2012A&A...543A.121V

X-ray binaries are usually divided in persistent and transient sources. For ultracompact X-ray binaries (UCXBs), the mass transfer rate is expected to be a strong function of orbital period, predicting persistent sources at short periods and transients at long periods. For 14 UCXBs including two candidates, we investigate the long-term variability and average bolometric luminosity with the purpose of learning how often a source can be expected to be visible above a given luminosity, and we compare the derived luminosities with the theoretical predictions. We use data from the RXTE All-Sky Monitor because of its long-term, unbiased observations. Many UCXBs are faint, i.e., they have a count rate at the noise level for most of the time. Still, information can be extracted from the data, either by using only reliable data points or by combining the bright-end variability behavior with the time-averaged luminosity. Luminosity probability distributions show the fraction of time that a source emits above a given luminosity. All UCXBs show significant variability and relatively similar behavior, though the time-averaged luminosity implies higher variability in systems with an orbital period longer than 40-min. There is no large difference in the statistical luminosity behavior of what we usually call persistent and transient sources. UCXBs with an orbital period below ∼30-min have a time-averaged bolometric luminosity that is in reasonable agreement with estimates based on the theoretical mass transfer rate. Around 40-min the lower bound on the time-averaged luminosity is similar to the luminosity based on the theoretical mass transfer rate, suggesting these sources are indeed faint when not detected. Above 50-min some systems are much brighter than the theoretical mass transfer rate predicts, unless these systems have helium burning donors or lose additional angular momentum.