1: \documentclass[12pt,preprint]{aastex}
2: %\documentclass[manuscript]{aastex}
3: \newcommand{\vdag}{(v)^\dagger}
4: \newcommand{\myemail}{lcz01@mails.tsinghua.edu.cn}
5:
6: \shorttitle{Spectral Variability in Cyg X-1}
7: \shortauthors{Liu \& Li}
8:
9: %% This is the end of the preamble. Indicate the beginning of the
10: %% paper itself with \begin{document}.
11:
12: %\received{2003 November 22}
13: \begin{document}
14: \title{X-RAY SPECTRAL VARIABILITY IN CYGNUS X-1}
15: \author{C.Z. Liu\altaffilmark{1} and T.P. Li\altaffilmark{1,~ 2}}
16: \altaffiltext{1}{Department of Engineering Physics \& Center for
17: Astrophysics, Tsinghua University, Beijing, China;
18: \myemail}
19: \altaffiltext{2}{Particle Astrophys.
20: Lab., Inst. of High Energy Phys., Chinese Academy of Sciences,
21: Beijing, China}
22:
23:
24: \begin{abstract}
25: Spectral variability in different energy bands of X-rays from
26: Cyg X-1 in different states is studied with $RXTE$ observations
27: and time domain approaches. In the hard tail of energy spectrum
28: above $\sim 10$\,keV, average peak aligned shots are softer than
29: the average steady emission and the hardness ratio decreases
30: when the flux increases during a shot for all states. In regard
31: to a soft band lower $\sim 10$\,keV, the hardness in the soft
32: state varies in an opposite way: it peaks when the flux of the
33: shot peaks. For the hard and transition states, the hardness
34: ratio in respect to a soft band during a shot is in general
35: lower than that of the steady component and a sharp rise is
36: observed at about the shot peak. For the soft state, the
37: correlation coefficient between the intensity and hardness ratio
38: in the hard tail is negative and decreases monotonically as the
39: timescale increases from 0.01\,s to 50\,s, which is opposite to that
40: in regard to a soft band. For the hard and transition states,
41: the correlation coefficients are in general negative and have a
42: trend of decrease with increasing timescale.
43: \end{abstract}
44:
45: \keywords {accretion, accretion disks --- black hole physics ---
46: stars: individual (Cygnus X-1) --- X-rays: stars}
47:
48:
49: \section{INTRODUCTION}
50: The intensity and spectral variabilities of the high energy
51: emission from X-ray binaries carry valuable information about
52: their production regions and mechanisms around compact objects.
53: Cyg X-1, one of the most early discovered and best-studied black
54: hole X-ray binaries \citep{Bow65}, is the brightest source in
55: the hard X-ray sky suitable for the study of rapid variability.
56: The compact object has a mass of $\sim$$10$\,$M_{\odot}$ and the
57: companion star HDE226868 discovered
58: in an optical observation by Walborn (1973) has a mass of
59: $\sim$18$M_{\odot}$ \citep{Her95}.
60: The X-ray emission of Cyg X-1 is very complex in a wide
61: time scale from months to milliseconds. In a long-term period, Cyg
62: X-1 generally shows three different states: hard, soft and
63: transition states.
64:
65: Many temporal and spectral properties
66: of X-rays from Cyg X-1 have been studied before, e.g., power
67: density spectra (PDS), time lags and coherences in different energy
68: bands and on different time scales. Most of these studies are
69: carried out in the frequency domain by using Fourier transformation.
70: Studies for different states of Cyg X-1 have been done with the
71: observation data of the Proportional Counter Array (PCA) on board the
72: {\sl Rossi X-ray Timing Explorer} ({\sl RXTE}), e.g.,
73: studies of temporal properties of Cyg X-1 in the soft state
74: \citep{cui97a}, during the spectral transitions \citep{cui97b},
75: and in the hard state \citep{now99}. The hard state PSD of Cyg X-1
76: in the Fourier frequency range between 0.001\,Hz and 100\,Hz can be
77: successfuly modeled with multiple energy-dependent Lorentzians -- distinct
78: broad noise components (Nowak 2000; Pottschmidt et al. 2003).
79:
80: A characteristic shape of a Fourier PDS or a structure in the PDS,
81: e.g. a quasi-periodic oscillation
82: (QPO) or a broad noise component, can be generated by different kinds
83: of process. For finally understanding the undergoing physical processes,
84: it is also necessary to study variabilities directly in the time domain.
85: On the subsecond timescale, the X-rays from Cyg X-1 show
86: large-amplitude chaotic fluctuations. Randomly occurring shots
87: may have significant contribution on the rapid variability (Terrel
88: 1972; Oda 1977; Sutherland, Weisskopf \& Kahn 1978; Nolan et al.
89: 1981; Meekings et al. 1984; Miyamoto et al. 1988, 1992; Lochner,
90: Swank \& Szymkowiak 1991).
91: To study the shot process
92: Negoro, Miyamoto \& Kitamoto (1994) constructed the average peak
93: aligned shot profile in the 1.2--58.4\,keV energy band and its
94: hardness ratios in (7.3--14.6\,keV)/(1.2--7.3\,keV) and (14.6--21.9\,keV)/(1.2--7.3\,keV)
95: with $Ginga$ observation data of Cyg X-1 in
96: the hard state. They found that the shot becomes first softer,
97: and then, just after the peak, harder than the average emission.
98: With $RXTE$/PCA data of Cyg X-1 and an improved shot detection
99: algorithm, Feng, Li \& Chen (1999) studied evolution of hardness
100: ratios in (13--60\,keV)/(2--6\,keV) in the hard, soft and transition
101: states. Their results confirmed what Negoro, Miyamoto \&
102: Kitamoto (1994) had found in the hard state, and they found that
103: in the soft state, shots are harder, and in the transition
104: states softer than the time average emission. Li, Feng \& Chen
105: (1999) used the correlation analysis technique to study the
106: relationship between temporal and spectral variabilities in Cyg
107: X-1 in different states and the results are consistent with that
108: from the average shot analysis mentioned above.
109:
110: The X-ray Spectrum of Cygnus X-1 can be decomposed into several
111: components : thermal
112: Comptonization, Compton reflection, and a soft excess.
113: We make use of $RXTE$/PSPC observations of Cyg X-1
114: to study spectral variability in different energy bands.
115: Poutanen (2001) in his review paper pointed out that
116: it would be of interest to see how the hardness in the energy
117: bands above $\sim 10$\,keV varies in the shots. This would
118: provide important information about spectral variability of the
119: hard (power-law like) tail.
120: We lay the emphasis on studying hardness evolution during shots and
121: the correlation coefficient between hardness and intensity in the hard tail
122: in this paper. Our results imply that the observed
123: behavior, and then the physical mechanism, of the hard component
124: is quite different from that of soft component.
125:
126: \section{SPECTRAL EVOLUTION DURING SHOTS}
127: A series of observations of Cyg X-1 were performed by $RXTE$ in
128: 1996. The All Sky Monitor on $RXTE$ revealed that Cyg X-1 started
129: a transition from the normal hard state to the soft state. It
130: stayed in soft state for nearly 3 months and then went back down
131: to hard state. The data used in our work are from the public
132: $RXTE$ archives, listed in the Table~\ref{tbl-1}. We extract the
133: PCA data with the version 4.2 of
134: standard $RXTE$ ftools for windows to get 1 ms time bin light
135: curves. In this process, the data were selected to use when the
136: source was observed at the elevation angle larger than
137: $10^{\circ}$, the offset pointing less than $0.02^{\circ}$, and
138: the number of PCUs turning ``ON'' equaling to 5. In hard
139: X-ray band, Cyg X-1 is a very bright source. The average of the
140: background contribution for 13--60\,keV is at 10\% level and less
141: than 1\% for lower bands (2--6\,keV \& 6--13\,keV). For the purpose of
142: studying the average features of shots, background can be negligible.
143: The $RXTE$ dead-time per event in one PCU is about 10\,$\mu$s \citep{Zhang99}.
144: For Cyg X-1, the effect of dead-time is about 1\% (Maccarone, Coppi \&
145: Poutanen 2000), which is neglected in our analysis. We search shots
146: from 1\,ms time bin light curves by using the algorithm proposed by
147: Feng, Li \& Chen (1998), which is a modified algorithm to the peak
148: detection technique of Negoro, Miyamoto \& Kitamoto (1994) and Li \& Fenimore (1996).
149: In order to suppress the effects of statistical fluctuation, a light curve
150: with 1\,ms time bin was first merged into larger time bin of 10\,ms.
151: Then the bin having more counts $C_p $ than the both sides
152: neighboring bins is selected as a candidate peak. In the
153: neighboring 1\,s on both sides of each candidate peak, we search
154: for the bins with counts $ C_1 $ and $ C_2 $ so that the condition
155: \begin{equation} C_p > C_{1, 2} + \alpha\sqrt{C_p+C_{1, 2}}
156: \end{equation} is satisfied with $\alpha$ being selected from [2,
157: 3]. If and only if the number of bins with counts $C_1$ and $C_2$
158: is larger than a certain criterion respectively, we affirm that
159: the count rate is significantly smaller around the candidate peak bin.
160: And then this candidate peak is selected as a potential shot peak. Then
161: each selected potential shot peak bin and its both sides
162: neighboring bins are divided into thirty bins with time bin of 1\,ms.
163: A shot is finally selected by the criteria that their count
164: should be 2 times larger than that of the mean count of the
165: observation and should be the maximum within the thirty bins.
166:
167: The shot detection process was performed respectively in each of
168: three energy bands 2--6\,keV, 6--13\,keV and 13--60\,keV (2--5\,keV,
169: 5--13\,keV and 13--60\,keV for the hard state). A shot peak is selected as
170: a true shot peak if it coincidences in all the three energy bands
171: within 30\,ms. We define the 400\,ms time interval neighboring a
172: shot peak as a shot period. For a studied pair of energy
173: bands $A$ and $B$, we calculate the evolution curve of hardness
174: ratios with $h = f_A/f_B$ in the shot period, where $f_A, f_B$ are
175: the count rate of band $A$ and $B$ respectively. Figure~1 shows
176: the total shot profile of 2--60\,keV energy band (dashed line) and
177: hardness variation (solid line) for different states and different
178: pairs of studied energy band. All the panels in Fig.~1 are
179: classified into four groups, a1-a4 for the soft state, b1-b4 for
180: the transition state of hard-to-soft, c1-c4 for the soft-to-hard
181: transition state, and d1-d4 for the hard state.
182:
183: From the panels a4, b4, c4, and d4 of Fig.~1, one can see that the
184: hardness ratios in the power-law tail, (16--60\,keV)/(13--16\,keV),
185: vary negatively correlated with the intensity, the hardness
186: variation during a shot is dominated by a negative peak. This
187: feature is almost completely opposite to what is seen in the
188: hardness ratios in respect to an energy band below $\sim 10$\,keV
189: for the soft state: The hardness ratios of (6--13\,keV)/(2--6\,keV),
190: (13--60\,keV)/(2--6\,keV), and (13--60\,keV/(2--13\,keV) peak when the
191: corresponding fluxes of the shots peak, shown in a1, a2, and a3 in
192: Fig.~1. Maccarone \& Coppi (2002a) studied the short timescale
193: correlations between line and continuum fluxes in Cyg X-1. They found
194: that the spectrum got harder as the source got brighter in soft state.
195: For the hard and transition states, the variation of hardness
196: ratio in regard to a energy band below $\sim 10$\,keV behaves with
197: a similar feature: the shot becomes first softer, and then, just
198: around the peak, rapidly harder.
199:
200:
201: \section{CORRELATION BETWEEN HARDNESS AND INTENSITY}
202: We here study X-ray spectral variability of Cyg X-1 in different
203: energy bands and different timescales through correlation
204: analysis. Li, Feng \& Chen (1999) proposed an algorithm to
205: calculate the correlation coefficient $r(h,f)$ between the
206: hardness ratio $h$ and the total intensity $f$ on a given
207: timescale $\Delta t$. The total observation data of Cyg X-1 in a
208: spectral state is divided into $M$ periods. For an observation
209: period $k~(k = 1,2,...,M)$ and a given time bin $\Delta t$, the
210: effective data in two studied energy bands are divided into
211: segments with a duration of $n\Delta t,~ n=10$. The correlation
212: coefficient \begin{equation} r(h,f) =
213: \frac{\displaystyle\sum_{i=1}^n{(h(i)-\bar h)(f(i)-\bar
214: f)}}{\sqrt{\displaystyle\sum_{j=1}^n{(h(i)-\bar
215: h)^2}\displaystyle\sum_{i=1}^n{(f(i)-\bar f)^2}}} \end{equation}
216: is calculated for each segment and then their average $\bar r(k)$
217: and standard deviation $\sigma(\bar r(k))$ are derived. The
218: correlation coefficient on a timescale $\Delta t$ is estimated by
219: \begin{equation} \langle \bar r \rangle=\sum_{k=1}^M \omega(k)\bar
220: r(k)/\sum_{k=1}^M\omega(k) ,
221: \end{equation} where \ \begin{equation}\omega(k)=1/\sigma^2[\bar r(k)]
222: \end{equation} and its uncertainty is given by
223: \begin{equation} \sigma=\sqrt{\sigma_1^2+\sigma_2^2} ,\end{equation}
224: where
225: \begin{displaymath}\sigma_1^2=1/\sum_{k=1}^M\omega(k)
226: \end{displaymath}
227: \begin{equation} \sigma_2^2=\sum_{k=1}^M[\bar r(k)- \langle \bar
228: r \rangle]^2/M(M-1). \end{equation} Finally, the result for the
229: observation can be presented as : $\bar r$ = $\langle \bar r
230: \rangle$ $\pm ~\sigma$.
231:
232: Usually we can use some convenient statistical methods based on the
233: normal distribution to make statistical inference, e.g. significance
234: test, on $\bar r$. For the case of short time scale
235: $\Delta t$, although the number of counts per bin may be small for it
236: to be assumed as a normal variable, it is easy to obtain a total number
237: of segments from a certain observation that is large enough that the
238: central limit theorem can be applied and that a normal distribution can
239: be assumed for estimating the uncertainty of $\bar r$.
240:
241: For each state, we calculate the correlation coefficients
242: between intensity and hardness ratio of (13--60\,keV)/(2--6\,keV)
243: and (16--60\,keV)/(13--16\,keV) on time scales 0.01\,s, 0.1\,s, 1\,s,
244: 10\,s, and 50\,s respectively. The results are shown in Table 2--3
245: and Figure 2.
246:
247: From upper-left panel of Fig.~2 for the soft state, we can see that
248: the correlation coefficients between the intensity and hardness ratio
249: in the 2--6\,keV band are positive
250: on all timescales between 0.01\,s and 50\,s. In contract, they are negative
251: for the 13--16\,keV band. The two kinds of correlation vary
252: with timescale in opposite directions:
253: for the hardness ratios in regard to the
254: soft band, the correlation coefficient $r(h,f)$ increases
255: monotonically and reaches near unity as the timescale increases
256: from 0.01\,s to 50\,s; on the other hand, for $h$ in the
257: hard band, the correlation coefficient $r(h,f)$ decreases
258: monotonically and reaches near $-1$. The hard and transition
259: states have the similar feature in hardness-intensity correlation:
260: the correlation coefficients
261: are in general negative and have a trend of decrease with
262: increasing timescale.
263:
264: Figure 3 shows example of light curve and hardness profiles in 50\,s
265: time bin in the soft and hard states. From Fig.3 we can see
266: that, consistent with the results of correlation analysis, the
267: hardness ratios in (13--60\,keV)/(2--6\,keV) are positively related to
268: the light curve and the hardness in (16--60\,keV)/(13--16\,keV)
269: negatively for the soft state, and a rather complex mixture of
270: positive and negative correlations for the hard states.
271:
272: \section{SUMMARY AND DISCUSSION}
273: The simplest single shot models assume that the observed
274: lightcurve consists of the superposition of individual uncorrelated
275: shots with a fixed profile and intensity. Lochner, Swank \& Szymkowiak
276: (1991) found that the simplest shot models cannot fit the overall
277: PSD shape of Cyg X-1 well. Recently Maccarone \& Coppi (2002b) found that
278: the single shot variability models also cannot produce the observed skewness
279: pattern. In this work we study the spectral evolution in different energy
280: bands of X-rays from Cyg X-1 in different states during the average peak
281: aligned shots in subsecond duration region, our results on shot are
282: in the average meaning, neither assuming the shots have a fixed profile
283: and nor assuming the shots can be responsible alone for the overall power spectrum.
284: The timescale analysis we performed for correlation between hardness and
285: intensity is not correlated with any special shot model.
286: Main features in spectral variability of Cyg X-1 X-rays
287: revealed by our time domain analysis can be summarized as follows:
288: (1) For any spectral state of Cyg X-1, on the average the hardness profile of
289: the hard spectral component above $\sim 10$\,keV
290: is in the shape of a valley bottomed out when flux peaks
291: during a shot,
292: which is different from that in respect to the softer component.
293: (2) During an average shot of Cyg X-1 in the soft state, the hardness
294: in regard to a softer band lower $\sim 10$\,keV peaks when
295: the flux peaks .
296: (3) For the hard and transition states, shots are softer than
297: the time average emission, a sharp rise is appeared at about the
298: shot peak in the average profile of hardness ratio in respect to
299: a soft band.
300: (4) The correlation coefficients between the intensity and hardness
301: ratio on timescales between 0.01\,s and 50\,s are negative or near zero
302: when Cyg X-1 is in the hard or transition
303: states. In the soft state, the correlation coefficient in the hard
304: band decreases monotonically and reaches near $-1$
305: as the timescale increases from 0.01\,s to 50\,s; by contrast, the correlation
306: in regard to a soft band is positive and increases when timescale
307: increases.
308:
309: The significant difference observed in Cyg X-1 of soft state
310: between the shot spectral evolution in the hard band above $\sim
311: 10$\,keV and that in respect to a softer band indicates that
312: different mechanisms dominate the shot processes in the two energy
313: bands. It is broadly believed that the process of producing the
314: hard power-law like spectrum is thermal Comptonization: hard
315: emission above $\sim 10$\,keV is produced by inverse Compton
316: scattering of soft seed photons by hot electrons of temperature
317: $kT_e\le 100$\,keV (e.g. Sunyaev \& Tr\"{u}mper 1979). When a seed
318: photon of energy $h\nu_i$ collides with a relativistic electron in
319: the hot corona with Lorenz factor $\gamma$, the energy, $h\nu$, of
320: the emitted photon is given by $h\nu \approx \gamma^2h\nu_i$ (e.g.
321: Lang 1999). After a collision from an electron of energy 100\,keV,
322: the energy of a seed photon of 10\,keV will be lifted to $\sim 14$\,keV.
323: Most scattered photons escaped from the hot corona undergo
324: only one collision and the hardness ratio in respect to the 13--16\,keV
325: band decreases when the number of seed and scattered photons
326: increase. Thus the anti-correlation between the ratio of counts in
327: (16--60\,keV)/(13--16\,keV) and the hard band intensity during shots
328: in any spectral state can be understood by the Comptonization
329: process in the hot corona. Recently, by using a timescale
330: analysis technique of shot width, Feng, Li \& Zhang (2004) found
331: that the energy dependence of shortest width of X-ray shots from
332: the black hole binaries Cyg X-1, XTE J1550-564 and GRO J1655-40 in
333: their hard state in the energy band below about 10--20\,keV is
334: opposite to that in higher energy region: below about 10--20\,keV
335: the shortest width decreases with increasing energy, and above
336: about 10--20\,keV it increases with increasing energy. It is
337: expected that detailed study of shot spectral evolution and other
338: energy resolved timing in the hard band can help to understand the
339: hot corona and the shot production and propagation processes.
340:
341: The evolution of shot hardness ratio in regard to a soft band below
342: $\sim 10$\,keV should depend on not only the hot gas, but also the
343: cold disk. It is natural to assume that shots are most probably
344: produced at the innermost region of the cold disk joined with the
345: hot corona which is a most turbulent region,
346: and the steady component around a shot is a global
347: average of emission from different regions of the hot corona
348: (Li, Feng \& Chen 1999).
349: For the soft state, the disk may extend down to the last stable
350: orbit and the hot cloud is restricted to the corona surrounding
351: the disk \citep{esi98}. As the inner disk, and then the corona
352: embedding it, has a higher temperature, the shot spectrum
353: from the innermost region should be harder. For the hard and
354: transition states, the optically thick disk is truncated at
355: larger distance and jointed a spherical corona around the black hole.
356: Shots should be softer than the steady emission as they produce
357: and Comptonize at the outer part of the corona
358: with temperature lower than the average of the total corona.
359: The difference between the profile of hardness ration in regard to
360: a soft band in the soft state and that in the other states
361: may reflect the difference of their corona geometries.
362: The presence of an uprush of the hardness profile around the shot peak
363: in the hard and transition states is probably an evidence that
364: a shock wave is produced along with shot propagating in the spherical corona.
365:
366: The correlation coefficients between the hardness ratio and intensity
367: obtained for different timescales and different states
368: of Cyg X-1 shown in Tables 2--3 and Figure 2 are qualitatively consistent
369: with the results from the shot analysis. The correlation analysis is
370: made for the temporal variability in general, not only for shots.
371: The correlation being weaker at shorter timescales indicates
372: the existence of other uncorrelated rapid variations.
373: For the soft state, the correlation coefficients varying
374: (increasing or decreasing) monotonically and reaching a perfect
375: correlation (or anti-correlation) along with the timescale from 0.01\,s to 50\,s
376: may indicate that the effect of uncorrelated noise being weakened on larger
377: timescales.
378:
379: The authors thank the referee for helpful comments and suggestions.
380: This work is supported by the Special Funds for Major State Basic Research
381: Projects and
382: the National Natural Science Foundation of China. The data analyzed in this
383: work are obtained through
384: the HEASARC on-line service provided by the NASA/GSFC.
385:
386: \clearpage
387:
388: \begin{thebibliography}{}
389:
390: \bibitem[Bowyer et al. 1965]{Bow65}
391: Bowyer, S., Byram, E. T., Chubb, T. A., \& Friedman, H. 1965,
392: Science, 147, 394
393: \bibitem[Cui et al. 1997a]{cui97a}Cui W., Heindl W.A., Rothschild R.E., Zhang S.N.,
394: Jahonda K. \& Focke W. 1997a, ApJ, 474, L57
395: \bibitem[Cui et al. 1997b]{cui97b}Cui W., Zhang S.N., Focke W. \& Swank J.H. 1997b, ApJ, 484, 383
396: \bibitem[Esin et al. 1998]{esi98}Esin A.A., Narayan R., Cui W., Grove J.E. \& Zhang S.N. 1998, ApJ, 505, 854
397: \bibitem[Feng, Li, \& Chen 1999]{feng99} Feng, Y. X., Li, T. P,
398: \& Chen, L. 1999, \apj, 514, 373
399: \bibitem[Feng, Li, \& Zhang 2004]{feng04} Feng, H., Li, T.P., \& Zhang, S.N.
400: 2004, \apj, 606 (astro-ph/0401205)
401: \bibitem[Herrero et al. 1995]{Her95}Herrero, A., Kudritzki, R.P.,
402: Gabler, R., Vilchez, J.M., \& Gabler, A. 1995, A\&A, 297, 556
403: \bibitem[Lang 1999]{lan99} Lang K.R. 1999, Astrophysical Formulae (Springer-Verlag)
404: \bibitem[Li, Feng, \& Chen 1999]{lit99} Li, T. P., Feng, Y.
405: X., \& Chen, L. 1999, \apj, 521, 789
406: \bibitem[Li 1996]{lif96} Li, H., \& Fenimore, E. E. 1996,
407: \apj, 469, L115
408: \bibitem[Lochner, Swank, \& Szymkowiak 1991]{loc91}Lochner, J.C., Swank, J.H.,
409: \& Szymkowiak, A.E. 1991, \apj, 375, 295
410: \bibitem[Maccarone, Coppi, \& Poutanen 2000]{mac00}Maccarone, T.J., Coppi, P.S., \& Poutanen J., 2000, \apj, 537, L107
411: \bibitem[Maccarone, Coppi, 2002a]{mac02a}Maccarone, T.J., Coppi, P.S., 2002a, MNRAS, 336, 817
412: \bibitem[Maccarone, Coppi, 2002b]{mac02b}Maccarone, T.J., Coppi, P.S., 2002b, MNRAS, 335, 465
413: \bibitem[Meekings et al. 1984]{mee84}Meekins, J.F., Wood, K.S., Hedler, R.L.,
414: Byram, E.T., Yentis, D.J., Chubb, T.A., \& Friedman, H. 1984,
415: \apj, 278, 288
416: \bibitem[Miyamoto et al. 1988]{miy88}Miyamoto, S., Kitamoto, S., Mitsuka, K.,
417: \& Dotani, T. 1988, \nat, 336, 450
418: \bibitem[Miyamoto et al. 1992]{miy92} Miyamoto, S., Kitamoto, S., Iga, S., Negoro, H., \& Terada, K. 1992, \apj, 391
419:
420: \bibitem[Negoro, Miyamoto, \& Kitamoto 1994]{neg94} Negoro, H.,
421: Miyamoto, S., \& Kitamoto, S. 1994 \apj, 423, L127
422:
423: \bibitem[Nolan et al. 1981]{nol81}Nolan, P.L., et al. 1981, \apj, 246, 494
424: \bibitem[Nowak et al. 1999]{now99}Nowak M.A., Vaughan B.A., Wilms J., Dove J.B. \& Begelman C. 1999,
425: ApJ, 510, 874
426: \bibitem[Nowak 2000]{now00}Nowak M.A. 2000, \mnras, 318, 361
427: \bibitem[Oda 1977]{oda77}Oda, M. 1977, Space Sci. Rev., 20, 757
428: \bibitem[Pottschmidt et al. 2003]{pot03}Pottschmidt K., Wilms J., Nowak M.A., Pooley G.G.,
429: Gleissner T., Heindl W.A., Smith D.M., Remillard R., \& Staubert R. 2003,
430: A\&A, 407, 1039
431: \bibitem[Poutanen 2001]{Pou01} Poutanen, J. 2001,
432: Adv. Space Res., 28, 267
433: \bibitem[Sutherland, Weisskopf, \& Kahn 1978]{sut78}
434: Sutherland, P.G., Weisskopf, M.C., \& Kahn, S.M. 1978, \apj, 219, 1029
435: \bibitem[Sunyaev \& Tr\"{u}mper 1979]{sun79}Sunyaev R. \& Tr\"{u}mper J. 1979,
436: Nature, 279, 506
437: \bibitem[Terrel 1972]{ter72}Terrel, N.J. 1972, \apj, 174, L35
438: \bibitem[Walborn 1973]{Wal73} Walborn, N. R. 1973 \apj, 179, L123
439: \bibitem[Zhang et al. 1999]{Zhang99} Zhang, W., Jahoda, K., Swank, J. H.,
440: Morgan, E. H., \& Giles, A. B. 1999, 449, 930
441:
442: \end{thebibliography}
443:
444:
445: \clearpage
446: \begin{figure}
447: \begin{center}
448: \epsscale{0.22}
449: \plotone{f1a.eps}
450: \plotone{f1b.eps}
451: \plotone{f1c.eps}
452: \plotone{f1d.eps}
453:
454: \plotone{f1e.eps}
455: \plotone{f1f.eps}
456: \plotone{f1g.eps}
457: \plotone{f1h.eps}
458:
459:
460: \plotone{f1i.eps}
461: \plotone{f1j.eps}
462: \plotone{f1k.eps}
463: \plotone{f1l.eps}
464:
465: \plotone{f1m.eps}
466: \plotone{f1n.eps}
467: \plotone{f1o.eps}
468: \plotone{f1p.eps} \caption{Flux and hardness ratio profiles of
469: average shots. {\it dashed line}: normalized shot flux profile.
470: {\it solid line}: hardness ratio profile. a1-a4: soft state;
471: b1-b4: hard to soft transition; c1-c4: soft to hard transition;
472: d1-d4: hard state. \label{fig1}}
473: \end{center}
474: \end{figure}
475:
476: \clearpage
477: \begin{figure}
478: \begin{center}
479: \epsscale{0.42}
480: \plotone{f2a.eps}
481: \plotone{f2b.eps}
482: \plotone{f2c.eps}
483: \plotone{f2d.eps} \caption{Correlation
484: coefficients between hardness and intensity vs. time scale. {\it
485: Solid line}: hardness in (13--60\,keV)/(2--5\,keV); {\it Dotted line}:
486: hardness in (16--60\,keV)/(13--16\,keV). {\it Upper-left panel}: soft
487: state; {\it Upper-right panel}: hard state; {\it Lower-left
488: panel}: hard to soft transition; {\it Lower-right panel}:soft to
489: hard transition. \label{fig2}}
490: \end{center}
491: \end{figure}
492:
493: \clearpage
494:
495: \begin{figure}
496: \begin{center}
497: \epsscale{0.44}
498: \plotone{f3a.eps} \epsscale{0.41}
499: \plotone{f3b.eps} \caption{Intensity(E for (13--60\,keV), D for
500: (13--16\,keV) in the figures) and hardness ratio profiles
501: in soft and hard states. {\sl Left panel}: soft state; {\sl
502: Right panel}: hard state. {\it Solid line}: 2--13\,keV light curve
503: in 50 s time bin in arbitrary unit. {\it dotted line}: hardness ratios in
504: (13--60\,keV)/(2--6\,keV); {\it dash-dotted line}: hardness ratios in
505: (16--60\,keV)/(13--16\,keV).\label{fig3}}
506: \end{center}
507: \end{figure}
508:
509: \clearpage
510:
511: \begin{deluxetable}{cccc}
512: \tablecaption{List of RXTE observation of Cyg X-1 used for the
513: analysis. Dates are presented in DD/MM/YY format.\label{tbl-1}}
514: \tablewidth{0pt} \tablehead{ \colhead{State} & \colhead{Obs. ID} &
515: \colhead{Start Time} & \colhead{Stop Time}} \startdata
516: Hard to Soft &10412--01--01--00 &23/05/96 14:14:05 &23/05/96 18:08:05 \\
517: &10412--01--03--00 &30/05/96 07:47:05 &30/05/96 08:45:05\\
518: \cline{1-4}
519: Soft &10512--01--08--00 &17/06/96 07:59:05 &17/06/96 09:08:05\\
520: &10512--01--08--02 &17/06/96 04:47:05 &17/06/96 05:44:05\\
521: \cline{1-4}
522: Soft to Hard&10412--01--05--00 &11/08/96 07:02:05 &11/08/96 08:25:05\\
523: &10412--01--07--00 &12/08/96 14:41:05 &12/08/96 15:59:05\\
524: \cline{1-4}
525: Hard&10236--01--01--03 &17/12/96 12:45:42 &17/12/96 13:25:13\\
526: &10236--01--01--04 &17/12/96 22:21:42 &17/12/96 00:42:13\\
527: \enddata
528: \end{deluxetable}
529: %\clearpage
530:
531: \begin{deluxetable}{c|ccccc}
532: \tablecaption{Average Correlation Coefficients between Hardness
533: (13--60\,keV)/(2--13\,keV) and Intensity of 2--60\,keV in Cyg X-1}
534: \label{tbl-2}\tablewidth{0pt} \tablehead{\colhead{State} & &
535: &\colhead{ Time Scale}
536: \cr
537: %\hspace{2cm}
538: &\colhead{0.01\,s} & \colhead{0.1\,s} &\colhead{1\,s}
539: &\colhead{10\,s} &\colhead{50\,s} } \startdata
540: Soft & 0.05$\pm0.002$ & 0.15$\pm0.01$ &0.34$\pm0.03$&0.60$\pm0.06$&0.81$\pm0.05$ \\
541: Hard & -0.007$\pm0.004$ & -0.05$\pm0.009$ & -0.37$\pm0.04$&-0.61$\pm0.08$ &-0.58$\pm0.30$ \\
542: Hard to Soft & -0.03$\pm0.006$ &-0.39$\pm0.02$ & -0.51$\pm0.08$&-0.51$\pm0.20$&-0.63$\pm0.13$ \\
543: Soft to Hard& -0.03$\pm0.005$ & -0.37$\pm0.03$ & -0.35$\pm0.11$ & -0.15$\pm0.19$ &-0.26$\pm0.61$\\
544: \enddata
545: \end{deluxetable}
546:
547: \begin{deluxetable}{c|ccccc}
548: \tablecaption{Average Correlation Coefficients between Hardness
549: (16--60\,keV)/(13--16\,keV) and Intensity of 13--60\,keV in Cyg X-1}
550: \label{tbl-3}\tablewidth{0pt} \tablehead{\colhead{State} & &
551: &\colhead{ Time Scale}
552: \cr
553: %\hspace{2cm}
554: &\colhead{0.01\,s} & \colhead{0.1\,s} &\colhead{1\,s}
555: &\colhead{10\,s} &\colhead{50\,s} } \startdata
556: Soft & -0.016$\pm0.01$ & -0.12$\pm0.01$ & -0.35$\pm0.03$&-0.78$\pm0.04$&-0.91$\pm0.004$ \\
557: Hard & -0.01$\pm0.003$ & -0.08$\pm0.01$ & -0.24$\pm0.05$&-0.41$\pm0.09$ &-0.31$\pm0.21$ \\
558: Hard to Soft & -0.01$\pm0.006$ &-0.13$\pm0.01$ & -0.20$\pm0.03$&-0.34$\pm0.09$&-0.59$\pm0.11$ \\
559: Soft to Hard& -0.005$\pm0.007$ & -0.13$\pm0.01$ & -0.19$\pm0.05$& -0.56$\pm0.24$&-0.74$\pm0.25$\\
560: \enddata
561: \end{deluxetable}
562:
563:
564: \end{document}
565: