The candidate 31.4 day periodicity from Cyg X-3 is an artifact of ASM data sampling

ATel #3437; John A. Tomsick (SSL/UC Berkeley), Osmi Vilhu (Univ. Helsinki), Joern Wilms (ECAP, Bamberg), Arash Bodaghee (SSL/UC Berkeley), Katja Pottschmidt (CRESST & NASA-GSFC), Jerome Rodriguez (CEA-SAp/Lab. AIM Saclay), Al Levine (MIT), Pasi Hakala (Univ. Turku, Finland), Diana Hannikainen (Aalto University Metsahovi Radio Observatory, Finland)
on 17 Jun 2011; 21:43 UT
Credential Certification: John A. Tomsick (jtomsick@ssl.berkeley.edu)

Tweet

Cyg X-3 is a High-Mass X-ray Binary (HMXB) with a 4.8 hour orbital period that is well-measured using the strong X-ray modulation exhibited by this system. Recently, a candidate third period at 31.4 days was reported in ATEL#3325 that was suggested to be caused by either a tertiary body in the Cyg X-3 system or a precessing disk. Both possibilities would have important implications for the interpretation of the gamma-rays detected from Cyg X-3 (Tavani et al. 2009, Nature, 462, 620; Abdo et al. 2009, Science, 326, 1512) as well as for HMXB evolution (e.g., Lommen et al. 2005, A&A, 443, 231).

We have re-analyzed the All-Sky Monitor (ASM) Cyg X-3 data from 15 years of the Rossi X-ray Timing Explorer mission. We used a slightly longer light curve than the one reported in ATEL#3325 (through 2011 March instead of 2010 October). As described in ATEL#3325, we used the DYNPSRCH software to produce a periodogram of the Observed-Computed (O-C) minima of the 4.8 hour period. While we reproduced the detection of a 31.3 day period with a power of 34.2, we note that the method for finding the observed minima (see ATEL#3325) is to pick the lowest ASM count rate measured in a phase range (-0.1 to 0.1) around the calculated minimum. In nearly half the cases (47%), there are only one or two ASM measurements in the phase range, making the observed minimum times highly susceptible to ASM data sampling.

After reproducing the result, we tested whether ASM data sampling could lead to such a period by re-calculating a periodogram with the same sampling cadence but with replacing the ASM count rates actually measured with random count rates selected from a Gaussian distribution with a mean of 13.0 c/s and a standard deviation of 3.0 c/s (the mean and standard deviation of the actual Cyg X-3 ASM light curve). The fact that we still find a signal with a period of 31.3 days and a power of 34.4 indicates that the signal is an artifact of data sampling.

An inspection of the O-C plot appears to show strong patterns before approximately MJD 52,050 (2001 May 21) that do not appear after MJD 52,050. Thus, we divided the light curve into these two parts and produced periodograms for these two segments. During the first segment, the power for the signal at 31.3 days increases to 86.2, while there are no significant signals in the periodogram from the second segment (the highest power is 9.2). These two periodograms as well as the others described above have been posted on-line (http://sprg.ssl.berkeley.edu/~jtomsick/).

In summary, from the first (random light curve) test, we conclude that the 31.3-31.4 day period is spurious and that it is likely caused by the ASM data sampling. The second test suggests that the ASM sampling was only obtained in this way until approximately 2001 May. A deeper understanding of the spurious signal could likely be obtained via an analysis of ASM surveying procedures.