We study the hardness-intensity correlation (HIC) in gamma-ray bursts (GRBs). In particular, we analyze the decay phase of pulse structures in their light curves. The study comprises a sample of 82 long pulses selected from 66 long bursts observed by the Burst And Transient Source Experiment (BATSE) on the Compton Gamma-Ray Observatory. We find that at least 57% of these pulses have HICs that can be well described by a power law. A number of the other cases can still be explained with the power law model if various limitations of the observations are taken into account.The distribution of the power law indices γ, obtained by modeling the HIC of pulses from different bursts, is broad with a mean of 1.9 and a standard deviation of 0.7. We also compare indices among pulses from the same bursts and find that their distribution is significantly narrower. The probability p of a random coincidence is shown to be very small (< 2 × 10 −5 ). In most cases, the indices are equal to within the uncertainties. These results demand a physical model to be able to reproduce multiple pulses with similar characteristics for an individual burst, but with a large diversity for pulses from an ensemble of bursts. This is particularly relevant when comparing the external versus the internal models.In our analysis, we also use a new method for studying the hardness-intensity correlation, in which the intensity is represented by the peak value of the EF E spectrum, where E is the energy and F E is the energy flux spectrum. We compare it to the traditional method in which the intensity over a finite energy range is used instead, which may be an incorrect measure of the bolometric intensity. This new method gives stronger correlations and is useful in the study of various aspects of the HIC. In particular, it produces a better agreement between indices of different pulses within the same burst. Also, we find that some pulses exhibit a track jump in their HICs, in which the correlation jumps between two power laws with the same index. We discuss the possibility that the track jump is caused by strongly overlapping pulses. Based on our findings, the constancy of the index is proposed to be used as a tool for pulse identification in overlapping pulses and examples of its application are given.

Abstract. Analyzing the archival data from the Rossi X-ray Timing Explorer (RXTE), we study the power density spectra (PDS) of Cygnus X-1 from 1996 to 2003 in the frequency range of 0.01-25 Hz. Using a model consisting of one or two Lorentzians and/or an exponentially cut-off power-law, we are able to achieve a good fit to the PDS during the observations. With our model we are also able to track the evolution of the Lorentzian components through all spectral states of the source. We confirm the relation between characteristic frequencies seen both in black hole candidate and neutron star sources, and show the changes in this relation during the transitional and soft states of the source. The connection between the Lorentzian components is investigated by analyzing similarities and differences in their behavior. We find that the spectral state of the source can be uniquely determined from the parameters of these components. The parameter correlations can all be described by continuous functions, which differ between components. We discuss our results in the context of relativistic precession model for the accretion disk, and show a remarkable agreement between the model prediction and the data in the hard state. We estimate a value for the specific angular momentum of a * = 0.49 (−0.57) in the case of prograde (retrograde) rotation and an estimate for the inner radius of 22 to 50 (25 to 55) gravitational radii. Additional assumptions are required to explain the soft state data, and attempting to invoke rotational reversal for state transitions shows that it is insufficient to explain the differences between the hard and soft state data.

The emission processes active in the highly relativistic jets of gamma-ray bursts (GRBs) remain unknown. In this paper we propose a new measure to describe spectra: the width of the EF E spectrum, a quantity dependent only on finding a good fit to the data. We apply this to the full sample of GRBs observed by Fermi/GBM and CGRO/BATSE. The results from the two instruments are fully consistent. We find that the median widths of spectra from long and short GRBs are significantly different (chance probability < 10 −6 ). The width does not correlate with either duration or hardness, and this is thus a new, independent distinction between the two classes. Comparing the measured spectra with widths of spectra from fundamental emission processes -synchrotron and blackbody radiation -the results indicate that a large fraction of GRB spectra are too narrow to be explained by synchrotron radiation from a distribution of electron energies: for example, 78% of long GRBs and 85% of short GRBs are incompatible with the minimum width of standard slow cooling synchrotron emission from a Maxwellian distribution of electrons, with fast cooling spectra predicting even wider spectra. Photospheric emission can explain the spectra if mechanisms are invoked to give a spectrum much broader than a blackbody.

Building on results from previous studies of Cygnus X-1, we analyze Rossi X-ray Timing Explorer (RXTE) data taken when the source was in the soft and transitional spectral states. We look at the power spectrum in the 0.01-50 Hz range, using a model consisting of a cut-off power-law and two Lorentzian components. We are able to constrain the relation between the characteristic frequencies of the Lorentzian components, and show that it is consistent with a power-law relation having the same index (1.2) as previously reported for the hard state, but shifted by a factor ∼2. Furthermore, it is shown that the change in the frequency relation seen during the transitions can be explained by invoking a shift of one Lorentzian component to a higher harmonic, and we explore the possible support for this interpretation in the other component parameters. With the improved soft state results we study the evolution of the fractional variance for each temporal component. This approach indicates that the two Lorentzian components are connected to each other, and unrelated to the power-law component in the power spectrum, pointing to at least two separate emission components.