poor constraint on $z$, in principle we
177: may still be able to measure accurate distances since the effect of
178: redshift on distance comes from dependence on (1+$z$).
179:
180: Implementation of the redshift marginalization involves several minor
181: modifications to any fitting process. The most important is that care
182: must be taken to place calculated fit probabilities for each redshift
183: on an equal footing, so that at each step they may be meaningfully
184: combined. A suitable prior for $z$ must also be applied. The most
185: obvious to consider is to scale by volume, which can be accomplished
186: using a uniform redshift step, weighting each by $z$. The same prior
187: could be implemented with a uniform step in log $z$, weighting by
188: $z^2$. A different prior to consider is one constant in $z$, rather
189: than volume, which could be obtained by a linear step in redshift
190: $without$ scaling by $z$. One then iterates over the range of $z$
191: contributing non-negligible values of probability, combining the
192: results to produce a final estimate of $d$.
193:
194: It is important to note that this redshift marginalization method is
195: $not$ simply a restatement of the photometric redshift procedure
196: typically used for galaxies. Photometric redshifts use the observed
197: colors of galaxies in order to produce a likelihood range for the
198: redshift based on template SEDs. Since galaxies are not standard
199: candles, only a very weak constraint on distance is possible by
200: considering magnitude. The redshift-independent implementation of a
201: SN Ia distance estimator uses observed light-curves in order to
202: produce a constraint on both distance and redshift by considering the
203: magnitude and light-curve shape as well as the color. This additional
204: information greatly strengthens the power of the method in comparison
205: to galaxy photometric redshifts.
206:
207: It is also important to stress that this redshift marginalization
208: procedure may be implemented with $any$ method for measuring distances
209: to Type Ia supernovae (stretch, $\Delta m_{15}$, MLCS, BATM, or any
210: other equivalent method). Rather than evaluating $d$ for the measured
211: spectroscopic redshift, derived probabilities should be measured for a
212: wide range of $z$, with $z$ then marginalized over as is done with
213: other fit parameters.
214:
215: \section{Accurate Redshift Independent Distances}
216: \label{batmloop-distances}
217:
218: The BATM method for calculating luminosity distances for SNe Ia was
219: briefly described by Tonry et al. (2003) and Barris et al. (2004),
220: with a more complete treatment to appear in Barris (2004). It
221: is based upon an idealized set of representative SN Ia light-curves
222: which are photometrically and spectroscopically well-sampled in time,
223: and for which accurate distances are known. With such a
224: spectrophotometric template set, predicted light-curves could be
225: produced to compare to observations. However, data of this quality
226: are extremely uncommon at present, so we use a set of light-curves
227: which have excellent temporal coverage over a range of wavelengths and
228: span a wide range of luminosity, and a large set of observed spectra.
229: For a given redshift, the SEDs are shifted and warped so that they
230: match the observed photometry of each template light-curve. BATM
231: treats the ``template'' and ``unknown'' in a fundamentally different
232: manner from the previous methods. The SEDs and light-curves are
233: shifted to the redshift of the SN to be measured, so that redshift
234: effects are $introduced$ to the template set rather than $removed$
235: from the observational data. The idea is to compare the observed SN
236: to what we would expect the template to look like at a given redshift,
237: as opposed to comparing the template to what the SN would look like
238: were it at the redshift of the template.
239:
240: Since in BATM it is the template data that is transformed, rather than
241: the observed data, the operation can naturally be performed over a
242: range of redshift, facilitating the implementation of the redshift
243: marginalization. It should be stressed again that $any$ method for
244: measuring distances may similarly be extended by merely iterating the
245: procedure over a range of redshifts, treating $z$ as an additional
246: free parameter and marginalizing it out.
247:
248: We have performed the redshift-marginalization procedure in
249: combination with the BATM method for a set of 60 SNe Ia taken from
250: Hamuy et al. (1996), Riess et al. (1999), and Jha (2002). This is a
251: well-observed sample for which we know relative distances via
252: spectroscopic redshifts. Results are shown on a Hubble diagram in
253: Figure ~\ref{hubcompare}. It is important to note that the redshifts
254: used to construct these Hubble diagrams were measured
255: spectroscopically, $not$ photometrically. The RMS scatter about the
256: best-fit line is 0.21 mags for the $z$-free BATM distances, compared
257: to 0.19 mags for distances taken from the recent compilation of Tonry
258: et al. (2003), who calculated distances by combining the results from
259: as many methods as possible. We observe no significant difference in
260: performance between the redshift priors mentioned in the previous
261: section, indicating that neither the redshift prior nor details of its
262: implementation are limiting factors for the method for this sample.
263: We have therefore used distances calculated with a linear step of
264: $z$=0.01 in redshift and weighted by $z$, beginning at $z$=0.01 and
265: typically truncated at $z$=0.20, where the probability has become
266: negligible.
267:
268: We have scaled the uncertainties for the $z$-free BATM distances in
269: order to produce $\chi^{2}/N_{dof} \sim 1$ for the set of 60 SNe. Our
270: initial uncertainties were overestimated in comparison to their
271: scatter about the Hubble line, as shown by a total $\chi^{2}$ value of
272: 40.0 for 60 objects. We expect that a ``training'' process for BATM
273: similar to that used to produce templates for other methods will both
274: alleviate this discrepancy and reduce the scatter about the Hubble
275: diagram. Such a procedure would determine how to optimally calculate
276: and combine distance estimates from each of the BATM template
277: light-curves, and has yet to be performed. At present it seems
278: sensible to rescale the uncertainties so that they more accurately
279: reflect the scatter and yield $\chi^{2}/N_{dof} \sim 1$.
280: Uncertainties in Figure ~\ref{hubcompare} reflect this rescaling. The
281: distance uncertainties from Tonry et al. (2003) appear to be
282: $under$estimated for this set of objects, but we have left them as
283: reported.
284:
285: The residuals with respect to a Hubble line for the two sets of
286: distances are compared in the inset of Figure ~\ref{hubcompare}.
287: There is a clear correlation between the two methods, indicating that
288: the residuals are due to intrinsic properties of the SNe rather than
289: some feature of the redshift marginalization or the BATM analysis.
290: This is further evidence that the redshift marginalization procedure
291: is recovering the same distance information as the established
292: $z$-specific techniques. It is natural to worry about additional
293: possible systematic biases that might be introduced by this method.
294: In Figure ~\ref{residscompare} we compare the Hubble diagram residuals
295: as a function of several SN properties. There are no evident biases
296: in the redshift-independent distances as a function of extinction,
297: redshift, or light-curve shape. It is interesting to note that for
298: SNe Ia at the high ends of the sample distribution of $A_V$, $z$, and
299: MLCS $\Delta$, the redshift-independent distances appear to be
300: slightly $less$ biased than those using redshift-specific methods.
301:
302: It should in principle also be possible to examine the ($z$, $d$)
303: probabilities without marginalizing over $z$. Figure
304: ~\ref{dzcontours} shows contours in ($z$, $d$) space for two of our
305: Hubble flow sample as well as for two high-$z$ SN Ia from Barris et
306: al. (2004). Since the $z$-free method is sensitive to (1+$z$), we get
307: very poor constraints on $z$ at low redshift, but for $z > 0.3$, the
308: constraints on $z$ and $d$ are quite similar, and good enough to begin
309: to differentiate between cosmological models. Much further
310: investigation into the correlations between these two parameters is
311: necessary before the accuracy in values of ($z$, $d$) is demonstrated
312: as we have done for $d$ alone.
313:
314: \section{Conclusions}
315: \label{batmloop-conclusions}
316:
317: We have described a procedure for measuring accurate luminosity
318: distances for SNe Ia independent of knowledge of redshift by
319: marginalizing over $z$. This procedure may be used in combination
320: with any distance measurement technique to remove the limitations on
321: supernova surveys imposed by the need for spectroscopic redshifts.
322: When applied as an extension to the BATM method, the redshift
323: marginalization formalism produces a Hubble diagram scatter of
324: $\sim$0.2 magnitudes, comparable to that using conventional
325: implementations of distance methods, in which knowledge of redshift is
326: required. Further refinement of the BATM technique through a
327: ``training'' process, or use with other methods, is likely to reduce
328: the scatter to values equal to that of redshift-specific methods.
329:
330: Distances measured with this method will allow investigation of
331: various properties of SNe as a function of distance, which is all that
332: is necessary to answer many fundamental questions. The rates of SNe
333: Ia as a function of redshift are still not well constrained (see Pain
334: et al. 1996, 2002; Tonry et al. 2003), and likely will remain so due
335: to the difficulty in collecting large numbers of spectroscopic
336: redshift. However, rates as a function of distance are just as
337: fundamental, and the lack of redshift knowledge will not prevent
338: investigations into the dependence of rates on numerous host-galaxy
339: properties, for instance. For investigations where cosmological
340: parameters may be considered as known, the conversion from distance to
341: redshift is determined, so accurate distances are all that is
342: necessary.
343:
344: The method may also be used to produce probability contours as a
345: function of ($d$, $z$) for high-$z$ SNe. This will allow cosmological
346: parameters to be measured based purely upon photometric observations.
347: Such a development will be an important step for cosmological studies
348: based on supernova surveys.
349: In order to accomplish this, it will be necessary to obtain a
350: sufficient number of observations to properly determine the
351: multi-wavelength light-curve shape to isolate the effects of ($d$,
352: $z$) from other fit parameters (particularly $A_V$ and $R_D$) and
353: sufficiently constrain the solution space. The achievable accuracy,
354: and the observational strategy required to reach it, is influenced by
355: both the intrinsic properties of SNe Ia (which are not equally
356: "standard candles" at all wavelengths---see Jha 2002) and the quality
357: of temporal and wavelength coverage of nearby SNe Ia used for distance
358: calibration.
359:
360: The method described here is more powerful than the photometric
361: redshift procedure used to estimate redshifts for galaxies based on
362: their observed colors, because it also takes into account the
363: magnitude and light-curve shape to produce constraints on both
364: redshift and distance. It promises to remove the limitations to SN
365: studies caused by the requirement for spectroscopic redshifts for the
366: determination of distances for SNe Ia. The ongoing ESSENCE (Smith et
367: al. 2002) and Canada-France-Hawaii Legacy Survey
368: (http://www.cfht.hawaii.edu/Science/CFHLS), and future projects such
369: as Pan-STARRS (Kaiser \& Pan-STARRS Team 2002) and the proposed
370: SuperNova Acceleration Probe (Nugent 2001) will also be able to
371: greatly increase their yields using this method, enhancing their
372: ability to investigate fundamental cosmological questions.
373:
374: \acknowledgments
375:
376: We thank the anonymous referee for numerous comments that greatly
377: improved the manuscript. We thank Saurabh Jha and Brian Schmidt for
378: their MLCS and dm15 fits. Financial support for this work was
379: provided by NASA through program GO-09118 from the Space Telescope
380: Science Institute, which is operated by the Association of
381: Universities for Research in Astronomy, Inc., under NASA contract NAS
382: 5-26555. Further support was provided by the National Science
383: Foundation through grant AST-0206329.
384:
385: \begin{thebibliography}{}
386:
387: \bibitem[Barris 2004]{barr04} Barris, B.J. 2004, Ph.D. thesis,
388: University of Hawaii
389:
390: \bibitem[Barris et al. 2002]{barr02} Barris, B., et al. 2002,
391: BAAS, 34, 1306
392:
393: \bibitem[Barris et al. 2004]{bare04} Barris, B.J., et al. 2004, ApJ,
394: 602, 571
395:
396: \bibitem[Hamuy et al. 1996]{hamy96} Hamuy, M., et al. 1996, AJ, 112,
397: 2408
398:
399: \bibitem[Jha 2002]{jha02} Jha, S. 2002, Ph.D. thesis, Harvard
400: University
401:
402: \bibitem[Kaiser 2002]{kais02} Kaiser, N. \& Pan-STARRS Team. 2002,
403: BAAS, 34, 1304
404:
405: \bibitem[Nugent 2001]{nuge01} Nugent, P. 2001, in Particle Physics and
406: Cosmology: Second Tropical Workshop, Edited by Nieves, J.F. (New York:
407: AIP), 263
408:
409: \bibitem[Pain et al. 2002]{pain02} Pain, R., et al. 2002, ApJ, 577, 120
410:
411: \bibitem[Pain et al. 1996]{pain96} Pain, R., et al. 1996, ApJ, 473, 356
412:
413: \bibitem[Perlmutter et al. 1997]{perl97} Perlmutter, S., et al. 1997,
414: ApJ, 483, 565
415:
416: \bibitem[Perlmutter et al. 1999]{perl99} Perlmutter, S., et al. 1999,
417: ApJ, 517, 565
418:
419: \bibitem[Phillips 1993]{phil93} Phillips, M.M. 1993, ApJ, 413, L105
420:
421: \bibitem[Riess et al. 1998]{ries98} Riess, A.G., et al. 1998, AJ, 116,
422: 1009
423:
424: \bibitem[Riess et al. 1999]{ries99} Riess, A.G., et al. 1999, AJ, 117,
425: 707
426:
427: \bibitem[RPK 96]{rpk96} Riess, A.G., Press, W.H., \& Kirshner,
428: R.P. 1996, ApJ, 473, 88
429:
430: \bibitem[Riess et al. 2004]{ries04} Riess, A.G., et al. 2004, ApJ,
431: 600, L163
432:
433: \bibitem[Smith et al. 2002]{smit02} Smith, R.C., et al. 2002, BAAS,
434: 34, 1232
435:
436: \bibitem[Tonry et al. 2003]{tonr03} Tonry, J.L., et al. 2003, ApJ,
437: 594, 1
438:
439: \bibitem[Wang et al. 2003]{wang03} Wang, L., et al. 2003, ApJ, 590,
440: 944
441:
442: \end{thebibliography}
443: \clearpage
444:
445: \begin{figure}
446: \epsscale{0.5}
447: %\includegraphics[scale=0.6,angle=-90]{hubcompare.ps}
448: \plotone{f1.ps}
449: \caption{Hubble Diagram plots for 60 nearby SNe Ia. Bottom line shows
450: distances taken from Tonry et al. (2003), with a scatter of 0.19 mags.
451: Top values are calculated with the redshift-independent
452: marginalization procedure in combination with the BATM method, with a
453: scatter of 0.21 mags. The x-axis values in both cases are
454: $spectroscopic$ redshifts, $not$ redshifts calculated from the SN
455: photometry. Plots of residuals relative to the Hubble diagrams are
456: shown in the inset, with x-axis values for Tonry et al. (2003)
457: distances and y-axis from $z$-free distances. There is a clear
458: correlation, indicating that the residuals are due to properties of
459: the SNe rather than some aspect of the BATM method or the redshift
460: marginalization procedure. Dotted diagonal line is $not$ a fit to the
461: data, but is to guide the eye. }
462: \label{hubcompare}
463: \end{figure}
464: \clearpage
465:
466:
467: \begin{figure}
468: \epsscale{0.5}
469: \plotone{f2.ps}
470: \caption{Hubble diagram residuals as a function of SN properties for
471: distances calculated with the redshift-independent marginalization
472: procedure in combination with the BATM method (top), and for those
473: calculated from various redshift-specific methods (bottom). There are
474: no apparent biases in the redshift-independent distances as a function
475: of extinction (values taken from Tonry et al. 2003), redshift, or
476: light-curve shape. In fact, for large values of $A_V$, $z$, and MLCS
477: $\Delta$, the redshift-independent distances appear to be slightly
478: $less$ biased than redshift-specific methods.}
479: \label{residscompare}
480: \end{figure}
481: \clearpage
482:
483:
484:
485: \begin{figure}
486: \epsscale{0.5}
487: %\includegraphics[scale=0.6,angle=-90]{dzcontours4a.ps}
488: \plotone{f3.ps}
489: \caption{Plot showing contours of constant probability in ($d$,$z$)
490: parameter space corresponding to 1-$\sigma$ for SN1996ab ($z$=0.124)
491: and SN1996C ($z$=0.028), from the Hubble-flow sample, as well as
492: SN2001iy ($z$=0.568) and SN2001jm ($z$=0.978) from Barris et
493: al. (2004). Also shown are cosmological models for flat universes
494: with ($\Omega_{M}$,$\Omega_{\Lambda}$)=(0.3, 0.7) and (1.0, 0.0), top
495: and bottom, respectively. Distance moduli are calculated with
496: $H_0$=72 km/s/Mpc. Contours for the two low-$z$ SN are artificially
497: truncated since our minimum value for redshift was $z$=0.01. With
498: further investigation, such contour plots may allow cosmological
499: density parameters to be determined based purely on photometric
500: observations of high-$z$ SNe.}
501: \label{dzcontours}
502: \end{figure}
503:
504: \end{document}