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

2: \documentclass[12pt,preprint]{emulateapj}

3: \usepackage{apjfonts}

4: 

5: \begin{document}

6: 

7: \journalinfo{Accepted to the Astrophysical Journal Letters}

8: \submitted{Received 2004, Oct. 6; Accepted 2005 Jan. 21}

9: 

10: \title{A Discovery of Rapid Optical Flares from Low-Luminosity

11: Active Nuclei in Massive Galaxies}

12: 

13: \author{Tomonori Totani$^1$, Takahiro Sumi$^2$, George Kosugi$^3$, Naoki

14: Yasuda$^4$, Mamoru Doi$^5$, and Takeshi Oda$^1$} 

15: 

16: \altaffiltext{1}{

17: Department of Astronomy, Kyoto University, Sakyo-ku, Kyoto, 606-8502,

18: Japan} 

19: 

20: \altaffiltext{2}{Princeton University Observatory, Peyton Hall,

21: Princeton, NJ 08544-1001, USA } 

22: 

23: \altaffiltext{3}{Subaru Telescope,

24: National Astronomical Observatory of Japan, Hilo, HI 96720, USA}

25: 

26: \altaffiltext{4}{Institute for Cosmic Ray Research, The University of

27: Tokyo, Kashiwa, Chiba 277-8582, Japan}

28: 

29: \altaffiltext{5}{Institute of Astronomy, The University of

30: Tokyo, Mitaka, Tokyo 181-1500, Japan }

31: 

32: %\author{Charlie Author}

33: % \homepage{http://www.Second.institution.edu/~Charlie.Author}

34: %\affiliation{

35: %Second institution and/or address\\

36: %This line break forced% with \\

37: %}%

38: 

39: %\date{\today}% It is always \today, today,

40:              %  but any date may be explicitly specified

41: 

42: \begin{abstract}

43: We report a serendipitous discovery of six very low-luminosity active

44: galactic nuclei (AGNs) only by optical variability in one-month

45: baseline.  The detected flux variability is $\sim$ 1--5\% of the total

46: luminosity of host galaxies.  Careful subtraction of host galaxy

47: components in nuclear regions indicates that the fractional variability

48: $\Delta F/F$ of the nuclei is of order unity. At least one of them is

49: showing a compelling flaring activity within just a few days, which

50: appears to be quite different from previously known AGN variability. We

51: obtained spectroscopic data for the one showing the largest flare and

52: confirmed that it is in fact an AGN at $z = 0.33$ with an estimated

53: black hole mass of $\sim 10^8 M_\odot$.  As a possible interpretation,

54: we suggest that these activities are coming from the region around the

55: black hole event horizon, which is physically similar to the recently

56: discovered near-infrared flares of our Galactic nucleus. It is

57: indicated that our Galaxy is not special, and that surprisingly rapid

58: flaring activity in optical/near-infrared bands may be commonly hidden

59: in nuclei of apparently normal galaxies with low Eddington ratios, in

60: contrast to the variability of well-studied luminous AGNs or quasars.

61: \end{abstract}

62: 

63: %\maketitle

64: 

65: \keywords{black hole physics --- galaxies: active}

66: 

67: \section{Introduction}  

68: Active galactic nuclei (AGNs) are generally variable, giving important

69: information on the activity of central super massive black holes

70: (SMBHs). In optical bands, most AGNs show significant variability on

71: time scales longer than months, but day-scale or shorter variability is

72: generally small (fractional amplitude of $\lesssim 10\%$) and rare (Ulrich,

73: Maraschi, \& Urry 1997; Webb \& Malkan 2000; Hawkins 2002). Therefore

74: the recent discovery (Genzel et al. 2003; Ghez et al. 2004) of

75: near-infrared (NIR) flares within one hour from Sgr A$^*$, an extremely

76: low-luminosity AGN of our Galaxy with a SMBH mass of $\sim 3 \times 10^6

77: M_\odot$, is surprising. Our Galaxy may be a peculiar AGN, or instead,

78: such variability may be common for very low-luminosity AGNs like Sgr

79: A$^*$, but it could not have been examined because of the difficulty in

80: observation for extragalactic sources.

81: 

82: In a deep search for faint transient objects using the Subaru

83: Telescope which has the largest field of view ($30'\times24'$) among 

84: existing 8m class telescopes, we serendipitously found several AGNs only

85: by variability on one month baseline, which is about 1--5\% of the total

86: luminosity of host galaxies. There have been several attempts to find

87: AGNs by their variability (Hawkins 1983; Bershady, Trevese, \& Kron

88: 1998; Trevese et al. 1989; Sarajedini et al. 2003), but most of such

89: surveys looked for bright AGNs compared with host galaxy luminosity on

90: time baselines of many (five to ten) years. Therefore our discovery

91: is unique in terms of the time scale and nuclear/host luminosity ratios, 

92: providing us with a new method to find low-luminosity AGNs.

93: Here we report this discovery and also suggest a possible (though not

94: exclusive) interpretation.

95: 

96: \section{The Observation and Variable Object Search}

97: 

98: \begin{figure*}%[h]

99: %\plotone{VIcolor-cap2.ps} 

100: \plotone{f1.eps} 

101: \caption{$V$ and $I$ color images of the six

102: variable AGNs found in the survey.  The first column shows the

103: reference images taken about one month prior to the main observing run,

104: and the reference-subtracted images of the consecutive four nights in

105: the main run are shown from the second to fifth columns.  The size of

106: each panel is $10'' \times 10''$, and the pixel scale is

107: $0''.2$/pixel. The upper four objects (1--4) brightened in a month,

108: while the lower two faded. The subtracted images of the lower two are

109: inverted for presentation. The $V$ and $I$ surface brightness profiles

110: of host galaxies are also shown in the right panels; photon counts are

111: in arbitrary unit and two-dimensional surface brightness data are

112: projected onto the primary axis of elliptical fitting.  }

113: \label{fig:VIimage}

114: \end{figure*}

115: 

116: A fixed field centered on the galaxy cluster Abell 2152 was consistently

117: monitored in $V$ and $I$ bands with similar exposure times in this

118: observation, with the primary scientific purpose of searching

119: microlensing events (Totani 2003).  A reference frame was observed using

120: a night on May 5, 2003, and the main-run observations during four

121: consecutive nights were performed about one month after the reference

122: observation (June 1--4).  The weather condition was good and seeing was

123: mostly stable; the effective seeing FWHM on the stacked images over each

124: night is 0.76, 0.68, 0.78, 0.56, and 0.62 arcsec for the reference frame

125: and the four nights of the main run, respectively.  After stacking these

126: frames for each night, we subtracted the reference frame from the

127: main-run frames by using the Alard \& Lupton (1998) algorithm.  We then

128: searched for any brightened objects in subtracted frames (and in their

129: inverted frames for faded objects) by a standard software, SExtractor

130: (Bertin \& Arnouts 1996), requiring $S/N>10$ of the SExtractor best

131: magnitude both in $V$ and $I$ bands (about 26.1 and 24.4 mag,

132: respectively) in either of the four nights. The CCD counts were saturated

133: in the

134: centers of the brightest stars and galaxies, and these regions were

135: removed from the analysis because the subtraction does not work

136: correctly.

137: 

138: Using this procedure we detected more than 20 objects, and most of them are

139: likely to be supernovae that are associated with faint host galaxies but

140: are offset from their centers. However, if we choose variable objects

141: associated with bright galaxies ($I < 20$), we found 6 variable

142: objects, all of which are located at the very centers of

143: well-resolved host galaxies (Fig.  \ref{fig:VIimage}), indicating that

144: they are very likely to be AGNs. Assuming that supernovae trace stellar

145: light, the chance probability of finding a supernova in the center of a

146: host galaxy within the angular resolution is about 3\%,

147: and hence the probability that all these six objects are supernovae by chance

148: is $\sim 2 \times 10^{-10}$.  The objects 1--4 brightened in one month,

149: but the other two faded.

150: 

151: \input{table1.tex}

152: 

153: To estimate the statistical significance of these detections, we

154: measured an aperture flux ($1.2''$ = 6 pixels diameter) of centers of

155: all $\sim$ 1,000 galaxies with $I < 20$ on the subtracted images.  The

156: distribution is fit by a Gaussian, and the $S/N$ ratios of the maximum

157: variability flux within the same aperture among the four nights are

158: estimated by using this empirical noise level. The results are shown in

159: Table 1. All objects are statistically significant using a criterion of

160: $S/N > 10$ in at least one band.  Three of the six objects show this

161: month-like timescale variability at such high levels of significance in

162: both bands.  In principle, we cannot exclude a possibility that these

163: are false detections created by some systematic failure of image

164: subtraction. However, failures of subtraction, which often happen in

165: saturated regions, can easily be recognized by complicated and clearly

166: artificial image profiles. The saturation effect is sensitive to seeing,

167: and if these detections were created by saturation, the variability flux

168: of all objects should show a similar dependence on the observed date,

169: which is not actually observed. Therefore we consider that these

170: detections of one-month scale variability are reliable. As shown below,

171: the spectroscopy of the object 2 shows some AGN features, which provides

172: a further support for the significance of the variability detection.

173: 

174: \section{Flux Variability Estimates}

175: 

176: \begin{figure}%[h]

177: \epsscale{1.5}

178: \plotone{f2.eps}

179: \caption{(Left panels) Variability flux (in $\mu$Jy)

180: of the six AGNs during the four

181: nights of the main observing run, relative to the reference frame

182: taken one month earlier. 

183: The filled and open circles are for $V$ and $I$ bands,

184: respectively. 

185: (Right panels) Spectral index $\alpha$ ($f_\nu \propto

186: \nu^\alpha$) of variability flux between $V$ and $I$ bands.

187: The error bars in the left panel and thick error bars

188: in the right panel are statistical 1$\sigma$ errors, while the thin

189: error bars in the right panel are the total including systematics

190: (see text). 

191: The upper four objects brightened in a month, but

192: the lower two faded and their flux is measured in inverted subtracted

193: images.  

194: }

195: \label{fig:lc}  

196: \end{figure}

197: 

198: The fluxes of the variable objects measured on the subtracted images during

199: the four days are shown as light curves in Fig. \ref{fig:lc}.  The host

200: galaxy luminosities are shown in Table \ref{table:agn}, and the

201: variability flux is typically about 1--5\% of these.  We estimated

202: detection efficiency and photometric errors by placing artificial point

203: sources on the surface brightness peaks of the $\sim$1,000 galaxies with

204: $I < 20$. Then we repeated the same procedure of image subtraction and

205: source detection. We found that the detection efficiency is about 50\%

206: for $f_\nu = $0.176 and 0.515 $\mu$Jy, or 25.8 and 24.1 mag for $V$ and

207: $I$ bands, respectively.

208: 

209: We made two different error estimates: one is ``statistical'', which is

210: the variance from the mean of the observed flux of an artificial point

211: source during the four nights, and the other is ``total'' including

212: systematics, which is the variance from the real flux of an artificial

213: point source. The significance of variability within the four days

214: should be estimated by the statistical errors. The chance probability of

215: getting the data from a constant source during the four days is shown as

216: $P_d$ for each object in Table \ref{table:agn}. The evidence for

217: variability within the four days is compelling for the object 2 and

218: strong for the object 4. The spectral indices between $V$ and $I$ bands

219: are also shown in Fig. \ref{fig:lc}, with the total error estimates. The

220: evidence of spectral variability within days is weak for all

221: objects. The values of the indices show a large scatter from object to

222: object, but strong conclusions cannot be derived because of the large

223: systematic errors.

224: 

225: The AGN component in the pre-subtraction images must be estimated to

226: know the fractional variability ($\Delta F / F$) of the nuclei. There is

227: no apparent central excess in surface brightness profiles of host

228: galaxies (Fig. \ref{fig:VIimage}), indicating that stellar light

229: dominates AGN luminosity. We estimate the host galaxy component in the

230: nuclear regions of the reference frame as follows.  We set an aperture

231: diameter of $1''.2$ as the nuclear region. The surface brightness

232: profiles outside this aperture are elliptically fitted by the

233: $r^{1/4}$-law as well as the exponential profiles after convolved with

234: the seeing.  Then we subtract them from the observed total nuclear

235: fluxes, to get an estimate of the AGN components in the reference frame

236: (Table \ref{table:agn}).  The use of the exponential profile is

237: conservative because it gives lower host galaxy contribution and hence

238: lower fractional variability estimates, compared with the $r^{1/4}$-law

239: that is often used for the central stellar light profile. To be even

240: more conservative, we also estimate the host galaxy contribution by a

241: constant surface brightness within the aperture, with a value estimated

242: by photon counts along the aperture annulus. The dominant error is that

243: of variability flux $\Delta F$, which can be inferred from $S/N$ ratios

244: in the table.  These results indicate fractional variabilities ($\Delta

245: F / F$) of order unity for the six objects.

246: 

247: \section{Spectroscopy and Redshift Estimates/Determination}

248: Although there are two galaxy clusters at $z = 0.04$ and 0.13 in the

249: observed field (Totani 2003), the locations of the six AGNs are not

250: concentrated to the cluster centers.  The morphology of these host

251: galaxies is apparently elliptical or early-type, with magnitude of $I

252: \sim $ 18.5--19.5. The field galaxy counts at $I \sim 19$ dominate the

253: expected number of the cluster galaxies. According to a galaxy evolution

254: model that is in good agreement with various data (Totani \& Yoshii 2000),

255: the peak of the redshift distribution of early type galaxies at this

256: magnitude is $z \sim$ 0.3--0.4, and observed $V - I$ colors ($\sim$

257: 1.6--2.2) are consistent with expected values for passively-evolving

258: early-type galaxies in this redshift range.  Therefore, it is likely

259: that these galaxies are field early-type galaxies at 

260: $z \sim$ 0.3--0.4, i.e.,  out of the two

261: clusters.

262: 

263: \begin{figure}[h]

264: \plotone{f3.eps}

265: \caption{The optical spectrum of the object 2. The redshift is $z=0.33$.}

266: \label{fig:obs2-spec}  

267: \end{figure}

268: 

269: We obtained an optical spectroscopic data by Subaru/FOCAS for the object

270: 2 (Fig. \ref{fig:obs2-spec}), and confirmed that the redshift is, as

271: expected, $z = 0.33$ and the host galaxy is a typical giant galaxy with

272: $M_B \sim -20$.

273: %\footnote{We have requested time for spectroscopic

274: %follow-up of the other sources during the next observing season.}

275: Several emission lines are detected in the nuclear region, whose FWHMs

276: ($\sim$ 300 km/s) are typical of narrow line regions.  The line ratios

277: are typical of the low ionization nuclear emission regions (LINERs),

278: confirming that this object is a low-luminosity AGN.  In the following,

279: we assume $z = 0.4$ for the other five AGNs for order-of-magnitude

280: discussions.  Using the empirical relation between bulge luminosity and

281: black hole mass (McLure \& Dunlop 2001), the SMBH mass of these AGNs

282: should be about $\sim 10^8 M_\odot$, and the nuclear luminosity becomes

283: $\sim 10^{-5}$--$10^{-4}$ in units of the Eddington ratio, i.e., $(\nu

284: L_\nu)/L_{\rm Edd}$, where $L_{\rm Edd}$ is the Eddington luminosity.

285: 

286: 

287: \section{Discussion}

288: For the first time, except for the recently discovered flares in Sgr

289: A$^*$, fractional AGN variability of $\sim$ 100\% within a few days in

290: optical/infrared bands is detected in one, and probably six, such

291: low-luminosity (low Eddington ratio) AGNs. The Sgr A$^*$ flares

292: have been interpreted as violent nonthermal phenomena 

293: at the innermost region of a radiatively inefficient accretion flow

294: (RIAF) around the SMBH (Yuan, Quataert, \& Narayan 2004). Such behavior

295: cannot be seen for luminous AGNs whose time variability has been studied

296: much better; the only exception is the rare population of blazars whose

297: jets are closely directed to the observer (Ulrich et al. 1997).

298: Although yet other interpretations might be possible, we here examine

299: these two interpretations (RIAFs and blazars).

300: 

301: The continuum-to-line luminosity ratio is a useful diagnostic to

302: discriminate between the two interpretations.  Since the continuum of

303: blazars is strongly beamed by relativistic jets, unbeamed emission lines

304: are generally weak and even difficult to see. On the other hand, if we

305: are observing less beamed continuum emission from accretion disks, we

306: expect a continuum-to-line ratio that is typical of normal (non-blazar)

307: AGNs. The H$\alpha$ emission ($2.6 \times 10^{40}$ erg/s) is clearly

308: detected for the object 2 and it argues against the blazar

309: interpretation. By using the empirical relations between line and

310: continuum luminosities (Ho \& Peng 2001; Wang, Staubert, \&

311: Ho 2002), the H$\alpha$ luminosity translates into the

312: expected optical continuum luminosity of $\nu f_\nu \sim 1.6 \times

313: 10^{41} $ and $7.8 \times 10^{43}$ erg/s for normal AGNs (Seyfert 1s)

314: and blazars, respectively. This should be compared with the variability

315: flux of the object 2: $\Delta (\nu f_\nu) \sim$ (0.9--3.2) $\times

316: 10^{42}$ erg/s in the $V$ band. The observed variability flux is

317: intermediate between the two interpretations, but considering that the

318: object 2 is likely in a flare phase, the mean flux appears to be closer

319: to the RIAF interpretation.

320: 

321: The number density of these six AGNs is about $\sim 10^{-4} \ \rm

322: Mpc^{-3}$, using the comoving volume out to $z = 0.4$. If the AGN

323: variability is from disk emission, this number can be compared with

324: published AGN luminosity functions in optical (Ho 2004) or

325: X-ray (Ueda et al. 2003) bands at a nuclear activity level of $ \nu L_\nu

326: \sim 10^{42}$ erg/s, which is found to be $\phi(>10^{42}{\rm erg/s})

327: \sim 10^{-3.5} \ \rm Mpc^{-3}$.  These numbers are not very different,

328: indicating that the six AGNs found in this work constitute a

329: considerable fraction of low-luminosity AGNs. On

330: the other hand, if variability is from blazar activity, the real

331: number density should be much larger than $\sim 10^{-4} \ \rm

332: Mpc^{-3}$, because of the collimation of blazar emission.  It is

333: believed that blazars are beamed by a factor of $\Delta \Omega /

334: (4\pi) \sim \Gamma^{-2}/4 \lesssim 1/400$, where $\Gamma \sim 10$ is

335: the jet Lorentz factor (Salvati, Spada, \& Pacini 1998), 

336: and hence the real number

337: density of AGNs like the objects found in this work 

338: should be $\gtrsim 10^{-2} \ \rm

339: Mpc^{-3}$.  However, the isotropic H$\alpha$ line luminosity indicates

340: that the nuclear activity must be larger than $\sim 10^{41}$ erg/s,

341: and the number density of such AGNs should not be larger than

342: $\sim 10^{-3} \ \rm Mpc^{-3}$ from the AGN luminosity functions. This

343: provides another argument against the blazar interpretation.

344: 

345: It is interesting to note that, if the variability reflects the scale of

346: emission region around SMBHs, the variability time scale of $\sim$ 1

347: hour for the $3 \times 10^6 M_\odot$ black hole of Sgr A$^*$ corresponds

348: to $\sim$ 1 day for $\sim 10^8 M_\odot$ black holes. Although the

349: Eddington ratio of our AGNs ($\sim 10^{-5}$--$10^{-4}$) is much larger

350: than that of the Sgr A$^*$ ($\sim 10^{-9}$), it is well below the border

351: ($\sim 10^{-2}$) separating the standard thin disk and RIAFs (Kato,

352: Fukue, \& Mineshige 1998). A RIAF model of NIR/X-ray flares of Sgr A$^*$

353: predicts that flare activity should disappear with increasing Eddington

354: ratio, but it remains up to $L/L_{\rm Edd} \sim 10^{-5}$ in optical/NIR

355: bands (Yuan et al. 2004). Therefore it seems a reasonable interpretation

356: that previously unknown violent activity in optical bands has been

357: hidden around the SMBH event horizon of very low Eddington-ratio AGNs, which

358: cannot be seen in luminous AGNs whose luminosity is close to the 

359: Eddington limit.

360: 

361: We would like to thank S. Mineshige and the anonymous

362: referee for useful comments. 

363: This work has been supported by the Grant-in-Aid for the 21st

364: Century COE ``Center for Diversity and Universality in Physics'' from

365: the Ministry of Education, Culture, Sports, Science and Technology

366: (MEXT) of Japan.

367: 

368: \begin{thebibliography}{}

369: 

370: \bibitem{subtraction} Alard, C. \& Lupton, R.H. 1998,

371: ApJ, 503, 325 

372: 

373: \bibitem{bershady}

374: Bershady, M.A., Trevese, D., \& Kron, R.G. 1998, ApJ, 496, 103

375: 

376: \bibitem{sex} Bertin, E. \& Arnouts, S.  1996,

377: % SExtractor: Software for source extraction.

378: A\&A, 117, 393

379: 

380: \bibitem{genzel} Genzel, R. et al. 2003, Nature, 425, 934 

381: 

382: \bibitem{ghez} Ghez, A.M. et al. 2004, ApJ, 601, L159 

383: 

384: \bibitem{hawkins-variability-survey}

385: Hawkins, M.R.S. 1983, MNRAS, 202, 571

386: 

387: \bibitem{hawkins} Hawkins, M.R.S. 2002, MNRAS, 

388: 329, 76 

389: 

390: \bibitem{AGN-LF-B} Ho, L.C. 2004, 

391: in Carnegie Observatories Astrophysics Series,

392: vol. 1: Coevolution of Black Holes and Galaxies, ed. L.C. Ho 

393: (Cambridge: Cambridge Univ. Press). Available at LANL server

394: (astro-ph/0401527).

395: 

396: \bibitem{line-Seyfert} Ho, L.C. \& Peng, C.Y. 2001, 

397: ApJ, 555, 650 

398: 

399: \bibitem{ADAF-review} 

400: Kato, S., Fukue, J., \&

401: Mineshige, S. 1998, 

402: ``Black-Hole Accretion Disks'', Kyoto University Press

403: 

404: %Narayan, R., Mahadevan, R., \& Quataert, E., in

405: %"The Theory of Black Hole Accretion Discs", eds. M. A. Abramowicz,

406: %G. Bjornsson, and J. E. Pringle (1998), available on LANL server

407: %(astro-ph/9803141)

408: 

409: \bibitem{m_BH-L_R} McLure, R.J. \& Dunlop, J.S. 2001, 

410: MNRAS 327, 199 

411: 

412: \bibitem{blazar-beaming} Salvati, M., Spada, M., \& Pacini, F. 1998, 

413: ApJ. 495, L19 (1998)

414: 

415: \bibitem{sarajedini}

416: Sarajedini, V.L., Gilliland, R.L., \& Kasm, C. et al. 2003, 

417: ApJ, 599, 173 

418: 

419: \bibitem{cc-lens} Totani, T. 2003, ApJ, 586, 735 

420: 

421: \bibitem{totani-yoshii} Totani, T. \& Yoshii, Y. 2000, 

422: ApJ,  540, 81

423: 

424: \bibitem{dario} Trevese, D., Pittella, G., Kron, R.G., 

425: Koo, D.C. \& Bershady, M. 1989, AJ, 98, 108

426: 

427: \bibitem{AGN-LF-X} Ueda, Y., Akiyama, M., Ohta, K., \& Miyaji, T. 2003,

428: ApJ, 598, 886

429: 

430: \bibitem{ulrich} Ulrich, M.-H., Maraschi, L., \& Urry, C.M. 1997, 

431: ARA\&A,  35, 445 

432: 

433: \bibitem{malkan} Webb, W. \& Malkan, M. 2000, 

434: ApJ, 540, 652

435: 

436: \bibitem{line-blazar} Wang, J.-M., Staubert, R., \& Ho, L.C.

437: 2002, ApJ,  579, 554 

438: 

439: \bibitem{yuan} Yuan, F., Quataert, E., \& Narayan, R. 2004, 

440: ApJ, 606, 894 

441: 

442: 

443: \end{thebibliography}

444: 

445: 

446: 

447: \end{document}

448: 

449: 

450: