1: \documentclass[]{emulateapj}
2: \submitted{Accepted for Publication in ApJ}
3:
4:
5: \newcommand{\Ha }{H$\alpha$}
6: \newcommand{\HI}{H{\sc i}}
7: \newcommand{\Dpak } {DensePak}
8: \newcommand{\kms} {km s$^{-1}$}
9:
10:
11: \shorttitle{LSB Galaxy Constraints on the NFW Potential}
12: \shortauthors{Kuzio de Naray, McGaugh, \& Mihos}
13:
14: \begin{document}
15: \title{Constraining the NFW Potential with Observations and Modeling of LSB Galaxy Velocity Fields}
16: \author{Rachel Kuzio de Naray\altaffilmark{1}}
17: \affil{Center for Cosmology, Department of Physics and Astronomy, University of California,
18: Irvine, CA 92697-4575}
19: \altaffiltext{1}{NSF Astronomy and Astrophysics Postdoctoral Fellow}
20: \email{kuzio@uci.edu}
21: \author{Stacy S. McGaugh}
22: \affil{Department of Astronomy, University of Maryland, College Park,
23: MD 20742-2421}
24: \email{ssm@astro.umd.edu}
25: \and
26: \author{J.~Christopher Mihos}
27: \affil{Department of Astronomy, Case Western Reserve University,
28: 10900 Euclid Avenue, Cleveland, OH 44106}
29: \email{mihos@case.edu}
30:
31:
32: \begin{abstract}
33: We model the NFW potential to determine if, and under what conditions, the
34: NFW halo appears consistent with the observed velocity fields of low surface
35: brightness (LSB) galaxies. We present mock \Dpak\ IFU velocity fields
36: and rotation curves of axisymmetric and non-axisymmetric potentials that are
37: well-matched to the spatial resolution and velocity range of our sample
38: galaxies. We find that the \Dpak\ IFU can accurately reconstruct the velocity
39: field produced by an axisymmetric NFW potential and that a tilted-ring fitting
40: program can successfully recover the corresponding NFW rotation curve. We
41: also find that non-axisymmetric potentials with fixed axis ratios change only
42: the normalization of the mock velocity fields and rotation curves and not their
43: shape. The shape
44: of the modeled NFW rotation curves does not reproduce the data: these
45: potentials are unable to simultaneously bring the mock data at both small and
46: large radii into agreement with observations. Indeed, to match the slow rise
47: of LSB galaxy rotation curves, a specific viewing angle of the
48: non-axisymmetric potential is required. For each of the simulated LSB
49: galaxies, the observer's line-of-sight must be along the minor axis of the
50: potential, an arrangement which is inconsistent with a random distribution of
51: halo orientations on the sky.
52: \end{abstract}
53:
54:
55: \keywords{dark matter --- galaxies: kinematics and dynamics}
56:
57:
58: \section{Introduction}
59: Cosmologically motivated numerical simulations of cold dark matter (CDM)
60: describe very specifically the properties of the dark matter halos that
61: should be observed in the universe. The simulations show that CDM halos
62: are cuspy, meaning that the density of the halo, regardless of its mass,
63: rises very steeply toward the center
64: \citep[e.g.][]{Dubinski94,NFW96,NFW97,Moore,Reed,Diemand}. The simulations
65: also dictate the range of permissible values of halo parameters based on the
66: assumed cosmology of the simulations. The concentration $c$ of a halo, for
67: example, depends on the density of the universe at the time the halo forms,
68: which in turn depends on the adopted values of $h$, $\Omega_{m}$,
69: $\sigma_{8}$, etc. \citep{NFW96,NFW97}. Due to this intimate connection to
70: cosmology, the values of halo parameters are not arbitrary.
71:
72: The most well-known description of CDM halo behavior is the cuspy NFW halo
73: where $\rho \sim r^{-1}$ \citep{NFW96,NFW97}. The rotation curves of these
74: halos are parameterized by two numbers: the concentration, $c$, and a
75: characteristic velocity, $V_{200}$. These two parameters cannot freely vary,
76: nor can they vary independently of the other; there is a correlation between
77: $c$ and $V_{200}$ \citep[e.g.][]{NFW97,Jing,Bullock01,Wechsler}. This
78: $c-V_{200}$ relation, combined with the cosmological constraints on $c$,
79: means that the expected rotation curve for a CDM halo of a given mass is
80: well-determined.
81:
82: Though the need for dark matter in disk galaxies has long been indicated by
83: flat rotation curves \citep[e.g.][]{Rubin,Bosma}, it has been less obvious
84: that the dark matter halos are consistent with cuspy CDM halos. Because low
85: surface brightness (LSB) galaxies are thought to be dark matter-dominated
86: down to small radii (de Blok \& McGaugh 1996; de Blok \& McGaugh 1997;
87: Borriello \& Salucci 2001, but see Fuchs 2003), their kinematics have been
88: used as probes of the density distribution of galaxy mass dark matter halos.
89: Rotation curves derived from \HI\ velocity fields and long-slit \Ha\
90: observations are frequently consistent with halos having a cored
91: $\rho \sim r^{0}$ density distribution
92: \citep[e.g.][]{Flores,DMV,dBB,Marchesini,Cote} rather than the steeper profile
93: of the NFW halo. This result has also been supported by rotation curves
94: derived from high-resolution two-dimensional velocity fields obtained with
95: integral field spectrographs
96: \citep[e.g.][]{Chemin04,Gentile05,Simon05,K06,K08}.
97:
98: In \citet[][hereafter K06 and K08, respectively]{K06,K08}, we presented
99: \Dpak\ Integral Field Unit (IFU) \Ha\ velocity fields, rotation curves, and
100: halo fits for a sample of 17 LSB galaxies. We fit both a cored
101: pseudoisothermal halo ($\rho \sim r^{0}$) and a cuspy NFW halo
102: ($\rho \sim r^{-1}$) to the data and found the halo central densities and
103: rotation curve shapes to be better described by the cored halo model.
104: The NFW fits to the \Dpak\ rotation curves were often found to have
105: concentrations lower than what is expected for galaxies in a $\Lambda$CDM
106: cosmology (see also Gentile et al.~2007, but see Swaters et al.~2003b for a
107: different conclusion) and to favor a power spectrum having a lower amplitude
108: on small scales (Zentner \& Bullock 2002, McGaugh et al.~2003, K08). We
109: found that the NFW rotation curves specified by the $c-V_{200}$ relation
110: (the rotation curves that our galaxies \textit{should} have according to
111: $\Lambda$CDM) are much more steeply rising than the
112: observed \Dpak\ rotation curves. In addition, these cosmologically
113: consistent halos show a cusp mass excess at the centers of the galaxies,
114: indicating that at least two times more mass is expected in the cuspy CDM
115: halos than is allowed by the data.
116:
117: \defcitealias{K06}{K06}
118: \defcitealias{K08}{K08}
119:
120: CDM halos must be both cuspy and follow the $c-V_{200}$ relation defined by
121: $\Lambda$CDM. The density profiles of the \citetalias{K06} and
122: \citetalias{K08} data are not well-described by cuspy halos, nor do the
123: galaxies fall on the $c-V_{200}$ relation. These \Dpak\ results are
124: consistent with many previous long-slit and \HI\ studies of LSB galaxies
125: \citep[e.g.][]{dBMR,Bolatto02,dBB,Swaters03a}, as well as similar \Dpak\
126: studies by \citet{Simon05}. That different observational techniques (with
127: different data reduction and analysis procedures as well as sources of error)
128: lead to similar conclusions suggests that perhaps the discrepancy
129: between the NFW halo and the observations does not arise at the telescope or
130: during data analysis, but rather is due to an incorrect assumption
131: about the specific form of the NFW halo potential.
132:
133: Our goal in this paper is to model the NFW halo to determine if, and under
134: what conditions, it appears consistent with the observed data. Starting
135: with an axisymmetric NFW potential, how must it be modified (e.g.,
136: introduction of an asymmetry) in order to appear consistent with both the
137: observed two-dimensional velocity field and the derived rotation curve?
138: We construct a model disk galaxy embedded in an NFW halo and then ``observe''
139: it in the same way as we have observed our sample of galaxies with \Dpak.
140: We then compare the mock velocity field and the derived mock rotation curve
141: to the real galaxy data.
142:
143: We adopt a numerical approach to investigating non-axisymmetric halo
144: potentials because once axisymmetry is broken, the data analysis
145: becomes much more complicated. Noncircular motions and asymmetries
146: are traditionally investigated by doing a higher-order Fourier decomposition
147: of the velocity field \citep[e.g.][]{Schoenmakers,Wong}. We have tried this
148: approach, but it was not sufficiently well-constrained for these difficult
149: LSB targets to give unique results. But we do find that useful constraints
150: can still be extracted by simulating what is expected to be observed for
151: various hypothesized halo potentials.
152:
153: The paper is organized as follows. In $\S$ 2 we describe the simulations.
154: The axisymmetric NFW potential is explored in $\S$ 3. In $\S$ 4 we describe
155: the mock velocity fields and rotation curves produced by a non-axisymmetric
156: NFW potential with a constant axis ratio. We determine in $\S$ 5 the
157: non-axisymmetric potentials that best describe the observed galaxy data.
158: We discuss our results and conclusions in $\S$ 6.
159:
160: \begin{deluxetable}{lcccc}
161: \tabletypesize{\footnotesize}
162: \tablecaption{Simulated NFW Halo Parameters}
163: \tablewidth{0pt}
164: \tablehead{
165: \colhead{} &\colhead{} &\colhead{} &\colhead{$V_{200}$} &\colhead{$R_{excess}$}\\
166: \colhead{Galaxy} &\colhead{} &\colhead{$c$}&\colhead{\kms} &\colhead{$\arcsec$}\\
167: \colhead{(1)} &\colhead{} &\colhead{(2)} &\colhead{(3)} &\colhead{(4)}}
168: \startdata
169: NGC 4395 & &8.6 &87 &59\\
170: DDO 64 & &9.2 &62 &40\\
171: UGC 4325 & &6.9 &249 &40\\
172: F583-1 & &8.7 &83 &37\\
173: F563-1 & &8.4 &101 &13\\
174: F583-4 & &9.1 &67 &22\\
175: UGC 5750 & &9.1 &67 &31\\
176: F563-V2 & &7.9 & 130 &22\\
177: F568-3 & &8.2 &110 &13\\
178: \enddata
179: \tablecomments{Columns 2 and 3 list the NFW halo parameters for the simulated galaxies. Listed in column 4 are the radii at which the observed \Dpak\ rotation curves and input NFW$_{constr}$ rotation curves overlap in the minimum disk case. These are the radii out to which the observed and mock rotation curves are compared.}
180: \end{deluxetable}
181:
182:
183:
184: \section{Description of Simulations}
185: $N$-body simulations show that CDM halos of all masses can be described
186: by the NFW potential \citep{NFW96, NFW97} and its variants
187: \citep[e.g.,][]{Diemand, Moore, Navarro2004, Reed}. These cuspy halo
188: potentials show a steep rise in the mass density toward the center of
189: the halo. Most theoretical estimates of the inner slope of the halo
190: mass density profile are as steep or steeper than that of NFW, so we
191: choose to simulate the NFW halo as the conservative case. If the
192: NFW potential predicts a dark matter halo with a steeper density profile
193: than is allowed by the observed galaxy data, then even more steeply rising
194: potentials are automatically excluded. In some formulations
195: \citep[e.g.,][]{Navarro2004}, there is no well-defined inner slope,
196: which continues to roll over to a value that asymptotes to a flatter value
197: than NFW. However, this is a small effect at small radii. The difference
198: between the original NFW profile and that of \citet{Navarro2004} is
199: too small to be detected observationally.
200:
201: In \citetalias{K06} and \citetalias{K08} we defined a constrained NFW halo,
202: NFW$_{constr}$. We required the halo to match the velocities at the outer
203: radii of each galaxy by choosing a value of $V_{200}$ which forced the NFW
204: velocities to agree with the data points at large radii with the minimum
205: requirement of falling within the errorbars of the data. We then used the
206: $c-V_{200}$ relation \citep{NFW97,dBBM} to determine the corresponding
207: cosmologically-consistent concentration. This is adjusted to the `vanilla'
208: cosmology of \citet{Tegmark} by subtracting 0.011 dex in concentration
209: \citep[see][]{McGaugh03}. According to $\Lambda$CDM, these
210: are the rotation curves that our galaxies should have.
211: The chief remaining uncertainty in the normalization of the $c-V_{200}$
212: relation is the power spectrum. Rotation curves data prefer lower
213: $\sigma_8$ \citep{McGaugh07}.
214:
215: \begin{figure*}[!ht]
216: \epsscale{0.65}
217: \plotone{f1_lowres.eps}
218: \caption{\textit{(a)} Observed \Dpak\ velocity field of UGC~4325. \textit{(b)} Mock \Dpak\ velocity field of the axisymmetric simulation. Both velocity fields have isovelocity contours at 10 \kms\ intervals. \textit{(c)} Residual velocity field showing the differences between the UGC~4325 data and the velocity field of the idealized (i.e.~no velocity dispersion), axisymmetric NFW$_{constr}$ halo. \textit{(d)} Same as \textit{(c)} but for the mock \Dpak\ velocity field. The residuals are large and obvious in \textit{(c)}. \textit{Bottom:} Observed and mock rotation curves. The solid points are the observed \Dpak\ rotation curve of UGC~4325, the solid line is the NFW rotation curve corresponding to the input NFW potential, and the open (red) circles are the rotation curve recovered from the mock velocity field. The arrow indicates the radius out to which the rotation curves are compared. The last three points of the recovered mock rotation curve are high because of a lack of fibers at large radii.}
219: \end{figure*}
220:
221: \begin{figure*}
222: \epsscale{0.72}
223: \plotone{f2_lowres.eps}
224: \caption{Same as Figure 1, but for the less well-resolved galaxy F583-4.}
225: \end{figure*}
226:
227:
228: Our goal is to compare mock \Dpak\ velocity fields and rotation curves of
229: NFW$_{constr}$ halos to the observed \Dpak\ velocity fields and rotation
230: curves of the LSB galaxies in \citetalias{K06} and \citetalias{K08}. We
231: model those galaxies that have well-sampled velocity fields and rotation
232: curves that are constrained at large radii by previous long-slit and/or \HI\
233: rotation curves, allowing NFW$_{constr}$ halo fits to be made. Of our 17
234: galaxies, 9 meet these criteria. The spatial resolution and \Dpak\ coverage
235: of these data vary. The parameters of the NFW$_{constr}$ halos in the limit
236: of minimum disk \citepalias{K06,K08} for each of the 9 modeled galaxies are
237: listed in Table 1. The galaxies are listed in order of decreasing spatial
238: resolution, from NGC~4395 ($\sim$ 20 pc/$\arcsec$) to F568-3
239: ($\sim$ 375 pc/$\arcsec$).
240:
241: We developed a code which does a fourth-order Runge-Kutta (RK4) test particle
242: integration of point masses moving in a two-dimensional rigid analytic NFW potential.
243: Specifically, the potential used is:
244: \begin{equation}
245: \Phi(R) = -\frac{GM_{200}\ln(1 + \frac{R}{R_{s}})}{Rf(c)},
246: \end{equation}
247: where
248: \begin{equation}
249: R = \sqrt{x^{2} + (y^{2}/q^{2})},
250: \end{equation}
251: \begin{equation}
252: R_{s} = \frac{R_{200}}{c},
253: \end{equation}
254: and
255: \begin{equation}
256: f(c) = \ln(1 + c) - \frac{c}{1 + c}.
257: \end{equation}
258: In these equations, $M_{200}$ is the enclosed halo mass at radius
259: $R_{200}$, $q$ is the axis ratio ($q=y/x$), and $c$ is the concentration
260: of the halo ($c=R_{200}/R_{s}$). The halo parameters are set to those of
261: the NFW$_{constr}$ halo determined for each galaxy in \citetalias{K06} and
262: \citetalias{K08} and are listed in Table 1. Each simulated galaxy is an
263: infinitely thin exponential disk of 10,000 test particles. For the nearby
264: galaxies UGC~4325 and DDO~64, the number of test particles was increased to
265: 100,000 to ensure sufficient sampling of particles to recreate the higher
266: resolution data. The two-dimensional disk is in the plane of the potential.
267: To wash out any numerical pattern noise of the initial conditions, we
268: integrate for 50 half-mass rotation periods having 500 timesteps each.
269: Each simulated galaxy is given the disk scale length, spatial resolution,
270: and inclination of the real galaxy.
271:
272: Each simulated galaxy is then ``observed'' by \Dpak. \Dpak\ is an integral
273: field spectrograph on the 3.5 m WIYN telescope at the Kitt Peak National
274: Observatory (KPNO). It is a 43$\arcsec\times28\arcsec$ fixed array of
275: 3$\arcsec$ fibers with 3.84$\arcsec$ separations. We model the 85 working
276: fibers, as well as the 5 missing or broken fibers, in the main bundle.
277: For the galaxies observed in \citetalias{K06} and \citetalias{K08}, the fiber
278: bundle orientation on the sky and the total number of pointings per galaxy
279: were tailored to each galaxy so that the critical central regions were
280: covered by the \Dpak\ fibers. We aim for obtaining roughly equivalent
281: coverage of the simulated galaxies by using similar numbers and alignments
282: of \Dpak\ pointings on the simulations. Prior to extracting a rotation curve,
283: these mock \Dpak\ velocity fields are given the velocity dispersion observed
284: in the real \Dpak\ galaxy velocity fields. We have defined the velocity
285: dispersion of the \Dpak\ data to be the fiber-to-fiber velocity variation;
286: to recreate this in the mock data, we randomly add the desired dispersion to
287: the fibers in the mock velocity field. Rotation curves were then derived
288: from the mock \Dpak\ observations by using the NEMO \citep{Teuben} program
289: ROTCUR \citep{Begeman}. ROTCUR treats the observed velocity field as an
290: ensemble of tilted rings and fits for the center, systemic velocity,
291: inclination, position angle, and rotation velocity in each ring. The reader
292: is referred to \citetalias{K06} and \citetalias{K08} for a more extensive
293: explanation of ROTCUR and its application to the \Dpak\ velocity fields.
294:
295:
296:
297: \section{Axisymmetric NFW Halos}
298: The most obvious and simple starting point is to assume an axisymmetric halo
299: potential. The axis ratio $q$ is equal to 1 and the test particles move on
300: circular orbits. With this straightforward potential, we can test whether
301: or not \Dpak\ observations are sufficient to detect the signature of NFW
302: halos in the velocity fields and/or whether the data analysis procedure with
303: ROTCUR also suffices to recover NFW rotation curves.
304:
305: In Figure 1 we model the NFW$_{constr}$ halo of UGC~4325 and ``observe'' the
306: simulation with 5 pointings of the \Dpak\ array. The pointings are arranged
307: to match the spatial coverage of the real galaxy as much as possible.
308: Observed and residual velocity fields for both UGC~4325 and the simulation
309: are shown. The simulated galaxy has the same fiber-to-fiber velocity
310: dispersion as the real galaxy: $\sigma$ = 9.0 \kms. UGC~4325 is one of the
311: most nearby ($D\approx 10$ Mpc) and well-resolved galaxies in our sample.
312: Diffuse \Ha\ emission was abundant in the galaxy and was detected in almost
313: all of the \Dpak\ fibers.
314:
315: Figure 1 demonstrates two important points. First, as
316: evidenced by the very small residuals of the mock \Dpak\ velocity field, the
317: \Dpak\ instrument \textit{is} able to successfully detect an NFW velocity
318: field. The residuals are generally $\lesssim$ 5 \kms\ across the entire
319: observed area. This means that observed \Dpak\ velocity fields are not
320: inconsistent with NFW halos because of an inadequacy of the experimental
321: design or analysis. Second, the observed \Dpak\ velocity field of UGC~4325
322: is not consistent with the axisymmetric NFW$_{constr}$ halo; most of the
323: residuals are $\sim$ 10 \kms, and there is a significant region of
324: $\sim$ 15 \kms\ residuals near the center. For UGC~4325, and the other LSB
325: galaxies in our sample, 15 \kms\ residuals are non-trivial. The observed
326: fiber-to-fiber velocity dispersions are $\sim$ 6-10 \kms\ \citepalias{K06},
327: and since mass scales as $\sigma^{2}$, the implied mass difference is
328: a factor of two or more. In addition, noncircular motions
329: caused by disk instabilities, such as spiral or bar modes, are expected to be
330: small in LSB galaxies. The low surface mass densities of the disks provide
331: little self-gravity to drive such modes, and
332: their high dark matter content provides a higher degree of stabilization
333: than in high surface brightness galaxies \citep{Mihos97}.
334:
335: At the bottom of Figure 1 are the observed and mock rotation curves derived
336: from the velocity fields. The recovered mock rotation curve is consistent
337: with the input rotation curve ($\chi^{2}_{r}=0.93$) out to
338: $R_{excess}\sim 40\arcsec$, where $R_{excess}$ is defined to be the radius at
339: which the observed \Dpak\ rotation curve and the input axisymmetric
340: NFW$_{constr}$ rotation curve begin to overlap. This shows that accurate
341: extraction of the rotation curve of an axisymmetric NFW halo with ROTCUR is
342: \textit{also} possible.
343:
344: In Figure 2 we show similar plots for F583-4. This galaxy has lower spatial
345: resolution ($D\approx 49$ Mpc) than UGC~4325 and only a single pointing of
346: \Dpak\ coverage. The velocity fields, residuals, and rotation curves show
347: that despite the reduced sampling and lower data quality, an axisymmetric
348: NFW halo can still be detected if present. While the differences between
349: the observed and mock velocity field residuals are not as pronounced as in
350: the UGC~4325 case, the input NFW rotation curve is successfully recovered by
351: ROTCUR ($\chi^{2}_{r}=0.32$).
352:
353: The velocity field and rotation curve data and simulations plotted in Figures
354: 1 and 2 together show that the \Dpak\ IFU and the tilted-ring fitting program
355: ROTCUR are able to successfully identify an axisymmetric NFW halo in data of
356: both high and low quality, if one is present. That the \citetalias{K06} and
357: \citetalias{K08} samples of \Dpak\ observations are inconsistent with NFW
358: halos suggests that if the underlying halo potential is NFW, it must not be
359: an axisymmetric NFW potential. This is perhaps not surprising, as CDM
360: simulations suggest that the halo potentials are triaxial
361: \citep*[e.g.][]{Hayashi07}.
362:
363: \begin{figure}
364: \epsscale{1.0}
365: \plotone{f3_color.eps}
366: \caption{Orientation ($\phi$) of the elongated axis of the two-dimensional
367: non-axisymmetric potential with respect to the observer's line-of-sight. In the $\phi=0\degr$ case, the elongated axis of the potential is along the observer's line-of-sight, whereas in the $\phi=90\degr$ case, it is perpendicular to the observer's line-of-sight. [{\it See the electronic edition of the Journal for a color version of this figure.}] }
368: \end{figure}
369:
370: \begin{figure*}
371: \epsscale{0.55}
372: \plotone{f4_lowres.eps}
373: \caption{Mock \Dpak\ velocity fields for the non-axisymmetric NFW simulations of UGC~4325. UGC~4325 is a well-resolved galaxy with multiple pointings of \Dpak\ coverage. Simulations with an axis ratio $q=0.98$ are shown in the left column, and simulations with $q=0.88$ are in the right column. The observer's veiwing angle changes from $\phi=0\degr$ in the top panels to $\phi=90\degr$ in the bottom panels. Note that the effect of viewing angle is much more pronounced in the $q=0.88$ case. For easy comparison, all of the velocity fields are on the same color/velocity scale, and isovelocity contours are drawn at 10 \kms\ intervals. }
374: \end{figure*}
375:
376: \begin{figure*}
377: \epsscale{0.52}
378: \plotone{f5_lowres.eps}
379: \caption{Same as Figure 4, but for the mock \Dpak\ velocity fields for the non-axisymmetric NFW simulations of UGC~5750. UGC~5750 is a more distant galaxy and has only a single pointing of \Dpak\ coverage. }
380: \end{figure*}
381:
382: \begin{figure}
383: \epsscale{0.87}
384: \plotone{f6_color.eps}
385: \caption{A comparison of three mock velocity fields to the observed UGC~4325 data using velocities measured along two slits placed parallel to, and offset from, the minor axis of the velocity fields. The solid (green) line is for the ($q,\phi$) = (0.88,~0$\degr$) mock velocity field, the long-dash (red) line is for the axisymmetric mock velocity field, the short-dash (blue) line is for the ($q,\phi$) = (0.88,~90$\degr$) mock velocity field, and the line+circles are the UGC~4325 data. At the same distance from the center of the velocity field, the ($q,\phi$) = (0.88,~90$\degr$) mock velocity field has velocities most similar to the observed data. A typical errorbar is shown in the lower left corner. [{\it See the electronic edition of the Journal for a color version of this figure.}] }
386: \end{figure}
387:
388:
389:
390: \section{Non-axisymmetric NFW Halos with a Fixed Axis Ratio}
391: We next consider non-axisymmetric two-dimensional NFW potentials with axis
392: ratios $q$~$<$~1 that are constant with radius. These 2D potentials are
393: equivalent to 3D prolate dark matter halos in which the long axis of the halo
394: coincides with the elongated axis of the disk. We simulate halos with axis
395: ratios $q$ = 0.98, 0.96, 0.94, 0.92, 0.90, 0.88, 0.86, and 0.84, similar to
396: the range of non-axisymmetry seen in the CDM simulations of \citet{Hayashi07}.
397: Because axisymmetry has been broken ($q\neq 1$), the test paticles are no
398: longer moving on circular orbits and not all lines of sight in the plane of
399: the disk are equivalent. This means that the observed mock \Dpak\ velocity
400: field and derived rotation curve are affected not only by the value of $q$,
401: but also by the orientation ($\phi$) of the potential's elongation with
402: respect to the observer's line-of-sight (see Figure 3). The potential is
403: elongated along the observer's line-of-sight in the $\phi=0\degr$ case,
404: whereas in the $\phi=90\degr$ case, the potential is elongated perpendicular
405: to the observer's line-of-sight. For 0$\degr<\phi< 90\degr$, the elongation
406: is at an intermediate viewing orientation. For each value of $q$, the
407: orientation of the potential is set to $\phi= 0\degr$, 30$\degr$, 45$\degr$,
408: 60$\degr$, and 90$\degr$.
409:
410:
411: \subsection{``Observed'' Mock \Dpak\ Velocity Fields}
412: In Figures 4 and 5 we show a series of mock \Dpak\ velocity fields for two
413: galaxies representative of the range of data presented in \citetalias{K06}
414: and \citetalias{K08}: UGC~4325 and UGC~5750. As previously mentioned,
415: UGC~4325 has high spatial resolution and extended \Dpak\ coverage. As in the
416: axisymmetric case, 5 pointings of the \Dpak\ array are overlayed on these new
417: non-axisymmetric simulations. In contrast, UGC~5750 is a more distant galaxy
418: ($D\approx 56$ Mpc) and both the real galaxy and the simulations have only
419: one pointing of \Dpak\ coverage.
420:
421: These are simulations of non-axisymmetric NFW halos that obey the cosmic
422: $c-V_{200}$ relation. The virial velocity $V_{200}$ has been chosen to match
423: each galaxy (the NFW$_{constr}$ halos of \citetalias{K06} and
424: \citetalias{K08}). In this section, we explore the effect of introducing a
425: non-axisymmetric potential with equal squashing $q$ at all radii.
426:
427: In the $q=0.98$ simulations, the potential is nearly circular. Throughout
428: their orbits, the particles maintain a roughly constant distance from the
429: center of the potential and as a result, have approximately constant orbital
430: speeds. The viewing angle therefore has little effect on the observed
431: velocity field. The $q=0.98,~\phi=0\degr \rightarrow 90\degr$ mock
432: velocity fields appear very similar, looking not only to be consistent with
433: different realizations of the same underlying potential, but also very much
434: like the mock velocity field of the axisymmetric potential.
435:
436: The same cannot be said for the mock velocity fields of the $q=0.88$
437: simulations. With orbits deviating significantly from circular, a particle's
438: orbital speed depends on its location, making the viewing angle quite
439: important. In the $\phi=0\degr$ orientation, the particles moving along the
440: observer's line-of-sight are traveling along the long axis of the potential
441: and are moving at the maximum orbital speed. These particles are moving
442: faster than particles on circular orbits at the same radius. The
443: minimum-maximum velocity range observed by \Dpak\ is larger than what is
444: observed in the axisymmetric case, and the derived rotation curves will
445: reach higher velocities. The opposite situation is happening in
446: the $\phi=90\degr$ orientation. In this case, the particles moving along the
447: observer's line-of-sight are traveling at the minimum orbital speed and the
448: minimum-maximum velocity range observed by \Dpak\ is smaller than what is
449: observed in the axisymmetric case. The rotation curves derived from these
450: data will therefore be suppressed. The difference between the observed
451: velocity ranges of the $\phi=0\degr$ and $\phi=90\degr$ velocity fields
452: becomes more exaggerated the more noncircular the potential becomes.
453:
454: Some ($q,\phi$) combinations can automatically be excluded as possible
455: descriptions of the observed \Dpak\ galaxy data based simply on the mock
456: velocity fields they produce. The observed \Dpak\ galaxy velocity fields put
457: constraints on the allowable velocity range of the mock velocity fields, as
458: well as the correlation between velocity and position. Regardless of how the
459: mock rotation curves may turn out, if the observed and mock velocity fields do
460: not match, the corresponding simulation is not a viable solution. For
461: example, the ($q,~\phi$) = (0.88,~0$\degr$) mock velocity field of UGC~4325
462: shown in Figure 4 can rule out that particular axis ratio/viewing orientation
463: combination for that galaxy. Overall, the mock velocity field covers a much
464: larger velocity range than the UGC~4325 data, and when the velocities at the
465: same positions in the two velocity fields are compared, they are inconsistent
466: over a large portion of the observed area.
467:
468: Because of the rapidly rising velocities at the centers of NFW halos, the
469: isovelocity contours of NFW velocity fields are pinched in the central regions
470: \citep{dBBM}. This pinch is a distinctive signature of the cuspy NFW halo.
471: If the velocity field is noisy or has high velocity dispersion, the
472: pinch is more difficult to see. We can quantify the pinch by measuring
473: velocities along slits that are offset from, and parallel to, the minor axis
474: of the velocity field. It is in this fashion that we compare the appearance
475: of the UGC~4325 \Dpak\ velocity field and three mock velocity fields. In
476: Figure 6, we have placed a 3$\arcsec \times 28\arcsec$ slit (the width of a
477: \Dpak\ fiber and the width of the \Dpak\ array, respectively) 6$\arcsec$ away
478: from each side of the minor axis (about the separation of two rows of \Dpak\
479: fibers) of the observed \Dpak\ velocity field of UGC~4325 and the axisymmetric
480: NFW mock velocity field (both shown in Figure 1), as well as the
481: ($q,\phi$) = (0.88,~0$\degr$~and~90$\degr$) mock velocity fields in Figure
482: 4. When we plot the average of the measured velocities as a function of
483: position along the slit, we can readily see that of the 3 mock velocity
484: fields, it is the one produced by the ($q,\phi$) = (0.88,~90$\degr$)
485: potential that is most like the data. At the same position in the velocity
486: fields, the axisymmetric, and especially the ($q,\phi$) = (0.88,~0$\degr$),
487: mock observations are detecting velocities much higher than the galaxy data.
488:
489: In the next section we derive mock rotation curves for all the mock \Dpak\
490: velocity fields and examine the effect that the asymmetry of the potential has
491: had on both the normalization and shape of the derived rotation curves.
492:
493: \begin{figure*}[!ht]
494: \epsscale{1.0}
495: \plotone{f7_color.eps}
496: \caption{Mock rotation curves (lines) for the non-axisymmetric NFW simulations of
497: the nearby galaxy UGC~4325 (points). Each panel is for a different value of the axis ratio $q$ (labeled in the lower right). The solid lines are for $\phi=0\degr$, 30$\degr$, 45$\degr$, and 60$\degr$. The dashed line is for $\phi=90\degr$. For all values of $q$, the $\phi=90\degr$ line has the most overlap with the data at small radii. The dotted line in the lower right panel is the input NFW$_{constr}$ rotation curve. The arrows indicate the radius out to which the rotation curves are compared. [{\it See the electronic edition of the Journal for a color version of this figure.}] }
498: \end{figure*}
499:
500: \begin{figure*}
501: \epsscale{1.0}
502: \plotone{f8_color.eps}
503: \caption{Same as Figure 7, but for UGC 5750. [{\it See the electronic edition of the Journal for a color version of this figure.}] }
504: \end{figure*}
505:
506: \begin{figure*}
507: \epsscale{1.0}
508: \plotone{f9_color.eps}
509: \caption{\textit{Top Row:} The solid line is the NFW$_{constr}$ rotation curve. The short dashed lines are the NFW rotation curves corresponding to the $\pm$1$\sigma$ scatter \citep{Bullock01} expected in the concentration in $\Lambda$CDM, and the long dashed lines correspond to the $\pm$1$\sigma$ scatter on $V_{200}$. The halo of this galaxy could plausibly be drawn from anywhere in this range. For example, the dotted line is the NFW rotation curve representing a low $V_{200}$ and the lowest corresponding concentration within the scatter. This halo provides a good initial match to the inner data at the expense of falling well short of the outer data. \textit{Lower panels:} Mock rotation curves for the non-axisymmetric NFW simulations of this low $V_{200}$ halo (dotted lines). Each panel is for a different value of the axis ratio $q$ (labeled in the upper left). The solid lines are for $\phi=0\degr$, 30$\degr$, 45$\degr$, 60$\degr$, and 90$\degr$. The arrows indicate the radius out to which the rotation curves are compared. [{\it See the electronic edition of the Journal for a color version of this figure.}] }
510: \end{figure*}
511:
512: \begin{figure}
513: \epsscale{1.0}
514: \plotone{f10_color.eps}
515: \caption{The viewing angle $\phi$ that minimizes the differences between the mock and observed galaxy rotation curves out to $R_{excess}$ for each value of $q$ for each galaxy. Nearly all of the galaxies fall on top of each other in the $\phi=85\degr$ to 90$\degr$ range for all values of $q$. The optimal viewing angle for F563-V2 (open squares) is $\phi \sim 55\degr$; there is a stellar bar at this position angle. The results for F563-1 and NGC~4395 are not shown because $\phi$ is unconstrained due to the radial extent of the data. Nevertheless, the fact that we detect the bar in F563-V2 is an encouraging confirmation of the method. [{\it See the electronic edition of the Journal for a color version of this figure.}] }
516: \end{figure}
517:
518: \begin{figure}
519: \epsscale{1.0}
520: \plotone{f11_color.eps}
521: \caption{The value of $\chi^{2}_{r}$ as a function of $\phi$ for the best-fitting $q$ for each galaxy. Although the $\phi \rightarrow 90\degr$ mock rotation curves are not formally good fits to the observed data, they are the minimum in $\chi^{2}$-space. The exception is F563-V2, for which $\chi^{2}_{r} \approx 1$ for the optimal viewing angle of $\phi\approx55\degr$. [{\it See the electronic edition of the Journal for a color version of this figure.}] }
522: \end{figure}
523:
524:
525: \begin{figure*}
526: \epsscale{0.80}
527: \plotone{f12_color.eps}
528: \caption{The ``best-fitting'' $\phi \rightarrow 90\degr$ rotation curves ($\phi \rightarrow 55\degr$ for F563-V2) for each galaxy. The solid (red) line is the average of the $\phi=85\degr \rightarrow 90\degr$ ($\phi=45\degr \rightarrow 60\degr$) rotation curves and the shaded (blue) band outlines the spread in those rotation curves. For comparison, the dotted line is one realization of the $\phi=60\degr$ ($\phi=90\degr$) rotation curve. The arrow indicates $R_{excess}$. From top to bottom, the galaxies are ordered by increasing distance. The simulations of UGC~4325 do not extend to the outermost observed rotation curve point. The inner 7$\arcsec$ of the mock rotation curves of UGC~5750 are poorly sampled, as in the real data. The results for F563-1 and NGC~4395 are not shown because $\phi$ is unconstrained.
529: [{\it See the electronic edition of the Journal for a color version of this figure.}] }
530: \end{figure*}
531:
532: \begin{figure}
533: \epsscale{0.72}
534: \plotone{f13_lowres.eps}
535: \caption{\textit{(a)} Observed \Dpak\ velocity field of UGC~4325. \textit{(b)} The ``best-fitting'' non-axisymmetric $\phi=90\degr$ mock NFW velocity field of UGC~4325.
536: \textit{(c)} Residual velocity field showing the difference between the observed and mock velocity fields. The velocity fields are similar in the central regions, but become more mismatched at large radii.}
537: \end{figure}
538:
539:
540: \subsection{Derived Mock Rotation Curves}
541: The LSB galaxies observed with \Dpak\ in \citetalias{K06} and
542: \citetalias{K08} surely have some level of noncircular motions, but we have
543: assumed only circular motion when deriving the rotation curves with ROTCUR;
544: we treat the mock observations the same way. The test particles are,
545: by construction, no longer on circular orbits in these non-axisymmetric
546: potentials, but because we did not correct for this in the real data,
547: we do not correct for it in the mock observations. Any errors in the
548: galaxy rotation curves which may have resulted from the assumption of
549: circular motion will be reproduced in the rotation curves of the mock
550: observations.
551:
552: In Figure 7 we show the mock rotation curves for the non-axisymmetric
553: simulations of the well-resolved galaxy UGC~4325. Each panel shows the
554: observed rotation curve of UGC~4325 along with the $\phi=0\degr$, 30$\degr$,
555: 45$\degr$, 60$\degr$, and 90$\degr$ mock rotation curves for a single value
556: of $q$. Similar to the trends seen in the mock velocity fields, we find that
557: as the axis ratio decreases and the potential becomes increasingly more
558: elongated, the influence of the viewing angle on the inferred rotation curves
559: becomes more striking: the mock rotation curves for the different values of
560: $\phi$ spread farther out in velocity space. We can quantitatively measure
561: and compare the shapes of these rotation curves using the ratio of the radii
562: containing 80\% and 50\% of the velocity at $R_{excess}$. The average values
563: of R$_{80}$/R$_{50}$ for the UGC~4325 $q=$ 0.98, 0.90, and 0.86 mock rotation
564: curves, for example, are 2.7$\pm$0.7, 2.8$\pm$0.8, and 3.1$\pm$0.7,
565: respectively. \textit{This indicates that the overall shapes of the mock
566: rotation curves are not changing significantly as $q$ or $\phi$ change; it is
567: the normalization of V(r), including V$_{max}$, that is shifting up or down.}
568: This is an important point to recognize, as it means that simply adopting a
569: different (lower) value of $V_{200}$ for the underlying NFW halo (which
570: cannot be done without ignoring cosmological constraints) will not reconcile
571: the observed and mock rotation curves; the mock NFW rotation curve will still
572: not fit the data properly. Though none of the mock rotation curves match
573: the shape of the entire observed rotation curve of UGC~4325, those that are
574: most consistent with the data at small radii are the $\phi=90\degr$ mock
575: rotation curves. This is true for all of the values of $q$ that were
576: simulated, but it is the $\phi$ = 90$\degr$ rotation curves in the
577: $q\lesssim0.90$ simulations that have the most overlap with the data.
578:
579: As a comparison to the well-resolved observations of UGC~4325, we show in
580: Figure 8 similar plots of the mock rotation curves for UGC~5750. Despite
581: the lower spatial resolution, we find the mock rotation curves of UGC~5750 to
582: behave in very much the same way as the mock rotation curves of UGC~4325.
583: \textit{We again see that the amplitude, not the shape, of the rotation
584: curve changes as $q$ and $\phi$ change, with the magnitude of the change
585: becoming more pronounced as the potential becomes more asymmetric.} We
586: also find that the $\phi=90\degr$ mock rotation curves are again the closest
587: to approaching the data, though substantial overlap at small and intermediate
588: radii does not occur until $q\lesssim 0.86$.
589:
590: We examine in Figure 9 how changing the input NFW halo parameters affects the
591: mock rotation curves, specifically exploring if $q$ and $\phi$ can change the
592: \textit{shape} of a more slowly rising input NFW rotation curve. For both
593: UGC~4325 and UGC~5750, we have simulated NFW halos with low values of
594: $V_{200}$ and the lowest corresponding concentration within the scatter of
595: \citet{Bullock01} (UGC~4325: $c=5.2$, $V_{200}=140$; UGC~5750: $c=6.9$,
596: $V_{200}=60$). These new input NFW rotation curves fall between the rotation
597: curves representing the -1$\sigma$ scatter on $c$ and $V_{200}$ of the
598: original NFW$_{constr}$ rotation curves. In addition, these slowly rising
599: rotation curves overlap the observed \Dpak\ data at small radii, in contrast
600: to the NFW$_{constr}$ rotation curves which match the data at large radii.
601: Essentially, we choose to match the inner rather than outer velocities with
602: halos drawn from the favorable edge of the plausible cosmological distribution.
603:
604: For both galaxies, we find the new mock rotation curves to behave similarly to
605: the mock rotation curves in Figures 7 and 8. As the axis ratio $q$ decreases,
606: the mock rotation curves for the different values of $\phi$ scatter about the
607: input NFW rotation curve, spreading farther out in velocity space. Even
608: though these new mock rotation curves overlap some of the observed \Dpak\ data
609: at small radii, none are formally acceptable fits ($\chi^{2}_{r}$ $\gg$ 1).
610: More importantly, the shapes of the mock rotation curves are not significantly
611: changing as $q$ and $\phi$ change: the average values of R$_{80}$/R$_{50}$
612: for the UGC~4325 $q=$ 0.98, 0.90, and 0.84 mock rotation curves are
613: 2.3$\pm$0.3, 2.8$\pm$0.4, and 3.1$\pm$1.2. The mock UGC~5750 rotation curves
614: are essentially flat between $\sim$10$\arcsec$ and $\sim$30$\arcsec$,
615: preventing useful measurements of R$_{80}$/R$_{50}$. From Figure 9 we can see
616: that regardless of the input NFW halo parameters, $q$ and $\phi$ change only
617: the normalization, not the radial behavior, of the mock rotation curves.
618:
619: It is also worth stressing that, given the behavior of the mock rotation
620: curves in Figures 7 and 8, as well as Figure 9, observers should see rotation
621: curves with a range of normalizations: there should be rotation curves both
622: above and below the nominal rotation curve expected from the $c-V_{200}$
623: relation (compare the mock rotation curves to the dotted lines in Figure 9 and
624: the lower right panels of Figures 7 and 8). But this is, in fact, not what is
625: observed in long-slit data \citep{MRdB,dBB}. LSB and NFW rotation curves
626: nearly always differ in the sense that the observed rotation curve velocities
627: at small radii must increase so that the data match the models, or
628: equivalently, the NFW rotation curve velocities must decrease so the models
629: match the data. LSB rotation curves which are higher than NFW rotation curves
630: are seldom found, if ever.
631:
632: There is a trade-off between $q$ and $\phi$ such that different combinations
633: of the two parameters can produce similar mock rotation curves. In the
634: following section, we explore what combination of ($q,\phi$) minimizes the
635: differences between the NFW halo and the \Dpak\ galaxy observations. We
636: simulate the NFW$_{constr}$ halos rather than low $V_{200}$ halos like those
637: in Figure 9 because the constrained halos were required to match the
638: velocities at the outer radii of each galaxy, a reasonable constraint since
639: dark matter must explain the high velocities at large radii where the
640: contribution of the baryons has fallen off. Parameter space is too large
641: to explore all possible initial halos. However, given that plausible
642: combinations of NFW $c$ and $V_{200}$ parameters give rather degenerate
643: rotation curves, and that $q$ and $\phi$ affect only the normalization, not
644: the shape of $V(R)$, our choice should lead to fairly general results.
645:
646:
647: \section{Minimizing the Cusp Mass Excess with $q$ and $\phi$}
648: In \citetalias{K08} we showed that there is a substantial cusp
649: mass excess near the centers of the galaxies when the NFW$_{constr}$
650: halo is used to describe the dark matter halo. Evaluating the difference
651: between the NFW$_{constr}$ rotation curve and the observed galaxy rotation
652: curve in terms of mass rather than velocity, we determined that interior to
653: the radius where the two rotation curves begin to overlap ($R_{excess}$), NFW
654: halos are at least twice as massive as the galaxy data will allow. In this
655: section, we are interested in determining for each \Dpak\ galaxy what
656: combination of ($q,\phi$) minimizes the differences between the observed
657: and mock rotation curves out to $R_{excess}$ where, in the limit of zero
658: stellar mass, the cusp mass excess is $\sim 0$ (see Table 1).
659: In this fashion, one can imagine a toy model in which the halo of a particular
660: galaxy is squashed to the best fit ($q,\phi$) within $R_{excess}$ while
661: outside of $R_{excess}$ we have a more nearly spherical, cosmologically
662: consistent NFW$_{constr}$ halo.
663:
664: Figures 7 and 8 showed that regardless of the axis ratio, the mock
665: $\phi=90\degr$ rotation curves came closest to the observed rotation
666: curves of UGC~4325 and UGC~5750. To confirm that the differences between the
667: rotation curves derived from the simulations and from the observed galaxy
668: data are really minimized at $\phi\approx 90\degr$ and not somewhere
669: between $\phi=60\degr$ and 90$\degr$, we ran additional simulations at
670: $\phi=75\degr$, 85$\degr$, 86$\degr$, 87$\degr$, 88$\degr$, and 89$\degr$.
671: We then determined for each combination of $q$ and $\phi$ how well, as
672: measured by $\chi^{2}_{r}$, the mock and observed rotation curves matched
673: out to $R_{excess}$.
674:
675: In Figure 10, we plot the best $\phi$ for each value of $q$ for each galaxy.
676: For nearly all of the 9 simulated galaxies, the mock rotation curves are the
677: most consistent with the \Dpak\ galaxy rotation curves when $\phi$ is
678: between 85$\degr$ and 90$\degr$ for all values of $q$. A
679: $\phi \rightarrow 90\degr$ means that the elongated axis of the NFW
680: potential is pointing perpendicular to our line-of-sight. This required
681: $\phi$ is completely inconsistent with a random distribution of halo
682: orientations on the sky.
683:
684: There is one galaxy in our sample, F563-V2, which has a preferred value of
685: $\phi$ \textit{other} than 90$\degr$. This galaxy has a bar in it
686: \citep{pildis}. We ran additional simulations for F563-V2 found that
687: $\phi \approx 55\degr$ is the optimal viewing angle for matching the mock
688: rotation curves to the observed rotation curves. This position angle matches
689: that of the bar. It would therefore seem that we have detected the expected
690: noncircular motion associated with the bar rather than the squashing of the
691: halo \citep[see also][]{Spekkens07}.
692:
693: This result for F563-V2 confirms that we \textit{are} able to detect the
694: presence and orientation of an asymmetry in a velocity field. If the other
695: \Dpak\ galaxies contain bars or are embedded in non-spherical NFW halos, we
696: would be able to detect the asymmetry. That $\phi \approx 90\degr$ for all
697: the other \Dpak\ galaxies demonstrates that either the data are inconsistent
698: with non-axisymmetric NFW halos, or that we must accept the unlikely
699: coincidence that all of these galaxies are oriented such that the elongated
700: axis of the potential is perpendicular to our line-of-sight. It is not
701: surprising that the effect goes in this sense as the rotation curves of LSB
702: galaxies are persistently measured to be shallower than expected for NFW halos.
703:
704: There are two galaxies whose results are not shown in Figure 10: F563-1 and
705: NGC~4395. The value of $\phi$ is unconstrained for both of these galaxies
706: due to the radial extent of the data. For F563-1, there are only a few data
707: points to compare between the observed and mock rotation curves. NGC~4395
708: is a very nearby galaxy ($D\approx 3.5$ Mpc), and although there are many
709: data points to compare in the rotation curves, the data probe a radius of less
710: than $\sim$800~pc.
711:
712: As discussed in \S~4.2, $q$ and $\phi$ can be used to change the amplitude of
713: the NFW rotation curve. They do not, however, alter the overall shape of
714: that rotation curve. This is reflected by high $\chi^{2}_{r}$ values for
715: the comparisons of the mock and observed rotation curves. Although the
716: $\chi^{2}_{r}$ values are typically greater than 1, the sharp decline in
717: $\chi^{2}_{r}$ as $\phi \rightarrow 90\degr$ indicates that
718: $\phi \sim 90\degr$ is truly the minimum (see Figure 11), even though the mock
719: rotation curves are not formally acceptable fits to the observed data.
720: In Figure 12 we plot the ``best-fitting'' $\phi \rightarrow 90\degr$ rotation
721: curves ($\phi \rightarrow 55\degr$ for F563-V2) over the observed galaxy data
722: for each galaxy. The two galaxies with the highest spatial resolution,
723: DDO~64 and UGC~4325, are clear examples of how the mock rotation curve has
724: shifted down in velocity such that the inner half of the mock rotation curve
725: is roughly consistent with the observed data, but the outer half of the mock
726: rotation curve falls below the observed data.
727:
728: Despite not being able to fully match the observed galaxy rotation curves
729: within $R_{excess}$, the important trend in Figures 10 and 11, and reinforced
730: by Figures 4-6, is that $\phi$ is being driven toward 90$\degr$ if one wants
731: to match the observed data at small radii where the cusp/core problem is most
732: severe. This means that the elongated axis of the NFW potentials for every
733: \Dpak\ galaxy (with the exception of the barred galaxy F563-V2) must point
734: perpendicular to our line-of-sight. This required $\phi$ is completely
735: inconsistent with a random distribution of halo orientations on the sky. A
736: non-axisymmetric potential with a fixed axis ratio may be able to bring
737: \textit{parts} of the NFW rotation curve into agreement with the observed
738: data for individual galaxies, but in general, very peculiar,
739: observer-dependent conditions must occur.
740:
741: It is worth mentioning here that the mismatches between the ``best-fitting''
742: mock rotation curves and the data, as seen in Figure 12 for example, are a
743: result of real differences in the velocity fields; information is not being
744: lost or suppressed as ROTCUR collapses all the data contained in the
745: two-dimensional velocity fields into a one-dimensional representation of the
746: rotation. As was shown in Figures 4 and 5, the lowest velocity portions of a
747: particle's orbit are being preferentially observed in the $\phi=90\degr$
748: viewing orientation, ensuring that the observed range of velocities detected
749: in the mock velocity fields is both small and slowly varying. But like the
750: rotation curves show, only some parts and not all of the observed and mock
751: data can be made to match. Shown as an example in Figure 13 are the observed
752: UGC~4325 \Dpak\ velocity field, the ``best-fitting'' $\phi=90\degr$
753: velocity field, and the residual velocity field showing the differences
754: between the two. While the residuals are relatively small in the central
755: regions of the velocity field, there are multiple areas at larger radii where
756: the residuals are in the range of $\sim$ 10-15 \kms, or more. As was
757: discussed in \S~3, in galaxies where the fiber-to-fiber velocity dispersions
758: are measured to be $\sim$ 6-10 \kms, 15\kms\ residuals imply a significant
759: mismatch.
760:
761:
762: \section{Discussion and Conclusions}
763: In this paper we have simulated the two-dimensional NFW halo and
764: tested several modifications to the potential in an attempt to
765: simultaneously reconcile both the NFW velocity field and rotation
766: curve with observed \Dpak\ galaxy data. We have found that it is
767: difficult to make the cuspy NFW halo appear consistent with core-like
768: data without violating the predicted range of NFW parameters expected
769: in the $\Lambda$CDM cosmological model.
770:
771: Beginning with simulations of an axisymmetric NFW potential, we found that
772: both the \Dpak\ IFU instrument and the rotation curve fitting program ROTCUR
773: are able to successfully identify the NFW potential. Observed galaxy data is
774: inconsistent with the velocity fields and rotation curves corresponding to
775: axisymmetric NFW potentials. The simulated observations show that our data
776: would detect the NFW cusp if it were present.
777:
778: We also tried a non-axisymmetric potential with a fixed axis ratio. We found
779: that if the parameters of the NFW halo (determined from the $c-V_{200}$
780: relation) are held constant and only the axis ratio and viewing orientation
781: are varied, parts of the mock velocity fields and mock rotation curves, but
782: not their entire area or length, could be made to roughly match the observed
783: galaxy data. The axis ratio and viewing orientation work to
784: \textit{change only the normalization, not the radial behavior,}
785: of the mock data. The shape of the predicted NFW rotation curve remains
786: distinct from the observations. This remains true even if a more slowly
787: rising NFW rotation curve is simulated.
788:
789: Only when the elongated axis of the mock galaxies is oriented perpendicular
790: to the observer's line-of-sight ($\phi \sim 90\degr$) are the critical
791: velocities at small radii in both the velocity fields and rotation curves
792: consistent with observations. Having all halos elongated perpendicular to
793: our line of sight is clearly not reasonable. This constraint on $\phi$ can
794: be relaxed if we start with an NFW halo having a lower $V_{200}$ (see Figure
795: 9), \textit{but changing the halo parameters requires that we disregard
796: cosmological constraints.} Numerical simulations have shown that scatter in
797: the $c-V_{200}$ relation exists ($\Delta$($\log$ c) = 0.18;
798: \citet{Bullock01}), but this is not enough to ``fix'' the mock NFW data, as
799: the values of the halo parameters would have to be outside the range of the
800: allowable dispersion. Furthermore, even though lowering $V_{200}$ allows for
801: more scatter in $\phi$, the problem with the shape of the mock NFW rotation
802: curve remains unresolved.
803:
804: In order to reconcile both the entire area of the NFW mock velocity fields and
805: the entire length of their derived rotation curves with galaxy data, we need
806: an asymmetry that preferentially suppresses velocities at small radii. The
807: asymmetric NFW potentials with fixed axis ratios that we have tested in this
808: paper have altered the velocities at \textit{all} radii, either suppressing
809: all the measured velocities (the $\phi=90\degr$ case) or boosting all of
810: them (the $\phi \rightarrow 0\degr$ cases). One possible way to address that
811: problem may be to invoke a non-axisymmetry that varies with radius
812: \citep{Hayashi07}. \citet{Hayashi07} have suggested that galaxy-sized CDM
813: halos are triaxial with radially varying axis ratios. They find the halo
814: potential to be highly elongated near the center ($b$/$a$ $\rightarrow$ 0.78
815: and $c$/$a$ $\rightarrow$ 0.72) and increasingly more spherical at large
816: radii. This is the general behavior that the results of our two-dimensional
817: simulations suggest is required.
818:
819: It remains to be seen though if such an asymmetry is a viable solution to the
820: problem. The simulations will always be constrained by the fact that LSB
821: galaxy velocity fields and rotation curves are slowly rising. This will lead
822: to the problem of a preferential viewing angle that we have already
823: encountered. Any potential that deviates significantly from axisymmetry at the
824: center will have to be viewed at an angle which lowers, not increases, the
825: observed velocities; the inner ellipsoid will need to be perpendicular to the
826: line-of-sight. On the other hand, if one insists that a radially-varying
827: asymmetry \textit{is} the correct solution and that the viewing orientation
828: is randomly distributed, both slowly and rapidly rising LSB galaxy rotation
829: curves should be observed. LSB galaxy rotation curves that are steeper than
830: NFW rotation curves are not generally found. Of the 50+ long-slit rotation
831: curves in the literature \citep[e.g.][]{Zackrisson06,Spekkens05,dBB,MRdB},
832: it is very common to see slowly rising rotation curves, but exceedingly rare
833: to see rotation curves that are in excess of the expectation for NFW halos
834: that obey the $\Lambda$CDM $c-V_{200}$ relation.
835:
836: The analysis of an NFW potential with a variable axis ratio is sufficiently
837: complex that if one is to do it right, it should be done in three dimensions
838: with interacting particles. If having a highly elongated potential is key to
839: ``fixing'' the NFW velocity field and rotation curve, then adiabatic
840: contraction should also be considered, as it will round out the inner halo
841: potential \citep{Dubinski94,Bailin07}. Adiabatic contraction also has the
842: effect of increasing the dark matter density of the halo over its initial
843: value and serves to worsen the concentration problem
844: \citep[e.g.,][]{Gnedin04, SellwoodMcGaugh}. We therefore defer the analysis
845: of a radially varying axis ratio to a forthcoming paper in which the
846: complexity of the simulations is increased by moving to three dimensions and
847: including gas physics.
848:
849:
850: \acknowledgements
851: We thank the referee for a thorough, detailed, and constructive report.
852: The work of R.~K.~D. and S.~S.~M. was supported by NSF grant
853: AST0505956. R.~K.~D. is also supported by an NSF Astronomy \&
854: Astrophysics Postdoctoral Fellowship under award AST0702496.
855: This paper was part of R.~K.~D.~'s Ph.~D. dissertation at
856: the University of Maryland. J.~C.~M. is supported by NSF grants AST0607526
857: and AST0707793.
858:
859: \begin{thebibliography}{}
860: \bibitem[Bailin et al.(2007)]{Bailin07} Bailin, J., Simon, J.~D., Bolatto, A.~D., Gibson, B.~K., \& Power, C. 2007, \apj, 667, 191
861: \bibitem[Begeman(1989)]{Begeman} Begeman, K. 1989, \aap, 223, 47
862: \bibitem[Bolatto et al.(2002)]{Bolatto02} Bolatto, A.~D., Simon, J.~D., Leroy, A., \& Blitz, L. 2002, \apj, 565, 238
863: \bibitem[Borriello \& Salucci(2001)]{Borriello} Borriello, A., \& Salucci,
864: P. 2001, \mnras, 323, 285
865: \bibitem[Bosma(1981)]{Bosma} Bosma, A. 1981, \aj, 86, 1825
866: \bibitem[Bullock et al.(2001)]{Bullock01} Bullock, J.~S., Kolatt, T.~S., Sigad, Y., Somerville, R.~S., Kravtsov, A.~V., Klypin, A.~A., Primack, J.~R., \& Dekel, A. 2001, \mnras, 321, 559
867: \bibitem[Chemin et al.(2004)]{Chemin04} Chemin, L., et al. 2004, IAUS, 220, 333
868: \bibitem[C\^{o}t\'{e}, Carignan, \& Freeman(2000)]{Cote} C\^{o}t\'{e}, S.,
869: Carignan, C., \& Freeman, K.C. 2000, \aj, 120, 3027
870: \bibitem[de Blok \& Bosma(2002)]{dBB} de Blok, W.~J.~G., \& Bosma, A.
871: 2002, \aap, 385, 816
872: \bibitem[de Blok et al.(2003)de Blok, Bosma, \& McGaugh]{dBBM} de Blok, W.~J.~G., Bosma, A., \& McGaugh, S.~S. 2003, \mnras, 340, 657
873: \bibitem[de Blok \& McGaugh(1996)]{dBM96} de Blok, W.~J.~G., \& McGaugh,
874: S.~S. 1996, \apj, 469, L89
875: \bibitem[de Blok \& McGaugh(1997)]{dBM97} ---------- . 1997, \mnras, 290, 533
876: \bibitem[de Blok et al.(2001)de Blok, McGaugh, \& Rubin] {dBMR} de Blok, W.~J.~G.,
877: McGaugh, S.~S., \& Rubin, V.~C. 2001, \aj, 122, 2396
878: \bibitem[de Blok et al.(1996)de Blok, McGaugh, \& van der Hulst]{DMV} de Blok,
879: W.~J.~G., McGaugh, S.~S., \& van der Hulst, J.~M. 1996, \mnras, 283, 18
880:
881: \bibitem[Diemand et al.(2005)]{Diemand} Diemand, J., Zemp, M., Moore, B.,
882: Stadel, J., \& Carollo, M. 2005, \mnras, 364, 665
883: \bibitem[Dubinski(1994)]{Dubinski94} Dubinski, J. 1994, ApJ, 431, 617
884: \bibitem[Flores \& Primack(1994)]{Flores} Flores, R.~A., \& Primack, J.~R.
885: 1994, \apjl, 427, L1
886:
887: \bibitem[Fuchs(2003)]{Fuchs} Fuchs, B. 2003, \apss, 284, 719
888: \bibitem[Gentile et al.(2005)]{Gentile05} Gentile, G., Burkert, A.,
889: Salucci, P., Klein, U., \& Walter, F. 2005, \apj, 634, L145
890: \bibitem[Gentile et al.(2007)]{Gentile07} Gentile, G., Salucci, P.,
891: Klein, U., \& Granato, G.~L. 2007, \mnras, 375, 199
892: \bibitem[Gnedin et al.(2004)]{Gnedin04} Gnedin, O.~Y., Kravtsov, A.~V.,
893: Klypin, A.~A., \& Nagai, D. 2004, \apj, 616, 16
894: \bibitem[Hayashi et al.(2007)Hayashi, Navarro, \& Springel]{Hayashi07} Hayashi, E.,
895: Navarro, J.~F., \& Springel, V. 2007, \mnras, 377, 50
896: \bibitem[Jing(2000)]{Jing} Jing, Y. 2000, \apj, 535, 30
897: \bibitem[Kuzio de Naray et al.(2006)]{K06} Kuzio de Naray, R.,
898: McGaugh, S.~S., de Blok, W.~J.~G., \& Bosma, A. 2006, \apjs, 165, 461 (K06)
899: \bibitem[Kuzio de Naray et al.(2008)Kuzio de Naray, McGaugh, \& de Blok]{K08} Kuzio de Naray, R., McGaugh, S.~S., \& de Blok, W.~J.~G. 2008, \apj, 676, 920 (K08)
900: \bibitem[Marchesini et al.(2002)]{Marchesini} Marchesini, D.,
901: D'Onghia, E., Chincarini, G., Firmani, C., Conconi, P., Molinari, E.,
902: \& Zacchei, A. 2002, \apj, 575, 801
903: \bibitem[McGaugh et al.(2003)]{McGaugh03} McGaugh, S.~S., Barker, M.~K., \&
904: de Blok, W.~J.~G. 2003, \apj, 584, 566
905: \bibitem[McGaugh et al.(2007)]{McGaugh07} McGaugh, S.~S., de Blok, W.~J.~G., Schombert, J.~S., Kuzio de Naray, R., \& Kim, J.~H. 2007, \apj, 659, 149
906: \bibitem[McGaugh, Rubin, \& de Blok(2001)]{MRdB} McGaugh, S.~S., Rubin,
907: V.~C., \& de Blok, W.~J.~G. 2001, \aj, 122, 2381
908: \bibitem[Mihos, McGaugh, \& de Blok(1997)]{Mihos97} Mihos, J.~C., McGaugh, S.~S., \& de Blok, W.~J.~G. 1997, \apjl, 477, L79
909: \bibitem[Moore et al.(1999)]{Moore} Moore, B., Quinn, T., Governato,
910: F., Stadel, J., Lake, G. 1999, \mnras, 310, 1147
911: \bibitem[Navarro et al.(1996)Navarro, Frenk, \& White]{NFW96} Navarro, J.~F., Frenk,
912: C.~S., \& White, S.~D.~M. 1996, \apj, 462, 563
913: \bibitem[Navarro et al.(1997)Navarro, Frenk, \& White]{NFW97} ---------- . 1997, \apj, 490, 493
914: \bibitem[Navarro et al.(2004)]{Navarro2004} Navarro, J.F., et al. 2004, \mnras, 349, 1039
915: \bibitem[Pildis et al.(1997)]{pildis} Pildis, R.~A.,
916: Schombert, J.~M., \& Eder, J.~A.\ 1997, \apj, 481, 157
917: \bibitem[Reed et al.(2003)]{Reed} Reed, D., Gardner, J., Quinn, T.,
918: Stadel, J., Fardal, M., Lake, G., \& Governato, F. 2003, \mnras, 346, 565
919: \bibitem[Rubin et al.(1980)]{Rubin} Rubin, V.~C., Thonnard, N., \& Ford, W.~K., Jr. 1980, \apj, 238, 471
920: \bibitem[Sellwood \& McGaugh(2005)]{SellwoodMcGaugh} Sellwood, J.~A.,
921: \& McGaugh, S.~S. 2005, \apj, 634, 70
922: \bibitem[Schoenmakers, Franx, \& de Zeeuw(1997)]{Schoenmakers} Schoenmakers, R.~H.~M., Franx, M., \& de Zeeuw, P.~T. 1997, \mnras, 292, 349
923: \bibitem[Simon et al.(2005)]{Simon05} Simon, J.~D., Bolatto, A.~D.,
924: Leroy, A., Blitz, L., \& Gates, E. 2005, \apj, 621, 757
925: \bibitem[Spekkens, Giovanelli, \& Haynes(2005)]{Spekkens05} Spekkens, K., Giovanelli, R., \& Haynes, M.~P. 2005, \aj, 129, 2119
926: \bibitem[Spekkens \& Sellwood(2007)]{Spekkens07} Spekkens, K., \&
927: Sellwood, J.~A. 2007, \apj, in press (astro-ph/0703688)
928: \bibitem[Swaters et al.(2003a)]{Swaters03a} Swaters, R.~A., Madore, B.~F.,
929: van den Bosch, F.~C., \& Balcells, M. 2003a, \apj, 583, 732
930: \bibitem[Swaters et al.(2003b)]{Swaters03b} Swaters, R.~A., Verheijen,
931: M.~A.~W., Bershady, M.~A., \& Andersen, D.~R. 2003b, \apj, 587, L19
932: \bibitem[Tegmark et al.(2004)]{Tegmark} Tegmark, M., et al. 2004, \prd, 69, 103501
933: \bibitem[Teuben(1995)]{Teuben} Teuben, P.J. The Stellar Dynamics
934: Toolbox NEMO, in: Astronomical Data Analysis Software and Systems
935: IV, ed. R. Shaw, H.E. Payne and J.J.E. Hayes. (1995), PASP Conf
936: Series 77, p398
937: \bibitem[Wechsler et al.(2002)]{Wechsler} Wechsler, R.~H., Bullock, J.~S., Primack, J.~R., Kravtsov, A.~V., \& Dekel, A. 2002, \apj, 568, 52
938: \bibitem[Wong, Blitz, \& Bosma(2004)]{Wong} Wong, T., Blitz, L., \& Bosma, A. 2004, \apj, 605, 183
939: \bibitem[Zackrisson et al.(2006)]{Zackrisson06} Zackrisson, E., Bergvall, N., Marquart, T., \& \"{O}stlin, G. 2006, \aap, 452, 857
940: \bibitem[Zentner \& Bullock(2002)]{Zentner02} Zentner, A.~R., \& Bullock,
941: J.~S. 2002, \prd, 66, 043003
942: \end{thebibliography}
943:
944:
945:
946:
947: \end{document}
948: