Abstract

Context. Some close binaries of the beta Lyrae type show photometric cycles longer than the orbital one, which are possibly related to changes in their accretion disks. Aims. We aim to understand the short- and long-scale morphologic changes observed in the light curve of the eclipsing system OGLE-BLG-ECL-157529. In particular, we want to shed light on the contribution of the disk variability to these changes, especially those related to the long cycle, occurring on timescales of hundreds of days. Methods. We studied I-band Optical Gravitational Lensing Experiment (OGLE) photometric times series spanning 18.5 years, constructing disk models by analyzing the orbital light curve at 52 different consecutive epochs. An optimized simplex algorithm was used to solve the inverse problem by adjusting the light curve with the best stellar-orbital-disk parameters for the system. We applied an analysis of principal components to the parameters to evaluate their dependence and variability. We constructed a description of the mass transfer rate in terms of disk parameters. Results. We find that the overall light variability can be understood in terms of a variable mass transfer rate and variable accretion disk. The system brightness at orbital phase 0.25 follows the long cycle and is correlated with the mass transfer rate and the disk thickness. The long-cycle brightness variations can be understood in terms of differential occultation of the hotter star by a disk of variable thickness. Our model fits the overall light curve during 18.5 years well, including epochs of reversal of main and secondary eclipse depths. The disk radius cyclically change around the tidal radius, decoupled from changes in the mass transfer rate or system brightness, suggesting that viscous delay might explain the non-immediate response. Although the disk is large and fills a large fraction of the hot star Roche lobe, Lindblad resonance regions are far beyond the disk, excluding viscous dissipation as a major source of photometric variability.