1: % This is to emulate the later appearance of the paper
2: \documentclass[apj]{emulateapj}
3: %\documentclass[12pt,preprint]{aastex}
4: \slugcomment{Draft version 01/07/2008}
5: \usepackage{graphicx,amssymb,amsmath,times}
6:
7: %\setcounter{page}{1}
8:
9: \def\Journal#1#2#3#4{{#4}, {#1}, {#2}, #3}
10: \def\swift{{\it Swift}/BAT}
11: \def\asm{{\it RXTE}/ASM}
12: \def\ariel{{\it Ariel\,V}/ASM}
13: \def\vela{{\it Vela\,5B}/ASM}
14: \def\numin{$\nu_1$}
15: \def\numax{$\nu_{m/2}$}
16: \def\nupeak{$\nu_\textrm{peak}$}
17: \def\cyg{Cygnus~X-1}
18: \def\lsi{LS\,I~\,+61~303}
19: \def\cenx{Centaurus~X-4}
20: \def\mod{$A$}
21: \def\batse{{\it CGRO}/BATSE}
22: \def\pretrial{${\cal P}_\textrm{pre}$}
23: \def\postrial{${\cal P}_\textrm{post}$}
24:
25: \begin{document}
26: \title{The variable super-orbital modulation of Cygnus X-1}
27:
28: \shorttitle{The variable super-orbital modulation of Cygnus X-1}
29: \shortauthors{Rico}
30:
31: \author{
32: Javier Rico\altaffilmark{1,2}%\altaffilmark{b}
33: }
34:
35: %\affiliation{
36: \altaffiltext{1} {Instituci\'o Catalana de Recerca i Estudis
37: Avan\c{c}ats (ICREA), E-08010 Barcelona, Spain. E-mail: jrico@icrea.cat}
38: \altaffiltext{2} {Institut de F\'{\i}sica d'Altes Energies (IFAE), Edifici Cn., Campus UAB, E-08193
39: Bellaterra, Spain. E-mail: jrico@ifae.es}
40: %}
41:
42:
43: %%%%%%%%%%%%%%%%%%%%
44: \begin{abstract} %%
45: %%%%%%%%%%%%%%%%%%%%
46: We study the super-orbital modulation present in the \cyg\ X-ray data,
47: usually attributed to the precession of the accretion disk and
48: relativistic jets. We find a new, strong, 326$\pm$2~d period
49: modulation starting in 2005, in \swift\ and \asm\ light curves
50: (LCs). We also investigate \vela\ and \ariel\ archival data and
51: confirm the previously reported $\sim$290~d periodic modulation, and
52: therefore confirming that the super-orbital period is not
53: constant. Finally, we study \asm\ LC before 2005 and find that the
54: previously reported $\sim$150~d period is most likely an artifact due
55: to the use of a Fourier-power based analysis under the assumption that
56: the modulation has a constant period along the whole data
57: sample. Instead, we find strong indications of several discrete
58: changes of the precession period, happening in coincidence with soft
59: and failed state-transition episodes. We also find a hint of
60: correlation between the period and the amplitude of the
61: modulation. The detection of gamma-rays above 100 GeV with MAGIC in
62: September 2006 happened in coincidence with a maximum of the
63: super-orbital modulation. The next maximum will happen between 2 and
64: 14 of July 2008, when the observational conditions of
65: \cyg\ with ground-based Cherenkov telescopes, such as MAGIC and
66: VERITAS, are optimal.
67: \end{abstract}
68:
69: \keywords{binaries: general --- X-rays: individual (Cygnus~X-1)}
70:
71:
72: %%%%%%%%%%%%%%%%%%%%%%%%%
73: \section{Introduction} %%
74: %%%%%%%%%%%%%%%%%%%%%%%%%
75:
76: Cygnus X-1\citep{Bowyer1965} is the best established candidate for a
77: stellar mass black-hole (BH). It is composed of a $21\pm
78: 8$~M$_\odot$ BH turning around an O9.7~Iab companion of $40\pm
79: 10$~M$_\odot$ \citep{Ziolkowski2005} in a circular orbit of 5.6 days
80: \citep{Brocksopp99a}. High
81: resolution radio imaging has unveiled the presence of a highly
82: colimated, relativistic \citep{Stirling2001}, radiatively inefficient
83: \citep{Gallo2005} jet. The X-ray source displays soft and hard states
84: and relatively frequent failed transitions between them. There are
85: strong evidences of a high energy non-thermal component extending up
86: to soft gamma-rays \citep{Mcconnell2002,Cadolle2006}. The steady
87: emission of gamma-rays above 100 GeV is strongly constrained by the
88: observations with MAGIC which has obtained, however, a very strong
89: evidence of an intense, fast flaring episode at these energies
90: \citep{Albert07}.
91:
92: A $\sim$5.6~d period modulation, attributed to the orbital motion of
93: the compact object around the companion, has been observed at various
94: wavelengths by numerous authors
95: \citep[e.g.,][]{Pooley99,Brocksopp99a,Brocksopp99b,LaSala98,Lachowicz06}.
96: On the other hand, a super-orbital $\sim$290~d period was claimed by
97: \citet{Priedhorsky83} on the soft X-ray data recorded by \vela\
98: (1969-1979) and \ariel\ (1974-1980). Later, a $\sim$150~d periodic
99: variability has been reported by various authors
100: \citep{Pooley99,Brocksopp99b,Ozdemir01,Benlloch01,Benlloch04,Lachowicz06}
101: using different data samples ranging between April 1991 and November
102: 2003. It must be noted, however, that other significant modulations
103: with periods $\sim$200~d and $\sim$420~d have been also found
104: \citep[e.g.][]{Benlloch01,Benlloch04,Lachowicz06}.
105:
106:
107: In this letter, we search the latest \cyg\ X-ray data for periodic
108: modulations. We also perform a critical revision of the previous
109: results obtained from archival X-ray data. Finally, we put our results
110: in the context of a multiwavelength description of the source.
111:
112:
113: %%%%%%%%%%%%%%%%%%%%%%%%%
114: \section{Data samples} %%
115: %%%%%%%%%%%%%%%%%%%%%%%%%
116: \label{sec:data}
117:
118: \begin{deluxetable*}{lcccccccc}
119: \tablewidth{0pt}
120: \tabletypesize{\footnotesize}
121: \tablecaption{Cygnus X-1 analyzed data\label{tab:data}}
122: \tablehead{
123: \colhead{Instrument/} & \colhead{Energy} & \colhead{Operation} & \colhead{Start} & \colhead{End} & \colhead{Time} & \colhead{Number of} & \colhead{\numin} & \colhead{\numax} \\
124: \colhead{subsample} & \colhead{[keV]} & \colhead{time} & \colhead{[MJD]} & \colhead{[MJD]} & \colhead{span [d]} & \colhead{points (m)} & \colhead{[d$^{-1}$]} & \colhead{[d$^{-1}$]}
125: }
126: \startdata
127: \vela & 3-12 & May 1969--Jun 1979 & 40368 & 44042 & 3675 & 1097 & 2.7$\times 10^{-4}$ & 0.15 \\
128: \ariel & 3-6 & Feb 1976--Feb 1980 & 42830 & 44292 & 1464 & 740 & 6.8$\times 10^{-4}$ & 0.25 \\
129: \asm\,1& 2-10 & Sep 1996--Sep 1999 & 50328 & 51444 & 1117 & 913 & 9.0$\times 10^{-4}$ & 0.41 \\
130: \asm\,2& 2-10 & Jun 2005--May 2008 & 53529 & 54592 & 1064 & 909 & 9.4$\times 10^{-4}$ & 0.43 \\
131: \swift &15-150 & Jun 2005--May 2008 & 53529 & 54598 & 1070 & 829 & 9.3$\times 10^{-4}$ & 0.39
132: \enddata
133: \end{deluxetable*}
134:
135: The data samples analyzed in this work are summarized in
136: Table~\ref{tab:data}. They are available through the High Energy
137: Astrophysics Archive Research Center (HEASARC). We use data from four
138: different instruments, namely: \vela, \ariel\, \asm\ and \swift. All
139: data are averaged into one-day bins, except when explicitely
140: stated. No periodic behavior is found in the X-ray data during the
141: soft state \citep{Wen99,Lachowicz06} and therefore we analyze data
142: corresponding to the hard state data only. The interval MJD
143: 42338--42829 is dominated by soft flare events \citep{Liang84} and
144: hence excluded from \vela\ and \ariel\ analyses. The soft state
145: periods during the operation of \asm\ are identified as those for
146: which the ratio of count rates in the bands C (5.0-12.1 keV) and A
147: (1.3--3.0 keV) is lower than 1.2 and the total count rate exceeds the
148: mean value by more than 4 standard deviations. The mean and standard
149: deviations are computed from the interval MJD 50660--50990
150: \citep{Lachowicz06}. This excludes from the analysis the following
151: periods (MJD): 50087--50327, 50645--50652, 51002--51026, 51369--51397,
152: 51445--51625, 51776--51952, 52093--52584, 52762--52878, 52982--53092,
153: 53198--53528, 53780--53872. After this, two long intervals dominated
154: by hard state (samples A and B in Table~\ref{tab:data}) are defined
155: and studied separately. Based on the results for \asm, the intervals
156: 53414--53528 and 53780--53872 are also removed from the analysis of
157: \swift\ LC.
158:
159: %%%%%%%%%%%%%%%%%%%%%
160: \section{Analysis} %%
161: %%%%%%%%%%%%%%%%%%%%%
162: \label{sec:analysis}
163:
164: We search the different data samples for periodic signals using the
165: Lomb-Scargle (L-S) test of uniformity \citep{Lomb76,Scargle82}. The
166: \emph{chance probability} is the probability of obtaining a certain
167: L-S test value ($z_0$) or larger out of a purely Gaussian noise
168: sample, and is given by ${\cal P}_\textrm{pre}(z>z_0) =
169: e^{-z_0}$. When several frequencies are inspected, the {\it
170: post-trial} probability, i.e. the probability to get a L-S test value
171: $z_0$ or higher for \emph{at least} one of the scanned frequencies, is
172: given then by ${\cal P}_\textrm{post}(z>z_0) = 1 - [1-{\cal
173: P}_\textrm{pre}(z>z_0)]^n$ where $n$ is the number of
174: \emph{independent} scanned frequencies.
175:
176: Given $m$ data points, there is a discrete finite set of $m/2$
177: independent frequencies. For the case of evenly spaced data there is a
178: natural set of frequencies: $\nu_k = \frac{k}{T}$ ($k=1,\dots,m/2$)
179: where $T$ is the time spanned by the data set. The values of the
180: Fourier transform powers for the natural frequencies are independent
181: of one another. The data set does not contain enough information to
182: search for periodicities below $\nu_1 = 1/T$ or above $\nu_{m/2} =
183: m/2T$. The time span, number of data points and the maximum and
184: minimum accessible frequencies for the different studied data samples
185: are shown in Table~\ref{tab:data}.
186:
187: For each investigated data sample we produce the \emph{periodogram},
188: where $-\log_{10}({\cal P}_\textrm{post})$ is represented as a
189: function of the frequency. A prominent periodic component in the data
190: is visible as a peak in the periodogram at the relevant frequency
191: (\nupeak). We consider as significant those peaks for which the
192: post-trial probability is lower than $10^{-6.5}$, equivalent to a
193: deviation of 5$\sigma$ from the Gaussian noise case. We scan all
194: natural frequencies, with an oversampling factor of 5. This means that
195: we scan $5m/2$ evenly spaced frequencies, from
196: \numin\ to \numax. The oversampling does not increase the number of
197: trials in the post-trial chance probability, since the number of
198: independent frequencies remains constant, but increases the precision
199: of \nupeak. To estimate the error ($\Delta$\nupeak), we use the
200: standard deviation of \nupeak\ over 100 random data samples obtained
201: by bootstrap \citep{Davison06} of the original LC. Then, we
202: fold the LC into a
203: \emph{phaseogram} using the period $1/$\nupeak. The phaseogram is
204: produced using 50 bins, to ensure a smooth description of the
205: waveform. The time of the phase 0 ($T_0$) is determined from a fit to
206: the LC using a Cosine function, where the value of the
207: frequency is fixed to \nupeak. In this way, $T_0$ corresponds to the
208: maximum of the fitting Cosine function (although not necessarily to
209: the maximum of the waveform). The modulation amplitude (\mod) is
210: defined as the ratio between the amplitude and the mean value of the
211: Cosine function obtained from the fit. Finally, we remove the periodic
212: component of frequency \nupeak\ from the LC (a process called
213: \emph{prewhitening}). This is done by subtracting the deviations of
214: the phaseogram from its mean value throughout the LC.
215:
216: We subsequently search for the next most prominent peak in the
217: prewhitened LC, and follow the whole process described above
218: in an iterative fashion. This process is stopped when the obtained
219: \nupeak\ has a post-trial chance probability larger then $10^{-6.5}$.
220:
221:
222: %%%%%%%%%%%%%%%%%%%%%%
223: \section{Results} %%
224: %%%%%%%%%%%%%%%%%%%%%%
225:
226: We first search for periodic signals in \swift\ and \asm\,2 data
227: samples, which correspond to the same epoch, and which are analyzed in
228: this work for the first time. The results are shown in
229: Table~\ref{tab:results}. The L-S periodograms for both LCs are shown
230: in Figure~\ref{fig:asm_swift_per}. A very strong, dominant periodic
231: signal with period 326$\pm$2~d is found in both data samples. The LCs
232: are shown in Figure~\ref{fig:asm_swift_lc}. The modulation is clearly
233: seen by eye, which is reflected by the extremely low values of
234: \postrial. The previously reported $\sim$150~d modulation is not found
235: in these data. The pre-trial chance probabilities for such a
236: modulation are $10^{-1.3}$ and $10^{-0.1}$ for \asm\ and
237: \swift\ data, respectively. Figure~\ref{fig:asm_swift_per} shows the
238: $\sim 150$~d modulation as reported by \citet{Lachowicz06} overlaid with
239: \asm\,2 LC, confirming that such a modulation does not describe well the
240: data. The phaseograms corresponding to the 326~d period are shown in
241: Figure~\ref{fig:phaseogram}. We see that hard and soft X-ray LCs are
242: strongly correlated, with Pearson's correlation factor $r=0.97$. A
243: second, also strong, component with period $\sim$1000~d is present in
244: both data samples. This corresponds to a long-term modulation of the
245: X-ray flux with respect to the 326~d oscillation, but cannot be
246: established as periodic since the period is similar to the total time
247: spanned by the observations. An alternative explanation to the
248: $\sim$1000~d period will be given below. Finally, a third component is
249: seen in \asm\,2 data sample at period $\sim$5.6~d, compatible with the
250: orbital modulation, visible in the periodogram
251: (Figure~\ref{fig:asm_swift_per}) even before prewhitening. This
252: modulation is not seen in \swift\ data, for which ${\cal
253: P}_\textrm{pre} \simeq 10^{-2}$. On a similar energy band,
254: \citet{Paciesas97} claimed a modulation compatible with the orbital
255: period in \batse\ LC between April 1991 and September 1996, but the
256: method used lacks of a mathematical
257: justification. \citet{Brocksopp99b} did not find any evidence for the
258: orbital modulation in \batse\ data between May 1996 and September
259: 1998. Finally, \citet{Lachowicz06}, using the whole BATSE light curve,
260: reported a deviation of $\sim 1\sigma$ from the Gaussian noise case,
261: insufficient to establish the presence of the orbital modulation in
262: the data.
263:
264: We have searched \vela\ and \ariel\ LCs for periodic
265: modulations and found peaks at P=(276$\pm$3)~d and P=(288$\pm$3)~d,
266: respectively (see Table~\ref{tab:results}), in agreement with the
267: results obtained by \citet{Priedhorsky83} and
268: \citet{Lachowicz06}. By comparison with the P=(326$\pm$3)~d present in
269: \swift and \asm\,2 LC, this shows that the period of the super-orbital
270: modulation is variable. The corresponding phaseograms are shown if
271: Figure~\ref{fig:phaseogram} (two lowermost panels). Even if the
272: relative dispersion of the points is larger due to the lower
273: sensitivity of \ariel\ and \vela, the waveform is still visible. They
274: follow a very similar shape as those found for \swift\ and \asm\,2,
275: albeit for a different period, as one expects if the underlying
276: physical process is the same. The correlation factor for \swift\ and
277: \ariel\ (\vela) phaseograms is $r=0.73$ ($r=0.57$). However, the
278: modulation amplitudes are significantly lower than for the case of
279: \asm. This could have a physical explanation, but it could also happen
280: if the periodic modulation was not present in part of the LCs, which
281: can be certainly not excluded. On the other hand, we do not find
282: evidence for the orbital period in
283: \vela\ or \ariel\ LCs\footnote{We note that shorter integration times
284: have been used for this search in the
285: \vela\ LC since, with 1-day bins, the minimum
286: accessible period is P=6.7~d, as shown in Table~\ref{tab:data}}. It
287: is worth noting that, given \vela\ and \ariel\ sensitivities, we do
288: not expect to detect an orbital modulation with an amplitude of
289: $\sim$5$\%$ as the one we see in \asm\,2 data. The mean relative
290: variance of the data points in the phaseogram (which is a good
291: estimate of the measurement error) are $46\%$ and $13\%$ for
292: \vela\ and \ariel\, respectively. Both values are well above the $5\%$
293: modulation which is hence difficult to detect. We have crosschecked
294: this by analyzing Monte Carlo (MC) simulated LCs for \vela\
295: and \ariel. We use the same sampling as the measured LCs and
296: simulate a $5\%$ amplitude modulation convolved with $46\%$ and $13\%$
297: point-to-point random fluctuations, respectively. The analysis of
298: these LCs yields no significant peak.
299:
300: \begin{deluxetable}{lr@{$\pm$}lrrl}
301: \tablewidth{0pt}
302: \tablecaption{Results for periodic signal search\label{tab:results}}
303: \tablehead{
304: \colhead{Sample} & \multicolumn{2}{c}{P$_\textrm{peak}$} & \colhead{$T_0$} & \colhead{\mod} & ${\cal P}_\textrm{post}$ \\
305: & \multicolumn{2}{c}{[d]} & \colhead{[MJD]} & \colhead{[$\%$]} &
306: }
307: \startdata
308: \swift & 326 & 2 & 54027.3 & 25 & $10^{-87}$ \\
309: & 1030 & 50 & 53707.2 & 6 & $10^{-10}$ \\
310: %\hline
311: \asm\,2 & 326 & 2 & 54032.2 & 29 & $10^{-96}$ \\
312: & 990 & 30 & 53670.7 & 13 & $10^{-41}$ \\
313: & 5.600 & 0.002 & 53670.3 & 5 & $10^{-11}$ \\
314: \asm\,1a & 248 & 9 & 50377.2 & 29 & $10^{-30}$ \\
315: \asm\,1b & 123 & 3 & 50761.1 & 11 & $10^{-10}$ \\
316: \asm\,1c & 168 & 4 & 51177.6 & 18 & $10^{-24}$ \\
317: \asm\,1 & 5.602 & 0.002 & 51117.1 & 4.1 & $10^{-7.1}$ \\
318: \ariel & 276 & 3 & 42865.8 & 14 & $10^{-16}$ \\
319: \vela & 288 & 3 & 40187.4 & 21 & $10^{-6.8}$
320: \enddata
321: \end{deluxetable}
322:
323: \begin{figure}
324: \epsscale{1}
325: \plotone{f1.eps}
326: \figcaption[f1.eps]{Periodograms for \swift\ and
327: \asm\,2 \cyg\ samples, showing the post-trial chance probability as a
328: function of the scanned frequency. The horizontal line marks the line
329: corresponding to a post-trial probability of $10^{-6.5}$. The arrow
330: marks the orbital frequency.
331: \label{fig:asm_swift_per}}
332: \end{figure}
333:
334: \begin{figure}
335: \epsscale{1}
336: \plotone{f2.eps}
337: \figcaption[f2.eps]{\swift\ and \asm\,2 LCs. The
338: shaded area shows an interval of soft state, identified by the
339: criteria exposed in Section~\ref{sec:data}, and not considered in the
340: analysis. The thick, red curves are the fits by Cosine functions to
341: each subsample (see Section~\ref{sec:analysis}). The thin, green curve
342: represents the $\sim 150$~d super-orbital modulation using the
343: ephemeris reported by
344: \citet{Lachowicz06}. The vertical, purple line marks the time of the
345: TeV signal reported by
346: \citet{Albert07}
347: \label{fig:asm_swift_lc}}
348: \end{figure}
349:
350: \begin{figure}
351: \epsscale{1}
352: \plotone{f3.eps}
353: \figcaption[f3.eps]{From top to bottom:
354: \swift, \asm\,2, \ariel\ and \vela\ phaseograms folded using
355: period P and time 0 values (P [d],$T_0 \textrm{[MJD]}$)= (326,
356: 54027), (326, 54027), (276, 42866) and (288, 40187), respectively.
357: The values of $T_0$ are obtained from the fit of a Cosine
358: function. Data points and error bars correspond, respectively, to the
359: mean count rate and variance measured within each phase bin. The
360: vertical, red, dashed line corresponds to the phase of the TeV signal
361: reported by \citet{Albert07}. \label{fig:phaseogram}}
362: \end{figure}
363:
364:
365: \begin{figure}
366: \epsscale{1}
367: \plotone{f4.eps}
368: \figcaption[f4.eps]{\asm\,1 light curve. The vertical
369: shaded lines show the intervals of soft or failed transition states,
370: identified by the criteria exposed in Section~\ref{sec:data}, which
371: delimit the three subsamples (1a, 1b and 1c) used for further
372: analysis. The red curves are the fits by Cosine functions to each
373: subsample (see Section~\ref{sec:analysis}).
374: \label{fig:asmA_lc}}
375: \end{figure}
376:
377:
378: Finally, we have searched \asm\,1 data sample for periodic
379: modulations, and found 5 significant peaks at P = (148$\pm$1),
380: (188$\pm$2), (310$\pm$11), (475$\pm$11) and (5.598$\pm$0.003)~d, all
381: with chance probabilities lower than $10^{-10}$. The latter
382: corresponds to the orbital modulation, whereas the other four seem to
383: denote a complex power spectrum. We stress that some of these peaks
384: have been found in previous studies of the \asm\ LC
385: \citep{Benlloch01,Benlloch04,Lachowicz06}. The understanding of the
386: super-orbital modulation can be greatly simplified if we consider that
387: the period can change along the observation time in a discrete way. We
388: have analyzed separately the data between each two consecutive soft or
389: failed transition states, i.e.\ three samples, namely
390: 1a=[50328--50644], 1b=[50653--51001] and 1c=[51027--51368] (see
391: Figure~\ref{fig:asmA_lc}). We obtain a single significant peak in each
392: of them (see Table~\ref{tab:results}). We prewhiten the leading
393: frequency in each of the subsamples and merge them to look for
394: sub-leading frequencies. The only remaining significant peak
395: corresponds to the orbital modulation (see
396: Table~\ref{tab:results}). Figure~\ref{fig:asmA_lc} shows \asm\,1 LC,
397: overlaid with the result of the fits to independent Cosine functions
398: to the three defined subsamples. The general agreement with the data
399: is remarkably good. We have also generated three MC samples
400: corresponding to the samplings of 1a, 1b and 1c subsamples, and pure
401: sinusoidal modulations with the periods found for each of them,
402: convolved with the measured point-to-point fluctuations. Then we have
403: merged the three samples together and analyzed the resulting LC. We
404: obtain significant peaks at P = (147$\pm$1), (190$\pm$2) and
405: (523$\pm$12)~d, in surprisingly good agreement with the results of
406: analyzing the real data. We note that this effect could be also
407: responsible of the $\sim$1000~d periodicity of the \swift\ and \asm\,2
408: data, since they also contain a soft state episode that might have
409: changed the period of the super-orbital modulation prior to MJD=53780.
410:
411:
412:
413: %%%%%%%%%%%%%%%%%%%%%%%%%
414: \section{Discussion} %%
415: %%%%%%%%%%%%%%%%%%%%%%%%%
416:
417: We find that \cyg\ displays a super-orbital modulation, with a period
418: that changes, probably in a discrete way and in coincidence with soft
419: or failed state-transition phases, over time scales ranging from a few
420: hundred days to several years. According to our findings, the very
421: much cited $\sim$150~d period is most probably an artifact of applying
422: a (sometimes biased) Fourier-transform based analysis to a data sample
423: where more than one consecutive period modulations are present. Since
424: 2005, \cyg\ shows a very powerful and stable super-orbital modulation
425: with a period of $326\pm2$~d.
426:
427: The super-orbital modulation is usually attributed to the precession of
428: the accretion disk~\citep{Priedhorsky83} and relativistic
429: jet~\citep{Romero02}, as a result of the tidal forces exerted by the
430: companion star on a tilted disk~\citep{Katz73}. A mechanism for
431: keeping the disk tilted can be provided by radiation pressure
432: warping~\citep{Petterson77,Pringle96,Wijers99,Ogilvie01}. In the
433: case of tidally forced precession, the expected period
434: $P_\textrm{prec}$ depends on the outer radius $R_o$ and inclination of
435: the disk $\delta$ as $P_\textrm{prec} \propto R_o^{-3/2}
436: \cos^{-1}\delta$ \citep{Larwood98}. Then, the longer the period the
437: larger the precession angle, and hence also a larger modulation
438: amplitude is expected. This is in agreement with our results for
439: \asm, where we have found four different super-orbital periods, which
440: follow this tendency (see Table~\ref{tab:results} and
441: Figure~\ref{fig:asmA_lc}).
442:
443: A different issue is why the precession of the disk produces a
444: modulation in the X-ray flux. Some authors
445: \citep[e.g.,][]{Lachowicz06,Ibragimov07} have considered the
446: possibility that the precession movement changes the optical thickness
447: along the line of sight. They reject this possibility since it seems
448: unlikely due to the fine tuning required to produce the observed
449: modulation amplitude, which in addition should depend on the energy,
450: which is not confirmed by our observations. The multiwavelenth data
451: seem to support a scenario where the emission itself is
452: anisotropic. The precession modulation is detected at similar times
453: with identical periods in radio, soft and hard X-rays during the hard
454: state. A unified picture, where the anisotropy is provided by the jet,
455: has been proposed by \citet{Brocksopp99b}. The soft X-ray emission is
456: produced in the disk via bremsstrahlung of thermal electrons, and are
457: then up-scattered to higher energies via Compton scattering in the hot
458: corona or at the base of the relativistic jet (which precesses with
459: the disk). The acceleration of electrons along magnetic field lines in
460: the jet produces the radio emission by synchrotron emission. During
461: the soft state, the jet and corona dissappear and no modulation is
462: observed. According to our findings, once the source goes back to the
463: hard state, the reconstructed disk and jet have different kinematical
464: properties, and the modulation period changes. It seems that failed
465: transitions produce a similar effect.
466:
467: MAGIC detected a fast and intense episode of emission of gamma-rays
468: above 100~GeV \citep{Albert07} during MJD=54003, albeit at the limit
469: of the telescope's sensitivity. This happened in coincidence with the
470: soft and hard X-ray maxima (Figures~\ref{fig:asm_swift_lc} and
471: \ref{fig:phaseogram}) and an unusually bright outburst detected with
472: {\it INTEGRAL} \citep{Malzac08}. It is interesting to note that,
473: according to the ephemeris shown in Table~\ref{tab:results}, the next
474: passage for the precession maximum will happen at MJD=54655$\pm$2,
475: i.e. between 6 and 10 of July 2008. The observational conditions of
476: \cyg\ with ground-based Cherenkov telescopes, such as MAGIC and
477: VERITAS, will be optimal during those days.
478:
479: %\vspace{0.2cm}
480: \acknowledgments I would like to warmly thank the help, discussions and comments
481: to the draft from Emma de O\~na-Wilhelmi, Roberta Zanin, Diego Torres,
482: Daniel Mazin, Juan Cortina and Miguel A. P\'erez-Torres.
483:
484:
485: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
486: \begin{thebibliography}{99} %%
487: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
488:
489:
490: \bibitem[Albert et al.\ (2007)]{Albert07}
491: Albert, J., et al.\ 2007, ApJ 665, L51
492:
493: \bibitem[Benlloch et al.\ (2001)]{Benlloch01}
494: Benlloch, S., et al., 2001, ESASP, 459,
495: 236
496: %Benlloch, S., Wilms, J., Staubert, R. \& Nowak, M., 2001, ESASP, 459,
497: %263
498:
499: \bibitem[Benlloch et al.\ (2004)]{Benlloch04}
500: Benlloch, S. et al., 2004, AIPC, 714, 61
501:
502: \bibitem[Bowyer et al.(1965)]{Bowyer1965}
503: Bowyer, S., Byram, E. T., Chubb, T. A., \& Friedman, H. 1965, Science,
504: 147, 394
505:
506: \bibitem[Brocksopp et al.\ (1999a)]{Brocksopp99a}
507: Brocksopp, C., et al. 1999a, \mnras, 309, 1063
508:
509: \bibitem[Brocksopp et al.\ (1999b)]{Brocksopp99b}
510: Brocksopp, C., et al., 1999b, A\&A 343, 861
511: %Brocksopp, C., Tarasov, A. E., Lyuty, V. M., Roche P., 1999b, A\&A 343, 861
512:
513: \bibitem[Cadolle Bel et al.(2006)]{Cadolle2006}
514: Cadolle Bel, M., et al. 2006, \aap, 446, 591
515:
516: \bibitem[Davison \& Hinkley (2006)]{Davison06}
517: Davison, A. C., Hinkley, D., 2006, ``Bootstrap Methods and their
518: Applications'' 8th edition, Cambridge: Cambridge Series in Statistical
519: and Probabilistic Mathematics.
520:
521: \bibitem[Gallo et al.(2005)]{Gallo2005}
522: Gallo, E., et al., 2005, \nat, 436, 819
523: %Gallo, E., Fender, R.~P., Kaiser, C., Russell, D., Morganti, R.,
524: %Oosterloo, R., \& Heinz, S.. 2005, \nat, 436, 819
525:
526: \bibitem[Ibragimov, Zdziarski \& Poutanen (2007)]{Ibragimov07}
527: Ibragimov, A., Zdziarski, A., \& Poutanen, J., 2007, \mnras, 381, 723
528:
529: \bibitem[Katz (1973)]{Katz73}
530: Katz, J. I., 1973, Nat.\ Phys.\ Sci.\, 246, 87
531:
532: \bibitem[Lachowicz et al.\ (2006)]{Lachowicz06}
533: Lachowicz, P., et al., 2006, \mnras, 368, 1025
534: %Lachowicz, P., Zdziarski, A. A., Schwarzenberg-Czerny, A., Pooley,
535: %G. G., \& Kitamoto, S., 2006, \mnras, 368, 1025L
536:
537: \bibitem[Larwood (1998)]{Larwood98}
538: Larwood, L., 1998, \mnras, 299, L32
539:
540: \bibitem[LaSala et al.\ (1998)]{LaSala98}
541: LaSala, J., et al., 1998, \mnras, 301, 285
542: %LaSala, J., Charles, P. A., Smith, R. A. D., Balucinska-Church, M., \&
543: %Church, M. J. 1998, \mnras, 301, 285L
544:
545: \bibitem[Liang \& Nolan (1984)]{Liang84}
546: Liang, E. P., Nolan, P. L., 1984 SSRv, 38, 353
547:
548: \bibitem[Lomb (1976)]{Lomb76}
549: Lomb, N. R., 1976, Astrophys. Space Sci., 39, 447
550:
551: \bibitem[Malzac et al. (2008)]{Malzac08}
552: Malzac, J., et al., 2008, A\&A submitted, arXiv:0805.4391v1 [astro-ph]
553:
554: \bibitem[McConnell et al.(2002)]{Mcconnell2002}
555: McConnell, M. L., et al. 2002, \apj, 572, 984
556:
557: \bibitem[Ogilvie \& Dubus (2001)]{Ogilvie01}
558: Ogilvie, G. I, \& Dubus, G., 2001, MNRAS, 320, 485
559:
560: \bibitem[Ozdemir \& Demircan (2001)]{Ozdemir01}
561: Ozdemir, S., \& Demircan, O., 2001, A\&SS, 278, 319
562:
563: \bibitem[Paciesas et al. (1997)]{Paciesas97}
564: Paciesas W. S et al. 1997, AIPC, 410, 834
565:
566: \bibitem[Petterson (1977)]{Petterson77}
567: Petterson, J. A, 1977, \apj, 216, 827
568:
569: \bibitem[Pooley et al. (1999)]{Pooley99}
570: Pooley, G. G., Fender, R. P., \& Brocksopp, C., 1999, \mnras, 302, L1
571:
572: \bibitem[Priedhorsky, Terrel \& Holt (1983)]{Priedhorsky83}
573: Priedhorsky, W. W., Terrel, J. \& Holt, S. S., 1983, \apj, 270, 233
574:
575: \bibitem[Pringle (1996)]{Pringle96}
576: Pringle, J. E., 1996, MNRAS, 281, 357
577:
578: \bibitem[Romero, Kaufman Bernad\'o \& Mirabel (2002)]{Romero02}
579: Romero, G. E., Kaufman Bernad\'o, M. M. \& Mirabel, F., 2002, A\&A,
580: 393, L61
581:
582: \bibitem[Scargle (1982)]{Scargle82}
583: Scargle, J. D., 1982, \apj, 263, 835
584:
585: \bibitem[Stirling et al.(2001)]{Stirling2001}
586: Stirling, A.~M., et al., 2001, \mnras, 327, 1273
587: %Stirling, A.~M., Spencer, R.~E., de la Force, C.~J., Garrett, M.~A.,
588: %Fender, R.~P., \& Ogley, R.~N.\ 2001, \mnras, 327, 1273
589:
590: \bibitem[Wen et al.\ (1999)]{Wen99}
591: Wen, L., Cui, W., Levine, A. M., Bradt, H. V., 1999, \apj, 525, 968
592:
593: \bibitem[Wijers \& Pringle (1999)]{Wijers99}
594: Wijers, R. A. M. J., \& Pringle, J. E., 1999, MNRAS, 308, 207
595:
596: \bibitem[Zi\'o{\l}kowski(2005)]{Ziolkowski2005}
597: Zi\'o{\l}kowski, J. 2005, \mnras, 358, 851
598:
599: \end{thebibliography}
600: %\pagebreak
601:
602: \end{document}
603:
604: