1:
2:
3: \documentclass{emulateapj}
4:
5: %\documentclass[manuscript]{aastex}
6:
7: %% preprint2 produces a double-column, single-spaced document:
8:
9: %% \documentclass[preprint2]{aastex}
10:
11: \usepackage{natbib}
12:
13: \usepackage{graphicx}
14:
15: %\usepackage{natbib}
16:
17:
18: \newcommand{\kpc}{\,\mathrm{kpc}}
19: \renewcommand{\d}{\mathrm{d}}
20:
21: \slugcomment{ApJ accepted}
22:
23: \shorttitle{Comparisons of mass profiles in strong lensing}
24:
25: \shortauthors{Shu et al.}
26:
27: \begin{document}
28:
29: \input epsf
30:
31: \title{Comparisons between isothermal and NFW mass profiles for strong-lensing galaxy clusters}
32:
33: \author
34: {Chenggang Shu\altaffilmark{1,2,6} Binglu Zhou\altaffilmark{1,2}, Matthias
35: Bartelmann\altaffilmark{2}, \\ Julia M. Comerford\altaffilmark{3}, J.-S. Huang\altaffilmark{4},
36: Yannick Mellier\altaffilmark{5}}
37:
38: \email{cgshu@shao.ac.cn}
39:
40: \altaffiltext{1} {Shanghai Normal University, 100 Guilin Road,
41: Shanghai 200234, China}
42:
43: \altaffiltext{2} {Zentrum f\"ur Astronomie der Universit\"at
44: Heidelberg, ITA, Albert-\"Uberle-Str. 2, 69120 Heidelberg, Germany}
45:
46: \altaffiltext{3} {Astronomy Department, 601 Campbell Hall,
47: University of California, Berkeley, CA 94720, USA}
48:
49: \altaffiltext{4} {Harvard-Smithsonian Center for Astrophysics, 60
50: Garden Street, Cambridge, MA 02138, USA}
51:
52:
53: \altaffiltext{5} {Institut d'Astrophysique de Paris, UMR 7095, 98bis
54: BD Arago, 75014 Paris, France}
55:
56: \altaffiltext{6} {Shanghai Astronomical Observatory, Chinese Academy
57: of Sciences, Shanghai 200030, China}
58:
59:
60: \begin{abstract}
61:
62: While both isothermal and NFW-based mass models for galaxy clusters
63: are widely adopted in strong-lensing studies, they cannot easily be
64: distinguished based solely on observed positions of arcs and
65: arclets. We compare the magnifications predicted for giant arcs
66: obtained from isothermal and NFW profiles, taking axially-symmetric
67: and asymmetric mass distributions into account. We find that arc
68: magnifications can differ strongly between the two types of density
69: profiles even if the image morphology is well reproduced.
70: Magnifications by lenses with NFW density profiles are usually
71: larger than those for lenses with singular or nearly singular
72: isothermal density profiles, unless the latter have large cores.
73: Asymmetries play an important role. We illustrate our results with
74: the two well-studied clusters MS~2137 and A~370. We confirm earlier
75: results showing that both isothermal and NFW mass models can very
76: well reproduce the observed arcs, radial arcs and other arclets.
77: While the mass model for MS~2137 is not very well constrained, the
78: two types of mass models produce strongly differing critical curves
79: and caustics for A~370. We find that the NFW mass profile is
80: preferred for A~370. We identify new arclet candidates in the field
81: of A~370. Redshift estimates allowed by the lens model are
82: consistent with results in the literature, where available. Three
83: newly found counter-images are suggested to arise from an active,
84: dust-enshrouded star-forming galaxy at $z\approx1.1$.
85:
86: \end{abstract}
87:
88: \keywords{dark matter - gravitational lensing - galaxies: clusters:
89: individual (MS~2137, A~370)}
90:
91: \section{Introduction}
92:
93: Strong lensing by galaxy clusters, giving rise to strongly distorted
94: arc-like images, plays an important role in studies of the mass
95: distributions of galaxy clusters. It is highly non-linear, weakly
96: affected by baryonic physics, and the data reduction is
97: comparatively simple because it focuses on the positions of arcs and
98: arclets (e.g.~Leonard et al 2007; Limousin et al 2006; Halkola,
99: Seitz \& Pannella 2006; Zekser et al 2006; Covone et al 2006;
100: Comerford et al 2006; Broadhurst et al 2005; Gavazzi 2005; Gavazzi
101: et al 2003; Smith et al 2005; Kneib et al 2003; Kneib et al 1993;
102: Mellier, Fort \& Kneib 1993; Hammer 1991). Cosmological applications
103: of strong lensing range from the use of clusters as gravitational
104: telescopes (Frye et al 2007; Ofek et al 2006; Chary, Stern \&
105: Eisenhardt 2005; Kneib et al 2004; Ellis et al 2001) to cosmological
106: constraints based on arc statistics (Fedeli et al 2008; Hamana et al
107: 2005; Li et al 2005; Meneghetti et al. 2005a,b; Puchwein et al 2005;
108: Dalal \& Holder, 2004; Wambsganss et al 2004; Bartelmann et al
109: 1998). Currently, about 100 clusters are known to contain large
110: arcs, including those newly discovered by Bolton et al (2006) in the
111: Sloan Digital Sky Survey data.
112:
113: One way of investigating cluster mass distributions with strong
114: lensing starts from an assumed mass profile, which is commonly
115: taken to be the well-known singular or non-singular isothermal
116: sphere (SIS or NIS, respectively)
117: \begin{equation}
118: \label{eq:SIS}
119: \rho_{\rm SIS}\propto \frac{1}{r^2}\quad\mbox{or}\quad
120: \rho_{\rm NIS}\propto \frac{1}{r^2+r_\mathrm{c}^2}\;,
121: \end{equation}
122: where $r$ is the radius from the cluster centre and $r_\mathrm{c}$
123: the core radius.
124:
125: Another mass model commonly used for strong lensing is based on the
126: high-resolution numerical simulations of dark-matter halos in the
127: $\Lambda$CDM framework by Navarro, Frenk \& White (1996, 1997;
128: hereafter NFW), who found the density profile
129: \begin{equation} \label{eq:NFW}
130: \rho_{\rm NFW}\propto\left(\frac{r}{r_\mathrm{s}}\right)^{-1}\,
131: \left[1+\left(\frac{r}{r_\mathrm{s}}\right)^2\right]^{-1}\;,
132: \end{equation}
133: parameterized by the scale radius $r_\mathrm{s}$. While simulations
134: and X-ray observations of galaxy clusters both support the NFW
135: profile and applications to strong-lensing studies have been
136: successful (Zekser et al 2006; Comerford et al 2006; Kneib et al
137: 2003), the SIS and NIS models can also reproduce observed arcs
138: remarkably well in many clusters (e.g.~Kneib et al 1993; Mellier et
139: al 1993; Miradal-Escud\'e 1995; Gavazzi et al 2003; Gavazzi 2005).
140:
141: So far, conclusive evidence that the density profiles of real and
142: simulated galaxy clusters agree with each other has not been found.
143: Unfortunately, it is virtually impossible to distinguish isothermal
144: and NFW density profiles based solely on the positions of observed
145: arcs because of the near-degeneracy between these two density
146: profiles close to the scale radius. If radial arcs or multiple
147: tangential arcs at different distances from the cluster centre are
148: available, the situation improves substantially because they probe
149: the density profile at different radii (Gavazzi et al 2003; Gavazzi
150: 2005; Tu et al 2008). While the positions of arcs and arclets are
151: determined by the first derivative of the lensing potential, their
152: magnifications and distortions depend on its second derivatives.
153: Different models reproducing the image geometry can thus yield
154: substantially different image magnifications.
155:
156: The scientific goal of this paper is to investigate how relative arc
157: and arclet magnifications can be used to break degeneracies between
158: different mass models for strongly lensing clusters when combined
159: with positional and photometric information.
160:
161:
162: Two well-studied clusters, A~370 and MS~2137, are used for
163: illustration. The code for fitting lens models is based on Comerford
164: et al (2006) and will be briefly described below. Wherever needed,
165: we adopt the ``concordance'' flat cosmology with
166: $\Omega_\mathrm{m}=0.3$, $\Omega_\Lambda=0.7$ and Hubble constant
167: $h=0.7$.
168:
169: The paper is structured as follows. Starting from a one-component
170: lens model, we compare in Sect.~2 the predicted magnifications of
171: the resulting arcs for the SIS, NIS, and NFW profiles, respectively,
172: starting from axial symmetry and taking deviations therefrom into
173: account. The results are easily extrapolated to lens systems with
174: multiple components since the distance between two lens components
175: is typically larger than their core or scale radii, $r_\mathrm{c}$
176: or $r_\mathrm{s}$. In Sect.~3, earlier work on MS~2137 and A~370 is
177: briefly summarized. We adapt different mass models to these
178: clusters, derive the arc magnifications and compare with
179: observations. Based on the high-quality images observed, we study
180: the arclets in the field of A~370 in detail. We present our
181: conclusions in Sect.~4.
182:
183:
184: \section{Theoretical perspectives}
185:
186: \subsection{Basic lensing notation}
187:
188: We first outline the notation to be used below. As usual, lenses are
189: assumed to be sheet-like perpendicular to the line-of-sight. Then,
190: the lens mapping from the lens to the source plane can be written as
191: \begin{equation} \label{eq:lens}
192: \vec S=\vec L-\nabla\psi\;,
193: \end{equation}
194: and the local distortion of the lensed images is determined by the
195: Jacobian matrix
196: \begin{equation} \label{eq:metric}
197: \mathcal{A}_{ij}=\frac{\partial S_i}{\partial L_j}\;,
198: \end{equation}
199: where $(S_1,S_2)$ and $(L_1,L_2)$ are dimensionless coordinates in
200: the source and lens planes, respectively, and $\psi$ is the
201: effective lensing potential, which characterizes the lensing mapping
202: completely for any given lens model. The magnification at a given
203: position $\vec L$ is
204: \begin{equation} \label{eq:mag}
205: \mu=\frac{1}{\det\mathcal{A}}\;.
206: \end{equation}
207: Giant arcs appear close to the critical curves when their sources
208: are close enough to the corresponding caustics. Since giant arcs
209: are the most pronounced structures due to highly non-linear strong
210: lensing, they provide sensitive constraints on the lens models.
211:
212: For simplicity, we adopt in the following subsections a
213: one-component lens model to illustrate the predicted magnifications
214: of expected giant arcs for the SIS, NIS, and NFW profiles,
215: respectively. Although real lenses typically have multiple
216: components, the main conclusions on the magnification will remain
217: valid because they are due to generic lensing behaviour near
218: critical curves or caustics.
219:
220: \subsection{Axial symmetry}
221:
222: If the lens is axially symmetric, i.e.~the ellipticity $\epsilon=0$,
223: we can obtain the radial gradient $(\d\det\mathcal{A}/\d s)|_
224: \mathrm{t}$ of the lens mapping Eq.~(\ref{eq:metric}) at the radius
225: $r_\mathrm{t}$ where giant tangential arcs appear. There,
226: $\det\mathcal{A}=0$. We introduce
227: \begin{equation}
228: s=r/r_\mathrm{s}
229: \end{equation}
230: throughout this section for easier comparison. We plot the gradient
231: $(\d \det\mathcal{A}/\d s)|_{\mathrm{t}}$ as a function of $s$ in
232: Fig.~\ref{sym_dadr}, where the solid, dotted and dashed lines show
233: the results for SIS, NIS, and NFW profiles, respectively.
234:
235: \begin{figure}
236: \includegraphics[width=\hsize]{f1}
237: \caption{The predicted gradient $(\d\det\mathcal{A}/\d s)|_t$ of the
238: Jacobian determinant at the position of the tangential critical
239: curve is shown as a function of the dimensionless radius $s$ with
240: the solid, dotted and dashed lines for the SIS, NIS, and NFW density
241: profiles, respectively. Dotted lines from top to bottom show the
242: results for the NIS model with $r_\mathrm{s}/r_\mathrm{c}=50, 10,
243: 5$, respectively, compared to the scale radius of the NFW model (see
244: the text for details).} \label{sym_dadr}
245: \end{figure}
246:
247: The radial gradient of the Jacobian determinant at the tangential
248: critical curve is smaller for NIS profiles with different core
249: radii than for the SIS profile, since the NIS density profile is
250: always flatter than that of the SIS, thus the projected mass
251: distribution is flatter for the NIS. With increasing core radii in
252: NIS profiles, i.e.~decreasing $r_\mathrm{s}/r_\mathrm{c}$, the
253: gradient decreases due to the flattening of the density profiles. If
254: the giant arc is located at a radius $r_\mathrm{t}\ga r_\mathrm{s}$,
255: i.e., $s\ga 1$, the gradient is larger for a NFW than for the SIS
256: and NIS profiles because it is steepens towards $\propto r^{-3}$
257: outside the scale radius. If the giant arc appears in the inner
258: region, i.e.~$r_\mathrm{t}\la r_\mathrm{s}$, the gradient for a SIS
259: or a NIS lens with a small core ($r_\mathrm{s}/r_\mathrm{c}\ga10$)
260: is larger than that for a NFW lens, while the gradient for a NIS
261: profile with a relative large core
262: ($r_\mathrm{s}/r_\mathrm{c}\la10$) is smaller than for a NFW profile
263: due to its flatter density profile.
264:
265: It is interesting to study the corresponding magnifications of the
266: giant arcs produced by these different mass profiles. The
267: determinant $\det\mathcal{A}$ of the lens mapping in
268: Eq.~(\ref{eq:metric}) at a distance $\delta r$ from the critical
269: curve can be written as
270: \begin{equation} \label{eq:dadr}
271: {\rm det}\mathcal{A}={\rm det}\mathcal{A}|_\mathrm{t}+
272: \left.\frac{\d{\,\rm det}\mathcal{A}}{\d r}\right|_\mathrm{t}\delta r=
273: \left.\frac{\d{\,\rm det}\mathcal{A}}{\d r}\right|_\mathrm{t}\delta r\;,
274: \end{equation}
275: where the subscript $t$ again denotes the position of the tangential
276: critical curve where $\det\mathcal{A}=0$. If the width of the arc is
277: $W$, its total magnification $\bar\mu_\mathrm{t}$ can be expressed
278: by
279: %\begin{equation}
280: \begin{eqnarray} \label{eq:mu_s}
281: {\bar \mu_\mathrm{t}}&=&\frac{\int\d S_i}{\int\d S_s}=
282: \frac{\int\d S_i}{\int\d S_i/\mu_i} \nonumber \\
283: &=&\frac{\int_{r_\mathrm{t}-{\rm W/2}}^{r_\mathrm{t}+{\rm W/2}}\;r\d r}
284: {\left.\d\det\mathcal{A}/\d r\right|_\mathrm{t}
285: \int_{r_\mathrm{t}-{\rm W/2}}^{r_\mathrm{t}+{\rm W/2}}\;(r-r_\mathrm{t})r\d
286: r} \nonumber \\
287: &=&\frac{12s_\mathrm{t}}{\d\det\mathcal{A}/\d s|_\mathrm{t}}\left(
288: \frac{r_\mathrm{s}}{W}
289: \right)^2
290: =\frac{12}{\d\det\mathcal{A}/\d s|_\mathrm{t}s_\mathrm{t}}\left(
291: \frac{r_\mathrm{t}}{W}
292: \right)^2\;.
293: \end{eqnarray}
294: If an observed giant arc appears at a radius $r_\mathrm{t} \sim
295: r_\mathrm{s} \sim15''$ from the cluster centre with a width of
296: $W\sim3''$, say, its estimated magnification can reach $\sim100$.
297:
298: The predicted total magnification $\bar \mu_\mathrm{t}$ scaled by
299: $(r_\mathrm{t}/W)^2$ for giant arcs as a function of their distance
300: $s$ from the lens centre is plotted in Fig.~\ref{sym_mag} with the
301: same meaning of the line types as in Fig.~\ref{sym_dadr}. The figure
302: shows that the expected magnification of an arc is always lower for
303: a SIS than for a NIS model, given the observed positions of giant
304: arcs and their widths. This is because the SIS profile is always
305: steeper than the NIS profile. If the arc appears at $r_\mathrm{t}\ga
306: r_\mathrm{s}$, the predicted magnification for a NFW profile is the
307: lowest due to the steepest density distribution. If the arc appears
308: at $r_\mathrm{t}\la r_\mathrm{s}$, the expected magnification for
309: the SIS or the NIS models with a small core ($r_\mathrm{s}/r_
310: \mathrm{c}\ga10$) is lower than that for a NFW model. Only when the
311: core of the NIS profile is large ($r_\mathrm{s}/r_\mathrm{c}\la10$),
312: comparable to the scale radius of a NFW profile, the expected
313: magnification is larger than that for a NFW model since its density
314: profile will be significantly flatter in the inner region. The
315: conclusions can thus be well understood in terms of the flatness of
316: the density profile near the arc locations. See also the discussion
317: in Meneghetti et al (2005b).
318:
319: \begin{figure}
320: \includegraphics[width=\hsize]{f2}
321: \caption{The expected magnification for giant arcs,
322: $\mu_\mathrm{t}$, scaled by $(r_\mathrm{t}/W)^2$, as a function of
323: the dimensionless radius $s$. As in Fig.~\ref{sym_dadr}, the solid,
324: dotted and dashed lines represent the SIS, NIS, and NFW density
325: profiles, respectively. From bottom to top, the dashed lines show
326: the results for a NIS profile with $r_\mathrm{s}/r_\mathrm{c}=50,
327: 10, 5$, respectively (see the text for more detail).}
328: \label{sym_mag}\end{figure}
329:
330: \subsection{Asymmetry}
331:
332: Since the gravitational potentials of real lensing clusters must be
333: asymmetric, as was already pointed out by Grossman \& Narayan
334: (1988), Kovner (1989), Mellier et al (1993) and Miralda-Escud\'e
335: (1995), we now turn to asymmetric models, which play an important
336: role in cluster-lensing studies (Bartelmann \& Meneghetti 2004). As
337: in Bartelmann \& Meneghetti (2004), we define the surface-mass
338: distribution of the cluster in terms of the projected elliptical
339: radius
340: \begin{equation} \label{eq:re}
341: r_\mathrm{e}=\left[
342: (r\cos\theta)^2(1-e)+(r\sin\theta)^2/(1-e)
343: \right]^{1/2}\;.
344: \end{equation}
345: Here, $\theta$ is the position angle counterclockwise from the $+y$
346: axis. The ellipticity
347: \begin{equation} \label{eq:e}
348: e=\frac{a-b}{a+b}
349: \end{equation}
350: of an ellipse with major and minor axes $a$ and $b$ refers to the
351: lensing potential rather than the projected density. Note that the
352: definition is chosen so as to conserve the total projected mass
353: within $r_\mathrm{e}$ independent of ellipticity, as pointed out by
354: Meneghetti et al (2005b).
355:
356: Projected ellipticities of real clusters are not very large; they
357: are always smaller than their real ellipticities in three
358: dimensions. We restrict our theoretical results to the first order
359: in $\epsilon$ in this subsection. Then, the critical curve
360: approximates an ellipse. Without restriction, we assume that its
361: major and minor axes are oriented along the $x$ and $y$ coordinates,
362: respectively. Giant arcs preferentially appear near positions with
363: either $x\approx0$ or $y\approx0$ from sources in the neighbourhood
364: of cusp points in the caustics.
365:
366: \begin{figure}
367: \includegraphics[width=\hsize]{f3}
368: \caption{Same as Fig.~\ref{sym_dadr}, but choosing different
369: ellipticities $\epsilon\ne0$. Giant arcs appearing near the $x$ and
370: $y$ axes are considered (see text for further detail).}
371: \label{asym_dadr}
372: \end{figure}
373:
374: The gradient of the Jacobian determinant $(\d\det\mathcal{A}/\d
375: s)|_\mathrm{t}$ of the lens mapping Eq. (\ref{eq:metric})
376: perpendicular to the critical curve at $x=0$ and $y=0$, where giant
377: tangential arcs are expected to appear, is shown in
378: Fig.~\ref{asym_dadr} as a function of the dimensionless radius $s$
379: for different ellipticities $\epsilon$. The line types are chosen as
380: in Figs.~\ref{sym_dadr} and \ref{sym_mag}. For a given $\epsilon$,
381: the gradient exhibits the same behaviour as in Fig.~\ref{sym_dadr}
382: for axially-symmetric SIS, NIS, and NFW density profiles for the
383: same reasons.
384:
385: Comparing with axially-symmetric lens models, the situation for
386: different ellipticities $\epsilon$ is more complicated. Two factors
387: will affect the gradient of the Jacobian determinant
388: $(\d\det\mathcal{A}/\d s)|_\mathrm{t}$ and its influence on a giant
389: arc at a given position: (1) the total mass enclosed by a circle
390: traced by the arc, and (2) the density gradient. Ellipticity
391: enhances the tidal field and thus shifts critical curves towards
392: larger radii. Thereby, it increases the gradient
393: $(\d\det\mathcal{A}/\d s)|_\mathrm{t}$ because of the increasing
394: steepness of the projected density profile for a given mass model.
395:
396: As in the previous subsection, we plot in Fig.~\ref{asym_mag} the
397: expected magnification for arcs near the $x$ and $y$ axes as a
398: function of their distance from the cluster centre for asymmetric
399: lens models. The figure shows that the arc magnifications show the
400: same general trends as in Fig.~\ref{sym_mag}, for the same reasons.
401: Note that the magnifications of axially-symmetric and
402: axially-asymmetric models differ significantly even if the
403: ellipticity $\epsilon$ is small. The predicted magnifications of
404: arcs for individual mass models decrease for increasing $\epsilon$
405: because of similar reasons as discussed above, especially for arcs
406: near $x = 0$, i.e.~near the major axis.
407:
408: In summary, the expected magnifications for giant arcs in lenses
409: with different mass models are different even though the arc
410: positions can be reproduced by all of them. Asymmetry of the mass
411: distribution plays an important role for strong lensing. The
412: resulting magnifications can change substantially even if the
413: system's ellipticity is small. Although the absolute magnifications
414: for individual arcs or arclets cannot be obtained observationally,
415: lens models will be additionally constrained by the relative
416: magnifications of arc or arclet pairs from the same sources within a
417: lens system. We shall use the well-studied clusters A~370 and
418: MS~2137 as illustrations in the next section.
419:
420: \begin{figure}
421: \includegraphics[width=\hsize]{f4}
422: \caption{Similar to Fig.~\ref{sym_mag}, but adopting different
423: ellipticities $\epsilon$ of the lensing potential. Results are shown
424: for giant arcs appearing near the $x$ and $y$ axes, respectively
425: (see the text for more detail).} \label{asym_mag}
426: \end{figure}
427:
428: \section{Illustration: A~370 and MS~2137}
429:
430: As discussed in the previous section, different mass models are
431: possible for individual galaxy clusters if they are only based on
432: the observed positions of giant arcs. Relative magnifications of arc
433: or arclet pairs from individual sources in a lens system can be used
434: to constrain the mass model further. We will investigate the arc
435: magnifications in NIS and NFW models, which can well reproduce the
436: arc configurations in A~370 and MS~2137, respectively. We choose
437: these two clusters for the following reasons:
438:
439: \begin{itemize}
440:
441: \item They are prototypical lensing clusters, with A~370 consisting of
442: two mass components and MS~2137 having one dominant component.
443:
444: \item Both of them have large tangential and radial arcs, which mass models are required to fit.
445:
446: \item Both of them have at least two arc or arclet pairs with
447: photometric data available in several bands, which allow us to use
448: relative magnifications (i.e.~flux ratios) of image systems from
449: individual sources to further constrain the mass models.
450:
451: \item High-quality HST images are available for both of them, which can
452: be used for detailed strong-lensing studies.
453:
454: \item Both clusters have been thoroughly studied earlier, which allows
455: detailed comparisons.
456:
457: \item For A~370 in particular, we have high-quality images in the $B$,
458: $V$, $R$, $I$, $z$, $J$, $H$ and $K'$ bands taken by Subaru with a
459: median seeing of $\sim0''.5$ (Cowie et al. 2008), and in four
460: infrared channels observed by IRAC on-board the \emph{Spitzer} space
461: telescope in addition to its HST archive images. The deep optical
462: and NIR images permit the identification of individual images of
463: lensed galaxies in the A370 region studied in the present paper. The
464: IRAC images have $\gtrsim2''$ resolution and will be used to study
465: in detail the spectral energy distributions (SEDs) for the sources
466: combined with the optical and NIR data in a forthcoming paper.
467:
468: \end{itemize}
469:
470: We shall very briefly summarize earlier work on these two clusters
471: in the following subsections. Different mass models will be applied
472: to the two clusters in order to reproduce the main observed features
473: of arcs and arclets. Since there are many arclets observed in the
474: field of A~370 and extensive observational data are available, we
475: will study this cluster in more detail.
476:
477: Our models improve the previous lens models for MS~2137 and A~370 in
478: the following way. Gavazzi et al (2003) studied the density profile
479: for MS~2137 in detail and Comerford et al (2006) successfully
480: applied the NFW density profile to this cluster. Both studies did
481: not explicitly consider the relative magnifications between image
482: pairs from individual sources as constrained by the photometric
483: observations, which we focus on in this paper. For A~370, we not
484: only employ more observed arclet pairs than Comerford et al (2006)
485: to constrain the mass model, but also compare their magnifications
486: with the observed photometry. Moreover, we study additional arclets
487: in more detail, including their estimated lens redshifts.
488:
489: We use the same code as in Comerford et al. (2006) to fit mass
490: models to observations. Although only NFW profiles were fit to
491: observations there, isothermal density profiles are also provided by
492: this code.
493:
494: Briefly, it proceeds as follows: Varying the density-profile
495: parameters, it strives (1) to make arc sources as small as possible,
496: (2) to make arcs and arclets produced from these sources match the
497: observed data as closely as possible, and (3) to avoid additional
498: images arising from the sources or make them faint enough to be
499: compatible with the observations.
500:
501: These three requirements are quantified by three contributions to a
502: figur-of-merit function, which is jointly minimized to find the
503: best-fitting model parameters. This function is not precisely a
504: $\chi^2$ function, as it cannot be normalised to the degrees of
505: freedom because the intrinsic source size is unknown. See Comerford
506: et al. (2006) for more detail. Relative magnifications, i.e.~the
507: flux ratios, between images in image systems of individual sources,
508: are not included in the figure-of-merit function when fitting mass
509: models to MS~2137 and A~370 in the following subsections. They are
510: used in the further discussions on how well the best-fitting models
511: can match the photometric observations.
512:
513: \subsection{MS~2137}
514:
515: MS~2137 is a rich cD cluster at $z=0.313$. Early studies by Mellier,
516: Fort \& Kneib (1993) and Miralda-Escud\'e (1995) found that
517: elliptical isothermal models can reproduce the observed arc system
518: remarkably well. Miralda-Escud\'e (2002) suggested that its mass
519: distribution should be highly elliptic. Comerford et al (2006)
520: showed that the cluster can also be modelled with NFW components.
521: Based on archival data from the HST and the ESO-VLT, Gavazzi et al
522: (2003) and Gavazzi (2005) investigated its mass profile in detail
523: and found that it can be fit by the isothermal and the NFW density
524: profiles alike. Based on a possible fifth image in the very cluster
525: centre and on stellar kinematical data, they concluded that the
526: isothermal model may be preferred.
527:
528: \begin{figure}
529: \includegraphics[width=\hsize]{f5}
530: \caption{Critical curves and caustics for the two best-fitting mass
531: models of MS~2137.} \label{ms2137 crit caus}
532: \end{figure}
533:
534: We fit both the NIS and the NFW mass density profiles to MS~2137,
535: attempting to reproduce the observed five arcs including the radial
536: arc. We adopt a single lens component at the position of the cD
537: galaxy, similar to Comerford et al. (2006). The predicted critical
538: curves and caustics of the different best-fitting models are shown
539: in Fig.~\ref{ms2137 crit caus}. They are almost identical for the
540: two types of model, which confirms the results of earlier studies,
541: since a one-component lens system cannot be tightly constrained with
542: arc systems at identical or similar redshifts.
543:
544: \begin{figure}
545: \includegraphics[width=\hsize]{f6}
546: \caption{The HST image (upper panel) and the reproduction of the
547: observed data (lower panel) by the two best-fitting mass models for
548: MS~2137, with (A0, A2, A4) and (A1, A5+A6) being identified as
549: counter-images of two different sources. Observed arcs, reproduced
550: images and the reconstructed sources are marked by $\diamond$,
551: $\times$, and $\ast$, respectively. The critical curves and caustics
552: of the models are marked in black.} \label{ms2137 arc}
553: \end{figure}
554:
555: The reproduction of the arc systems by the best-fitting mass models
556: are shown in Fig.~\ref{ms2137 arc} adopting the notation of Gavazzi
557: et al (2003), where (A0, A2, A4) and (A1, A5+A6) are identified as
558: counter-images of two different sources. The best-fitting model
559: parameters are given in Tab.~\ref{tab_fitting}. The figures-of-merit
560: obtained with the complete set of input data points read off the
561: observational data are also listed in the table. It appears that the
562: NFW profile is preferred based on the figures-of-merit of different
563: best-fitting mass models. Note that the figure-of-merit in this
564: paper adds the three contributions of sources, the observed and the
565: predicted images. Although it does not have the usual absolute
566: statistical meaning of a properly normalised $\chi^2$ function, it
567: can still serve to evaluate different models based on identical
568: input data (see Comerford et al (2006) for more detail).
569:
570: Since we use the same multiple-image systems as Comerford et al
571: (2006) to constrain our models, the model parameters obtained for
572: the NFW profile are similar except for a very small difference in
573: $\kappa_\mathrm{s}$, where we find $0.66$ rather than the $0.67$ of
574: Comerford et al (2006), which is perfectly within the expected
575: uncertainty.
576:
577: The parameter $\kappa_\mathrm{s}$ is the amplitude of the
578: convergence profile $\kappa(x)$, defined by
579: \begin{equation}
580: \kappa(x)=\frac{2\kappa_\mathrm{s}}{x^2-1}\left[
581: 1-\frac{2}{\sqrt{x^2-1}}\arctan\sqrt{\frac{x-1}{x+1}}
582: \right]
583: \label{kappa_NFW}
584: \end{equation}
585: for the NFW profile with $x=r/r_\mathrm{s}$ and by
586: \begin{equation}
587: \kappa(x)=\frac{\kappa_\mathrm{s}}{2}
588: \frac{x^2+2x_\mathrm{c}^2}{(x^2+x_\mathrm{c}^2)^{3/2}}
589: \label{kappa_NIS}
590: \end{equation}
591: for the NIS profile, where $x$ and $x_\mathrm{c}$ are the radius $r$
592: and the core radius $r_\mathrm{c}$ in units of the Einstein radius.
593: Both the NFW and the NIS models predict a fifth-image in the central
594: cluster region which was confirmed by Gavazzi et al (2003) after
595: subtracting the light produced by the cD galaxy.
596:
597: \begin{table}
598: \caption{Parameters of the best-fitting models for MS~2137 and
599: A~370, with Column (1): name of the cluster; (2): label of lenses
600: where applicable; (3): convergence amplitude $\kappa \mathrm{s}$;
601: (4) scale radius for NFW, core radius for NIS models; (5):
602: ellipticity; (6): position angle; (7): averaged figure-of-merit for
603: the total number of data points; (8): adopted mass model.}
604: \begin{tabular}{llrrrrrl}
605: \hline\hline cluster & lens & $\kappa_\mathrm{s}$ & $r_\mathrm{c/s}$
606: & $\epsilon$ & $\theta$ & $\chi^2$ & model \\
607: & & & ($\mathrm{kpc}/h$) & & (deg) & &\\
608: \hline
609: MS~2137 & -- & $0.66$ & $64$ & $0.11$ & $95.7$ & 0.19 & NFW \\
610: & -- & $3.12$ & $17$ & $0.11$ & $95.7$ & 0.82 & NIS \\
611: \hline
612: A~370 & G1 & $0.16$ & $254$ & $0.28$ & $78.0$ & 0.16 & NFW \\
613: & G2 & $0.17$ & $212$ & $0.07$ & $-13.4$ & &NFW \\
614: \hline
615: & G1 & $50.0$ & $1$ & $0.20$ & $75.0$ & 2.5& NIS \\
616: & G2 & $30.0$ & $1$ & $0.05$ & $-10.0$ & & NIS \\
617: \hline\hline
618: \end{tabular}
619: \label{tab_fitting}
620: \end{table}
621:
622: The predicted absolute magnifications for the individual arcs are
623: shown in Fig.~\ref{compare 2137} for the two different best-fitting
624: mass models for MS~2137, with the solid diagonal indicating where
625: the two magnifications for individual arcs would agree. As the
626: figure shows, the arc magnifications predicted by the two models
627: agree well. This implies that further constraints on the density
628: profile for MS~2137 based on relative image magnifications of arcs
629: from individual sources cannot be expected because MS~2137 is a
630: relatively simple lens.
631:
632: \begin{figure}
633: \includegraphics[width=\hsize]{f7}
634: \caption{Predicted magnifications of arcs and arclets in the fields
635: of MS~2137 based on the NIS model, compared to those based on the
636: NFW model, with (A0, A2, A4) and (A1, A5+A6) being identified as
637: image systems of two different sources. The solid line illustrates
638: where they would agree.} \label{compare 2137}
639: \end{figure}
640:
641: Observationally, we can obtain the relative magnifications for three
642: arc or arclet pairs, i.e., A0/A2 and A0/A4 for the giant arc and
643: A1/(A5+A6) for the radial arc, resulting from two sources based on
644: the photometric data in seven bands ($U$, $B$, $V$, $R$, $I$, $J$
645: and $K$; Gavazzi et al 2003). The averaged results obtained in seven
646: bands are shown in Fig.~\ref{obs_2137} with their photometric
647: errors. We find that the model predictions by both the NIS and the
648: NFW model are roughly in agreement with the averaged relative
649: magnifications for these three pairs within the observational
650: errors, although the NFW model again appears slightly preferred. We
651: conclude that both the NIS and the NFW mass profiles work well for
652: MS~2137, even if the magnification, which is the second order of the
653: derivative of the potential, is considered.
654:
655: Moreover, with the exception of the radial arc, the predicted
656: magnifications are somewhat higher for the NIS than for the NFW
657: model, which can be understood as follows. According to the
658: parameters of the best-fitting models for MS~2137 given in
659: Tab.~\ref{tab_fitting}, $r_\mathrm{s}/r_\mathrm{c}\sim4$, showing
660: that the core radius of the NIS model is comparable to the scale
661: radius of the NFW profile. This yields higher predicted
662: magnifications for the arcs reproduced by the NIS compared to the
663: NFW model (cf. Fig.~\ref{asym_mag}). It should be pointed out that
664: the predicted magnification of the radial arc is calculated
665: according to Eq.~(\ref{eq:mag}) based on its observed image
666: location, which straddles the corresponding critical curve. The
667: resulting magnification mainly depends on its observed size, which
668: is difficult to measure accurately since it is located in the very
669: central region of the lens.
670:
671: \begin{figure}
672: \includegraphics[width=\hsize]{f8}
673: \caption{Comparison of the predicted relative magnifications of arcs
674: and arclets pairs in the fields of MS~2137 with the observations.
675: Points from left to right denote R(A1)/A5+A6, A0/A2 and A0/A4 with
676: the photometric errors attached. The solid diagonal indicates where
677: they would agree.} \label{obs_2137}
678: \end{figure}
679:
680: \subsection{A~370}
681:
682: Abell~370 ($z=0.375$) was the first cluster in which gravitational
683: arcs were observed (Soucail et al 1987). This two-component,
684: strongly lensing system provides an interesting case for us to study
685: the magnifications of individual observed arcs and arclets in
686: detail, and to compare with previous work.
687:
688: This cluster has frequently been studied. For instance, Kneib et al
689: (1993) reconstructed the mass distribution of A~370 based on the
690: observed giant arc (A0) and the arc pair (B2,B3) to conclude that
691: the dark matter appears to follow the light distribution for this
692: cluster. Bartelmann (1996) showed that, as in MS~2137, the radial
693: arc can be reproduced by an NFW model, which was recently confirmed
694: with a detailed model by Comerford et al (2006). Using a deep HST
695: image, B\'{e}zecourt et al (1999) constructed a mass model for A~370
696: based on a truncated pseudo-isothermal distribution and found that
697: the observed lensed images can be well reproduced. Using A~370 as a
698: natural telescope, Hu et al (2002) found a galaxy at $z\approx6.56$
699: and Chary, Stern \& Eisenhardt (2005) argued that its star formation
700: rate could be $\ga140\,M_\odot\,\mathrm{yr}^{-1}$, thus providing
701: useful clues to the galaxy-formation models at high redshift.
702:
703: We adopt the notation of B\'{e}zecourt et al (1999) and Kneib et al
704: (1993) for the different arcs and arclets in the field of A~370 to
705: facilitate comparisons with earlier studies. The additional arclet
706: candidates we identify are jointly labeled by ``Z'' below. For
707: clarity, we only show one colour image of A~370 combined from the
708: $B$, $R$ and $K'$ bands rather than separate images in individual
709: bands, with the individual arcs and arclets identified in
710: Fig.~\ref{a370color}. Note that A5 appears near the bottom since it
711: appeared near the boundary of the $K'$-band image, but it can be
712: very clearly seen in the $B$ band.
713:
714: \begin{figure}
715: \includegraphics[width=\hsize]{f9}
716: \caption{Composite image of A~370 taken in the $B$, $R$ and $K'$
717: bands with the arcs and arclets marked.} \label{a370color}
718: \end{figure}
719:
720: Since there are many arcs and arclets in the field of A~370, we list
721: below the image systems produced by individual sources for clarity,
722: except the well defined giant arc A0, the radial arc R and the arc
723: systems B, C, D and E (B\'{e}zecourt et al 1999). In the labeled
724: ``A" sequence, (A1, A2) and (A3, A4) are arc pairs produced by two
725: different sources. A5 is a single arc. For the newly found arcs
726: labeled by ``Z", we suggest in the next subsection that Z1, Z3, and
727: Z6 be arcs produced by three different sources and (Z2, Z4, Z5)
728: result from a single source. Z7 and Z8 (also labeled as W3 and W1
729: respectively in Fig. \ref{a370color}) with W2 are identified as
730: counter-images of another source (see next subsection).
731:
732: As for MS~2137, we first construct both a NIS and a NFW mass model
733: for A~370 reproducing the giant arc A0, the radial arc R, the
734: arclets (A1, A2), (B2, B3), and A5. An extra arclet pair labeled
735: (A3, A4) is also included, which is easily seen in
736: Fig.~\ref{a370color} especially in the $B$-band. In total, we use
737: nine relatively bright arcs and arclets produced by six sources to
738: constrain the mass model for A~370. The lens components are fixed at
739: the centres-of-light of the two bright cD galaxies.
740:
741: The best-fitting model parameters are listed in
742: Tab.~\ref{tab_fitting} together with the averaged figures-of-merit
743: of the complete set of data points. The corresponding critical
744: curves, caustics and the reproduced images are shown in
745: Figs.~\ref{a370 crit caus} and \ref{a370 9arc}. Our results for the
746: NFW model are very close to those found by Comerford et al (2006)
747: where the six arcs and arclets A0, R, (A1, A2) and (B2, B3) were
748: taken into account. This suggests that (A1, A2) and (B2, B3) play
749: important roles in the mass configuration for A370 (Kneib et al
750: 1993). We point out that the high value of $\kappa_\mathrm{s}$ and
751: the low value of $r_\mathrm{c}$ in the best-fitting NIS model imply
752: that the arc in A~370 can be very well reproduced by nearly-singular
753: isothermal lens components.
754:
755: Since the redshifts of individual sources can also be considered as
756: free parameters during the model fitting, we can obtain
757: gravitational-lensing estimates for individual source redshifts,
758: using the measured redshift of the giant arc A0 as a reference for
759: the relative lensing efficiency at different redshifts. This will be
760: discussed further below in this subsection.
761:
762: \begin{figure}
763: \includegraphics[width=\hsize]{f10}
764: \caption{The critical curves and caustics for the best-fitting mass
765: models for A~370.} \label{a370 crit caus}
766: \end{figure}
767:
768: \begin{figure}
769: \includegraphics[width=\hsize]{f11}
770: \caption{The best-fitting results for A~370, using the same notation
771: as in Fig.~\ref{ms2137 arc}.} \label{a370 9arc}
772: \end{figure}
773:
774: Note that the critical curves and caustics of the best-fitting
775: models are very different for the two mass models, although both of
776: them reproduce the observed positions of individual arcs and arclets
777: very well (cf.~Fig.~\ref{a370 9arc}). Since the selected arcs and
778: arclets can well constrain the mass configuration for A~370, this
779: implies once again that mass profiles in clusters cannot be clearly
780: distinguished based exclusively on the image positions observed in
781: strong lensing.
782:
783: Although both best-fitting models succeed in reproducing the main
784: features of the observed arcs and arclets, there are four extra
785: images predicted that were not yet found observationally. The image
786: near the top in Fig.~\ref{a370 9arc} is the counter-image predicted
787: for B2 and B3, which matches the observed counter-image labeled B4
788: in the figures of the next subsection. The other three are the
789: predicted counter-images of the radial arc. These deviations arise
790: because (1) the lens model constructed for A~370 only has two
791: components centred on the cD galaxies without taking perturbations
792: by additional lens components centred on less massive cluster
793: galaxies into account; (2) the three additional images are too faint
794: to be observed which is consistent with the predictions that their
795: magnifications are small.
796:
797: \begin{figure}
798: \includegraphics[width=\hsize]{f12}
799: \caption{Predicted magnifications of arcs and arclets in the field
800: of A~370 based on the NIS model, compared to those based on the NFW
801: model. The solid diagonal indicates where they would agree.}
802: \label{compare_370}
803: \end{figure}
804:
805: \begin{figure}
806: \includegraphics[width=\hsize]{f13}
807: \caption{Comparison of the predicted relative magnifications of arcs
808: and arclets pairs in the field of A~370 with the observations.
809: Points from left to right denote A3/A4 , A1/A2 and B2/B3 with the
810: observational errors added, respectively. The solid diagonal
811: indicates where they would agree.} \label{obs 370}
812: \end{figure}
813:
814: The predicted magnifications of the arcs and arclets selected to
815: constrain the mass model for A~370 are calculated for both the NIS
816: and the NFW density profiles and displayed in Fig.~\ref{compare_370}
817: for comparison. The expected magnifications of these nine arcs and
818: arclets are always larger for the NFW than for the NIS profile, with
819: the magnification of the giant arc being larger by a factor of
820: $\sim2.5$. This is easily understood based on the theoretical
821: results shown in Fig.~\ref{asym_mag}. Since the ratio
822: $r_\mathrm{s}/r_\mathrm{c}$ is very large for the best-fitting
823: models of A~370, the magnifications by the NFW profile must exceed
824: those by the NIS profile.
825:
826: Since there are many arc and arclet pairs in A370 and high-quality
827: images are available in eight optical and NIR bands, we can compare
828: the expectations for the relative magnifications of the confirmed
829: arclet pairs (B2, B3), (A1, A2) and (A3, A4) with the observations
830: in individual bands, to determine which density profile may be
831: preferred. Photometric data reduction is done using SExtractor
832: (Bertin \& Arnouts 1996) for optical and NIR. Here, we only show the
833: results, while the detailed photometry will be described in a
834: forthcoming paper studying the SEDs of the lensed galaxies in A~370.
835:
836: In Fig.~\ref{obs 370}, we plot the relative magnifications (i.e.,
837: flux ratios) of individual arclet pairs averaged over eight
838: independent bands against the corresponding predicted values. Unlike
839: in MS~2137, we find that the \emph{relative} magnifications expected
840: from the NFW profile can match the observations much better than
841: those based on the NIS profile. The pair (B2, B3) plays an important
842: role here. Mellier et al (1991) had already pointed out that (B2,
843: B3) is very sensitive to the lens model. We confirm that the
844: best-fitting NFW profile achieves a lower figure-of-merit than the
845: NIS profile. According to the above analysis and current numerical
846: simulations, we thus suggest that the NFW profile for A~370 matches
847: the strong-lensing observations better than the NIS profile, and
848: adopt it in the remainder of this paper.
849:
850: As discussed above, the redshifts of individual sources can be seen
851: as input parameters in the model fitting because lensing by a fixed
852: mass distribution depends on a distance factor (see Comerford et al
853: 2006) for sources at different redshifts. This allows redshift
854: estimates for individual sources required by the best-fitting
855: models. We proceed as follows. We choose the giant arc A0 as a
856: reference, which has the spectroscopic redshift $z_{\rm A0}=0.724$
857: (Soucail et al 1988). By varying the distance factors of the other
858: eight arcs and arclets and optimising the model, we estimate their
859: redshifts. The error bars are estimated by the redshift interval
860: within which the figure-of-merit increases by a factor of two from
861: the minimum.
862:
863: We find that the source redshifts obtained in this way for the
864: best-fitting NFW and NIS models are similar. They are shown in the
865: upper part of Table~\ref{tab_z}. Redshifts from spectroscopy,
866: photometry and lensing estimates for the same sources obtained in
867: earlier studies are listed in Column 4 where available. The table
868: shows that the redshifts found here are consistent with the earlier
869: results except for the arc pair (A1, A2), which is suggested to have
870: a lensing redshift $z\approx1.4\pm0.2$ (B\'{e}zecourt et al 1999).
871: According to our photometric data in 12 bands, we find that A1 and
872: A2 are very blue (cf.~Fig.~\ref{a370color}) and are not detected by
873: IRAC. In qualitative agreement with our lensing-based estimate, this
874: suggests (A1, A2) are at lower redshift.
875:
876: \begin{table}
877: \caption{Redshift estimates $z_\mathrm{l}$ (column 3) for arcs in
878: A~370 (column 1) obtained from distance factors (DF, column 2) in
879: the best-fitting lens model, compared to spectroscopic, photometric
880: or lensing-estimated redshifts, $z_\mathrm{s}$, $z_\mathrm{p}$ or
881: $z_{\rm l}$ (column 4), where available from the literature (column
882: 5).}
883: \begin{tabular}{llllc}
884: \hline arc & DF & $z_\mathrm{l}$ & $z_\mathrm{s}$, $z_\mathrm{p}$ or
885: $z_{\rm l}$ &
886: ref. \\
887: \hline \hline
888: A0 & 1.00 & -- & 0.724 (s) & [1] \\
889: R & 1.46$\pm$0.10 & 1.31$\pm$0.12 & 1.312 (p) & [2] \\
890: & & & 1.7$\pm$0.2 (l) & [3]\\
891: B2,3 & 1.10$\pm$0.09 & 0.80$\pm$0.09 & 0.806 (s) & [3] \\
892: A1,2 & 1.15$\pm$0.11 & 0.85$\pm$0.11 & 1.4$\pm$0.2 (l) & [3] \\
893: & & & $>$ 1.2 (p) & [4]\\
894: A3,4 & 1.30$\pm$0.12 & 1.02$\pm$0.11 & $>$ 0.2 (p) & [4] \\
895: A5 & 1.46$\pm$0.06 & 1.30$\pm$0.12 & 1.306 (s) & [5] \\
896: \hline \hline
897: B4 & 1.11 & 0.80 & -- & -- \\
898: C1,2,4 & 1.05$\pm$0.13 & 0.76$\pm$0.12 & 0.75$\pm$0.1 (l) & [3]\\
899: & & & 0.60 - 0.92 (p)& [4]\\
900: D1,2 & 1.10$\pm$0.11 & 0.80$\pm$0.11 & 0.85$\pm$0.1 (l) & [3] \\
901: & & & $<$ 0.4 (p) & [4]\\
902: E1,2,3 & 1.30$\pm$0.11 & 1.02$\pm$0.18 & 1.3$\pm$0.1 (l) & [3] \\
903: \hline \hline
904: Z1 & 1.70$\pm$0.05 & 2.28$\pm$0.29 & -- & -- \\
905: Z2 & 0.95$\pm$0.14 & 0.69$\pm$0.10 & -- & -- \\
906: Z3 & 1.60$\pm$0.06 & 1.73$\pm$0.21 & -- & -- \\
907: Z4 & 0.95$\pm$0.15 & 0.69$\pm$0.11 & -- & -- \\
908: Z5 & 0.95$\pm$0.14 & 0.69$\pm$0.10 & -- & -- \\
909: Z6 & 1.00$\pm$0.11 & 0.72$\pm$0.09 & -- & -- \\
910: Z7 & 1.34$\pm$0.08 & 1.08$\pm$0.11 & -- & -- \\
911: Z8 & 1.34$\pm$0.08 & 1.08$\pm$0.11 & -- & -- \\
912: W2 & 1.34$\pm$0.08 & 1.08$\pm$0.11 & -- & --\\
913: \hline
914: \end{tabular}
915:
916: {DF: distance factor (Comerford et al 2006); $z_{\rm l}$: lens
917: redshift; $z_{\rm s}$: spectroscopic redshift; $z_{\rm p}$:
918: photometric redshift; (s): spectroscopy; (p): photometry; (l): lens. }
919:
920: {Refs: [1] Soucail et al (1988); [2] Smail et al (1996); [3]
921: B\'{e}zecourt et al (1999); [4] Kneib et al (1994); [5] Mellier et
922: al (1991)} \label{tab_z}
923: \end{table}
924:
925: \subsection{Further study of the arclets in A~370}
926:
927: The high-quality and deep images in eight bands from $B$ to $K'$
928: taken by Subaru enable us to apply the best-fitting NFW lens model
929: obtained in the preceding subsection to some arclets that were
930: already found and labeled by Kneib et al (1993) and B\'{e}zecourt et
931: al (1999), and to some new arclet candidates (labelled ``Z") in
932: order to test whether the best-fitting model is consistent with them
933: (see Fig.~\ref{a370color}).
934:
935: We thus apply the best-fitting NFW lens model obtained in the
936: previous subsection to the arclet pair candidates labeled B, C, D
937: and E by first optimising the model under the assumption of minimal
938: source sizes, only using their redshifts as free parameters. The
939: results of this fit for individual arclet pairs are shown in
940: Fig.~\ref{a370 bcez}, and their estimated redshifts are listed in
941: the middle part of Table~\ref{tab_z}.
942:
943: \begin{figure}
944: \includegraphics[width=\hsize]{f14}
945: \caption{The best-fitting results for the arclet pairs B, C, D, E in
946: A~370.} \label{a370 bcez}
947: \end{figure}
948:
949: Figure~\ref{a370 bcez} shows that the best-fitting NFW model can
950: fairly reproduce the observed images of the arclet pairs B, C, and
951: D. Note that the additional image produced by (B2, B3) in
952: Fig.~\ref{a370 9arc} matches well their possible counter image B4.
953: Their estimated redshifts are consistent with previous photometric
954: measurements. Small differences of the image positions between the
955: model predictions and the observations for E3 in the right panel of
956: the figure may arise because we ignore perturbations due to lensing
957: by less luminous galaxies within A~370. Such differences in the
958: image positions measured and reproduced by the model also occur for
959: C4. For the pair (D1, D2), the model predicts an extra counter-image
960: very close to D1 that is not clearly observed, possibly because the
961: image was yet too faint to be seen.
962:
963: \begin{figure}
964: \includegraphics[width=\hsize]{f15}
965: \caption{The best-fitting results for the new arclet candidates in
966: A~370.} \label{a370_more_arcs}
967: \end{figure}
968:
969: In addition to the arclet candidates labeled by Kneib et al (1993)
970: and B\'{e}zecourt et al (1999), we identify new arclet candidates in
971: the field of A~370 labeled ``Z1'' to ``Z8'' and shown in
972: Fig.~\ref{a370color}, with the measured ellipticities of individual
973: objects larger than 0.6 in $R$ band by SExtractor (Bertin \&
974: Arnouts 1996) and checking by eye in the other bands. Similar to B,
975: C, D and E, the best-fitting NFW model is also applied to them,
976: taking their redshifts as free parameters. Best-fitting results for
977: them are shown in Fig.~\ref{a370_more_arcs} with the expected
978: lensing redshifts for individual new arclets being listed in the
979: final part of Table~\ref{tab_z}. A detailed study of their SEDs will
980: be presented in a separate paper.
981:
982: Based on these results for the new arclets, we suggest that Z6 may
983: be a lensed galaxy at low redshift. Z2, Z4 and Z5 could be
984: counter-images because their recovered sources are close to each
985: other in the sky, and they have similar lensing redshifts. Although
986: the predicted separations of the source positions between Z4 and Z2,
987: and between Z4 and Z5 are possibly too large, this may be due to
988: model uncertainties. Moreover, two candidates of distant objects, Z1
989: and Z3, are found with lensing redshifts of $\sim2.3$ and
990: $\sim1.73$, respectively, which need to be spectroscopically or
991: photometrically confirmed.
992:
993: Finally, the new candidates Z7 and Z8 are suggested to be
994: counter-images of a source at redshift near 1.3. Based on the
995: best-fitting NFW model, we also found their predicted counter image
996: close to B2, labeled W2 in Figs.~\ref{a370_more_arcs} and
997: \ref{a370color}. Photometry confirms that it has the same colour as
998: Z7 and Z8. We note that this newly found arclet pair had already
999: been detected by Bernard Fort and collaborators several years ago,
1000: but remained unpublished (B.~Fort, private communication). Like (B2,
1001: B3), these counter images are also very sensitive to the mass model
1002: of A~370. Since they are extremely red, the study of their SEDs in
1003: the next paper suggests that they arise from an active star-forming
1004: galaxy heavily enshrouded by dust.
1005:
1006: \section{Conclusions}
1007:
1008: Since it is virtually impossible to distinguish mass models with
1009: different density profiles for galaxy clusters based exclusively on
1010: the observed arc and arclet positions, which depend on the first
1011: derivative of the effective lensing potential, we have studied in
1012: detail the image magnifications expected from lens models assembled
1013: from components with isothermal and NFW density profiles. We
1014: considered both axially-symmetric and asymmetric models and conclude
1015: that the magnifications for different mass models can be
1016: substantially different. The ellipticity of a mass model must be
1017: taken into account because it plays an indispensable role in
1018: strong-lensing studies.
1019:
1020: Given a lens model reproducing the position of a giant arc, its
1021: predicted magnification is usually smaller for singular than for
1022: non-singular isothermal profiles. Isothermal models with small cores
1023: predict magnifications smaller than models with NFW mass profiles.
1024: Isothermal mass models with large cores, whose density gradient in
1025: the inner region is nearly flat, predict magnifications increasing
1026: with the core radius and exceeding those for NFW density profiles.
1027: These effects are more pronounced for increasingly asymmetric
1028: models.
1029:
1030: We illustrate our results with the two well-studied clusters MS~2137
1031: and A~370, which we model using the code described by Comerford et
1032: al (2006). Comparing with previous studies, the main improvements in
1033: this paper are based on (1) the photometry for obtaining relative
1034: magnifications of individual counter-image pairs to constrain the
1035: lens models; and (2) further studies of additional arclets in A~370
1036: including lens redshifts of newly found arclets.
1037:
1038: The observational data were taken with Subaru, HST and Spitzer for
1039: A~370 and taken from Gavazzi et al (2003) for MS~2137. We find that
1040: both isothermal and NFW models can reproduce the observed arcs and
1041: arclet positions in MS~2137 and A~370, which is consistent with the
1042: results of Comerford et al (2006), Gavazzi (2005), Gavazzi et al
1043: (2003), Miralda-Escud\'e (2002, 1995), B\'{e}zecourt et al (1999),
1044: Mellier, Fort \& Kneib (1993), and Kneib et al (1993).
1045:
1046: Relative magnifications of individual arc and arclet pairs are
1047: adopted to adjust the mass model by comparing with observations.
1048: Since MS~2137 is a relatively simple, single-component system, it is
1049: hard to distinguish isothermal from NFW mass models. On the other
1050: hand, A~370 has two major lens components with several arc and
1051: arclet pairs, which allow us to adopt the relative magnifications of
1052: the confirmed arclet pairs to conclude that the NFW profile is more
1053: realistic than the isothermal profile for A~370.
1054:
1055: The best-fitting NFW mass model for A~370 is applied to several
1056: arclet candidates found by Kneib et al (1993) and B\'{e}zecourt et
1057: al (1999) that were not used for constraining the mass model. We
1058: find that this best-fitting model can well reproduce the observed
1059: arclets.
1060:
1061: The high-quality images of A~370 in twelve bands from $B$ to
1062: infrared will allow us to recover SEDs for individual sources in the
1063: field of A~370 in a forthcoming paper. We also find eight new arclet
1064: candidates, apply our best-fitting NFW lens model to them and show
1065: that we can reproduce their observed features fairly well. We obtain
1066: redshift estimates from lensing for individual arcs and arclets in
1067: A~370 that are consistent with spectroscopic and photometric results
1068: where available from earlier studies. Other redshift predictions
1069: from lensing need to be confirmed by future observations. Finally,
1070: we found a new arclet pair labeled Z7, Z8 and W2 and suggest that
1071: they arise from an active star-forming galaxy heavily enshrouded by
1072: dust.
1073:
1074: \acknowledgements
1075:
1076: We thank Raphael Gavazzi for very helpful discussions. CS and BZ
1077: acknowledge the financial support of the \emph{Deutsche
1078: Forschungsgemeinschaft} for a visit to Heidelberg University and are
1079: grateful to the Institute for Theoretical Astrophysics for its
1080: hospitality during the visit. This project is partly supported by
1081: the Chinese National Science Foundation No. 10333020, 10528307,
1082: 10778725, 973 Program No. 2007CB815402, Shanghai Science Foundations
1083: and Shanghai Municipal Education Commission No.05DZ09. JMC
1084: acknowledges support by a National Science Foundation Graduate
1085: Research Fellowship.
1086:
1087: %\clearpage
1088:
1089: \begin{thebibliography}{}
1090: \bibitem{} Bartelmann, M., Meneghetti, M., 2004, A\&A, 418, 413
1091:
1092: \bibitem{} Bartelmann, M., Meneghetti, M., Perrotta, F., Baccigalupi, C.,
1093: Moscardini, L., 2003, A\&A, 409, 449
1094:
1095: \bibitem{} Bartelmann, M., Huss, A., Colberg, J. M., Jenkins, A., Pearce, F.
1096: R., 1998, A\&A, 330, 1
1097:
1098: \bibitem{} Bartelmann, M., 1996, A\&A, 313, 697
1099:
1100: \bibitem{} Bertin, E., Arnouts, S., 1996, A\&A,
1101:
1102: \bibitem{} B\'{e}zecourt, J., Kneib, J.-P., Soucail, G., Ebbels, T. M. D., 1999, A\&A,
1103: 347, 21
1104:
1105: \bibitem{} Bolton, A. S., Burles, S., Koopmans, L. V. E., Treu, T., Moustakas,
1106: L., A., 2006, ApJ, 638, 703
1107:
1108: \bibitem{} Broadhurst, T., Ben\'{i}tez, N., Coe, D., Sharon, K., Zekser, K., White, R., Ford, H., et al, 2005,
1109: ApJ, 621, 53
1110:
1111: \bibitem{} Chary, R. -R., Stern, D., Eisenhardt, P., 2005, ApJ, 635, L5
1112:
1113: \bibitem{} Comerford, J. M., Meneghetti, M., Bartelmann, M., Schirmer, M.,
1114: 2006, ApJ, 642, 39
1115:
1116: \bibitem{} Covone, G., Kneib, J. -P., Soucail, G., Richard, J., Jullo, E.,
1117: Ebeling, H, 2006, A\&A, 456, 409
1118:
1119: \bibitem{} Cowie, L. L., et al, 2008 in preparation
1120:
1121: \bibitem{} Dalal, N., Holder, G., 2004, ApJ, 609, 50
1122:
1123: \bibitem{} Ellis, R., Santos, M. R., Kneib, J. -P., Kuijken, K., 2001, ApJ,
1124: 560, L119
1125:
1126: \bibitem{} Fedeli, C., Bartelmann, M., Meneghetti, M., Moscardini, L., 2008, A\&A submitted, preprint astro-ph/arXiv:0803.0656
1127:
1128: \bibitem{} Frye, B., L., Coe, D., Bowen, D. V., Ben\'{i}tez, N., Broadhurst, T., Guhathakurta, P., Illingworth, G., et al, 2007,
1129: ApJ, 665, 921
1130:
1131: \bibitem{} Gavazzi, R., 2005, A\&A, 443, 793
1132:
1133: \bibitem{} Gavazzi, R., Fort, B., Mellier, Y., Pello, R., Dantel-Fort, M.,
1134: 2003, A\&A, 403, 11
1135:
1136: \bibitem{} Grossman, S., A., Narayan, R., 1988, ApJ, 324, L37
1137:
1138: \bibitem{} Halkola, A., Seitz, S., Pannella, M., 2006, MNRAS, 372,
1139: 1425
1140:
1141: \bibitem{} Hamana, T., Bartelmann, M., Yoshida, N., Pfrommer, C., 2005,
1142: MNRAS, 357, 1407
1143:
1144: \bibitem{} Hammer, F., 1991, ApJ, 383, 66
1145:
1146: \bibitem{} Hu, E. M., Cowie, L. L., McMahon, R. G., Capak, P., et
1147: al, 2002, ApJ, 568, 75
1148:
1149: \bibitem{} Kneib, J. -P., Ellis, R. S., Santos, M. R., Richard, J., 2004,
1150: ApJ, 607, 697
1151:
1152: \bibitem{} Kneib, J. -P., Hudelot, P., Ellis, R., Treu, T., Smith, G. P., Marshall, P., et al, 2003, ApJ, 598, 804
1153:
1154: \bibitem{} kneib, J. -P., Mathez, G., Fort, B., Mellier, Y., Soucail, G., Longaretti,
1155: P.-Y., 1994, A\&A, 286, 701
1156:
1157: \bibitem{} Kneib, J.P., Mellier, Y., Fort, B., Mathez, G. 1993, A\&A, 273, 367
1158:
1159: \bibitem{} Kovner, I., 1989, ApJ, 337, 621
1160:
1161: \bibitem{} Leonard, A., Goldberg, D. M., Haaga, J. L., Massey, R.,
1162: 2007, ApJ, 666, L51
1163:
1164: \bibitem{} Li, G. -L., Mao, S., Jing, Y. P., Bartelmann, M., Kang, X.,
1165: Meneghetti, M., 2005, ApJ, 635, 795
1166:
1167: \bibitem{} Limousin, M., Richard, J., Jullo, E., Kneib, J. -P., Fort, B.,
1168: Soucail, G., et al, 2006, preprint(astro-ph/0612165)
1169:
1170: \bibitem{} Mellier, Y., Fort, B., Kneib, J. -P., 1993, ApJ, 407, 33
1171:
1172: \bibitem{} Mellier, Y., Fort, B., Soucail, G., Mathez, G., Cailloux, M., 1991,
1173: ApJ, 380, 334
1174:
1175: \bibitem{} Meneghetti, M., Jain, B., Bartelmann, M., Dolag, K., 2005a, MNRAS,
1176: 362, 1301
1177:
1178: \bibitem{} Meneghetti, M., Bartelmann, M., Dolag, K., Moscardini, L.,
1179: Perrotta, F., Baccigalupi, C., Tormen, G., 2005b, A\&A, 442, 413
1180:
1181: \bibitem{} Miradal-Escud$\acute{e}$, J., 1995, ApJ, 438, 514
1182:
1183: \bibitem{} Miradal-Escud\'e, J., 2002, ApJ, 564, 60
1184:
1185: \bibitem{} Navarro, J. F., Frenk, C. S., White, S. D. M., 1996, ApJ, 462, 563
1186: (NFW)
1187:
1188: \bibitem{} Navarro, J. F., Frenk, C. S., White, S. D. M., 1997, ApJ, 490, 493
1189: (NFW)
1190:
1191: \bibitem{} Ofek, E. O., Maoz, Dan., Rix, Hans-Walter, Kochanek, C. S.; Falco,
1192: E. E., 2006, ApJ, 641, 70
1193:
1194: \bibitem{} Puchwein, E., Bartelmann, M., Dolag, K., Meneghetti, M., 2005, A\&A,
1195: 442, 405
1196:
1197: \bibitem{} Schimd, C., Tereno, I., Uzan, J. -P., Mellier, Y., van Waerbeke,
1198: L., Semboloni, E., Hoekstra, H., Fu, L., Riazuelo, A., 2006,
1199: preprint(astro-ph/0603158)
1200:
1201: \bibitem{} Smith, G. P., Kneib, J. -P., Smail, I., Mazzotta, P., Ebeling, H.,
1202: Czoske, O., 2005, MNRAS, 359, 417
1203:
1204: \bibitem{} Smail, I., Dressler, A., Kneib, J. -P., Ellis, R., Couch, W. J., Sharples, R. M., Oemler, A.
1205: Jr., 1996, ApJ, 469, 508
1206:
1207: \bibitem{} Soucail, G., Fort, B., Mellier, Y., Picat, J. P., 1987, A\&A, 187,
1208: L1
1209: \bibitem{} Soucail, G., Mellier, Y., Fort, B., Cailloux, M., 1988
1210: A\&AS, 73, 471
1211:
1212: \bibitem{} Tu, H., Linmonsin, M., Fort, B., Shu, C., et al, 2008,
1213: MNRAS accepted (astro-ph/0710.2246)
1214:
1215: \bibitem{} Wambsganss, J., Bode, P., Ostriker, J.P. 2004, ApJ, 606, L93
1216:
1217: \bibitem{} Zekser, K. C., White, R. L., Broadhurst, T. J., Ben\'{i}tez, N., Ford, H. C., Illingworth, G. D., et al., 2006, ApJ, 640, 639
1218:
1219: \end{thebibliography}
1220:
1221: \clearpage
1222:
1223: \end{document}
1224: