1: \documentclass[12pt,preprint]{aastex}
2: %\documentclass{emulateapj}
3: \usepackage{amsmath}
4: \usepackage{amssymb}
5: \usepackage{graphicx}
6:
7: \voffset 0.5truecm
8:
9: \newcommand\bmath[1] {\mbox{\boldmath$\rm #1$}}
10:
11: %\bibliographystyle{mn2e}
12: \usepackage{natbib}
13: \bibliographystyle{aa}
14:
15: % Standard abbreviations
16: \newcommand\cf{{cf.}} %.......confer
17: \newcommand\eg{{e.g.}} %......exempli gratia
18: \newcommand\ie{{i.e.}} %......id est
19: \newcommand\etal{{et al.}} %..and others
20: \newcommand\etc{{etc}} %......et cetera
21:
22: % Unit abbreviations
23: %%% Temperature
24: \newcommand\K{{\rm K}} %...........Kelvin
25: %%% Time
26: \newcommand\s{{\rm s}} %...........seconds
27: \newcommand\ns{{\rm n}\s} %........nanoseconds
28: \newcommand\mus{\mu\s} %...........microseconds
29: \newcommand\ms{{\rm m}\s} %........milliseconds
30: \newcommand\ks{{\rm k}\s} %........kiloseconds
31: \newcommand\yr{{\rm yr}} %.........years
32: \newcommand\Myr{{\rm M}\yr} %......megayears
33: \newcommand\Gyr{{\rm G}\yr} %......gigayears
34: \newcommand\hr{{\rm hr}}
35: %%% Frequency
36: \newcommand\Hz{{\rm Hz}} %.........Hertz
37: \newcommand\muHz{\mu\Hz} %.........microhertz
38: \newcommand\mHz{{\rm mHz}} %.......millihertz
39: \newcommand\MHz{{\rm MHz}} %.......Megahertz
40: \newcommand\GHz{{\rm GHz}} %.......Gigahertz
41: \newcommand\THz{{\rm THz}} %.......Terahertz
42: %%% Length
43: \newcommand\m{{\rm m}} %...........meters
44: %\newcommand\fm{{\rm f}\m} %........femtometers
45: \newcommand\nm{{\rm n}\m} %........nanometers
46: \newcommand\mum{\mu\m} %...........micrometers
47: \newcommand\mm{{\rm m}\m} %........millimeters
48: \newcommand\cm{{\rm c}\m} %........centimeters
49: \newcommand\km{{\rm k}\m} %........kilometers
50: % Angnstroms are already \AA
51: \newcommand\pc{{\rm pc}} %.........parsecs
52: \newcommand\kpc{{\rm k}\pc} %......kiloparsecs
53: \newcommand\Gpc{{\rm G}\pc} %......gigaparsecs
54: \newcommand\au{{\rm au}} %.........astronomical units
55: %%% Mass
56: \newcommand\g{{\rm g}} %...........grams
57: \newcommand\mug{\mu\g} %...........micrograms
58: \newcommand\mg{{\rm m}\g} %........milligrams
59: \newcommand\kg{{\rm k}\g} %........kilograms
60: %\newcommand\Ms{M_\odot} %...........solar masses
61: %%% Energy
62: \newcommand\eV{{\rm eV}} %.........electron volts
63: \newcommand\keV{{\rm k}\eV} %......kiloelectron volts
64: \newcommand\MeV{{\rm M}\eV} %......megaelectron volts
65: \newcommand\GeV{{\rm G}\eV} %......gigaelectron volts
66: \newcommand\TeV{{\rm T}\eV} %......teraelectron volts
67: \newcommand\PeV{{\rm P}\eV} %......petaelectron volts
68: \newcommand\XeV{{\rm X}\eV} %......exaelectron volts
69: \newcommand\ZeV{{\rm Z}\eV} %......zetaelectron volts
70: \newcommand\erg{{\rm erg}} %.......ergs
71: \newcommand\J{{\rm J}} %...........joules
72: %%% Magnetic Fields
73: \newcommand\G{{\rm G}} %...........gauss
74: \newcommand\nG{{\rm n}\G} %........nanogauss
75: \newcommand\muG{\mu\G} %...........microgauss
76: \newcommand\mG{{\rm m}\G} %........milligauss
77: \newcommand\kG{{\rm k}\G} %........kilogauss
78: \newcommand\T{{\rm T}} %...........tesla
79: \newcommand\muT{\mu\T} %...........microtesla
80: \newcommand\mT{{\rm m}\T} %........millitesla
81: %%% Angular Resolution
82: % ^\circ is already degrees
83: % ' is alrady minutes
84: % `` is already seconds
85: \newcommand\mas{{\rm mas}} %.......milli-arcseconds
86: \newcommand\muas{\mu{\rm as}} %....micro-arcseconds
87: \newcommand\nas{{\rm nas}} %.......nano-arcseconds
88: %%% Spot & Disk Flux densities
89: %\newcommand\Fspot{{\rm F}_{spot}}
90: %\newcommand\Fdisk{{\rm F}_{disk}}
91:
92: \renewcommand\d{{\rm d}}
93: \newcommand\e{{\rm e}}
94: \newcommand\sinc{{\rm sinc}}
95:
96: \newcommand\SgrA{Sgr~A*}
97: \newcommand\EGS{J1745$-$2820}
98:
99: \newcommand\kms{km~s$^{-1}$}
100: \newcommand\uvplane{$(u,v)$-plane}
101:
102: \newcommand\Msun{M$_\odot$}
103: \newcommand\Ro{R_0}
104: \newcommand\hho{H$_2$O}
105:
106:
107:
108: %This is how to have an approximate sign under < or > :
109: \newbox\grsign \setbox\grsign=\hbox{$>$} \newdimen\grdimen \grdimen=\ht\grsign
110: \newbox\laxbox \newbox\gaxbox
111: \setbox\gaxbox=\hbox{\raise.5ex\hbox{$>$}\llap
112: {\lower.5ex\hbox{$\sim$}}}\ht1=\grdimen\dp1=0pt
113: \setbox\laxbox=\hbox{\raise.5ex\hbox{$<$}\llap
114: {\lower.5ex\hbox{$\sim$}}}\ht2=\grdimen\dp2=0pt
115: \newcommand{\gax}{\mathrel{\copy\gaxbox}}
116: \newcommand{\lax}{\mathrel{\copy\laxbox}}
117: %
118:
119:
120: \begin{document}
121:
122: \title{Limits on the Position Wander of Sgr A*}
123:
124: \author{Mark J. Reid$^1$, Avery E. Broderick$^{1,2}$, Abraham Loeb$^1$,
125: Mareki Honma$^3$, Andreas Brunthaler$^4$}
126: \affil{1. Harvard-Smithsonian Center for Astrophysics, 60 Garden Street,
127: Cambridge, MA 02138, USA\\
128: 2. Canadian Institute for Theoretical Astrophysics, 60 St. George Street,
129: Toronto, ON Canada\\
130: 3. Mizusawa VERA Observatory, NAOJ, Mitaka, Tokyo 181-8588, Japan\\
131: 4. Max-Planck-Institute f\"ur Radioastronomie, 69 Auf dem H\"ugel,
132: Bonn-53121, Germany}
133: \shorttitle{}
134: \shortauthors{Reid et al.}
135:
136:
137: \begin{abstract}
138: We present measurements with the VLBA of the variability in the
139: centroid position of \SgrA\ relative to a background quasar at $7\,\mm$
140: wavelength. We find an average centroid wander of $71\pm 45\,\muas$ for
141: time scales between 50 and $100\,\min$ and $113\pm50\,\muas$ for timescales
142: between 100 and $200\,\min$, with no secular trend. These are
143: sufficient to begin constraining the viability of the accretion
144: hot-spot model for the radio variability of \SgrA. It is possible to
145: rule out hot spots with orbital radii above $15\,G M_{\rm Sgr A*}/c^2$
146: that contribute more than 30\% of the total $7\,\mm$ flux. However,
147: closer or less luminous hot spots remain unconstrained. Since
148: the fractional variability of \SgrA\ during our observations was
149: $\sim20$\% on time scales of hours, the hot-spot model for \SgrA's
150: radio variability remains consistent with these limits.
151: Improved monitoring of \SgrA's centroid position has the
152: potential to place significant constraints upon the existence and
153: morphology of inhomogeneities in a supermassive black hole accretion flow.
154: \end{abstract}
155:
156: \keywords{accretion disks -- astrometry -- black hole physics --
157: Galaxy: center -- gravitational lensing --
158: instrumentation: high angular resolution}
159:
160: \maketitle
161:
162: \section{Introduction}
163: There is now overwhelming evidence that \SgrA\ is a supermassive
164: black hole at the center of the Milky Way. Many stars are
165: observed to orbit about a common focal position, requiring
166: an unseen mass of $\approx4\times10^6$~\Msun\ contained within a
167: radius of less than 100~AU \citep{Schoedel:2002,Ghez:2003},
168: for a distance to the center of 8.0~kpc \citep{Reid:1993}.
169: Accurate registration of the infrared and radio reference frames
170: \citep{Menten:1997,Reid:2003} reveal that the common
171: orbital focal position is coincident with \SgrA\ to within
172: measurement uncertainty of $\approx10$~mas. Finally, the
173: absence of intrinsic motion of \SgrA\ at levels near that
174: expected for a $4\times10^6$~\Msun\ object
175: \citep{Reid-Brun:2004}, coupled with a size less than
176: ~1~AU \citep{Bower:2004}, provide a lower limit on mass
177: density of $\sim10^{22}$~\Msun~pc$^{-3}$, which is only two orders of
178: magnitude less than the density of a $4\times10^6$~\Msun\ non-rotating
179: black hole within its innermost stable orbit. There can now be little
180: doubt that \SgrA\ is a supermassive black hole.
181:
182: \citet{Reid-Brun:2004} present measurements of
183: the position of \SgrA\ relative to a compact extragalactic radio source
184: (\EGS, also refered to as J1745-283 in earlier publications).
185: These measurements were conducted with the NRAO
186: \footnote{The National Radio Astronomy Observatory is operated by Associated
187: Universities, Inc., under a cooperative agreement with the National Science
188: Foundation.}
189: Very Long Baseline Array (VLBA) over a period of $8\,\yr$ at a wavelength of
190: $7\,\mm$ ($43\,\GHz$) and have been used to determine the
191: apparent proper motion of \SgrA. Over time scales of months or longer,
192: \SgrA's {\it apparent} motion is dominated by the
193: effects of the orbit of the Sun about the center of the Galaxy.
194: The component of the Sun's orbit in the Galactic plane is uncertain
195: at roughly the 10\% level, and this limits estimation of any intrinsic
196: motion of \SgrA\ at about the $\pm20$~\kms\ level. However, the component of the
197: motion of the Sun out of the Galactic plane is known to high accuracy
198: [$7.16\pm0.38$~\kms\ toward the North Galactic Pole; \citet{Dehnen-Binney:1998}].
199: After removing the effects of the Sun's motion, the residual motion of
200: \SgrA\ perpendicular to the Galactic plane is very small,
201: $\lesssim 1\,\km\,\s^{-1}$,
202: as expected for a supermassive black hole (SMBH) at the dynamical center of
203: the Galaxy.
204: While previous work concentrated on the long-term motion of
205: \SgrA, here we analyze its short-term position ``wander'' on time scales
206: of hours to weeks.
207:
208: Short-timescale motion of the centroid position of \SgrA\
209: would be expected if a portion of the emission comes from material
210: orbiting about the SMBH. The degree of centroid variability would
211: necessarily depend upon the brightness of the orbiting material, the
212: degree to which its emission is nonuniform, and the orbital radius
213: dominating the total flux. We use a simple hot-spot model to relate the
214: constraint from the observed short-term position wander of
215: \SgrA\ (\S \ref{OoCM}) to a constraint upon the presence of strong
216: inhomogeneities in the accretion flow onto the SMBH as a function of
217: hot-spot luminosity and orbital period (\S \ref{Constraints}).
218: Finally, concluding remarks are contained in \S \ref{C}.
219:
220:
221: \section{Observations of Centroid Motion} \label{OoCM}
222: \citet{Reid-Brun:2004} describe the observations and data calibration
223: methods in detail. Briefly, we obtained position data as follows:
224: the VLBA antennas switched between \SgrA\ and a compact
225: extragalactic source (\EGS) every 15 seconds in order provide
226: interferometer phase differences rapidly enough to cancel the
227: effects of short term atmospheric fluctuations.
228: The stronger source, \SgrA, was used as
229: the phase reference to calibrate data from the weaker source, \EGS.
230:
231: Astrometric imaging of \SgrA\ at 7-mm wavelength is best
232: accomplished with only the five inner-VLBA antennas (FD, KP, LA, OV
233: and PT). These antennas produce interferometer baselines with
234: lengths of up to $1500$~km, resulting in synthesized beams
235: typically about $2.5 \times 0.7$~mas (FWHM) elongated north-south.
236: Longer baselines (\eg\ involving the Washington (BR)
237: and Iowa (NL) state antennas)
238: are not generally useful for precise astrometry, as it is difficult to
239: detect \SgrA\ with the 8-sec on-source integrations afforded by
240: rapid switching, coupled with low fringe visibilities on long baselines
241: owing to the large, scatter-broadened, image of \SgrA. Also, the sources
242: are mutually visible with the inner five antennas for only a short time
243: period for antennas far from the inner ones.
244:
245: Our most accurate astrometry was obtained with atmospheric path-delay
246: calibration using ``geodetic'' blocks \citep{Reid-Brun:2004}.
247: This involves short periods of observations of quasars with
248: a wide spanned-bandwidth and scheduled to deliver
249: a wide range of source elevations. These geodetic blocks were placed
250: before the start, at the middle, and after the end of the \SgrA\
251: observations. Analysis of these data yield estimates of the zenith
252: atmospheric path-delay at each antenna accurate to $\sim0.5$ to 1~cm
253: (or about 1 wavelength).
254:
255:
256: \subsection{Position wander: hours}\label{section:hours}
257:
258: For analysis of short-term wander, we selected only our highest
259: quality data, requiring both high accuracy atmospheric path-delay
260: calibration using geodetic blocks (which was started in 2003) and
261: data from all five inner-VLBA antennas.
262: Data from VLBA programs BR84 on 2003 April 5 and 25 and
263: BR124 on 2007 April 5 and 11 satisfied these requirements.
264:
265: The position wander of \SgrA\ over time scales of hours was determined
266: by dividing the calibrated interferometer data into hourly bins.
267: The data were Fourier transformed to make images and intensity
268: centroid positions were determined.
269: In practice, we measured the background source, \EGS, which had
270: been phase-referenced to \SgrA, but we interpret any position changes
271: as owing to changes in \SgrA.
272: In Fig.~\ref{hourly_positions} we show the East-West (EW) position
273: offsets as a function of Greenwich sidereal time,
274: after removing an average position for each day's data.
275: The North-South (NS) positions are intrinsically less accurate by a
276: factor of about 3, as the NS projections of the interferometer baselines
277: are correspondingly shorter than the EW projections.
278:
279: \begin{figure*}
280: \begin{center}
281: \begin{tabular}{cc}
282: \includegraphics[width=0.75\textwidth]{f1.eps}
283: \\
284: \end{tabular}
285: \end{center}
286: \caption{Hourly measurements of the centroid position of Sgr~A*
287: relative to an extragalactic source. Data from 2003 April 05
288: ({\it red triangles}) and 25 ({\it cyan squares}) and 2007 April 05
289: ({\it green pentagons}) and 11 ({\it blue hexagons}) are plotted.
290: All positions are relative to their daily average position.
291: {\it Dotted} and {\it solid lines} indicate expected $\approx1\sigma$
292: and $\approx2\sigma$
293: uncertainties, based on Monte Carlo simulations of the effects
294: of uncompensated atmospheric path-delays (0.5~cm rms) at each antenna.
295: Data outside the vertical {\it dashed lines} were excluded from analysis
296: owing to their large expected position errors.
297: \label{hourly_positions}
298: }
299: \end{figure*}
300:
301: At the start and end of the observing tracks, the sources are
302: at low elevations and susceptible to large interferometer phase
303: shifts caused by uncompensated atmospheric path-delays. We
304: estimated the effects of zenith path-delay errors with 1000
305: independent simulations.
306: Each simulation started by randomly selecting zenith path-delay
307: errors (0.5~cm rms) for each antenna. We then calculated the expected
308: interferometer phase shifts, dependent on time-varying source
309: zenith angles, at half-hour time steps. Position shifts
310: were estimated as the product of the phase shifts (in turns)
311: multiplied by an approximate projected interferometer fringe
312: spacing. Finally, we calculated a weighted position shift for all
313: baselines and all times when the source was above $15^\circ$ elevation.
314: The resulting expected $\pm1\sigma$ and $\pm2\sigma$ position
315: error envelopes are plotted in Fig.~\ref{hourly_positions}.
316: Based on the observed and simulated increase in position scatter for
317: Greenwich sidereal times (GST) before $-0.8$~hr and after $+2.6$~hr,
318: we only use data within this time range for our study of hourly
319: position wander.
320:
321: We differenced the EW and NS positions
322: for all pairs of measurements separated by less than $10^3$~min;
323: the magnitudes of these position differences are shown on the left-hand
324: side of Fig.~\ref{wander}, along with 2 binned weighted averages.
325: The average difference in position for measurements separated by 50 to
326: $100$~min is $71\pm45~\muas$ and for measurements separated
327: by 100 to 200~min is $113\pm50~\muas$.
328: Neither average position difference deviates significantly from
329: zero, and the position differences are consistent with systematic errors
330: owing predominantly to mis-modeling zenith propagation path-delays through
331: the Earth's atmosphere at the $\approx1$~cm level.
332: Thus, we use these measurements as upper limits for the
333: position wander of the centroid of \SgrA's emission.
334:
335: \begin{figure*}
336: \begin{center}
337: \begin{tabular}{cc}
338: \includegraphics[width=0.65\textwidth]{f2.eps}
339: \\
340: \end{tabular}
341: \end{center}
342: \caption{Position wander of \SgrA. Plotted are East-West ({\it top panel})
343: and North-South ({\it bottom panel}) position differences as a function of
344: time separation. Also shown are binned average position differences
345: as large points with error bars. These position differences are likely
346: caused by uncompensated atmospheric path-delay errors and should be
347: considered as upper limits to the position wander of \SgrA.
348: \label{wander}
349: }
350: \end{figure*}
351:
352: \subsection{Position wander: days to weeks}
353: The position wander of \SgrA\ on time scales of days to weeks was
354: determined from ``daily'' position measurements, obtained from
355: $\approx5$ hours of data when source elevations were above
356: $\approx20^\circ$ at most of the five inner-VLBA antennas. These positions
357: were plotted in Fig.~3 of \citet{Reid-Brun:2004}, after removing
358: the long-term proper motion of \SgrA\ of $6.379$~\mas~yr$^{-1}$
359: along a position angle of $209.60^\circ$ East of North.
360: We differenced the EW and NS positions
361: for all pairs of measurements separated by less than $10^5$~min;
362: the magnitudes of these position differences are shown in Fig.~\ref{wander},
363: along with weighted averages for two time-bins.
364:
365: The average difference in EW position for measurements separated by
366: $\approx6000$~min ($\approx4$~days) is about $400~\muas$.
367: This difference is likely not a property of the emission of \SgrA;
368: instead it is probably caused by mis-modeling large-scale
369: propagation delays through the Earth's atmosphere.
370: Most of the data used in this analysis were collected before
371: we started to measure atmospheric path-lengths in 2003.
372: Without this calibration, residual zenith path-length errors of $\sim5$~cm
373: are typical for the VLBA correlator model at 7-mm wavelength.
374: Since propagation delays can be correlated with large-scale weather patterns,
375: which have characteristic time scales of several days, one
376: expects to see such a residual signature in our position differences.
377:
378: \section{Constraints upon hot-spot models of Sgr A* variability} \label{Constraints}
379: Limits upon the variability in the centroid position of high resolution
380: images of Sgr A* imply a constraint upon orbiting hot-spot models
381: for Sgr A*'s radio variability. This will necessarily be a function
382: of the hot-spot orbit and the hot-spot/disk flux ratio. In this
383: section we derive the expected variability for two simple hot-spot
384: models. The first is an optically thin Newtonian hot-spot, while the
385: second fully incorporates general relativity and the opacity of a disk
386: constructed such that it reproduces the observed spectrum of Sgr A*.
387:
388: \subsection{Idealized Newtonian hot-spot centroid variability}
389: For the idealized Newtonian hot-spot we assume that there is no
390: lensing by the black hole, the disk is completely optically thin, and
391: the hot-spot is a point (the first two are physically appropriate for
392: spots at large orbital radii). In this case the hot spot's orbit
393: (which we assume to be circular with radius $r$) is simply given by
394: Kepler's law:
395: \begin{equation}
396: x(t) = r \cos(\Omega t + \phi)\,\quad
397: y(t) = r \cos i \sin(\Omega t + \phi)\,
398: \end{equation}
399: where $i$ is the orbital inclination,
400: \begin{equation}
401: \Omega = \sqrt{\frac{GM_{\rm Sgr A*}}{r^3}}\,,
402: \end{equation}
403: $M_{\rm Sgr A*}$ is the mass of Sgr A*, and $\phi$ is an arbitrary phase.
404:
405: The image centroid, $\bmath{X}_C$ is constructed by integrating the
406: source emission over a time $T$, which need not be small in comparison
407: to the orbital period, and thus the motion of the hot-spot will
408: generally be important. Explicitly, if we set the centroid of the
409: disk emission to be at the origin,
410: \begin{equation}
411: \bmath{X}_C =
412: \frac{\int_0^T F_{\rm spot}(t) \bmath{x}(t) dt}
413: {\int_0^T \left[ F_{\rm disk} + F_{\rm spot}(t) \right] dt}\,.
414: \label{eq:avg_cent}
415: \end{equation}
416: This will generally be a function of the initial phase of the orbit,
417: $\phi$, and the integration time $T$. If we make the simplifying
418: assumption that the hot-spot flux may be treated as roughly constant,
419: then this reduces to
420: \begin{equation}
421: \bmath{X}_C =
422: \eta
423: \overline{\bmath{x}}(\phi;T)
424: \quad\text{where}\quad
425: \eta \equiv \frac{F_{\rm spot}}{F_{\rm disk} + F_{\rm spot}}
426: \,,
427: \end{equation}
428: and $\overline{x}(\phi;T)$ is the average spot position over some
429: time $T$ with initial orbital phase\footnote{While this
430: assumption may appear to be manifestly unjustified, for an optically
431: thick hot-spot in the Rayleigh-Jeans limit, the final expression is
432: formally accurate! The reason is that the relativistic beaming of the
433: hot-spot emission is precisely countered in this instance by the
434: combination of the Doppler-boosting and time-of-flight delays
435: \citep{Brod-Loeb:2005}.} $\phi$. It is straightforward to show that
436: \begin{eqnarray}
437: \displaystyle \overline{x}(\phi;T)
438: &=&
439: \displaystyle r~\sinc\left(\frac{\Omega T}{2}\right) \cos\phi \nonumber \\
440: \displaystyle \overline{y}(\phi;T)
441: &=&
442: \displaystyle r~\sinc\left(\frac{\Omega T}{2}\right) \cos i \sin\phi\,,
443: \end{eqnarray}
444: where a factor of $\Omega T/2$ has been subsumed into the arbitrary
445: phase $\phi$. This corresponds to a single observation of the
446: position of \SgrA.
447:
448: To study the position wander, we must compare two such measurements
449: separated by some time $\Delta T$. This results in a change in the
450: centroid position of
451: \begin{eqnarray}
452: &&\displaystyle \left|\Delta \bmath{X}_C \right|(\phi;T,\Delta T)
453: =
454: \eta
455: \left|\overline{\bmath{x}}(\phi+\Omega\Delta T;T)-\overline{\bmath{x}}(\phi;T)\right| \nonumber\\
456: &&\qquad=
457: \displaystyle \eta~
458: r~\sinc\left(\frac{\Omega T}{2}\right)
459: \sin\left(\frac{\Omega\Delta T}{2}\right) \nonumber \\
460: &&\qquad\qquad\qquad\displaystyle\times\sqrt{\cos^2\phi + \cos^2 i \sin^2 \phi}\,.
461: \end{eqnarray}
462: Maximizing the position wander over the arbitrary initial phase simply
463: removes the terms in the radical:
464: \begin{equation}
465: \left| \Delta \bmath{X}_{C,\rm max} \right| (T,\Delta T)
466: =
467: \eta~r~\sinc\left(\frac{\Omega T}{2}\right) \sin\left(\frac{\Omega\Delta T}{2}\right)\,.
468: \label{eq:dr1D}
469: \end{equation}
470: While it may be unlikely that any given observation of the position
471: wander will allow detection of the maximum centroid displacement,
472: this quantity places a strict constraint upon hot-spots.
473: Alternatively, one may wish
474: to compare the observations to the predicted RMS centroid variability,
475: which in this case is roughly $70\%$ of
476: $\left| \Delta \bmath{X}_{C,\rm max} \right| (T,\Delta T)$.
477:
478: For fixed $T$ and $\Delta T$,
479: $\left|\Delta\bmath{X}_{C,\rm max}\right|$ vanishes when an integral
480: number of periods is commensurate with either the integration time
481: (and thus $\overline{\bmath{x}}(\phi;T)$ vanishes identically) or the
482: delay between measurements. These are evident in Figure
483: \ref{fig:drs}, in which the thin blue line shows
484: $\left|\Delta\bmath{X}_{C,\rm max}\right|$ as a function of orbital
485: period for a number of values of $\eta$ with $T=1\,\hr$ and $\Delta
486: T=4\,\hr$. However, by choosing many different $T$ and $\Delta T$, it
487: is possible fill in the nulls associated with each individual
488: choice. There will be constraints upon what values of $T$ and $\Delta
489: T$ are possible, e.g., due to sensitivity and imaging requirements
490: (see \S \ref{OoCM}). The upper-envelope of the observed centroid
491: displacement for $1\,\hr \le T \le 4\,\hr$ and
492: $1\,\hr \le \Delta T \le 4\,\hr$ is shown in Fig. \ref{fig:drs} by the
493: thick, solid blue line. For reference we also show, with dotted
494: blue lines, the maximum deviation possible, corresponding to the
495: case when all of the flux is due to the hot spot.
496:
497: Generally, the observed displacement has a maximum near orbital
498: periods of roughly $\pi\sqrt{T\Delta T}$. That such a maximum must
499: exist can be inferred from the facts that (i) at large orbital radii
500: the Keplerian velocity decreases and thus beyond some distance the hot
501: spot will move sufficiently slow as to produce no detectable change in
502: it's position over the time $\Delta T$, and (ii) at small orbital
503: radii the intrinsic variability in the spot position is smaller and
504: the hot-spot makes many complete orbits in the integration time
505: $T$, and thus the variable portion of the centroid position is
506: dominated by a small fraction of the integrated flux,
507: yielding again a small observable change in the centroid position.
508: However, we may also show this explicitly by considering the
509: asymptotic expansions of eq. (\ref{eq:dr1D}). For small $\Omega$
510: (large period $P$), the $\sinc$-term is roughly unity and
511: \begin{equation}
512: \left| \Delta \bmath{X}_{C,\rm max} \right| (T,\Delta T)
513: =
514: \eta r \frac{\Omega \Delta T}{2}
515: \propto
516: r^{-1/2}
517: \,.
518: \end{equation}
519: In contrast, for large $\Omega$,
520: $\left| \Delta \bmath{X}_{C,\rm max} \right| (T,\Delta T)$ is
521: strongly oscillatory. Nevertheless, it is bounded from above by the
522: $\sinc$ term, and thus
523: \begin{equation}
524: \left| \Delta \bmath{X}_{C,\rm max} \right| (T,\Delta T)
525: \le
526: \frac{2 \eta r}{\Omega T}
527: \propto
528: r^{5/2}
529: \,.
530: \end{equation}
531: Setting these limiting expressions equal to each other gives the
532: desired condition that a maximum observed displacement occurs near
533: $\Omega \simeq 2/\sqrt{T\Delta T}$. Note that this is true for the
534: envelope obtained by varying $T$ and $\Delta T$ in the prescribed
535: ranges as well, and serves as a simple estimate of the sensitivity of
536: these types of measurements.
537:
538:
539: \subsection{Fully relativistic, hot-spot in an optically thick disk}
540: We now consider a more realistic model in which a hot spot is
541: embedded in an accretion disk, including the relativistic beaming,
542: Doppler boosting, strong gravitational lensing and the opacity of the
543: disk and hot spot. This is necessarily a more complicated model, and
544: thus we address it numerically via the ray-tracing, radiative-transfer
545: code described in \citet{Brod-Loeb:2006} (to which we direct the
546: reader for more information, the model only being summarized below).
547:
548: Due to its ability to shield the hot spot from view, the structure of
549: the background disk is of particular importance. In the absence of an
550: unambiguous prediction from existing accretion flow theory, we have
551: modeled it as a self-similar Radiatively Inefficient Accretion Flow
552: (RIAF) following \citet{Yuan-Quat-Nara:2003}. Specifically, the
553: accretion flow is characterized by a Keplerian velocity distribution,
554: a population of thermal electrons with density and temperature
555: \begin{equation}
556: n_{e,\rm th} = n^0_{e,\rm th} r^{-1.1} \e^{-z^2/2\rho^2}
557: \quad\text{and}\quad
558: T_{e} = T^0_{e} r^{-0.84}\,,
559: \end{equation}
560: respectively, a population of non-thermal electrons
561: \begin{equation}
562: n_{e,\rm nth} = n^0_{e,\rm nth} r^{-2.9} \e^{-z^2/2\rho^2}\,,
563: \end{equation}
564: and spectral index $\alpha_{\rm disk} = 1.25$
565: (defined as $S\propto\nu^{-\alpha}$),
566: and a toroidal magnetic field in approximate ($\beta=10$) equipartition
567: with the ions (which produce the majority of the pressure), i.e.,
568: \begin{equation}
569: \frac{B^2}{8\pi}
570: =
571: \beta^{-1} n_{e,\rm th} \frac{m_p c^2}{6 r}\,.
572: \end{equation}
573: In all expressions $r$, $\rho$ and $z$ are measured in units of
574: $G M_{\rm Sgr A*}/c^2$. The power-laws are taken from
575: \citet{Yuan-Quat-Nara:2003},
576: and the three coefficients ($n^0_{e,\rm th}$, $T^0_{e}$ and
577: $n^0_{e,\rm nth}$) are set by fitting the radio, sub-mm and
578: near-infrared spectrum of Sgr A*. While this is not a unique model
579: for the accretion flow around Sgr A*, it is representative of the
580: general class of RIAFs, and we expect that our results will be quite
581: generic.
582:
583: The hot-spot is modeled by a spherical (in it's frame) Gaussian
584: over-density of non-thermal electrons, with
585: a radial scale of $1.5 G M_{\rm Sgr A*}/c^2$. The
586: primary distinction between the systems shown in the panels of
587: Fig. \ref{fig:drs} is the size of the photosphere of the hot-spot
588: ($\delta r$), larger hot-spots necessarily being more luminous.
589:
590: The radiative transfer is assumed to be dominated by synchrotron
591: emission and absorption. We follow \citet{Jone-ODell:1977} for the power-law
592: electron distributions in the accretion flow and the hot spot, cutting
593: the electron distributions off below Lorentz factors of $10^2$,
594: roughly in agreement with the assumptions in
595: \citet{Yuan-Quat-Nara:2003}. We treat the radiative transfer of the
596: thermal component of the accretion flow in a similar fashion as that
597: discussed in \citet{Yuan-Quat-Nara:2003}, appropriately altered to
598: account for the relativistic nature of the bulk motion (see, e.g.,
599: \citealt{Brod-Blan:2004} for more detail on how this may be done).
600:
601: The centroid displacements were computed by (i) generating a
602: sequence of images associated with an entire orbit, (ii) determining
603: the instantaneous image centroid for each, (iii) inserting these into
604: eq.~(\ref{eq:avg_cent}) to determine the time-integrated centroid
605: position for a given $T$, and finally (iv) given a $\Delta T$ varying
606: the orbital phase until the maximum
607: $\left| \Delta \bmath{X}_{C,\rm max} \right| (T,\Delta T)$
608: was obtained. The result for $T=1\,\hr$ and $\Delta T=4\,\hr$ is
609: shown by the thin green line in Fig. \ref{fig:drs}, and exhibits
610: many of the features found in the idealized Newtonian case. This
611: procedure was repeated a number of times for randomly chosen
612: $T \in [1\,\hr,4\,\hr]$ and $\Delta T \in [1\,\hr,4\,\hr]$ to produce
613: the upper-envelope of the centroid displacements, and is shown by the
614: thick green line in Fig. \ref{fig:drs}.
615:
616: The primary effect is the suppression of the centroid variability due
617: to the opacity of the background accretion flow. This suppression
618: becomes substantial at radii near the location of the photosphere,
619: which is approximately $12.5 G M_{\rm Sgr A*}/c^2$ at $7\,\mm$ wavelength.
620: In contrast, at large orbital radii the idealized Newtonian model fits
621: quite well. This interpretation is further supported by the fact that the
622: idealized Newtonian model fits quite well once it has been modified to
623: include opacity. Specifically, as a rough approximation, we reduce the
624: hot-spot flux by a factor of $\e^{-\tau}$, where
625: \begin{equation}
626: \tau = \left(\frac{r+\delta r}{12.5}\right)^{-2.99}
627: \propto r n_{e,\rm nth} B^{\alpha_{\rm disk} + 3/2}
628: \end{equation}
629: is the optical depth associated with the non-thermal electrons in the
630: accretion flow. This is shown by the magenta lines in
631: Fig. \ref{fig:drs}. While we should also include the thermal
632: electron component, at $7\,\mm$~wavelength the non-thermal electrons appear to
633: dominate the opacity.
634:
635: The free parameter $\delta r$ is associated with the finite extent of
636: the hot spot. That is, it is possible for the spot to be visible even
637: if the hot-spot center is inside of the accretion-flow photosphere.
638: It is for this reason that the centroid variability is more similar to
639: the idealized Newtonian value for high-luminosity spots than for
640: low-luminosity spots. Setting $\delta r$ using the hot-spot
641: photosphere radius alone gives the dotted magenta line, which
642: underestimates the centroid variability considerably at small radii
643: for large (bright) hot spots. This is due to the failure of the
644: idealized Newtonian calculation to account for the strong lensing of
645: large spots in small orbits (with $r_{\rm spot}\simeq r$). However,
646: simply employing a larger $\delta r$ (64\% larger at the largest
647: disk/hot-spot flux ratio we consider) results in a substantially
648: improved fit.
649:
650: \begin{figure*}
651: \begin{center}
652: \begin{tabular}{cc}
653: \includegraphics[width=0.4\textwidth]{f3a.eps}
654: &
655: \includegraphics[width=0.4\textwidth]{f3b.eps}\\
656: \includegraphics[width=0.4\textwidth]{f3c.eps}
657: &
658: \includegraphics[width=0.4\textwidth]{f3d.eps}\\
659: \end{tabular}
660: \end{center}
661: \caption{Maximum observed physical displacement of the centroid as a
662: function of the orbital period for the idealized Newtonian case
663: (blue) and the fully general relativistic case appropriate for Sgr
664: A* (green) for a number of spot-disk flux ratios.
665: The period and angular offset scales assume a distance of 8.0 kpc and
666: M$_{Sgr A*} = 4 \times 10^6$~\Msun.
667: The thin blue and green lines show the variation associated with an
668: observations integrated over $T=1\,\hr$ and separated by $\Delta
669: T=4\,\hr$. The thick blue and green lines show the envelope for all
670: observation strategies with $T\ge 1\,\hr$ and $\Delta T\le 4$. In
671: all cases the thin dotted blue line shows the maximum deviation
672: possible, corresponding to when all of the flux is due to the hot
673: spot. The
674: solid black lines show the asymptotic power-law behavior at both
675: short and long periods. The dotted and solid magenta lines shows
676: the opacity modified idealized Newtonian case without and with
677: lensing, respectively. Finally, the regions excluded by $7\,\mm$ VLBI
678: are shown by the red hatched regions. For convenience the orbital
679: radii and angular scale of the centroid displacements are shown on
680: the top and right axes, respectively. \label{fig:drs}}
681: \end{figure*}
682:
683: \section{Discussion} \label{C}
684: Possible reasons for position wander include intrinsic
685: variations in the position of the emitting plasma
686: (\eg, variations in the accretion flow or perhaps in a jet)
687: or extrinsic processes such as refractive interstellar scattering.
688: \SgrA\ is observed to be diffractively scattered to a
689: size of $\theta_{sc} \sim 0.5 (\lambda/0.7~{\rm cm})^2$~mas,
690: where $\lambda$ is the observing wavelength.
691: Flux density fluctuations are modest and decrease
692: in strength with increasing wavelength; thus strong refractive
693: scintillations are not indicated \citep{Gwinn:1991}.
694: Any refractive position wander should be
695: $\ll \theta_{sc}$ and should occur on time scales
696: $>\theta_{sc}D/v$, where $D$ is the distance and $v$
697: is the transverse velocity of the scattering ``screen''
698: relative to the observer \citep{Romani:1986}.
699: For $D\approx\Ro\approx8$~kpc \citep{Reid:1993} and
700: $v\sim100$~\kms, characteristic of material in the inner
701: $\sim100$~pc of the Galaxy where large scattering sizes
702: are observed, the refractive time scale is $>10^3$~hours.
703: Thus, we would not expect a significant contribution to the
704: short-term wander of \SgrA\ from refractive scattering.
705: For comparison, \citet{Gwinn:1988}, using VLBI observations
706: of the Sgr~B2(N) \hho\ masers near the Galactic center,
707: find a wander limit of $<18~\muas$ over timescales of months
708: for maser spots, which are diffractively scattered to a
709: comparable size (at 22 GHz) as \SgrA\ (at 43 GHz).
710: Of course, our results provide an observation limit to any
711: refractive position wander.
712:
713: Since extrinsic sources of position wander (scattering) are
714: unlikely to be dominant, we now discuss the implications for
715: intrinsic wander from variations in brightness within an accretion
716: disk given in \S\ref{Constraints}.
717: Our observations of the lack of short-term wander of the centroid
718: position of \SgrA\ presented in \S\ref{section:hours}
719: give an upper limit of $\approx100~\muas$ for time scales of
720: $\approx1~{\rm to}~4$ hours.
721: This translates to an upper limit
722: on the wander versus orbital period plots in Fig.~\ref{fig:drs}
723: as indicated by the horizontal red line and hatched region.
724: (In the very unlikely event that the accretion disk inclination is both
725: near $90^\circ$ and oriented nearly North-South on the sky, we would need
726: to use our NS limits, which are a factor of three weaker.)
727: Our EW limit is below the dotted blue line in Fig.~\ref{fig:drs}, which is
728: for hot spot flux density dominating over disk (or possible
729: jet) emission, for orbital periods exceeding 120~min (corresponding
730: to orbital radii larger than $15\,G M_{\rm Sgr A*}/c^2$ for
731: $M_{\rm Sgr A*}= 4\times10^6$ \Msun). For cases
732: in which the hot spot flux density is weaker than that of the disk,
733: somewhat longer periods are allowed. For example, for
734: $F_{\rm spot}/F_{\rm disk} = 0.37$, orbital periods longer than 5~hr are
735: excluded.
736:
737: In practice, the limits placed by current $7\,\mm$ VLBI are
738: significantly weaker. The limited sensitivity to hot spots on
739: compact orbits is primarily due to two reasons:
740: (1) ``long'' integration times ($T\gtrsim 1\,\hr$) average much
741: of the short time variability out, and
742: (2) the opacity of the accretion flow itself makes it difficult
743: to view hot-spots on compact orbits at $7\,\mm$.
744: The integration time is limited by the sensitivity issues and
745: the small number of antennas yielding interferometer baselines
746: $<1500$~km afforded by the current VLBA; higher bandwidth recording
747: in the future should help alleviate this problem.
748: The optical depth is a property of Sgr A* itself, and can only be
749: addressed by observations at shorter wavelengths.
750: However, even in the absence of an optically thick accretion flow,
751: it is not possible to increase the centroid variability by more than
752: an order of magnitude due to the intrinsically small orbital radii,
753: as seen by comparing the blue and green limits in the lower-left
754: panel of Fig. \ref{fig:drs}.
755:
756: Nevertheless, high-resolution astrometry is reaching sensitivities and
757: resolutions sufficient to begin to test the hot-spot model for bright
758: Sgr A* flares. Unfortunately, the typical fractional variability at
759: $7\,\mm$ during our observations was roughly $\pm20\%$, implying
760: that significant improvement in positional accuracy will be required
761: to constrain such events. Since the observed centroid
762: wander is consistent with systematic errors, owing predominantly to
763: centimeter-scale errors in the modeling of the atmospheric
764: path-delays, substantially increasing accuracy
765: will require better calibration techniques.
766: However, for the somewhat rare instances in which
767: the spot is substantially brighter \citep{Zhao:2001}, the VLBA at
768: 7, or possibly 3,~mm wavelength appears poised to provide significant
769: limits upon the existence and morphology of inhomogeneities in the
770: accretion flow surrounding \SgrA.
771:
772: Ultimately, observations at $\sim1$~mm wavelength with VLBI techniques
773: or at infrared wavelengths with an instrument like GRAVITY
774: \citep{GRAVITY} may be necessary to image the region within
775: $\sim3$ Schwarzschild radii on the short time scales needed to test
776: the hot-spot model.
777:
778: \vskip 1truecm
779: A.B. is supported by the Priority Programme 1177 of
780: the Deutsche Forschungsgemeinschaft.
781: \vskip 0.5truecm
782: {\it Facilities:} \facility{VLBA}
783: \begin{thebibliography}
784:
785: \bibitem[see, \eg, Bower \etal (2004)]{Bower:2004}
786: Bower, G. C., Falcke, H., Herrnstein, R. M., Zhao, J.-H.,
787: Goss, W. M. \& Backer, D. C. 2004, Science, 304, 704
788:
789: \bibitem[Broderick \& Blandford (2004)]{Brod-Blan:2004}
790: Broderick, A. \& Blandford, R., 2004, \mnras, 349, 994
791:
792: \bibitem[Broderick \& Loeb (2005)]{Brod-Loeb:2005}
793: Broderick, A.~E. \& Loeb, A., 2005, \mnras, 363, 353
794:
795: \bibitem[Broderick \& Loeb (2006)]{Brod-Loeb:2006}
796: Broderick, A.~E. \& Loeb, A., 2006, \mnras, 267, 905
797:
798: \bibitem[Dehnen \& Binney (1998)]{Dehnen-Binney:1998}
799: Dehnen, W. \& Binney, J.~J 1998, \mnras, 298,387
800:
801: \bibitem[Ghez \etal (2003)]{Ghez:2003}
802: Ghez, A. M. \etal 2003, \apj, 586, L127
803:
804: \bibitem[Gillessen \etal (2006)]{GRAVITY}
805: Gillessen \etal 2006, SPIE, 6268, 33
806:
807: \bibitem[Gwinn \etal (1988)]{Gwinn:1988}
808: Gwinn, C.~R., Moran, J.~M., Reid, M.~J. \& Schneps, M.~H.
809: 1988, \apj, 330, 817
810:
811: \bibitem[Gwinn \etal (1991)]{Gwinn:1991}
812: Gwinn, C.~R., Danen, R.~M., Middleditch, J., Ozernoy, L.~M.
813: \& Tran, T. KH. 1991, \apj, 381, L43
814:
815: \bibitem[Jones \& O'Dell (1977)]{Jone-ODell:1977}
816: Jones, T.~W. \& O'Dell, S.~L., 1977, \apj, 214, 522
817:
818: \bibitem[Menten \etal (1997)]{Menten:1997}
819: Menten, K.~M., Reid, M.~J., Eckart, A. \& Genzel, R.
820: 1997, \apj, 475, L111
821:
822: \bibitem[Reid (1993)]{Reid:1993}
823: Reid, M. J. 1993, \araa, 31, 345
824:
825: \bibitem[Reid \etal (1999)]{Reid-etal:1999}
826: Reid, M.~J., Readhead, A.~C.~S., Vermeulen, R.~C. \&
827: Treuhaft, R.~N. 1999, \apj, 524, 816
828:
829: \bibitem[Reid \etal (2003)]{Reid:2003}
830: Reid, M.~J., Menten, K.~M., Genzel, R., Ott, T.,
831: Sch\"odel, R. \& Eckart, A. 2003, \apj, 587, 208
832:
833: \bibitem[Reid \& Brunthaler (2004)]{Reid-Brun:2004}
834: Reid, M.~J. \& Brunthaler, A. 2004, \apj, 616, 872
835:
836: \bibitem[Romani, Narayan \& Blandford (1986)]{Romani:1986}
837: Romani, R. W., Narayan, R. \& Blandford, R. 1986,
838: \mnras, 220, 19
839:
840: \bibitem[Sch\"odel \etal (2002)]{Schoedel:2002}
841: Sch\"odel, R. \etal 2002, \nat, 419, 694
842:
843: \bibitem[Yuan et al.~(2003)]{Yuan-Quat-Nara:2003}
844: Yuan, F., Quataert, E. \& Narayan, R., 2003, \apj, 598, 301
845:
846: \bibitem[Zhao, Bower \& Goss (2001)]{Zhao:2001}
847: Zhao, J.-H., Bower, G. C. \& Goss, W. M. 2001, \apj, 547, L29
848:
849: \end{thebibliography}
850:
851: \end{document}
852: