2 Observations and data analysis

Figure 1: The light curve of V4641 Sgr in 1999 according to observations of RXTE and BeppoSAX satellites. In the lower panels we present the source ASM hardness ratio (5-12.2 keV)/(1.3-3.0 keV) during the period around the bright X-ray flares ($\sim$Sep. 14-17, 1999). Filled circles represent the RXTE/PCA data of pointed observations, crosses - the RXTE/PCA scanning observations, open circles - RXTE/ASM data, filled squares - BeppoSAX data.

In our analysis we used publicly available RXTE data of Sgr. This includes 11 pointed observations in Sep., 1999, 61 sets of the observations scanning over the Galactic bulge region, performed from Feb. 5, 1999 till Oct. 24 1999, and three sets of scans over the Galactic bulge in Oct. 2000. In addition to this we used light curves of V4641 Sgr obtained by RXTE All Sky Monitor (ASM) in three energy channels (1.2-3-5-12.2 keV). The RXTE/ASM lightcurves of V4641 Sgr were provided by the RXTE/ASM team (http://xte.mit.edu/ASM_lc.html).

The data reduction of PCA was performed with the help of standard procedures from FTOOLS/LHEASOFT 5.1 package. In order to avoid possible additional problems with the xenon edge at 4.8 keV (that became more pronounced with Epoch 4 response matrix) for our spectral analysis we used only first layer data from all Proportional Counter Units (PCUs). To be convinced that our procedure allow us to perform the spectral analysis correctly we extracted the PCA spectrum of Crab Nebulae from the observation 1999 Sep. 13. Here we also used only the first layer data of all PCUs. The spectrum then was fitted by the conventional model: the power law with a photon index $\Gamma\sim 2.06$ with interstellar absorption ( $N_{\rm H}L$ was fixed at the value of $4\times10^{21}~{\rm cm}^{-2}$). Residuals do not exceed the value of about 1%. Therefore in subsequent analysis we used the value 1% as a rough estimation of systematic uncertainties in the energy band 3-20 keV. In reality it appears that the uncertainties in the response matrix do not exceed 10% at energies up to $\sim$40-50 keV therefore in some cases we used PCA data to illustrate approximate behavior of the source at these energies. For the estimation of the PCA background we used VLE-based model when the source is sufficiently bright (>40 cnts/s/PCA) and L7_240 model when the source is weak. The nominal accuracy of the background estimation ($\sim$1-2% of the background value, see e.g. http://lheawww.gsfc.nasa.gov/~keith/dasmith/Epoch4/systematics_l7_240.htm) allows us to follow a source behavior down to a level of the order of 0.1-0.5 mCrab. At such low level of the source activity the Galactic diffuse emission can also become important. We estimated the contribution of the diffuse emission basing on the results of e.g. Yamauchi & Koyama (1993), Yamasaki et al. (1997) and found it small (<0.1-0.5 mCrab) at the position of the source (l=6.774, b=-4.789).

For the reduction of HEXTE data we used standard tasks of FTOOLS/LHEASOFT 5.1 package. For the HEXTE spectral analysis we used Archive mode (64 energy channels for total energy band 15-250 keV, 16 s time resolution). We analyzed Cluster A and Cluster B data separately and found that in each case their results are consistent with each other. During fitting procedures we left normalizations of HEXTE clusters to be free parameters. Then, for plotting purposes the HEXTE different clusters points were averaged. Note here that absolute normalizations of HEXTE spectra on figures were adjusted to match the PCA spectra.