1: \documentclass[12pt,preprint]{aastex}
2: %\documentclass[preprint]{/scr0/gtr/aastex/aastex}
3: %\documentclass{aastex}
4: %usepackage{setspace}
5: \usepackage{natbib}
6: \citestyle{aa}
7:
8: \topmargin 0.5cm
9:
10: %\received{}
11: %\accepted{}
12: %\journalid{}{}
13: %\articleid{}{}
14: \newcommand{\AAA}{$\rm{\AA} \;$}
15: \newcommand{\kms}{km~s$^{-1}$}
16: \newcommand{\kmss}{km~s$^{-1} \;$}
17: \newcommand{\Myr}{$M_{\sun}$ yr$^{-1}$}
18: \newcommand{\Sersic}{S${\rm \acute{e}}$rsic}
19: \newcommand{\MSun}{M_{\odot}}
20: \newcommand{\Msun}{M_{\odot}}
21: %\newcommand{\R90}{$R_{90}$}
22: %\newcommand{\R50}{$R_{50}$}
23: \newcommand{\Msunkpc}{\; \MSun~{\rm kpc}^{-2}}
24: \newcommand{\MSunkpc}{\; \MSun~{\rm kpc}^{-2}}
25: %\newcommand{\a1i}{A_1^i}
26: %\newcommand{\a2i}{A_2^i}
27: %\newcommand{\a3i}{A_3^i}
28: %\newcommand{\a4i}{A_4^i}
29: %\newcommand{\a5i}{A_5^i}
30: %\newcommand{\da2i}{\Delta A_2^i}
31: %\newcommand{\da3i}{\Delta A_3^i}
32: %\newcommand{\da4i}{\Delta A_4^i}
33: %\newcommand{\da5i}{\Delta A_5^i}
34: %\newcommand{\farcs}{\mbox{\ensuremath{.\!\!^{\prime\prime}}}}
35:
36: \slugcomment{April 2007 draft}
37:
38: \author{
39: Timothy A. Reichard\altaffilmark{1}, Timothy
40: M. Heckman\altaffilmark{1}, Gregory Rudnick\altaffilmark{2,5}, Jarle
41: Brinchmann\altaffilmark{3}, Guinevere Kauffmann\altaffilmark{4}}
42:
43:
44: \altaffiltext{1}{Department of Physics and Astronomy, The Johns Hopkins University, 3400 North Charles Street, Baltimore, MD 21218-2686.}
45: \altaffiltext{2}{National Optical Astronomy Observatory (NOAO), 950 N. Cherry Ave., Tucson, AZ 85719, USA}
46: \altaffiltext{3}{Centro de Astrofisica da Universidade do Porto, Rua das Estrelas, 4150-762 Porto, Portugal.}
47: \altaffiltext{4}{Max-Planck-Institut f\"{u}r Astrophysik, D-85748 Garching, Germany.}
48: \altaffiltext{5}{Goldberg Fellow}
49: %\setlength{\oddsidemargin}{0in}
50: %\setlength{\evensidemargin}{0in}
51: %\setlength{\textwidth}{6.5in}
52: \begin{document}
53:
54: \title{The Lopsidedness of Present-Day Galaxies: Results from the Sloan Digital Sky Survey}
55:
56: \begin{abstract}
57: Large-scale asymmetries in the stellar mass distribution in galaxies
58: are believed to trace non-equilibrium situations in the luminous
59: and/or dark matter component. These may arise in the aftermath of
60: events like mergers, accretion, and tidal interactions. These events
61: are key in the evolution of galaxies. In this paper we quantify the
62: large-scale lopsidedness of light distributions in 25155 galaxies at
63: $z < 0.06$ from the Sloan Digital Sky Survey Data Release 4 using the
64: $m = 1$ azimuthal Fourier mode. We show that the lopsided
65: distribution of light is primarily due to a corresponding lopsidedness
66: in the stellar mass distribution. Observational effects, such as
67: seeing, Poisson noise, and inclination, introduce only small errors in
68: lopsidedness for the majority of this sample. We find that
69: lopsidedness correlates strongly with other basic galaxy structural
70: parameters: galaxies with low concentration, stellar mass, and stellar
71: surface mass density tend to be lopsided, while galaxies with high
72: concentration, mass, and density are not. We find that the strongest
73: and most fundamental relationship between lopsidedness and the other
74: structural parameters is with the surface mass density. We also find,
75: in agreement with previous studies, that lopsidedness tends to
76: increase with radius. Both these results may be understood as a
77: consequence of several factors. The outer regions of galaxies and
78: low-density galaxies are more susceptible to tidal perturbations, and
79: they also have longer dynamical times (so lopsidedness will last
80: longer). They are also more likely to be affected by any underlying
81: asymmetries in the dark matter halo.
82:
83: \end{abstract}
84: \keywords{galaxies: structure, galaxies: interactions, galaxies: general}
85:
86: \section{Introduction}
87:
88:
89:
90:
91: %\citet{rz95}: 18 galaxies. Galaxy disks exhibit a wide variety of shapes visible in the near-IR. They are due to distortions in surface density distributions, not mass-to-light ratios. Stellar velocities in nonaxisymmetric galaxies differ by 3-6\% from those in symmetric ones.
92:
93: %\citet{zr97}: 60 galaxies, magnitude-limited. 30\% of galaxies have $A_1 > 0.2$. Lopsided mass distributions remain long enough to indicate past interactions, not just ongoing ones. Often the companion is not obvious, either merged or fled.
94:
95: %\citet{rr98}: 54 early-type disk galaxies. Low SFRs to reduce possibility of asymmetric stellar distributions and increase possibilty of light->mass asymmetry tracing. Traces mass dependence because VRI bands show similar lopsidedness and it's OK to go shorter than I and K bands. 20\% of galaxies have lopsidedness above 0.19.
96:
97:
98: %\citet{rrk00}: Correlations between lopsidedness and recent SF, and current SF were found, lopsided = star-forming. Significant fractions of stellar content can be created in a short time from minor mergers and on a similar timescale (1 Gyr).
99:
100: It has long been recognized that galaxies show large-scale asymmetries
101: in their structure \citep{bl+80}. Lopsided galaxies have such
102: asymmetries where one side of their disk is more massive and/or more
103: extended than the opposite side. This “lopsidedness” can be traced in
104: the spatial structure of the stars \citep{rz95} and/or the HI gas
105: \citep{rs94} and/or in the large-scale kinematics of this material
106: \citep{ss+99}.
107:
108: There are a variety of mechanisms or events that have been proposed to
109: produce the observed lopsidedness. All of them involve a
110: time-dependent non-equilibrium dynamical state, in most cases
111: triggered through an external process. Such external processes are a
112: natural consequence of the standard Lambda Cold Dark Matter
113: cosmological framework. This implies that galaxies assemble
114: hierarchically (a process that is on-going). Examples that can lead to
115: lopsidedness include a minor merger (\citealt{wm+96};
116: \citealt{zr97}), the tidal interaction resulting from a close
117: encounter between roughly equal-mass galaxies \citep{kl+02},
118: and the asymmetric accretion of intergalactic gas into the disk
119: (\citealt{bc+05}; \citealt{kk+05}). Other mechanisms involve the
120: dark matter halo: stars and gas orbiting in a lopsided dark matter
121: halo (\citealt{w94}; \citealt{j97}; \citealt{j99}) or a stellar/gas
122: disk that is offset with respect to the center of the dark matter halo
123: (\citealt{ls98}; \citealt{ns+01}). These
124: also involve past tidal interactions and/or mergers that have
125: perturbed the dark matter halo, but such perturbations may be quite
126: long-lived. Finally, dynamical processes internal to the disk that
127: lead to mildly lopsided distributions have also been investigated (\citealt{st+90}; \citealt{st96}; \citealt{mt97}).
128:
129: A variety of programs to study lopsidedness have been undertaken over
130: the past decade. Most of these investigations have studied the
131: lopsided distribution of the stellar component through analysis of
132: optical and near-infrared images. \citet{zr97} studied a
133: magnitude-limited sample of 60 field spiral galaxies. They measured
134: lopsidedness as the radially averaged, azimuthal $m=1$ Fourier
135: amplitude $A_1$ of the light (see Section \ref{sec:error} below) and
136: computed lopsidedness between 1.5 and 2.5 scale lengths in the
137: galactic disks. The value of $A_1$ indicates the typical large-scale
138: variation in mass density from side to opposite side at the same
139: distance from the galactic center. The mass density typically varies
140: from between $1 \pm A_1$ times the average density at the same radius.
141: They found that $\sim 30$\% of field spiral galaxies exhibited
142: significant lopsidedness ($A_1 > 0.2$). \citet{rr98} followed up this
143: work by studying lopsidedness in 54 early-type galaxies and found that
144: $\sim 20$\% had $A_1 > 0.19$.
145: \citet{cbj00} studied a sample of 113 $z < 0.01$ galaxies (elliptical, spiral,
146: and irregular) but used a 180-degree rotational asymmetry measure
147: $A_{180}$. They found that asymmetry is strongly dependent on
148: morphological type, with lower asymmetry in elliptical and lenticular
149: galaxies and higher asymmetry in late-type disk and irregular
150: galaxies. More recently, \citet{bc+05} have measured the Fourier $A_1$
151: parameter for 149 galaxies in the Ohio State University Bright Galaxy
152: Survey. They confirmed that a large fraction of galaxies have
153: significant lopsidedness in their stellar disks, with late-type
154: galaxies being more lopsided.
155:
156: Lopsidedness in the {\it light} distribution can be produced by either
157: a corresponding asymmetry in the underlying {\it mass} distribution in
158: the stellar population or by large-scale variations in the
159: mass-to-light ratio (e.g., from star formation and dust
160: obscuration). \citet{rz95} investigated this issue with a sample of 18
161: face-on spiral galaxies imaged in the K$^\prime$ (2.2$\mu$ m) band
162: where the effect of young stars or dust is minimized. They found that
163: about a third of the sample showed significant lopsidedness (similar
164: to results from optical investigations). Similarly, \citet{rr98} found
165: that lopsidedness in early-type disk galaxies is nearly identical when
166: observed in the $V$, $R$, and $I$ bands. They concluded that an
167: asymmetric mass distribution then accounts for the majority of the
168: asymmetry in the light distribution in these galaxies.
169:
170: Lopsidedness has also been studied in the distribution of HI
171: gas. Since the HI can frequently be traced to significantly larger
172: radii than the stars, these investigations are highly complementary to
173: the optical image analysis. Due to the time-consuming nature of HI
174: interferometric mapping, only modest size samples have been analyzed
175: in this way \citep{ss+99}. On the other hand, \citet{rs94} have examined the global HI line profiles for roughly 1700
176: galaxies, and shown that at least 50\% are significantly asymmetric
177: (confirming that the large-scale HI distribution is frequently
178: lopsided). HI maps also show that – apart from a lopsided distribution
179: of the gas – the HI rotation curves are often asymmetric \citep{ss+99}. The connection between the phenomena of structural and
180: kinematic lopsidedness in galaxies is not yet clear \citep{ss+99}.
181:
182: Despite these diverse investigations and the abundance of proposed
183: models, the origin of lopsidedness remains unsettled. For models
184: involving tidal interactions or minor mergers, there is an expected
185: link between lopsidedness and the local environment. The evidence in
186: this regard has been mixed (e.g., \citealt{wp04};
187: \citealt{bc+05}; \citealt{aj+06,aj+07}; \citet{dc+07}).
188:
189: %\citet{dc+07} have undertaken
190: %the most comprehensive investigation so far of this issue. Using the
191: %rotational asymmetry measure $A_{180}$ to study a sample of over 3000
192: %galaxies, they find that close pairs of galaxies are more asymmetric
193: %than other galaxies and that the asymmetry increases as the pair
194: %separation decreases. They conclude that these global asymmetries
195: %trace recent tidal interactions or mergers.
196:
197: The investigations summarized above have all involved relatively small
198: samples of galaxies, making it difficult to assess the overall
199: distribution of asymmetry or lopsidedness as a function of the basic
200: parameters that characterize the structure of galaxies. This is the
201: first of three papers in which we use the wealth of data available
202: from the Sloan Digital Sky Survey (SDSS) to extend these studies of
203: small samples (of-order one hundred galaxies) to large samples (tens
204: of thousands). In this paper, we describe our sample selection and
205: methodology. We also relate lopsidedness to the basic structural
206: properties of the galaxies. In Paper II we will investigate the
207: connection between lopsidedness and both star formation and black hole
208: growth in galaxies. Finally, in Paper III we will examine the
209: connection between lopsidedness and the local galaxy environment.
210:
211: In \S\ref{sec:data}, we begin by presenting an initial low-redshift
212: sample from the SDSS and describe the observations and properties for
213: its galaxies. Next, we explain our lopsidedness calculation. In
214: \S\ref{sec:syserr}, we address the major data quality issues that
215: limit the reliability of the measurements for portions of the
216: sample. On this basis, we apply cuts on the observational parameters
217: to weed out the problematic cases for our subsequent analysis. Next,
218: \S\ref{sec:lopprops} describes the lopsidedness of galactic light
219: distributions in different optical/near-IR bands, its correspondence
220: with lopsided mass distributions, and its radial dependence. We then
221: examine the relationship between lopsidedness and the basic structural
222: properties of galaxies in \S\ref{sec:lopsfh}. Finally, we summarize
223: our findings in \S\ref{sec:summary}.
224:
225: \section{Data \& Basic Methodology \label{sec:data}}
226:
227: \subsection{Initial Sample}
228:
229:
230: The initial sample of galaxies was taken from the Sloan Digital Sky Survey
231: \citep{y+00,s+02}, a large survey of photometric and spectroscopic
232: data across $\pi$ sr. of the northern sky. The sample is derived from SDSS Data Release 4 \citep{dr4}. The survey's dedicated 2.5-m telescope \citep{g+06} at Apache Point Observatory uses a unique CCD camera
233: \citep{g+98} and drift-scanning to obtain $u$-, $g$-, $r$-, $i$-, and
234: $z$-band photometry \citep{f+06,h+01,i+04,sm+02,t+06}. The pixel scale
235: is 0\farcs396/px. Fiber spectroscopy is obtained using
236: 3\arcsec~fibers and results in wavelength coverage between
237: 3800$-$9200~\AA~at a resolution $R=\lambda/\delta\lambda = 1850-2200$.
238:
239: As we will show below, meaningful measurements of lopsidedness impose
240: requirements on the angular size and signal-to-noise in the galaxy
241: image. These criteria are not met by the full SDSS Main galaxy sample
242: (median redshift $\sim0.1$ and magnitude $r <$ 17.8; citealt{swl+02}). Accordingly, our
243: initial galaxy sample was selected from the Main sample with a simple
244: redshift cut ($z \le 0.06$). It contains 67107 galaxies. The SDSS
245: photometric pipeline, PHOTO \citep{lgi+01}, provides Petrosian
246: apparent magnitudes and half- and 90\%-light radii ($R_{50}$ and
247: $R_{90}$) in each of the five bands, along with seeing conditions and
248: other photometric and structural properties. We use the methodology
249: described in \citet{khw+03a} to use the spectral information to derive
250: $z$-band mass-to-light ratios and hence stellar masses ($M_*$). These
251: stellar masses and the $z$-band half-light radii are then used to
252: measure stellar surface mass densities ($\mu_*$, defined as the mean
253: mass per unit area inside the half-light radius; see
254: \citealt{khw+03b}).
255:
256:
257: \subsection{Measuring Lopsidedness \label{sec:error}}
258:
259: Lopsidedness in galactic light distributions has been computed in
260: recent years by two useful approaches. One approach, used by
261: \citet{rz95} and in later work, is to perform an azimuthal Fourier
262: decomposition of galaxy light. The first mode, $A_1$, quantifies the
263: large-scale overabundance of light in one side of a galaxy with a
264: corresponding under-abundance in the opposite side. The first mode
265: quantifies a galaxy-wide lopsidedness that can be quickly computed for
266: large numbers of galaxies. In the second approach, detailed in
267: \citet{avg+96}, a different asymmetry index $A_{180}$ was devised by
268: subtracting a 180$^\circ$ rotated image from the original galaxy image
269: and summing the residual light compared to the total galaxy light.
270: This lopsidedness measure is sensitive to both large- and small-scale
271: variations in symmetry. To convert $A_{180}$ into a measure of only
272: large-scale lopsidedness, a smoothing filter can be applied beforehand
273: to the image \citep{c03}. Our interest is to examine lopsidedness in
274: tens of thousands of galaxies, and we have determined that the modal
275: lopsidedness $A_1$ algorithm is computationally less expensive than
276: $A_{180}$. We will therefore take the radially averaged first mode
277: strength, $A_1$, as the measure of lopsidedness for the galaxy as a
278: whole in each band, as has been done in previous studies
279: \citep{rz95,zr97,rr98,rrk00}. The details of the calculation will fill out this section, but we first summarize the general method.
280:
281: The galaxy light is binned into radial and azimuthal bins, and then a
282: finite Fourier transform is applied to the surface brightness $\mu$ in
283: each radial bin. The transform for the $k$th radial bin is
284: \begin{equation}
285: \mu(r_k, \phi) = \sum\limits_{m = 0}^{m_{max}} b_m (r_k) e^{im(\phi - \phi_m (r_k))}. \label{eqn:decomp}
286: \end{equation}
287: The transform gives the mode magnitudes $b_m(r_k)$ and phases
288: $\phi_m(r_k)$ for each mode $m$. The magnitudes are divided by the
289: zeroth mode (average surface brightness) and radially averaged to calculate
290: $A_m$ for that galaxy and band, including the lopsidedness $A_1$.
291:
292: We compute this average over
293: the radial range between $R_{50}$ and $R_{90}$. These are the
294: practical limits: we can not go much inside $R_{50}$ because of the
295: seeing effects discussed below, and can not go much outside $R_{90}$
296: because the data become unacceptably noisy. As we will show below, the
297: average radial variation in $A_1$ beyond is modest, and the mean value
298: for $A_1$ does not depend significantly on the precise choices of the
299: inner and outer radii.
300:
301: The transform also yields values of the higher-order modes. The second
302: mode $A_2$ represents a combination of effects: ellipticity,
303: inclination, two-arm spiral structure, and barred structure. Since we
304: do not deproject the images to a face-on orientation, the amplitude of
305: the second mode primarily measures the ellipticity of the
306: image. Higher-order modes ($m > 2$) may include multi-arm spiral
307: structure as well as Fourier ``ringing'' of features also present in
308: lower-order modes. For example, a galaxy with a high ellipticity will
309: have a strong second mode with weaker even-order modes also present.
310: Similarly, a lopsided galaxy will have a strong first mode with weaker
311: odd-order modes present as well.
312:
313: To begin our process, the galaxy center was determined. Precision centering is
314: important because moving the center by more than $0.5$ px can increase
315: the lopsidedness significantly. The inner region of the galaxy was
316: smoothed in a 3 px radius to reduce the effects of Poisson noise in
317: determining the center.
318: \footnote{\citet{cbj00} have employed a
319: different centering technique with their $A_{180}$ asymmetry measure.
320: They found that the best center to use for the $A_{180}$ calculation
321: was the center that minimizes $A_{180}$. In our multi-mode
322: calculation, we have found that the center that minimizes the strength
323: of one mode does not minimize the strengths of other modes. Thus the
324: minimization technique would require a different center for each mode.
325: Since the modes are calculated simultaneously in the Fourier
326: transform, minimization is not feasible.}
327: We assume that the center point of the galaxy is within 3 px of the
328: brightest pixel of the inner region of the galaxy. That brightest
329: pixel is our initial estimate of the center point. To improve this
330: estimate, the first moment of light was computed in a 3 $\times$ 3 px
331: box centered on the brightest pixel. The first moment is the origin
332: that minimizes the second moment and gives a central position in
333: fractional units of a pixel that improves upon the brightest pixel
334: estimate.
335: %In low-$S/N$ galaxies, Poisson noise can have an important effect
336: %in centering and introduces more uncertainty into both the center
337: %position and the modes. The effect is strongest in the innermost
338: %region of the galaxy. To minimize this effect, we always mask the
339: %inner 3.0 px (1.2$\arcsec$), even if $R_{50}$ is smaller than this
340: %radius.
341:
342: Our centering method was used on each image without reference to the
343: centers determined for the same object in other bands. This means
344: that each object was given a separate center point for each band.
345: Nonetheless, the agreement between these independent center points is good.
346: Of the three pairs of distances between centers in the
347: three bands, the angular distance between the $g$- and $i$-band
348: centers is the greatest. The $1\sigma$ variation between these bands
349: is $0\farcs12^{+0.11}_{-0.07}$ ($0.32^{+0.29}_{-0.17}$ px). Some of
350: the scatter may be attributable to the offset pixel locations between
351: images of different bands, i.e., the coordinates of the center of the
352: central pixel in one band will not be the located in the center of a
353: pixel in another band. The rest of the scatter is due to variations
354: in the central structure between the different bands. Our tests show
355: that the mode strengths are significantly increased in a given image
356: if the center point is moved more than 0.5 px from the best center.
357: Thus the scatter of center points between the different bands should
358: have no significant affect on the subsequent lopsidedness computation.
359:
360:
361: The value of the sky background was calculated using the DAOPHOT
362: package in the IDL Astronomy Library. For each galaxy and each band,
363: the sky-subtracted image was partitioned into a polar grid of
364: logarithmic radial bins between the Petrosian radii $R_{50}$ and
365: $R_{90}$ and centered at the determined galaxy center. We selected
366: the minimum $R_{50}$ and maximum $R_{90}$ determined in the $g$-,
367: $r$-, and $i$-band images as the inner and outer radii of the grid,
368: and the same grid size is used in the lopsidedness computation in all
369: three bands. The radial bins were further partitioned into equal
370: azimuthal bins. The number of radial bins was allowed to vary with
371: the size of the galaxies by setting the innermost bin size as close to
372: but no smaller than 1.0 pixel. Exactly 12 azimuthal bins were used in
373: each grid so that the first six Fourier modes ($m = 0, 1, 2, \ldots,
374: 5$) could be determined. Each pixel that was split by the grid into
375: more than one bin was divided into a $9
376: \times 9$ array of sub-pixels, and bilinear interpolation of
377: neighboring pixels' surface brightness assigned each sub-pixel with a
378: value of surface brightness. Higher-order interpolation, such as using
379: bicubic splines, showed no significant improvement and thus was not
380: worth the computational expense. The sub-pixels were then included in
381: the appropriate polar bin.
382:
383:
384: Galaxies whose observed light distributions are contaminated with
385: light from other galaxies and foreground stars pose a problem for this
386: Fourier decomposition. The contaminating light results in a set of
387: strong modes that describe the combination of galaxy and overlapping
388: star or galaxy rather than the desired galaxy by itself. Foreground
389: stars can be masked out with some ease once they are located in
390: images. We use the DAOPHOT routines of the IDL Astronomy Library to
391: locate them and then estimate radii where their light drops to half the
392: level of the $1\sigma$ sky background noise. For intervening
393: galaxies, we queried the SDSS PHOTO catalog for galaxy positions and
394: Petrosian $R_{90}$. Then a mask image was created that was the same
395: size as the galaxy image, but each pixel accepted values of either 0
396: (pixel is masked) or 1 (pixel is ``good''). The Fourier decomposition
397: calculation uses only the ``good'' pixels after the stars and galaxies
398: are masked in the image. Any pixels that lay within a star, another
399: galaxy, or other extraneous light source were ignored.
400:
401:
402: After masking out the unwanted pixels, the surface brightness was
403: computed in each bin of the grid. An azimuthal Fourier transform
404: (Eq.~\ref{eqn:decomp}) was performed in each radial bin to determine a
405: magnitude $b_m (r_k)$ and phase $\phi_m (r_k)$ for each mode $m$. The
406: transform was formulated as a general linear least squares problem
407: \citep{p02}. Errors on the magnitudes and phases were deduced from
408: the diagonal elements of the covariance matrix. The phase $\phi_m$ of
409: the $m$th mode is physically identical to phases $\phi_m + 2\pi K/m$
410: for integers $K$, so $K$ was chosen for each radial bin to minimize
411: the radial dependence of the phases. The magnitudes $b_m (r_k)$
412: were then divided by the zeroth mode to give relative magnitudes $a_m
413: (r_k) = b_m (r_k)/b_0 (r_k)$. These values $a_m (r_k)$ give a profile
414: of lopsidedness ($m = 1$) and higher modes ($m > 1$), each normalized
415: to the average brightness at radius $r_k$.
416:
417:
418: For simplicity, we prefer to use a single value of the Fourier modes
419: in each band rather than a radial profile. We calculate the average of
420: $a_m$ at each radius between $R_{50}$ and $R_{90}$, weighted by its
421: error at each radius, to give a single, global measure $\left <a_m
422: \right>$ for each galaxy in each band. The weighting naturally counts
423: the dimmer, outer reaches of the galaxy less than the brighter, inner
424: region and discounts any radial bin that is too contaminated with
425: light from other sources. At radii where an extraneous light source
426: has been masked out, the azimuthal Fourier transform can fail. Such
427: radial bins are excluded from the weighted average.
428:
429:
430: Since $\left <a_m \right>$ is a positive definite quantity, random errors would
431: preferentially overestimate the quantity. We therefore correct the
432: average in the manner used in the past (e.g., \citealt{rr98}) and
433: adopt
434: \begin{equation}
435: A_m = \sqrt{\left <a_m \right >^2 - (\delta\left <a_m \right >)^2}
436: \label{eqn:posdef}
437: \end{equation}
438: as the strength of the $m$th mode, where $\delta \left <a_m \right >$
439: is the random error in the weighted mean mode. We will discuss
440: systematic and random errors in $A_1$ in the next section
441: (\S\ref{sec:syserr}).
442:
443: As an example, we show the Fourier decomposition of the nearly
444: face-on, barred spiral galaxy SDSS J125416.38-020204.4 in
445: Fig.~\ref{fig:sampleim}. The first three panels show the $g-$, $r-$,
446: and $i-$band images of this galaxy, a three-color image reconstructed
447: from Fourier modes, and a three-color image of the $m = 1$ through 5
448: Fourier modes. The core of the galaxy is omitted in the
449: three-color images as the decomposition is not attempted in that
450: region. However, for this illustration, we have extended the Fourier
451: decomposition to radii interior to $R_{50}$. The central bar is
452: described by the strong even-order modes in all three bands at low
453: radii $2\arcsec < r < 5\arcsec$. The bar weakens at larger radii, and
454: the even-order modes then decrease with increasing radii. At $r >
455: 8\arcsec$, the brightness of the galaxy drops until sky noise becomes
456: significant. The lopsidedness in this galaxy is easily seen outside
457: the bright center of the galaxy. The bar's brightness falls off more
458: slowly with radius in the upper-right direction than in the lower-left
459: direction. This lopsidedness is described in all three bands by the
460: significant first mode strength, which rises from near 0.1 to 0.4
461: between 2$\arcsec$ and 8$\arcsec$. The radially-averaged
462: lopsidedness, as computed between $R_{50}$ and $R_{90}$ in the
463: $i$-band, is 0.29.
464:
465: \begin{figure}[ht]
466:
467: \epsscale{1.0}
468: %\plotone{/home/tar/agn/asymm/imreconst/0338-51694-360.ps}
469: \plotone{fig1.eps}
470: \caption{Images of the Fourier decomposition of SDSS J125416.38-020204.4 with $A_1^i = 0.29$. {\em Upper row, left to right:} SDSS $g$-, $r$-, and $i$-band images. {\em Middle row, left:} Combined $gri$ image reconstructed from Fourier modes. Light from the $g$-band is colored blue, $r$-band green, and $i$-band red. {\em Middle row, center and right:} Combined $gri$ images of the first and second modes alone. {\em Lower row, left to right:} Combined $gri$ images of the third, fourth, and fifth modes alone.}
471: \label{fig:sampleim}
472: \end{figure}
473: \clearpage
474:
475: Next, in Fig,~\ref{fig:mosaic}, we show twelve example late-type
476: galaxies along with their lopsidedness values in a progression from
477: symmetric to lopsided. The three galaxies
478: shown in the top row each exhibit a regular and symmetric appearance
479: and have low values of lopsidedness $A_1 \le 0.04$. The next 5
480: galaxies have moderate $A_1$ values of 0.08$-$0.13, and minor
481: distortions are visible in each case. In cases of two-arm spiral
482: galaxies, one or both arms are disturbed so that their shape and
483: brightness are not identical at equal distances from the galactic
484: center. In the case of SDSS J084845.62+001729.5 (third row, left), a
485: second arm on the right has no clear counterpart on the left,
486: and the lopsidedness of this galaxy is 0.11$-$0.12 depending on the
487: band. The final four galaxies shown have clear asymmetry in
488: brightness and shape and have $A_1 > 0.16$ in each band.
489:
490: \begin{figure}
491: \centering
492: \leavevmode
493: \columnwidth=.30\columnwidth
494: \begin{minipage}[b]{1.6in}
495: %\includegraphics[width=1.6in]{/home/tar/agn/text/lopmethod/mosaic/904-403-52381.ps}
496: \includegraphics[width=1.6in]{fig2a.eps}
497: %{\scriptsize J101620.36+542847.1
498: %
499: %$A_1 = 0.04, 0.04, 0.02$}
500: \end{minipage}
501: \hfill
502: \begin{minipage}[b]{1.6in}
503: %\includegraphics[width=1.6in]{/home/tar/agn/text/lopmethod/mosaic/844-571-52378.ps}
504: \includegraphics[width=1.6in]{fig2b.eps}
505: %{\scriptsize J121935.76+055048.3
506: %
507: %$A_1 = 0.04, 0.03, 0.03$}
508: \end{minipage}
509: \hfill
510: \begin{minipage}[b]{1.6in}
511: %\includegraphics[width=1.6in]{/home/tar/agn/text/lopmethod/mosaic/894-551-52615.ps}
512: \includegraphics[width=1.6in]{fig2c.eps}
513: %{\scriptsize J083104.39+393721.9
514: %
515: %$A_1 = 0.03, 0.04, 0.04$}
516: \end{minipage}
517: \hfill
518: \begin{minipage}[b]{1.6in}
519: %\includegraphics[width=1.6in]{/home/tar/agn/text/lopmethod/mosaic/1334-128-52764.ps}
520: \includegraphics[width=1.6in]{fig2d.eps}
521: %{\scriptsize J155547.64+423625.9
522: %
523: %$A_1 = 0.08, 0.07, 0.06$}
524: \end{minipage}
525: \hfill
526: \begin{minipage}[b]{1.6in}
527: %\includegraphics[width=1.6in]{/home/tar/agn/text/lopmethod/mosaic/1748-136-53112.ps}
528: \includegraphics[width=1.6in]{fig2e.eps}
529: %{\scriptsize J103705.4+124614
530: %
531: %$A_1 = 0.08, 0.09, 0.08$}
532: \end{minipage}
533: \hfill
534: \begin{minipage}[b]{1.6in}
535: %\includegraphics[width=1.6in]{/home/tar/agn/text/lopmethod/mosaic/724-215-52254.ps}
536: \includegraphics[width=1.6in]{fig2f.eps}
537: %{\scriptsize J225422.3-101025.8
538: %
539: %$A_1 = 0.11, 0.10, 0.10$}
540: \end{minipage}
541: \hfill
542: \begin{minipage}[b]{1.6in}
543: %\includegraphics[width=1.6in]{/home/tar/agn/text/lopmethod/mosaic/467-147-51901.ps}
544: \includegraphics[width=1.6in]{fig2g.eps}
545: %{\scriptsize J084845.62+001729.5
546: %
547: %$A_1 = 0.12, 0.12, 0.11$}
548: \end{minipage}
549: \hfill
550: \begin{minipage}[b]{1.6in}
551: %\includegraphics[width=1.6in]{/home/tar/agn/text/lopmethod/mosaic/469-212-51913.ps}
552: \includegraphics[width=1.6in]{fig2h.eps}
553: %{\scriptsize J085747.32-001159.9
554: %
555: %$A_1 = 0.13, 0.13, 0.13$}
556: \end{minipage}
557: \hfill
558: \begin{minipage}[b]{1.6in}
559: %\includegraphics[width=1.6in]{/home/tar/agn/text/lopmethod/mosaic/746-156-52238.ps}
560: \includegraphics[width=1.6in]{fig2i.eps}
561: %{\scriptsize J232923.56+141215.4
562: %
563: %$A_1 = 0.18, 0.16, 0.16$}
564: \end{minipage}
565: \hfill
566: \begin{minipage}[b]{1.6in}
567:
568: %\includegraphics[width=1.6in]{/home/tar/agn/text/lopmethod/mosaic/336-13-51999.ps}
569: \includegraphics[width=1.6in]{fig2j.eps}
570: %{\scriptsize J124944.67-025826.4
571: %
572: %$A_1 = 0.20, 0.20, 0.22$}
573: \end{minipage}
574: \hfill
575: \begin{minipage}[b]{1.6in}
576: %\includegraphics[width=1.6in]{/home/tar/agn/text/lopmethod/mosaic/608-567-52081.ps}
577: \includegraphics[width=1.6in]{fig2k.eps}
578: %{\scriptsize J144930.14+610654.2
579: %
580: %$A_1 = 0.27, 0.27, 0.25$}
581: \end{minipage}
582: \hfill
583: \begin{minipage}[b]{1.6in}
584: %\includegraphics[width=1.6in]{/home/tar/agn/text/lopmethod/mosaic/769-473-52282.ps}
585: \includegraphics[width=1.6in]{fig2l.eps}
586: %{\scriptsize J094706.75+541842.8
587: %
588:
589: %$A_1 = 0.26, 0.27, 0.26$}
590: \end{minipage}
591: \hfill
592: \caption{Twelve late-type SDSS galaxies shown with their $g$-, $r$-, and $i$-band lopsidedness. They are arranged by increasing lopsidedness toward the right in each row, least lopsided in the uppermost row and most lopsided in the lowermost row. Clear distortions in shape and/or brightness are visible in the galaxies with $A_1 > 0.08$. \label{fig:mosaic}}
593: \end{figure}
594:
595: \begin{deluxetable}{l|l|lrrr}
596: \tabletypesize{\small}
597: \tablecolumns{6}
598: \tablewidth{0pc}
599: \tablecaption{Example Galaxies in Figure~\ref{fig:mosaic}}
600: \tablehead{
601: \colhead{Row} &
602: \colhead{Column} &
603: \colhead{Name} &
604: \colhead{$A_1^g$} &
605: \colhead{$A_1^r$} &
606: \colhead{$A_1^i$} }
607: \startdata
608: & Left & J101620.36+542847.1 & 0.04 & 0.04 & 0.02 \\
609: 1st & Middle & J121935.76+055048.3 & 0.04 & 0.03 & 0.03 \\
610: & Right & J083104.39+393721.9 & 0.03 & 0.04 & 0.04 \\
611: \hline
612: & Left & J155547.64+423625.9 & 0.08 & 0.07 & 0.06 \\
613: 2nd & Middle & J103705.40+124614.0 & 0.08 & 0.09 & 0.08 \\
614: & Right & J225422.30$-$101025.8 & 0.11 & 0.10 & 0.10 \\
615: \hline
616: & Left & J084845.62+001729.5 & 0.12 & 0.12 & 0.11 \\
617: 3rd & Middle & J085747.32$-$001159.9 & 0.13 & 0.13 & 0.13 \\
618: & Right & J232923.56+141215.4 & 0.18 & 0.16 & 0.16 \\
619: \hline
620: & Left & J124944.67$-$025826.4 & 0.20 & 0.20 & 0.22 \\
621: 4th & Middle & J144930.14+610654.2 & 0.27 & 0.27 & 0.25 \\
622: & Right & J094706.75+541842.8 & 0.26 & 0.27 & 0.26
623: \enddata
624: \label{tab:mosaic}
625: \end{deluxetable}
626:
627:
628:
629: \section{Final Sample Selection: The Effect of Systematic Errors \label{sec:syserr}}
630:
631: Lopsidedness is ideally measured from an image of a bright, well
632: resolved, and face-on galaxy without any overlapping background or
633: foreground sources. In such an image, the Poisson noise of the
634: detected light is negligible, and the size of the point-spread
635: function is small compared to the characteristic size scale of the
636: lopsidedness. Neither dust extinction nor inclination would
637: significantly alter the light distribution. In real data, these
638: conditions are not all met, and the resulting effects can
639: systematically change the measurements of the Fourier modes.
640:
641: Below we address in turn how various sources of systematic error
642: affect the strengths of Fourier modes. We then use this information to refine our sample selection and define a final sample of galaxies whose images are only negligibly affected by these systematic errors. These will form the basis of our scientific analysis.
643:
644: \subsection{Ellipticity and Inclination \label{sec:inclination}}
645:
646: We employ a circular polar grid to bin light from each galaxy, even if
647: the galaxy appears elliptical and/or inclined on the image. The
648: measured lopsidedness of elliptically projected galaxies can be
649: underestimated by using the circular grid if the deprojected bright side of the
650: galaxy coincides with the {\em minor} axis of the galaxy. This occurs
651: because the increment and decrement of light along the minor axis due
652: to lopsidedness is compared to the average light along a circular ring
653: that also intersects the brighter, inner region at the major axis.
654: Similarly, the lopsidedness will be overestimated if the bright and dim
655: sides of the galaxy are aligned with the {\em major} axis. The
656: resulting systematic error is diminished if the phase angle of the
657: first Fourier mode wraps at least a quarter of the way around the
658: galaxy. In this case, the first mode strength would be overestimated
659: at some radii and underestimated at other radii, and the radially
660: averaged $A_1$ would have a smaller systematic error.
661:
662: Our sample has galaxies imaged in random orientations, and the over-
663: and under-estimates of lopsidedness average to cancel each other
664: except for highly inclined galaxies. We show this result in
665: Fig.~\ref{fig:lopba}, where we have plotted the distribution of
666: $A_1^i$ (and also of $A_2^i$) as a function of $b/a$ as measured in
667: the $i$-band (similar results are seen in the $g$ and $r$-bands).
668:
669: The relationship between $b/a$ and $A_2^i$ is clear. Face-on galaxies
670: have a small second mode, typically 0.1-0.3, while highly inclined
671: galaxies have a much higher second mode that often exceeds unity for
672: $b/a < 0.4$. Though not shown, higher-order even modes show a similar
673: trend with $b/a$ as $A_2^i$ but with weaker magnitude.
674:
675: Lopsidedness exhibits an increase at low $b/a$ but is mainly
676: independent of $b/a$ for $b/a > 0.4$. The rise of observed
677: lopsidedness for increasingly inclined galaxies arises for a few
678: reasons. First, dust lanes appear more optically thick when a galaxy
679: is viewed nearly edge-on. A lopsided, edge-on distribution of dust may
680: obscure the light in a symmetric galaxy to make the galaxy light
681: appear lopsided. Second, the systematic error in using a circular
682: grid for an elliptically projected galaxy becomes more severe at low
683: $b/a$ ratio. Nonetheless, $A_1^i$ shows little dependence on $b/a$
684: for $b/a > 0.4$, so our use of a circular polar grid and a
685: finite-order transform on slightly elongated galaxies produces no
686: significant systematic error when in reference to a large population
687: of galaxies. The distributions of $b/a$ as measured in the $g$-, $r$-,
688: and $i$-bands are given in the lower panel of Fig.~\ref{fig:lopba}. A
689: cut at $b/a = 0.4$ in all three bands eliminates the 22\% of the
690: sample, and 52194 galaxies are retained. This cut on inclination is very similar to that adopted by \citet{bc+05}.
691:
692:
693: \begin{figure}
694: \epsscale{1.1}
695: %\plottwo{/home/tar/agn/asymm/lophists/meddist_baratio_a1i.ps}{/home/tar/agn/asymm/lophists/meddist_baratio_a2i.ps}
696: \plottwo{fig3a.eps}{fig3b.eps}
697: %\plottwo{/home/tar/agn/asymm/lophists/meddist_baratio_a3i.ps}{/home/tar/agn/asymm/lophists/meddist_baratio_a4i.ps}
698: %\plottwo{/home/tar/agn/asymm/lophists/meddist_baratio_a5i.ps}{/home/tar/agn/asymm/lophists/meddist_baratiogri_hist.ps}
699: \epsscale{0.5}
700: %\plotone{/home/tar/agn/asymm/lophists/banddist_baratio.ps}
701:
702: \plotone{fig3c.eps}
703: \caption{ {\em (Top two panels)} Distributions of the Fourier modes $A_1$ and $A_2$ as functions of $b/a$ in the $i$-band. The 5th, 25th, 50th, 75th, and 95th percentiles are shown. Lopsidedness is independent of inclination for $b/a > 0.4$. On the other hand, $A_2$, an indicator of ellipticity, inclination, and two-arm spiral patterns, decreases significantly with $b/a$ along the full range of $b/a$. {\em (Lower right)} The similar distributions of $b/a$ as measured in the $g$- {\em (green)}, $r$- {\em (orange)}, and $i$-bands {\em (red)}. A three-color cut at $b/a = 0.4$ eliminates only a small portion of the sample.}
704: \label{fig:lopba}
705: \end{figure}
706: \clearpage
707:
708:
709: \subsection{Spatial Resolution \label{sec:blur}}
710:
711: Lopsided galaxies systematically appear more symmetric when observed
712: in conditions of poor seeing. Light is smeared from the brighter
713: regions of the galaxy into surrounding, dimmer regions, reducing the
714: contrast of the brighter and dimmer sides of the galaxy. The effect
715: is more pronounced when the point-spread function is wide compared to
716: the physical size of the galaxy. It is useful to define a seeing
717: resolution
718: \begin{equation}
719: S_b \equiv \frac{2R_{50,b}}{{\rm PSF\,\; FWHM}_b}
720: \end{equation}
721: as the relative size of the galaxy compared to the FWHM of the PSF in
722: band $b$.
723:
724: To demonstrate how poor seeing reduces lopsidedness, we
725: have selected from our galaxy sample a subsample of 13500 isolated,
726: well resolved galaxies with $b/a > 0.5$ and $S_i > 5.0$ and convolved
727: the $i$-band images with a circular Gaussian PSF of 20 varying widths.
728: We calculated $A_1^i$ of each galaxy at each level of blurring. The
729: subsample was binned by $A_1^i$ of the original (not blurred) images.
730: Fig.~\ref{fig:psfblur} shows how the median $A_1^i$ of the blurred
731: images changes with seeing resolution in each bin of the original
732: $A_1^i$. Post-blurring $A_1^i$ exhibits little change for highly
733: resolved $S > 5$ galaxies. At moderate resolutions $2 < S < 5$,
734: lopsidedness decreases with the seeing resolution. Barely resolved
735: galaxies with $S < 2$ approach $A_1^i \sim 0.05$ regardless of the
736: lopsidedness at high resolution.
737:
738: \begin{figure}[ht]
739: \epsscale{1.0}
740: %\plotone{/home/tar/agn/asymm/degrade.hist_res_a1bins.ps}
741: \plotone{fig4.eps}
742: \caption{The $i$-band lopsidedness of galaxies in images simulating various degrees of seeing resolution. A sample of isolated, well-resolved galaxies was divided into bins by $A_1$ (measured before blurring) and the images were then blurred. For each bin, the relation between median $A_1$ after blurring and the ratio of the half-light and blurred PSF diameters ($S$) is shown above. Lopsided and symmetric galaxies can be distinguished for $S > 3$ but the range of $A_1$ is diminished for $S \lesssim 4$. }
743: \label{fig:psfblur}
744: \end{figure}
745: \clearpage
746:
747: The same effect is seen in our primary sample of galaxies without
748: convolving the images with a point-spread function.
749: Fig.~\ref{fig:psfdist} shows the distribution of $A_1^i$ at different
750: seeing resolutions along with the distributions of $S_g$, $S_r$, and
751: $S_i$ for the whole sample (before the inclination cut is made). At
752: low seeing resolution $S < 3$, the sample distribution of $A_1^i$
753: shifts to lower lopsidedness (0.07 median), and few galaxies have
754: $A_1^i > 0.2$. At high resolution $S > 7$, the distribution has a
755: median that levels off around 0.12, and high lopsidedness $A_1^i >
756: 0.20$ is found in 1 of every 4 galaxies. The 5th percentile increases
757: with resolution to $A_1^i = 0.05$ at $S = 7$.
758:
759: Our simulations of poor seeing suggest that a distinction between
760: symmetric and lopsided galaxies can be identified by the $A_1$ measure
761: even if $A_1$ is diminished by moderately poor seeing. The relative
762: lopsidedness of well-resolved galaxies is preserved as the seeing is
763: worsened down to $S \sim 3$, i.e., the most symmetric galaxies with $3
764: < S < 7$ are likely to also be the most symmetric galaxies if they
765: were observed instead at $S > 7$, and likewise for the more lopsided
766: galaxies. However, the extremely lopsided galaxies may not be able to
767: identified without higher resolution, perhaps $S > 10$. Trends at
768: extreme lopsidedness $A_1 > 0.3$ may not be reliable because only well
769: resolved galaxies can lie in this lopsidedness range. At $S = 4$,
770: seeing reduces the lopsidedness of 75\% of galaxies by less than a
771: factor of 1.5. We cut our sample at this seeing value and keep only
772: galaxies better resolved than $S = 4$ in the $g$, $r$, and $i$ bands.
773: The cut alone reduces the sample size by $\sim 37\%$ to 42558
774: galaxies. The more lopsided galaxies $A_1 > 0.2$ are underrepresented
775: due to this systematic effect.
776:
777:
778:
779: %Because
780: %information about lopsidedness is not lost for moderately resolved
781: %galaxies, we can attempt a correction to $A_1$. In
782: %Fig.~\ref{fig:blurcorr}, we show, as a function of $S$, the ratio of
783: %the observed lopsidedness after blurring to the average lopsidedness of
784: %simulations of the same galaxy with $S > 8$. While there is some
785: %scatter in the lopsidedness ratio, we can use the median ratio as a
786: %correction to $A_1$. [**I'm presenting the correction for now. I
787: %haven't used it anywhere else yet. Do I even need to show it?]
788:
789:
790:
791: \begin{figure}[ht]
792: \epsscale{1.0}
793: %\plottwo{/home/tar/agn/asymm/lophists/meddist_a1i_resi.ps}{/home/tar/agn/asymm/lophists/banddist_res.ps}
794: \plottwo{fig5a.eps}{fig5b.eps}
795: \caption{{\em Left:} Distribution of $i$-band lopsidedness for galaxies observed at different seeing resolution. The 5th, 25th, 50th, 75th, and 95th percentiles are shown. {\em Right:} Distribution of seeing resolution as measured in the $g$-band {\em (green)}, $r$-band {\em (orange)}, and $i$-band {\em (red)}. Blurring from poor seeing systematically reduces the measured lopsidedness. A cut in seeing resolution at $S = 4$ removes 37\% of the sample while allowing a distinction between symmetric and asymmetric galaxies.}
796: \label{fig:psfdist}
797: \end{figure}
798: \clearpage
799:
800: %\begin{figure}[ht]
801: %\epsscale{1.0}
802: %\plotone{/home/tar/agn/asymm/degrade.correction.ps}
803: %\caption{The correction factor for moderately resolved galaxies. The vertical axis shows the ratio of the lopsidedness of a blurred galaxy to the lopsidedness the same galaxy has when less blurring is applied ($S > 8$).}
804: %\label{fig:psfdist}
805: %\end{figure}
806: %\clearpage
807:
808:
809:
810:
811:
812: \subsection{Flat-fielding error}
813:
814: Poor flat-fielding in a galaxy image can increase the apparent
815: lopsidedness of the galaxy. Poor correction of the CCD's spatial
816: gradients in sensitivity can introduce a brighter sky level on one
817: side of the galaxy and a dimmer sky level on the other side. Because
818: a constant sky value is subtracted from the image in the lopsidedness
819: computation, the gradient in sky brightness will increase the strength
820: of the odd-order Fourier modes and overestimate the galaxy's true
821: lopsidedness. We have calculated flat-field errors in our images by
822: calculating the sky level in regions in each of the four corners of
823: the image and adopting the standard deviation of those sky levels as
824: our flat-fielding error. We find that flat-fielding errors are
825: typically negligibly small, $\sim 1$\% of the sky value, and only
826: become significant at large radii from a galactic center. In our
827: calculation of $A_m$, the contribution of the modes at large radii is
828: weighted less, and so flat-fielding errors do not significantly affect
829: the lopsidedness values that we have computed. We therefore do not
830: impose a cut based on flat-fielding error for our SDSS-detected
831: galaxies.
832:
833: \subsection{Random Error \label{sec:randerr}}
834:
835:
836: Noisy data from low-surface-brightness galaxies can bias measurements
837: of lopsidedness in a systematic way. The uncertainty in the correct
838: light-weighted center will be higher in a dim galaxy than in a bright
839: one, and a shifted center leads to an overestimate of the
840: lopsidedness.
841:
842:
843: We have undertaken two tests to determine the effect of noise on our
844: measurements. First, we have compared the values of $A_1$ measured in
845: the $r$- and $i$-bands. Lopsidedness values have been
846: determined before to be mainly independent of wavelength in this range
847: \citep{rr98}, and we will confirm this result for our sample as a
848: whole in \S\ref{sec:lightmass}. In the left panel of Fig.~\ref{fig:lopri},
849: we show the discrepancies between $A_1$ in the two bands as a rough
850: indicator of random error on the $A_1$ measurement in general. Here
851: the $S/N$ ratio is computed for light within the $R_{50}$-to-$R_{90}$
852: annulus in which $A_1$ is computed. The interquartile (25th-75th
853: percentile) range is small at high $S/N$ ($\lesssim 0.04$ at $S/N =
854: 300$) and gradually increases at low $S/N$ as the distribution spreads
855: out. At $S/N = 30$, the interquartile range has doubled to $0.09$, and the
856: median is skewed only slightly toward negative values of $A_1^r-A_1^i
857: \sim -0.01$.
858:
859: We have applied another cut to our sample to remove dim, noisy
860: galaxies with $S/N < 30$ in any of the three $gri$ bands. The right
861: panel of Fig.~\ref{fig:lopri} shows the distribution of $S/N$ in three
862: bands. The cut removes a 5\% of the sample. The sample retains 25155
863: galaxies after this cut and also the inclination and seeing cuts have
864: been made.
865:
866:
867: \begin{figure}[ht]
868: \epsscale{1.0}
869: %\plottwo{/home/tar/agn/asymm/lophists/meddist_sni_a1ri.ps}{/home/tar/agn/asymm/lophists/banddist_sn.ps}
870: \plottwo{fig6a.eps}{fig6b.eps}
871: \epsscale{1.1}
872: \caption{{\it (Left)} Distributions of the difference in lopsidedness in the $r$ and $i$ bands as a function of $i$-band S/N. The 5th, 25th, 50th, 75th, and 95th percentiles are shown. The difference is typically small ($|A_1^r-A_1^i| < 0.04$) for most of the sample with $S/N_i > 30$. Since lopsidedness is systematically similar in these two bands, random errors in $A_1^r$ and $A_1^i$ should be of the similar size as the difference $|A_1^r-A_1^i|$. {\it (Right)} The distribution of $S/N$ in the $g$- (green), $r$- (orange), and $i$-band (red). A cut at $S/N < 30$ in all three bands eliminates a small fraction of the sample. }
873: \label{fig:lopri}
874: \end{figure}
875: \clearpage
876:
877:
878:
879:
880:
881:
882: To further evaluate the effect of noise in calculating lopsidedness,
883: we looked to the full SDSS DR4 dataset to retrieve a $z < 0.06$ sample
884: of galaxies that have been observed 3 or more times. We narrowed this
885: sample down to a smaller sample where inclination, seeing resolution,
886: and $S/N$ met the same cuts as our main sample ($b/a > 0.4$, $S > 4$,
887: and $S/N (R_{50} < r < R_{90}) > 30$). We also required that the
888: fractional RMS variations between different observations in seeing
889: resolution and $S/N$ were at most 10\%. This latter requirement
890: ensures that the repeated observations had similar observing conditions.
891: The resulting sample contained 328 galaxies with repeated
892: observations. We calculated the RMS differences in $A_1$ in the
893: repeated observations of these galaxies and adopted this measure as
894: the error $\delta A_1$ in lopsidedness for these galaxies.
895:
896: Fig.~\ref{fig:randerr} shows the distribution of $\delta A_1$
897: vs. $A_1$ for the repeatedly observed galaxies. About a quarter of
898: the galaxies show small errors in lopsidedness, $\delta A_1 < 0.01$,
899: and median errors are typically $\sim 0.1A_1$. The largest errors are
900: $\sim 0.5A_1$ but affect only 5\% of the sample. The third quartile
901: rises as $\sim 0.2A_1$ but can be as high as 0.02 at lopsidedness as
902: low as 0.05. Typical errors can be expected to be the larger of 0.02
903: and $0.10A_1$. The majority of the sample has $A_1 < 0.20$, and so
904: 0.02 can be taken as the typical error for most of the sample, with
905: larger errors for the most lopsided galaxies. This result is
906: consistent with our estimate above based on comparing the $r$-band and
907: $i$-band values for a much larger sample.
908:
909: \begin{figure}[ht]
910: \epsscale{1.0}
911: %\plotone{/home/tar/agn/asymm/lophists/reobs_lopc_dlop.ps}
912: \plotone{fig7.eps}
913: \epsscale{1.1}
914: %\plottwo{/home/tar/agn/asymm/lophists/reobs_a2_da2.ps}{/home/tar/agn/asymm/lophists/reobs_a3_da3.ps}
915: %\plottwo{/home/tar/agn/asymm/lophists/reobs_a4_da4.ps}{/home/tar/agn/asymm/lophists/reobs_a5_da5.ps}
916: \caption{Errors in lopsidedness for galaxies with repeated observations. The 25th, 50th, and 75th percentiles are over-plotted. Typical lopsidedness errors are $\sim 0.02$ for the more symmetric galaxies and $\sim 0.1A_1$ for the more lopsided galaxies. }
917: \label{fig:randerr}
918:
919: \end{figure}
920: \clearpage
921:
922:
923:
924:
925:
926: \subsection{A Sample Suitable for Calculating Lopsidedness \label{sec:cuts}}
927:
928: We have discussed the systematic errors that skew our calculation of
929: lopsidedness. Here we summarize the proposed cuts to weed out cases
930: where the systematic effects give unphysical values of $A_1$. After
931: applying these cuts, we are left with a significant sample that is
932: suitable for studying correlations between lopsidedness and other
933: global properties of galaxies.
934:
935: \begin{itemize}
936:
937: %\item {\em Overlapping galaxies.} We eliminate all galaxies in our sample that overlap another galaxy brighter than 6 apparent magnitudes fainter than our sample galaxy. Galaxies with only faint overlapping galaxies and isolated galaxies remain in the sample.
938:
939: \item {\em Inclined galaxies.} We eliminate all galaxies
940: with $b/a < 0.4$ in any of the three bands to remove inaccurate $A_1$
941: values computed using a circular aperture on an elliptically projected
942: galaxy. This cut also removes cases of obscuration from optically thick
943: dust lanes on edge-on galaxies.
944:
945: \item {\em Poorly resolved galaxies.} We eliminate all galaxies
946: with seeing resolution $S < 4$ in any of the three bands in order to
947: remove galaxies whose lopsidedness is diminished due to poor seeing.
948:
949: \item {\em Dim galaxies.} We eliminate all galaxies with $S/N < 30$
950: in the $R_{50}$-to-$R_{90}$ annulus (the region where $A_1$ is
951: calculated) in any of the three bands to remove cases where
952: lopsidedness is augmented by Poisson noise and poor centering.
953:
954: \end{itemize}
955:
956: Table~\ref{tab:cuts} shows the census of galaxies retained after each
957: cut is applied separately. The resolution cut alone removes the
958: largest portion of the sample, and the noise cut removes the least
959: amount. After all three cuts are applied to the initial sample of
960: 67107 galaxies, 25155 (37\%) are retained. Unless otherwise specified,
961: we have employed these cuts in our sample for all analysis presented
962: below. These cuts depend on parameters linked to the observation of
963: the galaxies but may also depend on physical properties. We next look
964: at the structural properties of the sample and compare them to the
965: larger DR4 sample from which it was drawn.
966:
967: \begin{deluxetable}{lrr}
968: \tabletypesize{\small}
969:
970: \tablecolumns{3}
971: \tablewidth{0pc}
972: \tablecaption{Sample Cuts}
973: \tablehead{
974: \colhead{} &
975: \colhead{Percentage} &
976: \colhead{Galaxies} \\
977: \colhead{Cut} &
978: \colhead{Retained} &
979: \colhead{Retained} }
980: \startdata
981: Initial Sample & 100\% & 67107 \\
982: $b/a > 0.4$ & 78\% & 52194 \\
983: $S > 4$ & 63\% & 42558 \\
984: $S/N (R_{50} < r < R_{90}) > 30$ & 95\% & 63434 \\
985: Final Sample after all 3 cuts & 37\% & 25155 \\
986: \enddata
987: \label{tab:cuts}
988: \end{deluxetable}
989:
990:
991: The main structural parameters we will utilize in the analysis below
992: are the stellar mass $M_*$, the effective stellar surface mass density
993: $\mu_*$ (the mean stellar density interior to the z-band half light
994: radius), and the concentration $C_i$ (defined as the ratio of
995: $R_{90}/R_{50}$ in the i-band). See \citet{khw+03a} for a detailed
996: description.
997:
998: In Fig.~\ref{fig:lopsamplecut}, we show the distributions of these
999: structural parameters $M_*$, $\mu_*$, and $C_i$ after we apply each
1000: cut in succession to the sample. The upper curve in each panel shows
1001: the distribution from the initial $z < 0.06$ DR4 sample. Then the
1002: inclination cut is applied (2nd distribution from the top), followed
1003: by the resolution cut (3rd distribution from the top) and the $S/N$
1004: cut (lower distribution). The inclination cut reduces the galaxy
1005: counts without any strong correlation with structural properties. The
1006: resolution cut rejects more massive, high-mass-density, and
1007: concentrated galaxies, leaving the sample with a preference of
1008: late-type galaxies. Finally, the $S/N$ cut causes little change in
1009: the relative proportions of massive, high-mass-density, or
1010: concentrated galaxies.
1011:
1012: The final sample contains 25155 galaxies spanning 3 orders of
1013: magnitude in stellar mass ($10^8-10^{11} \MSun$), 3 orders of
1014: magnitude of stellar mass density ($10^{6.5}-10^{9.5} \MSunkpc$), and
1015: a wide range of $i$-band concentration ($1.5-3.5$), with a larger
1016: proportion of late-type galaxies than early-type galaxies.
1017:
1018:
1019: \begin{figure}[ht]
1020: \epsscale{1.1}
1021: %\plottwo{/home/tar/agn/asymm/lophists/banddist_m_cuts.ps}{/home/tar/agn/asymm/lophists/banddist_mu_cuts.ps}
1022: \plottwo{fig8a.eps}{fig8b.eps}
1023: \epsscale{0.5}
1024: %\plotone{/home/tar/agn/asymm/lophists/banddist_conci_cuts.ps}
1025: \plotone{fig8c.eps}
1026: \caption{Distributions of the structural parameters $C_i$, $M_*$, and $\mu_*$ as the observational cuts were applied. From top to bottom, in each panel: full DR4 $z<0.06$ sample before cuts were applied; $b/a > 0.4$ cut is applied; $S > 4$ cut is applied; $S/N > 30$ cut is applied. The resulting sample has similar numbers of low- and high-mass galaxies, low- and high-stellar-density galaxies, but more a larger proportion of late-type galaxies than early-type galaxies.}
1027: \label{fig:lopsamplecut}
1028: \end{figure}
1029: \clearpage
1030:
1031:
1032:
1033:
1034:
1035:
1036:
1037:
1038:
1039: \section{Properties of Lopsidedness \label{sec:lopprops}}
1040:
1041:
1042:
1043: \subsection{Light vs. Mass Distributions \label{sec:lightmass}}
1044:
1045: The Fourier modes describe the two-dimensional {\em light}
1046: distribution of galaxies. We wish to link these modes, specifically
1047: the lopsidedness, with a description of the two-dimensional {\em mass}
1048: distribution as seen along the same line of sight. If the contrast
1049: between the bright and dim halves of a galaxy is significantly different in the
1050: $g$-band than in the $i$-band, the lopsidedness measure may be
1051: indicating an asymmetry in mass-to-light ratios from asymmetrically distributed star formation and/or dust extinction. On the other hand, similar values in
1052: the two bands would suggest similar mass-to-light ratios and a
1053: corresponding lopsidedness in surface mass density. To see which scenario is
1054: prevalent, we can look at the colors and magnitude differences between
1055: galaxies of different star formation histories and hence mass/light ratios. Then we can compare those differences to
1056: those of the bright and dim halves of the lopsided galaxies.
1057:
1058:
1059: We start with a set of pairs of stellar population models. We have
1060: taken 32000 simulations of stellar populations with identical mass and
1061: varying star formation histories from \citet{khw+03a} and randomly
1062: paired them to look at their relative colors. The models were
1063: generated from a wide range of superimposed continuous and bursty star
1064: formation histories with varying metallicity, such that bursty and
1065: continuously star-forming models each contribute about half of the
1066: models. We plot in the upper left panel of Fig.~\ref{fig:relcolor}
1067: the distributions of the difference in color $\Delta(g-i)$ and in
1068: magnitude $\Delta i$ within the pairs. We find a tight relation
1069: between the difference in brightness and color within the pair
1070: ($\Delta(g-i) = 0.45 \Delta i$ in the median).
1071:
1072:
1073: We next perform the same comparison using 113000 pairs of low-redshift
1074: ($z < 0.06$) observed galaxies. The pairs were selected from the SDSS
1075: DR4 galaxy sample and have been matched in redshift with $\Delta z <
1076: 0.001$ and stellar mass with $\Delta\log_{10} M_* < 0.01$. The
1077: relation between relative color and magnitude is shown in the upper
1078: right panel of Fig.~\ref{fig:relcolor}. The brighter galaxy in the
1079: pair is the bluer galaxy, in agreement with the models. However the
1080: relation is somewhat shallower ($\Delta(g-i) = 0.29 \Delta i$ in the
1081: median). The difference in these relations may be due to a wider range
1082: of SFHs, especially the more extreme, bursty SFHs, in the models than
1083: in the observed galaxies. It may also reflect differences in the
1084: effects of dust extinction and reddening in the models vs. the data.
1085:
1086: We now compare these paired color-magnitude relations of the galaxies and the models to that determined from the bright and dim sides of the galaxies in our lopsided galaxy sample. Having determined the position angle that maximized the light asymmetry, we then measured the $g$, $r$, and $i$ magnitudes for each galaxy half. Stars and intervening galaxies were masked out
1087: and excluded, as before, but for each masked
1088: pixel here, the corresponding pixel at the same radius and $180^\circ$
1089: away is also excluded. Light is thus summed over the same area in
1090: each half.
1091:
1092: We show in the lower panel of Fig.~\ref{fig:relcolor} the difference
1093: in color $\Delta (g-i)$ between the contrasting halves of galaxies in
1094: the DR4 sample as a function of their difference in $i$ magnitude. The
1095: median color difference is zero and has little dependence on relative brightness. This is completely different from the behavior of the models or real galaxies as the mass/light ratio is varied, and suggests that the lopsidedness is not due to variations in mass/light ratio. In fact, \citet{k+07} have used SDSS galaxy spectra to show that the $(g-i)$ fiber color is an excellent proxy for the stellar mass/light ratio. The fact that there is no systematic difference in $(g-i)$ color between the brighter and dimmer sides of the lopsided galaxies then implies that the lopsided light distribution is primarily tracing a lopsided stellar mass distribution.
1096:
1097: While there is no systematic offset in color with magnitude, we note that the spread in $\Delta (g-i)$ increases as $\Delta i$ increases. This implies that there are large-scale spatial variations in the mass/light ratio in lopsided galaxies (presumably due to enhanced star formation and/or dust extinction).
1098:
1099: \begin{figure}[ht]
1100: \epsscale{1.1}
1101: %\plottwo{/home/tar/agn/asymm/plothalfmag/meddist_avgdi_avgdgi.ps}{/home/tar/agn/asymm/plothalfmag/meddist_rpdi_rpdgi.ps}
1102: \plottwo{fig9a.eps}{fig9b.eps}
1103: \epsscale{0.5}
1104: %\plotone{/home/tar/agn/asymm/plothalfmag/meddist_actdi_actdgi.ps}
1105: \plotone{fig9c.eps}
1106: \caption{Differences in $g - i$ color and $i$ magnitude for pairs of model stellar populations {\em (upper left)}, for pairs of observed galaxies {\em (upper right)}, and for bright and dim halves of observed galaxies {\em (lower panel)}. The pairs of stellar populations and observed galaxies have random star formation histories and exhibit different mass-to-light ratios in the $g$ and $i$ bands. The contrasting halves of observed galaxies instead show no correlation between color and magnitude differences, implying similar average mass-to-light ratios in each half.}
1107: \label{fig:relcolor}
1108:
1109: \end{figure}
1110: \clearpage
1111:
1112: %\begin{figure}[ht]
1113: %\epsscale{1.1}
1114: %\plottwo{/home/tar/agn/asymm/plothalfmag/rp_di_dgi_count.ps}{/home/tar/agn/asymm/plothalfmag/deltagi_deltai_count.ps}
1115: %\plottwo{/home/tar/agn/asymm/plothalfmag/rp_dr_dgr_count.ps}{/home/tar/agn/asymm/plothalfmag/deltagr_deltar_count.ps}
1116: %\plottwo{/home/tar/agn/asymm/plothalfmag/rp_di_dri_count.ps}{/home/tar/agn/asymm/plothalfmag/deltari_deltai_count.ps}
1117: %\caption{{\em Left panels:} Distributions of $\Delta $color vs. $\Delta$ apparent magnitude for pairs of real galaxies matched in stellar mass and redshift. {\em Right panels:} Same distribution but for halves of galaxies in the sample of this paper.}
1118: %\label{fig:2drelcolor}
1119: %\end{figure}
1120: %\clearpage
1121:
1122: %Show how lopsidedness and the higher modes change with radius in galaxies
1123: %and give the best choice for a radially averaged measurement of
1124: %lopsidedness. The radial profiles should show how symmetric the inner
1125: %region is, due in part to the PSF smearing of the center and the
1126: %inherent symmetry therein.
1127:
1128: %Plot: A few radial profiles, indicating $\sigma_{psf}, R_{50}, R_{90}$
1129: %(or other characteristic radii), to show how galaxies tend to have
1130: %symmetric centers even though PSF blurring symmetrizes the centers.
1131: %Perhaps overplot the $a_1$ profiles in all 5 bands and draw
1132: %conclusions in the next section.
1133:
1134:
1135: %Plot: similar plots for higher modes, in particular, for $a_2$, but
1136: %with less emphasis than for $a_1$.
1137:
1138: %Plot: compare lopsidedness in inner and outer regions, i.e.,
1139: %$A_1^{inner}$ vs. $A_1^{outer}$. Determine appropriate inner and
1140: %outer radii between the symmetric central region and the low-$S/N$
1141: %outer regions (now using R50 to R90)
1142:
1143: %Plot: a few objects' images (with masked stars and superimposed polar
1144: %grid), the images reconstructed from Fourier modes, and the residual
1145: %image. Residuals may show knots that require higher Fourier modes to
1146: %model well.
1147:
1148: %Show how lopsidedness varies between the different bands. Devise a
1149: %final lopsidedness measure from the band-dependent lopsidedness values?
1150: %Or pick out the band whose lopsidedness is most representative.
1151:
1152: As another way of addressing the importance of variations in the the mass/light ratio in causing lopsidedness, we have compared the distribution of $A_1$ in three different bands in Fig.~\ref{fig:modesgri}. The distribution measured from $g$-, $r$-, and $i$-band images are shown in green, orange, and red, respectively.
1153:
1154: The distributions of $A_1^r$
1155: and $A_1^i$ are nearly identical, but the $A_1^g$ distribution is
1156: skewed toward slightly higher values. We have already
1157: shown that the lopsidedness of the light distribution primarily traces
1158: the lopsidedness of the underlying mass distribution. Here we see the
1159: weaker, secondary effect. Newly formed stars are not uniformly
1160: distributed, and so the lopsidedness of the light distribution
1161: includes a small contribution from the lopsidedness of the
1162: distribution of star formation.
1163:
1164:
1165: \begin{figure}[ht]
1166: \epsscale{1.1}
1167: %\plottwo{/home/tar/agn/asymm/lophists/banddist_a1.ps}{/home/tar/agn/asymm/lophists/banddist_a2.ps}
1168: \plottwo{fig10a.eps}{fig10b.eps}
1169: \caption{Distributions of the first and second azimuthal Fourier modes in the $g-$ (green), $r-$ (orange), and $i-$ (red) bands between $R_{50}$ and $R_{90}$. The $r$- and $i$-band distributions are similar. The $g$-band distribution is also similar but is skewed toward slightly higher values of lopsidedness. The lopsidedness of the light distributions is mainly tied to a lopsidedness in the mass distribution, but there is a small contribution from lopsidedness in the distribution of star formation.}
1170: \label{fig:modesgri}
1171: \end{figure}
1172: \clearpage
1173:
1174: \subsection{Radial Dependence of Lopsidedness \label{sec:raddep}}
1175:
1176: Previous studies have shown that asymmetry in galaxies shows a clear
1177: radial dependence. \citet{rr98} used a sample of 54 face-on,
1178: early-type disk galaxies and found that an increase in $A_1$
1179: lopsidedness with radius was normal. \citet{cbj00} studied radial
1180: profiles of both elliptical and disk galaxies in a sample of 113 from
1181: the \citet{fgg+96} sample. They found that asymmetry typically
1182: increases at larger radii $r > R_{50}$ in disk galaxies but peaks at
1183: smaller radii in elliptical and lenticular galaxies. One would expect
1184: our larger sample of galaxies to confirm the dependence of
1185: lopsidedness on both radius and Hubble type. In what follows, we use
1186: concentration as a measure of Hubble type.
1187:
1188:
1189:
1190: \begin{figure}[ht]
1191: \epsscale{1.1}
1192: %\plottwo{/home/tar/agn/asymm/lophists/prof_r_a1_lowa1.ps}{/home/tar/agn/asymm/lophists/prof_r_a1_mida1.ps}
1193: \plottwo{fig11a.eps}{fig11b.eps}
1194: %\plottwo{/home/tar/agn/asymm/lophists/prof_r_a1_hia1.ps}{/home/tar/agn/asymm/lophists/prof_r_a1_etg.ps}
1195: \plottwo{fig11c.eps}{fig11d.eps}
1196: \caption{Radial profiles of lopsidedness shown as distributions of $a_1^i$ as a function of radius. The first three panels show distributions for symmetric, average, and lopsided late-type galaxies ($C_i < 2.6$), and the lower right panel shows the distribution for all early-type galaxies ($C_i > 2.6$). The 5th, 25th, 50th, 75th, and 95th percentiles are shown. Lopsidedness increases with radius in symmetric and lopsided, late-type and early-type galaxies alike.}
1197: \label{fig:lopradprof}
1198: \end{figure}
1199: \clearpage
1200:
1201: In Fig~\ref{fig:lopradprof}, we show radial profiles as the
1202: distribution of the first Fourier mode (lowercase) $a_1^i(r)$ at radii
1203: extending outward to $R_{90}$. The first three panels show profiles
1204: for late-type galaxies ($C_i < 2.6$) partitioned into bins of global
1205: lopsidedness (capitalized $A_1^i$), and the last panel shows
1206: early-type ($C_i > 2.6$) galaxy profile. $R_{50}$ corresponds to
1207: roughly $0.38R_{90}$ to $0.60R_{90}$ for these late-type galaxies. In
1208: all these profiles, lopsidedness increases steadily at radii larger
1209: than $\sim0.5R_{90}$. Moderately and highly lopsided ($A_1^i > 0.08$)
1210: late-type galaxies also show a gradual increase in lopsidedness with
1211: radius at smaller radii. The early-type and more symmetric late-type
1212: galaxies show a minor decline in lopsidedness at small radii and an
1213: increase at larger radii. The decline at small radii may be an effect
1214: originating from minor centering errors near the galactic center,
1215: causing a small overestimation of $a_1^i (r)$ at the smallest radii.
1216: This effect diminishes with radius. At large radii, the increase in
1217: $a_1^i (r)$ is tied to a real increase in light asymmetry, though
1218: there could also be some unphysical enhancement due to the low $S/N$.
1219:
1220:
1221:
1222:
1223: %\subsection{Catalog of Most Lopsided Galaxies}
1224:
1225: %Create an electronic catalog of galaxies that reports their
1226: %lopsidedness and other modes at varying radii and with other pertinent
1227: %quantities.
1228:
1229:
1230: \section{Structural Properties of Lopsided Galaxies \label{sec:lopsfh}}
1231:
1232:
1233: %We now investigate how the structural parameters stellar mass $M_*$,
1234: %stellar mass density $\mu_*$, and concentration ($C$ = Petrosian $R_{90}/R_{50}$)
1235: %depend on the Fourier modes.
1236:
1237: %We have divided our galaxy sample into
1238: %five subsamples according to $A_1^i$. The subsamples' are summarized
1239: %in Tab.~\ref{tab:5sub} with their $A_1^i$ ranges.
1240:
1241: %\begin{table}
1242: %Table~\ref{tab:5sub}. Subsamples of galaxies partitioned by $A_1^i$ \\
1243: %\begin{tabular}[b]{lcrl}
1244: %Subsample & $A_1^i$ range & No. of Obj. & Plot Color \\
1245: %Full Sample & All & 11874 & Black \\
1246: %Very Symmetric & $<$ 0.055 & 2357 & Dark Blue \\
1247: %Symmetric & 0.055$-$0.080 & 2647 & Light Blue \\
1248: %Average & 0.080$-$0.105 & 2168 & Green \\
1249: %Lopsided & 0.105$-$0.170 & 2635 & Orange \\
1250: %Very Lopsided & $>$ 0.170 & 2067 & Red
1251: %\label{tab:5sub}
1252: %\end{tabular}
1253: %\end{table}
1254:
1255: %In the first column of Fig.~\ref{fig:lopstruct}, we show the
1256: %distributions of $M_*$, $\mu_*$, and $C_i$ for the full sample and for
1257: %each subsample according to the colors given in Tab.~\ref{tab:5sub}.
1258: %We show the distributions as a function of $A_1^i$ in the second
1259: %column. The most symmetric galaxies center about a median mass of $2
1260:
1261: %\times 10^{10} \MSun$ with 90\% falling within $3 \times 10^9$ and
1262: %$9 \times 10^{10} \MSun$. As $A_1^i$ increases to 0.15, the median
1263: %mass drops by a factor of 7 to $3 \times 10^9 \MSun$ and the fifth
1264: %percentile drops to $3 \times 10^8 \MSun$, widening the distribution
1265: %toward lower masses. The mass distribution for very lopsided
1266: %galaxies, with $A_1^i > 0.15$, shows little dependence on $A_1^i$
1267: %aside from a slight widening of the distribution to include galaxies
1268: %with $10^8 \Msun$. Low-mass galaxies with $M_* < 10^9 \Msun$ are
1269: %rarely found to be more lopsided than $A_1^i = 0.05$. High-mass
1270: %galaxies with $M_* > 10^{10} \Msun$ are found to be symmetric or
1271: %lopsided, though most are more symmetric than $A_1^i = 0.10$.
1272:
1273: In \S\ref{sec:cuts}, we showed that our working sample was drawn from
1274: a full sample of $z < 0.06$ galaxies and was selected based on cuts on
1275: several observational parameters. The proportion of high-mass-density
1276: and highly concentrated galaxies was reduced, but a significant number
1277: of these galaxies was also retained. Our sample thus allows us to study
1278: lopsidedness of galaxies over a wide range of the basic structural properties of the galaxies, namely their concentration, stellar mass, and stellar mass density.
1279:
1280: The concentration is a rough proxy for Hubble type, with higher values
1281: corresponding to earlier types. The correspondence between $C_i$ and
1282: Hubble Type has been considered by \citet{si+01} and \citet{s+01}. The
1283: correspondence is not tight, but the value $C_i$ = 2.6 is the rough
1284: dividing line between early- and late-type galaxies (see also
1285: \citealt{khw+03a}. The top-left panel of Fig.~\ref{fig:lopstruct}
1286: shows the distribution of the global $A_1^i$ as a function of
1287: concentration. At $C_i = 2.6$, $A_1^i$ has a moderate value near 0.10.
1288: For $C_i > 3.1$ (typical of elliptical galaxies), the lopsidedness is
1289: typically too small to reliably measure. Below $C_i = 2.6$ the
1290: lopsidedness rises systematically with decreasing concentration (later
1291: Hubble types). Lopsidedness is thus commonplace in the late-type field
1292: galaxies and the galaxies in low-density environments that dominate
1293: our sample, in agreement with previous studies (\citealt{mv+98};
1294: \citealt{cbj00}; \citealt{bc+05}). However, this result might not extend to other environments \citep{aj+06,aj+07}.
1295:
1296: The top-right panel of Fig.~\ref{fig:lopstruct} shows the distribution of lopsidedness as a function of stellar mass. \citet{khw+03b} showed that the local galaxy population is bimodal, with the transition from late type star forming galaxies to early type old galaxies occurring at a value of $M_* \sim 10^{10.5} \Msun$. We see that lopsidedness is only significant for the low mass population, and increases systematically with decreasing mass.
1297:
1298: Finally, we show that there is a similar but even stronger trend
1299: between lopsidedness and stellar mass density, as shown in the lower
1300: panel of Fig.~\ref{fig:lopstruct}. \citet{khw+03b} showed that the
1301: transition between late type star forming galaxies and early type old
1302: galaxies occurs at a value $\mu_* \sim 10^{8.5} \Msunkpc$. We find
1303: that lopsidedness is only commonplace among the low density
1304: population. The lopsidedness increases very strongly with decreasing
1305: density in this regime. Indeed at the lowest densities ($\mu_* <
1306: 10^{7.5} \MSunkpc$) typical galaxies are significantly lopsided
1307: ($A_1^i > 0.20$), and few galaxies in this density range are less
1308: lopsided than 0.10.
1309:
1310:
1311:
1312:
1313:
1314:
1315: \begin{figure}[ht]
1316: \epsscale{1.1}
1317: %\plottwo{/home/tar/agn/asymm/lophists/meddist_a1i_conci.ps}{/home/tar/agn/asymm/lophists/meddist_a1i_m.ps}
1318: \plottwo{fig12a.eps}{fig12b.eps}
1319: \epsscale{0.5}
1320: %\plotone{/home/tar/agn/asymm/lophists/meddist_a1i_mu.ps}
1321: \plotone{fig12c.eps}
1322: \caption{Distributions of $i$-band lopsidedness as functions of the structural parameters $C_i$, $M_*$, and $\mu_*$. The 5th, 25th, 50th, 75th, and 95th percentiles are shown in each panel. Lopsided light distributions are more common in low-mass, low-mass-density, and low-concentration galaxies. Massive, dense, and concentrated galaxies tend to be considerably more symmetric.}
1323: \label{fig:lopstruct}
1324: \end{figure}
1325: \clearpage
1326:
1327: The above structural parameters of SDSS galaxies ($C, M_*, \mu*$) are
1328: highly correlated with one another (e.g., \citealt{khw+03a}). From
1329: the above plots, it is not immediately clear whether there is a
1330: separate underlying physical correlation between lopsidedness and each
1331: of the structural parameters, or whether some of the apparent
1332: correlations are induced by mutual correlations between the
1333: parameters. To explore this, we now turn to Fig.~\ref{fig:lop2dstr}
1334: and compare the three structural properties, two at a time, with
1335: lopsidedness. First, we have plotted $\mu_*$ and $C_i$ against $M_*$
1336: in two-dimensional bins and color-coded each bin holding at least 5
1337: galaxies by its median $A_1^i$. We see immediately that while
1338: lopsidedness does indeed simultaneously correlate with all these
1339: structural parameters, the correlation with $\mu_*$ is the strongest
1340: and most fundamental. In the plots of $\mu_*$ {\it vs.} $M_*$ and
1341: $C_i$ we see that the lopsidedness at a given value of $\mu_*$ is
1342: essentially independent of either $M_*$ or $C_i$. Conversely, for
1343: given values of $M_*$ or $C_i$, there is a systematic increase in
1344: lopsidedness with decreasing $\mu_*$. The plot of $C_i$ {\it vs.}
1345: $M_*$ implies that both parameters are correlated with lopsidedness
1346: (neither is clearly the more fundamental).
1347:
1348: We have used linear partial correlation analysis to see which pairs of
1349: these four structural properties show more fundamental correlations.
1350: We calculated the linear (Pearson) correlation coefficients for each of
1351: the six combinations of the four parameters. We use $\log M_*$, $\log
1352: \mu_*$, $\log A_1^i$, and $C_i$ for this analysis because the
1353: relationships are more linear if logarithms are used for three of the
1354: parameters. For each pair, we also removed the dependence of the
1355: remaining two parameters and calculated the partial correlation
1356: coefficient. The coefficients are listed in Table~\ref{tab:parcor}.
1357: Lopsidedness correlates moderately with all three of the other
1358: parameters but most strongly with stellar mass density
1359: (corr. coeff. $= -0.56$), then concentration ($-0.47$) and mass
1360: ($-0.46$). Once the correlations with stellar mass and concentration
1361: are removed, the correlation between lopsidedness and mass density is
1362: reduced in magnitude to $-0.20$. The partial correlations between
1363: lopsidedness and the other two structural parameters are weaker
1364: ($0.00$ with $C_i$ and $-0.12$ with log $M_*$). In agreement with
1365: Fig.~\ref{fig:lop2dstr}, we see that the most fundamentral correlation
1366: between lopsidedness and a structural parameter is with the surface
1367: mass density.
1368:
1369: \begin{deluxetable}{cclr}
1370: \tabletypesize{\small}
1371: \tablecolumns{4}
1372: \tablewidth{0pc}
1373: \tablecaption{Correlation Coefficients of Structural Parameters}
1374: \tablehead{
1375: \colhead{} &
1376: \colhead{} &
1377: \colhead{Dependence} &
1378: \colhead{(Partial)} \\
1379: \colhead{Par. 1} &
1380: \colhead{Par. 2} &
1381: \colhead{Removed} &
1382: \colhead{Corr. Coeff.} }
1383: \startdata
1384: $\log A_1^i$ & $\log \mu_*$ & \nodata & $-$0.56 \\
1385: %$A_1^i$ & $\mu_*$ & $M_*$ & $-$0.34 \\
1386:
1387: %$A_1^i$ & $\mu_*$ & $C_i$ & $-$0.37 \\
1388: $\log A_1^i$ & $\log \mu_*$ & $\log M_*$ \& $C_i$ & $-$0.20 \\
1389:
1390: \hline
1391: $\log A_1^i$ & $\log M_*$ & \nodata & $-$0.46 \\
1392: %$A_1^i$ & $M_*$ & $\mu_*$ & 0.04 \\
1393: %$A_1^i$ & $M_*$ & $C_i$ & $-$0.30 \\
1394: $\log A_1^i$ & $\log M_*$ & $\log \mu_*$ \& $C_i$ & 0.00 \\
1395: \hline
1396: $\log A_1^i$ & $C_i$ & \nodata & $-0.47$ \\
1397: %$A_1^i$ & $C_i$ & $\mu_*$ & $-0.12$ \\
1398: %$A_1^i$ & $C_i$ & $M_*$ & $-0.30$ \\
1399: $\log A_1^i$ & $C_i$ & $\log \mu_*$ \& $\log M_*$ & $-0.12$ \\
1400: \hline
1401: $\log \mu_*$ & $\log M_*$ & \nodata & $0.87$ \\
1402: %$\mu_*$ & $M_*$ & $A_1^i$ & $0.83$ \\
1403: %$\mu_*$ & $M_*$ & $C_i$ & $0.84$ \\
1404: $\log \mu_*$ & $\log M_*$ & $\log A_1^i$ \& $C_i$ & $0.82$ \\
1405: \hline
1406: $\log \mu_*$ & $C_i$ & \nodata & $0.70$ \\
1407: %$\mu_*$ & $C_i$ & $A_1^i$ & $0.60$ \\
1408: %$\mu_*$ & $C_i$ & $M_*$ & $0.63$ \\
1409: $\log \mu_*$ & $C_i$ & $\log A_1^i$ \& $\log M_*$ & $0.59$ \\
1410: \hline
1411: $\log M_*$ & $C_i$ & \nodata & $0.49$ \\
1412: %$M_*$ & $C_i$ & $A_1^i$ & $0.37$ \\
1413: %$M_*$ & $C_i$ & $\mu_*$ & $-0.32$ \\
1414: $\log M_*$ & $C_i$ & $\log A_1^i$ \& $\log \mu_*$ & $-0.32$ \\
1415: \enddata
1416: \label{tab:parcor}
1417: \end{deluxetable}
1418:
1419:
1420: \begin{figure}[ht]
1421: \epsscale{1.1}
1422: %\plottwo{/home/tar/agn/asymm/lophists/2d_m_mu_lopi.ps}{/home/tar/agn/asymm/lophists/2d_m_conci_lopi.ps}
1423: \plottwo{fig13a.eps}{fig13b.eps}
1424: \epsscale{0.5}
1425: %\plotone{/home/tar/agn/asymm/lophists/2d_mu_conci_lopi.ps}
1426: \plotone{fig13c.eps}
1427: \caption{Stellar mass-mass density and concentration-mass relationships colored by median $i-$band lopsidedness. Galaxies that are massive, dense, and concentrated tend to have symmetric light distributions, while low-mass, low-mass-density, and low-concentrated galaxies tend to be lopsided.}
1428: \label{fig:lop2dstr}
1429: \end{figure}
1430: \clearpage
1431:
1432:
1433: %\begin{figure}[ht]
1434: %\epsscale{1.1}
1435: %\plottwo{/home/tar/agn/asymm/lophists/2d_m_mu_lopi.ps}{/home/tar/agn/asymm/lophists/2d_m_conci_lopi.ps}
1436:
1437: %\caption{Stellar mass-mass density and concentration-mass relationships colored by median $i-$band lopsidedness.}
1438: %\label{fig:lop2dstr}
1439: %\end{figure}
1440: %\clearpage
1441:
1442: %The higher odd-order modes reveal another Fourier property of massive,
1443: %high-mass-density, and concentrated galaxies. We plot the stellar
1444: %mass, mass density, and $i$-band concentration against relative mode
1445: %strengths $\Delta A_3^i$ and $\Delta A_5^i$ in
1446: %Fig.~\ref{fig:relmodestr}, where the median values of the modes of
1447: %similarly lopsided galaxies are subtracted from the individual mode
1448:
1449: %strengths. All three structural parameters peak where $\Delta A_3^i$
1450: %and $\Delta A_5^i$ are -0.02 to 0.00, indicating that galaxies whose
1451: %third and fifth modes are typical of their lopsidedness tend to be
1452: %massive, dense, and concentrated elliptical galaxies. The result is
1453: %more striking when the relative modes are considered simultaneously. In
1454: %Fig.~\ref{fig:relmodestr2}, we bin and plot $\Delta A_3^i$ against
1455: %$\Delta A_5^i$ and color the bins by median stellar mass, mass
1456: %density, and $i$-band concentration.
1457:
1458:
1459:
1460: %\begin{figure}[ht]
1461: %\epsscale{1.1}
1462: %\plottwo{/home/tar/agn/asymm/lophists/meddist_da3i_m.ps}{/home/tar/agn/asymm/lophists/meddist_da5i_m.ps}
1463: %\plottwo{/home/tar/agn/asymm/lophists/meddist_da3i_mu.ps}{/home/tar/agn/asymm/lophists/meddist_da5i_mu.ps}
1464: %\plottwo{/home/tar/agn/asymm/lophists/meddist_da3i_conci.ps}{/home/tar/agn/asymm/lophists/meddist_da5i_conci.ps}
1465: %\caption{Distributions of structural parameters as functions of the excess strengths of $A_3^i$ and $A_5^i$ compared to typical strengths of galaxies with the same lopsidedness $A_1^i$.}
1466: %\label{fig:relmodestr}
1467: %\end{figure}
1468: %\clearpage
1469:
1470: %\begin{figure}[ht]
1471: %\epsscale{1.1}
1472: %\plottwo{/home/tar/agn/asymm/modeplots/2dm_da5i_da3i_m.ps}{/home/tar/agn/asymm/modeplots/2dm_da5i_da3i_mu.ps}
1473: %\plottwo{/home/tar/agn/asymm/modeplots/2dm_da5i_da3i_conci.ps}{}
1474: %\caption{Distributions of structural parameters as functions of the excess strengths of $A_3^i$ and $A_5^i$ compared to typical strengths of galaxies with the same lopsidedness $A_1^i$.}
1475: %\label{fig:relmodestr2}
1476: %\end{figure}
1477: %\clearpage
1478:
1479:
1480:
1481: %\begin{figure}[ht]
1482: %\epsscale{1.0}
1483: %\plottwo{/home/tar/agn/asymm/profiles/0338-51694-360.a.ps}{/home/tar/agn/asymm/profiles/0338-51694-360.phi.ps}
1484: %\caption{$A_m$ and $\phi_m$ profiles for the Fourier decomposition of SDSS J125416.38-020204.4 for modes $m = 0, 1, \ldots 5$. Profiles are shown for the $g$-band in green, $r$-band in orange, and $i$-band in red. Phase angles $\phi_m$ are measured counterclockwise from an axis pointing to the right in the images of Fig.~\ref{fig:sampleim}.}
1485: %\label{fig:sampleprofile}
1486: %\end{figure}
1487: %\clearpage
1488:
1489:
1490:
1491: %\subsection{Color and Star Formation History}%
1492:
1493: %[**Section intro, tie structure into SFH as a transition]
1494: %THIS MAY NEED TO GO BACK TO PAPER TWO!
1495: %We use two stellar spectral indices to provide information about star
1496: %formation history. The 4000~\AA~break, $D_{4000}$, indicates current
1497: %stellar age and any ongoing star formation, while the H$\delta$
1498: %absorption index, $H\delta_A$, traces recent star formation from 0.1-1
1499: %Gyr ago. We use the narrow definition of $D_{4000}$ as defined by
1500: %\citet{b83}: the ratio of the flux density $F_{\nu}$ in the ranges
1501: %4050$-$4250 and 3750$-$3950 \AA. Reddening affects this narrow
1502: %definition less than other, wider definitions. Younger stellar
1503: %populations produce almost no metal absorption just longward of 4000
1504: %\AA, giving $D_{4000} \sim 1$, while older populations have strong
1505: %metal line absorption, and $D_{4000}$ increases toward 2. We use the
1506: %$H\delta_A$ index as defined by \citet{wo97} where an equivalent width
1507: %is calculated between two pseudocontinuum bandpasses. H$\delta$
1508: %absorption is strong ($H\delta_A > 3$ \AA) for young stellar
1509: %populations dominated by O- and B-type stars that are $< 1$ Gyr old.
1510: %The index $H\delta_A$ peaks near 9 \AA~for stars between 0.1 and 1 Gyr
1511: %in age, where A-type stars dominate the absorption in starbursts, and
1512: %the index weakens to negative values between 1 and 10 Gyr, when the
1513: %stellar population is dominated by redder stars. For a comparison of
1514: %stellar spectral indices to stellar ages via different models, see
1515: %\citet{khw+03a}. Both $D_{4000}$ and $H\delta_A$ are spectroscopic
1516: %quantities that are measured from spectra taken within the
1517: %3\arcsec~fiber aperture used by the SDSS.
1518:
1519: %Fig.~\ref{fig:lopsfh} shows a plot of both star formation indicators
1520: %color-coded by median lopsidedness. The least star forming galaxies
1521: %(high $D_{4000}$ and low $H\delta_A$) are typically the most
1522: %symmetric. Lopsided galaxies are the norm in the star-forming end of
1523:
1524: %the locus. $D_{4000}$ correlates more strongly with lopsidedness
1525: %because at any fixed value of $H\delta_A$, lopsidedness decreases
1526: %significantly with increasing $D_{4000}$, but at fixed $D_{4000}$,
1527: %$A_1^i$ varies increases only slightly with $H\delta_A$. The most
1528: %lopsided region ($A_1^i > 0.15$) of the plot wraps around the upper
1529: %left part of the locus. This region was shown by
1530: %\citet{khw+03a} to be dominated by galaxies with recent (within the
1531: %past 0.1 Gyr) and older (more than 0.1 Gyr ago) bursts of star
1532: %formation. Median lopsidedness peaks at 0.25$-$0.30 near $D_{4000} <
1533: %1.1$, and this is the region dominated by older bursts.
1534:
1535:
1536:
1537: %\begin{figure}[ht]
1538: %\epsscale{0.55}
1539: %\plotone{/home/tar/agn/asymm/lophists/2d_d4000_hdelta_lopi.ps}
1540: %\caption{The relationship between $H\delta_A$ absorption index and the 4000-\AA break $D_{4000}$ color-coded by median $A_1^i$. }
1541: %\label{fig:lopsfh}
1542: %\end{figure}
1543: %\clearpage
1544:
1545:
1546: %We have seen that symmetric galaxies tend to have higher stellar mass,
1547: %higher stellar mass density, and higher concentration than lopsided
1548: %galaxies. These structural properties of the symmetric galaxies are
1549: %known to correlate with redder, older stellar populations that have
1550: %little current or recent star formation. On the other hand, lopsided
1551: %galaxies have the structural properties of bluer, younger stellar
1552: %populations. It should then be no surprise to find a similar division
1553: %of stellar populations among symmetric and lopsided galaxies.
1554:
1555: %In Fig.~\ref{fig:lopcolors}, we plot the distributions of lopsidedness
1556: %against the four $ugriz$ colors for our sample. All four colors show
1557: %similar correlations with lopsidedness. The lopsidedness distribution
1558: %moves toward high lopsidedness as each color becomes bluer and toward
1559:
1560: %low lopsidedness at redder colors. Symmetric galaxies (median $<
1561: %0.10$) are common at colors $u - g > 1.4$, $g - r > 0.6$, $r - i >
1562: %0.3$, or $i - z > 0.2$, which comprise approximately the redder half
1563: %of each color distribution.
1564:
1565:
1566: %\begin{figure}[ht]
1567: %\epsscale{1.1}
1568: %\plottwo{/home/tar/agn/asymm/lophists/meddist_a1i_ugcolor.ps}{/home/tar/agn/asymm/lophists/meddist_a1i_grcolor.ps}
1569: %\plottwo{/home/tar/agn/asymm/lophists/meddist_a1i_ricolor.ps}{/home/tar/agn/asymm/lophists/meddist_a1i_izcolor.ps}
1570:
1571: %\caption{Distributions of $A_1^i$ as functions of colors. The 5th, 25th, 50th, 75th, and 95th percentiles are shown.}
1572: %\label{fig:lopcolors}
1573: %\end{figure}
1574: %\clearpage
1575:
1576:
1577: %The bluer colors of the lopsided galaxies may be attributed, at least
1578: %in part, to recent and ongoing star formation. We look next at how
1579: %lopsidedness relates to two star formation indicators, $D_{4000}$ and
1580: %$H\delta_A$, in Fig.~\ref{fig:lopsfh}. In the top left panel, we plot
1581: %the lopsidedness distribution as a function of $H\delta_A$. There is
1582: %a nearly linear trend between median $A_1^i$ and $H\delta_A$ for
1583: %nearly the entire range of $H\delta_A$, with $A_1^i$ rising from near
1584: %0.05 at $H\delta_A = -1$ to 0.20 at $H\delta_A = 9$. Because high
1585: %values of $H\delta_A$ indicate bursts of star formation ending in the
1586: %past Gyr, we can connect the most lopsided galaxies as having
1587: %significant amounts of recent star formation. Galaxies without strong
1588: %$H\delta$ absorption and therefore with little or no recent star
1589: %formation compose a population of relatively symmetric galaxies.
1590:
1591: %The $D_{4000}$ index tells the relative population of old stars and
1592: %therefore can indicate a relative proportion of new stars. We plot
1593: %the distribution of $A_1^i$ against $D_{4000}$ in the lower left panel
1594: %of Fig.~\ref{fig:lopsfh}. For currently star-forming galaxies
1595: %($D_{4000} < 1.3$), few are more symmetric than $A_1^i = 0.05$, and
1596: %half are very lopsided ($A_1^i > 0.2$). The dependence is strong in
1597: %at low $D_{4000}$, but weakens toward higher $D_{4000}$, where smaller
1598: %values of lopsidedness become more typical. Lopsidedness diminishes
1599: %in galaxies with progressively older populations and more pronounced
1600: %4000 \AA breaks.
1601:
1602: %Star formation rates mimic these correlations between star formation
1603: %and lopsidedness. In the lower panel of Fig.~\ref{fig:lopsfh}, we
1604:
1605: %show the distribution of lopsidedness as a function the specific star
1606: %formation rate (star formation rate per solar mass of stars in the
1607: %galaxy) as computed by \citet{bcw+04}. Unsurprisingly, the galaxies
1608: %forming stars the fastest are typically quite lopsided. The
1609: %dependence is strong in this regime, with median $A_1^i$ increasing
1610: %from 0.1 to over 0.3 over a factor-of-10 increase in specific star
1611: %formation rate from $10^{-10}$ to $10^{-9} \MSun/$yr. Galaxies with
1612: %lower SFR$/M_*$ much more gradually become more symmetric.
1613: %Lopsidedness drops from 0.1 to 0.05 when SFR$/M_*$ decreases from
1614: %$10^{-10}$ to $10^{-12}$ $\MSun/$yr.
1615:
1616:
1617: %In sum, there is a strong trend between lopsidedness and star
1618: %formation apparent in all three measures of stellar population age.
1619: %Highly lopsided galaxies are most prevalent when current and recent
1620: %star formation rates are highest, and symmetric galaxies tend to have
1621: %old stellar populations with little recent or ongoing star formation.
1622: %The colors match this trend, with blue galaxies typically being
1623: %symmetric and red galaxies lopsided.
1624:
1625:
1626:
1627:
1628: %\begin{figure}[ht]
1629: %\epsscale{1.1}
1630: %\plottwo{/home/tar/agn/asymm/lophists/meddist_a1i_hdeltaa.ps}{/home/tar/agn/asymm/lophists/meddist_a1i_d4000.ps}
1631: %\epsscale{0.5}
1632: %\plotone{/home/tar/agn/asymm/lophists/meddist_a1i_sfrm.ps}
1633: %\caption{The distribution of $A_1^i$ as functions of the star formation indicators $D_{4000}$ and $H\delta_A$. The 5th, 25th, 50th, 75th, and 95th percentiles are shown. }
1634: %\label{fig:lopsfh}
1635: %\end{figure}
1636: %\clearpage
1637:
1638: %We next compare the relationships between specific star formation rate
1639: %and structural properties with lopsidedness in
1640: %Fig.~\ref{fig:lop2dsfh}. We see first that the specific SFR
1641: %correlates more strongly with lopsidedness than does either stellar
1642: %mass or concentration. At a fixed value of $SFR_{tot}/M_*$, $A_1^i$
1643: %decreases slightly as $M_*$ or $C_i$ increases. The trend of
1644: %increasing lopsidedness with decreasing mass may be mainly a
1645: %reflection of the dominance of fast SFRs at low mass and of
1646: %slow-to-moderate SFRs at high mass. The high-mass, concentrated, less
1647: %star-forming galaxies are typically very symmetric, while the
1648: %low-mass, unconcentrated, quickly star-forming galaxies are usually
1649: %lopsided.
1650:
1651: %The stellar mass density correlates more strongly with lopsidedness
1652: %than the other structural parameters, and when specific SFR is
1653: %included simultaneously as a third parameter, lopsidedness depends
1654: %strongly on both density and specific SFR (upper right panel). The
1655: %least dense ($\mu_* \sim 10^{7} \Msunkpc$) and fastest star-forming
1656: %(SFR$_{tot}/M_* \sim 10^{-9} \Msun$/yr) galaxies exhibit typically
1657: %large $A_1^i > 0.25$, which is not far from the value $A_1^i = 1/3$,
1658: %which corresponds to a factor of 2 in light and mass between the
1659: %opposing bright and dim regions of the galaxy.
1660:
1661: %\begin{figure}[ht]
1662: %\epsscale{1.1}
1663:
1664: %%\plottwo{/home/tar/agn/asymm/lophists/meddist_a1i_sfr.ps}{/home/tar/agn/asymm/lophists/hist_sfr.ps}
1665: %\caption{{\em Left panels:} Distribution of star formation rates and indicators as functions of $A_1^i$. The 5th, 25th, 50th, 75th, and 95th percentiles are shown. {\em Right panels:} Distribution of the specific star formation rate.}
1666: %\label{fig:lopsfr}
1667: %\end{figure}
1668: %\clearpage
1669:
1670:
1671:
1672: %\begin{figure}[ht]
1673: %\epsscale{1.1}
1674: %\plottwo{/home/tar/agn/asymm/lophists/2d_m_sfrm_lopi.ps}{/home/tar/agn/asymm/lophists/2d_mu_sfrm_lopi.ps}
1675: %\epsscale{0.5}
1676: %\plotone{/home/tar/agn/asymm/lophists/2d_conci_sfrm_lopi.ps}
1677: %\caption{The specific star formation rate vs. structural property relationships color-coded by median $i-$band lopsidedness.}
1678:
1679: %\label{fig:lop2dsfh}
1680: %\end{figure}
1681: %\clearpage
1682:
1683:
1684: %\subsection{Metallicity}
1685:
1686: %***Intro about metallicity in galaxies, Tremonti's method of
1687: %calculating it...
1688:
1689:
1690:
1691: %There are stronger correlations between lopsidedness and forbidden
1692: %oxygen lines, as shown in Fig.~\ref{fig:lopolines}. The distributions
1693: %of lopsidedness are shown as function of oxygen line fluxes
1694:
1695: %[\ion{O}{2}] $\lambda$3726, [\ion{O}{2}] $\lambda$3729, and
1696: %[\ion{O}{3}] $\lambda$5007 as well as metallicity $12 + \log_{10}
1697: %($O/H$)$. Median lopsidedness increases monotonously from near 0.10
1698: %in oxygen-poor galaxies to near 0.20 in oxygen-rich galaxies for each
1699:
1700:
1701: %of the three oxygen lines. While lopsidedness increases with oxygen
1702: %content, we see that it decreases with metallicity $12 + \log_{10}
1703:
1704: %($O/H$)$ in the lower left panel of the figure.
1705:
1706: %In Fig.~\ref{fig:2dmetal}, we show the relations between metallicity,
1707: %lopsidedness, and each of stellar mass, stellar mass density, $i$-band
1708: %concentration, and specific star formation rate. In the
1709:
1710: %mass-metallicity relation, high-mass, high-metallicity galaxies are
1711: %typically more symmetric than low-mass, low-metallicity galaxies. At
1712: %a fixed mass, galaxies of increasing metallicity are typically more
1713: %symmetric, and at a fixed metallicity, galaxies of increasing mass
1714: %become less symmetric.
1715:
1716: %\begin{figure}[ht]
1717: %\epsscale{1.1}
1718: %\plotone{/home/tar/agn/asymm/lophists/meddist_a1i_oh.ps}
1719: %\caption{{\em Left panels:} Distributions of [\ion{O}{2}] and [\ion{O}{3}] line fluxes and $12 + \log_{10}$(O/H) metallicity as functions of $A_1^i$. The 5th, 25th, 50th, 75th, and 95th percentiles are shown. {\em Right panels:} Distributions of [\ion{O}{2}], [\ion{O}{3}], and $12 + \log_{10}$(O/H)}
1720: %\label{fig:lopolines}
1721: %\end{figure}
1722: %\clearpage
1723:
1724:
1725:
1726:
1727:
1728: %\subsection{Nuclear Activity}
1729:
1730:
1731: %\begin{figure}[ht]
1732: %\epsscale{1.1}
1733: %\plottwo{/home/tar/agn/asymm/lophists/meddist_a1i_loiiicorr.ps}{/home/tar/agn/asymm/lophists/hist_loiiicorr.ps}
1734: %\epsscale{0.55}
1735: %\plotone{/home/tar/agn/asymm/lophists/2d_bptx_bpty_lopi.ps}
1736: %\caption{{\em Upper left:} The distribution of $A_1^i$ as a function of [O III] luminosity. The 5th, 25th, 50th, 75th, and 95th percentiles of $A_1^i$ are shown. Inactive galaxies with $L$[O III]$ = 0$ are plotted in the leftmost bin. Active galaxies have $L$[O III] $> 3$, as can be seen in the distribution of $L$[O III] {\em (upper right)}. {\em Lower panel:} The BPT diagram color coded by median $i-$band lopsidedness. The starburst galaxy wing of the diagram shows that starburst galaxies have above-average lopsidedness, while the AGN wing shows that AGN are typically hosted by symmetric host galaxies.}
1737: %\label{fig:lopagn}
1738: %\end{figure}
1739: %\clearpage
1740:
1741:
1742:
1743: \section{Conclusions \& Discussion \label{sec:summary}}
1744:
1745: We have measured large-scale galactic asymmetry for a large sample of
1746: low-redshift ($z <$ 0.06) galaxies drawn from the Sloan Digital Sky
1747: Survey. Our use of lopsidedness, a radially averaged $m=1$ azimuthal
1748: Fourier mode, has proven useful for a large fraction of the sample.
1749: Images of a minority of galaxies in the sample have poor observational
1750: properties that cause significant systematic errors in the lopsidedness
1751: calculation, and these galaxies were removed from the sample via cuts
1752: on angular size, signal-to-noise, and ellipticity/inclination. Those
1753: cuts removed a higher fraction of the high-mass, high-mass-density,
1754: and high-concentration galaxies than those with low values of these
1755: structural properties. Nonetheless, the resulting sample is well
1756: represented by the galaxies of the same range of structural properties
1757: as the original sample.
1758:
1759: We find that there are no systematic differences between the $(g-i)$
1760: colors of the brighter and fainter sides of lopsided galaxies. This
1761: implies that there is no systematic difference in the mass/light ratio
1762: \citep{k+07}, and hence that the lopsided light
1763: distributions are primarily caused by lopsided distributions in the
1764: stellar mass. We have verified this through analysis of the
1765: relationship between color and mass/light ratio for both model galaxy
1766: spectral energy distributions and SDSS galaxy data. However, for our
1767: sample the lopsidedness in the $g$-band tends to be slightly greater
1768: than in the $r$- and $i$-bands. Thus, some of the lopsidedness in the
1769: light does arise from the effects of star-formation and/or dust
1770: extinction (which will more strongly affect the $g-$band light).
1771:
1772: Lopsidedness is a structural property that depends strongly on other
1773: structural properties. Galaxies with progressively lower
1774: concentration, stellar mass, or stellar mass density tend to have
1775: progressively higher lopsidedness. We show that the strongest and most
1776: fundamental correlation is between lopsidedness and stellar mass
1777: density. We also find that lopsidedness increases systematically with
1778: increasing radius, particularly for late-type galaxies.
1779:
1780: Lopsidedness can be induced through tidal stress associated with
1781: interactions with a companion galaxy or through accretion or minor
1782: mergers (e.g. \citealt{zr97}; \citealt{bc+05}). Galaxies with low
1783: density will be most affected by tidal stress, and the effects of a
1784: tidal perturbation will last longer in such systems due to the longer
1785: dynamical times. The same arguments pertain to the outer parts of
1786: galaxies. Thus, the two above results make good physical
1787: sense. Alternatively, if the dark matter halo is lopsided, its effects
1788: on the structure of the stellar disk will be more pronounced in the
1789: outer region and in galaxies with low mass and low density (where dark
1790: matter is more dynamically important). The relatively large values of
1791: lopsidedness we measure to be commonplace ($A_1 > 0.1$) appear to be
1792: too large to be generated by internally generated dynamical processes
1793: (e.g., \citealt{mt97}).
1794:
1795: Our overall goal in this investigation has been to use lopsidedness as
1796: a way of quantifying the signature of moderate or weak global
1797: dynamical perturbations. The next step will be to determine the
1798: connections between such perturbations and both the on-going/recent
1799: star formation and the growth of supermassive black holes in
1800: galaxies. These connections can help constrain the processes and
1801: conditions that guide the formation and evolution of the galaxies. In
1802: future papers we will address these questions using the present sample
1803: of galaxies.
1804:
1805: %are also thought to induce bursts of star formation. Using two well
1806: %studied SFH indicators, we have shown that lopsidedness is typically
1807: %associated with bursty and fast star formation, and quiescent star
1808:
1809: %formation is typical in symmetric galaxies. We will extend this
1810: %discussion in future work.
1811:
1812: %Interactions are also thought to induce increased star formation rates
1813: %in galaxies. We have found that lopsided galaxies produce stars at
1814: %much faster rates per unit stellar mass than symmetric galaxies.
1815: %Lopsidedness is then another way to qualitatively describe the stellar
1816: %population age of galaxies. Taking lopsidedness as a signature of a
1817: %recent interaction, the link between star formation and lopsidedness
1818: %may be that they are symptoms of the same interactions. We will
1819: %extend this discussion in future work.
1820:
1821:
1822: JB acknowledges the receipt of a FCT post-doctoral grant
1823: BPD/14398/2003. We would like to thank Vivienne Wild for reading a
1824: draft of the manuscript.
1825:
1826:
1827:
1828: Funding for the SDSS and SDSS-II has been provided by the Alfred
1829: P. Sloan Foundation, the Participating Institutions, the National
1830: Science Foundation, the U.S. Department of Energy, the National
1831: Aeronautics and Space Administration, the Japanese Monbukagakusho, the
1832: Max Planck Society, and the Higher Education Funding Council for
1833: England. The SDSS Web Site is http://www.sdss.org/.
1834:
1835: The SDSS is managed by the Astrophysical Research Consortium for the
1836: Participating Institutions. The Participating Institutions are the
1837: American Museum of Natural History, Astrophysical Institute Potsdam,
1838: University of Basel, University of Cambridge, Case Western Reserve
1839: University, University of Chicago, Drexel University, Fermilab, the
1840: Institute for Advanced Study, the Japan Participation Group, Johns
1841: Hopkins University, the Joint Institute for Nuclear Astrophysics, the
1842: Kavli Institute for Particle Astrophysics and Cosmology, the Korean
1843: Scientist Group, the Chinese Academy of Sciences (LAMOST), Los Alamos
1844: National Laboratory, the Max-Planck-Institute for Astronomy (MPIA),
1845: the Max-Planck-Institute for Astrophysics (MPA), New Mexico State
1846: University, Ohio State University, University of Pittsburgh,
1847: University of Portsmouth, Princeton University, the United States
1848: Naval Observatory, and the University of Washington.
1849:
1850:
1851:
1852:
1853:
1854: \begin{thebibliography}
1855:
1856: \bibitem[Abraham et al.(1996)]{avg+96} Abraham, R.~G., van den Bergh, S., Glazebrook, K., Ellis, R.~S., Santiago, B.~X., Suva, P., \& Griffiths, R.~E.\ 1996, \apjs, 107, 1
1857: \bibitem[Adelman-McCarthy et al.(2006)]{dr4} Adelman-McCarthy, J.~K., et al.\ 2006, \apjs, 162, 38
1858: \bibitem[Angiras et al.(2006)]{aj+06} Angiras, R.~A., Jog, C.~J., Omar, A., \& Dwarakanath, K.~S.\ 2006, \mnras, 369, 1849
1859: \bibitem[Angiras et al.(2007)]{aj+07} Angiras, R.~A., Jog, C.~J., Dwarakanath, K.~S., \& Verheijen, M.~A.~W.\ 2007, \mnras, 378, 276 \bibitem[Arp(1966)]{a66} Arp, H.\ 1966, \apjs, 14, 1
1860: \bibitem[Baldwin et al.(1980)]{bl+80} Baldwin, J.~E., Lynden-Bell, D., \& Sancisi, R.\ 1980, \mnras, 193, 313
1861: \bibitem[Bournaud et al.(2005)]{bc+05} Bournaud, F., Combes, F., Jog, C.~J., \& Puerari, I.\ 2005, \aap, 438, 507
1862: \bibitem[Bournaud \& Duc(2006)]{bd+06} Bournaud, F., \& Duc, P.-A.\ 2006, \aap, 456, 481
1863: \bibitem[Brinchmann et al.(2004)]{bcw+04} Brinchmann, J., Charlot, S., White, S.~D.~M., Tremonti, C., Kauffmann, G., Heckman, T., \& Brinkmann, J.\ 2004, \mnras, 351, 1151
1864: \bibitem[Bruzual A.(1983)]{b83} Bruzual A., G.\ 1983, \apj,
1865: 273, 105
1866: \bibitem[Charlot \& Longhetti(2001)]{cl01} Charlot, S., \&
1867: Longhetti, M.\ 2001, \mnras, 323, 887
1868: \bibitem[Charlot et al.(2002)]{ckl+02} Charlot, S., Kauffmann,
1869: G., Longhetti, M., Tresse, L., White, S.~D.~M., Maddox, S.~J., \& Fall,
1870: S.~M.\ 2002, \mnras, 330, 876
1871: \bibitem[Conselice et al.(2000)]{cbj00} Conselice, C.~J., Bershady, M.~A., \& Jangren, A.\ 2000, \apj, 529, 886
1872: \bibitem[Conselice(2003)]{c03} Conselice, C.~J.\ 2003,
1873: \apjs, 147, 1
1874: \bibitem[De Propris et al.(2007)]{dc+07} De Propris, R., Conselice, C.~J., Driver, S.~P., Liske, J., Patton, D., Graham, A., \& Allen, P.\ 2007, ArXiv e-prints, 705, arXiv:0705.2528
1875: \bibitem[Frei et al.(1996)]{fgg+96} Frei, Z., Guhathakurta,
1876: P., Gunn, J.~E., \& Tyson, J.~A.\ 1996, \aj, 111, 174
1877: \bibitem[Fukugita et al.(1996)]{f+06} Fukugita, M., Ichikawa, T., Gunn, J.~E., Doi, M., Shimasaku, K., \& Schneider, D.~P.\ 1996, \aj, 111, 1748
1878: \bibitem[Gunn et al.(1998)]{g+98} Gunn, J.~E., et al.\ 1998,
1879: \bibitem[Gunn et al.(2006)]{g+06} Gunn, J.~E., et al.\ 2006,
1880: \aj, 131, 2332 \aj, 116, 3040
1881: \bibitem[Hogg et al.(2001)]{h+01} Hogg, D.~W., Finkbeiner, D.~P., Schlegel, D.~J., \& Gunn, J.~E.\ 2001, \aj, 122, 2129
1882: \bibitem[Ivezi{\'c} et al.(2004)]{i+04} Ivezi{\'c}, {\v Z}., et al.\ 2004, Astronomische Nachrichten, 325, 583
1883: \bibitem[Jog(1997)]{j97} Jog, C.~J.\ 1997, \apj, 488, 642
1884: \bibitem[Jog(1999)]{j99} Jog, C.~J.\ 1999, \apj, 522, 661
1885: \bibitem[Kauffmann \& Haehnelt(2000)]{kh00} Kauffmann, G., \& Haehnelt, M.\ 2000, \mnras, 311, 576
1886: \bibitem[Kauffmann et al.(2003a)]{khw+03a} Kauffmann, G., et al.\ 2003a, \mnras, 341, 33
1887: \bibitem[Kauffmann et al.(2003b)]{khw+03b} Kauffmann, G., et al.\ 2003b, \mnras, 341, 54
1888: \bibitem[Kauffmann et al.(2003c)]{kht+03c} Kauffmann, G., et al.\ 2003c, \mnras, 346, 1055
1889: \bibitem[Kauffmann et al.(2004)]{kwh+04} Kauffmann, G., White, S.~D.~M., Heckman, T.~M., M{\' e}nard, B., Brinchmann, J., Charlot, S., Tremonti, C., \& Brinkmann, J.\ 2004, \mnras, 353, 713
1890: \bibitem[Kauffmann et al.(2007)]{k+07} Kauffmann, G., et al.\ 2007, ArXiv Astrophysics e-prints, arXiv:astro-ph/0609436
1891: \bibitem[Kere{\v s} et al.(2005)]{kk+05} Kere{\v s}, D.,
1892: Katz, N., Weinberg, D.~H., \& Dav{\'e}, R.\ 2005, \mnras, 363, 2
1893: \bibitem[Kornreich et al.(2002)]{kl+02} Kornreich, D.~A., Lovelace, R.~V.~E., \& Haynes, M.~P.\ 2002, \apj, 580, 705
1894: \bibitem[Levine \& Sparke(1998)]{ls98} Levine, S.~E., \& Sparke, L.~S.\ 1998, \apjl, 496, L13
1895: \bibitem[Lupton et al.(2001)]{lgi+01} Lupton, R., Gunn, J.~E.,
1896: Ivezi{\'c}, Z., Knapp, G.~R., Kent, S., \& Yasuda, N.\ 2001, ASP
1897: Conf.~Ser.~238: Astronomical Data Analysis Software and Systems X, 238, 269
1898: \bibitem[Masset \& Tagger(1997)]{mt97} Masset, F., \& Tagger, M.\ 1997, \aap, 318, 747
1899: \bibitem[Matthews et al.(1998)]{mv+98} Matthews, L.~D., van Driel, W., \& Gallagher, J.~S., III 1998, \aj, 116, 1169
1900: \bibitem[Mihos \& Hernquist(1996)]{mh96} Mihos, J.~C., \& Hernquist, L.\ 1996, \apj, 464, 641 \bibitem[Press(2002)]{p02} Press, W.~H.\ 2002, Numerical recipes in C++ : the art of scientific computing by William H.~Press.~xxviii, 1,002 p.~: ill.~; 26 cm.~ Includes bibliographical references and index.~ISBN : 0521750334,
1901: \bibitem[Noordermeer et al.(2001)]{ns+01} Noordermeer, E., Sparke, L.~S., \& Levine, S.~E.\ 2001, \mnras, 328, 1064
1902: \bibitem[Richter \& Sancisi(1994)]{rs94} Richter, O.-G., \& Sancisi, R.\ 1994, \aap, 290, L9
1903: \bibitem[Rix \& Zaritsky(1995)]{rz95} Rix, H., \& Zaritsky, D.\ 1995, \apj, 447, 82
1904: \bibitem[Rudnick \& Rix(1998)]{rr98} Rudnick, G., \& Rix, H.\ 1998, \aj, 116, 1163
1905: \bibitem[Rudnick et al.(2000)]{rrk00} Rudnick, G., Rix, H., \& Kennicutt, R.~C.\ 2000, \apj, 538, 569
1906: \bibitem[Sanders \& Mirabel(1996)]{sm96} Sanders, D.~B., \& Mirabel, I.~F.\ 1996, \araa, 34, 749 \bibitem[Springel et al.(2005)]{sdh05} Springel, V., Di Matteo, T., \& Hernquist, L.\ 2005, \apjl, 620, L79
1907: \bibitem[Shimasaku et al.(2001)]{s+01} Shimasaku, K., et al.\ 2001, \aj, 122, 1238
1908: \bibitem[Shu et al.(1990)]{st+90} Shu, F.~H., Tremaine, S.,
1909: Adams, F.~C., \& Ruden, S.~P.\ 1990, \apj, 358, 495
1910: \bibitem[Smith et al.(2002)]{sm+02} Smith, J.~A., et al.\ 2002, \aj, 123, 2121
1911: \bibitem[Springel \& Hernquist(2005)]{sh05} Springel, V., \& Hernquist, L.\ 2005, \apjl, 622, L9 \bibitem[Stoughton et al.(2002)]{s+02} Stoughton, C., et
1912: al.\ 2002, \aj, 123, 485
1913: \bibitem[Strateva et al.(2001)]{si+01} Strateva, I., et al.\ 2001, \aj, 122, 1861
1914: \bibitem[Strauss et al.(2002)]{swl+02} Strauss, M.~A., et al.\ 2002, \aj, 124, 1810
1915: \bibitem[Swaters et al.(1999)]{ss+99} Swaters, R.~A.,
1916: Schoenmakers, R.~H.~M., Sancisi, R., \& van Albada, T.~S.\ 1999, \mnras,
1917: 304, 330
1918: \bibitem[Syer \& Tremaine(1996)]{st96} Syer, D., \& Tremaine, S.\ 1996, \mnras, 281, 925
1919: \bibitem[Toomre \& Toomre(1972)]{tt72} Toomre, A., \& Toomre, J.\ 1972, \apj, 178, 623
1920: \bibitem[Tucker et al.(2006)]{t+06} Tucker, D.~L., et al.\ 2006, Astronomische Nachrichten, 327, 821
1921: \bibitem[Walker et al.(1996)]{wm+96} Walker, I.~R., Mihos, J.~C., \& Hernquist, L.\ 1996, \apj, 460, 121
1922: \bibitem[Weinberg(1994)]{w94} Weinberg, M.~D.\ 1994, \apj, 421, 481
1923: \bibitem[Wilcots \& Prescott(2004)]{wp04} Wilcots, E.~M., \& Prescott, M.~K.~M.\ 2004, \aj, 127, 1900
1924: \bibitem[Worthey \& Ottaviani(1997)]{wo97} Worthey, G., \&
1925: Ottaviani, D.~L.\ 1997, \apjs, 111, 377
1926: \bibitem[York et al.(2000)]{y+00} York, D.~G., et al.\ 2000,
1927: \aj, 120, 1579
1928: \bibitem[Zaritsky \& Rix(1997)]{zr97} Zaritsky, D., \& Rix, H.\ 1997, \apj, 477, 118
1929:
1930:
1931: \end{thebibliography}
1932:
1933:
1934:
1935: \end{document}
1936:
1937: