Binary Source Lensing and the Repeating OGLE EWS Events

A microlensing event may exhibit a second brightening when the source and/or the lens is a binary star. Previous study revealed 19 such repeating event candidates among 4120 investigated microlensing light curves of the Optical Gravitational Lensing Experiment (OGLE). The same study gave the probability ~ 0.0027 for a repeating event caused by a binary lens. We present the simulations of binary source lensing events and calculate the probability of observing a second brightening in the light curve. Applying to simulated light curves the same algorithm as was used in the analysis of real OGLE data, we find the probability ~ 0.0018 of observing a second brightening in a binary source lensing curve. The expected and measured numbers of repeating events are in agreement only if one postulates that all lenses and all sources are binary. Since the fraction of binaries is believed to be <= 50%, there seems to be a discrepancy.

💡 Research Summary

The paper addresses the puzzling occurrence of “repeating” microlensing events—light curves that display a second brightening episode—within the OGLE Early Warning System (EWS) data set. Earlier work identified 19 candidate repeating events among 4,120 examined OGLE light curves and estimated that binary lenses (two stars acting together as the gravitational lens) would produce such repeats with a probability of roughly 0.0027 per event. However, this probability alone cannot account for the observed 19 cases, suggesting that another mechanism may be at work.

To explore this possibility, the authors performed a comprehensive suite of simulations in which the source, rather than the lens, is binary. In these “binary source” simulations, two stars are placed behind a single lens (or a binary lens) with a range of brightness ratios, time offsets, and impact parameters that mimic realistic stellar populations. The simulated light curves incorporate the same cadence, photometric noise characteristics, and seasonal gaps as the actual OGLE observations. Crucially, the same automated detection algorithm used in the original OGLE analysis—designed to flag two distinct magnification peaks—was applied to the simulated data to assess how often a second brightening would be recognized as a repeat event.

The results show that binary sources generate a detectable second brightening with a probability of about 0.0018, roughly 30 % lower than the binary‑lens estimate. The reduction stems from two main effects: (1) the secondary source is often significantly fainter than the primary, making its magnification peak less prominent, and (2) the time separation between the two peaks can be short enough that the algorithm merges them into a single, broader event rather than two distinct ones. Consequently, even when both binary‑lens and binary‑source mechanisms are combined, the expected number of repeats remains below the 19 observed.

The authors point out that the observed repeat count would match theoretical expectations only if all lenses and all sources were binary—a scenario that conflicts with well‑established stellar multiplicity statistics, which place the binary fraction at ≤ 50 % for both lens and source populations. This discrepancy suggests that either (a) the detection efficiency for secondary peaks is higher than the simulations predict, (b) the underlying binary fractions for the specific Galactic bulge populations probed by OGLE are anomalously high, or (c) additional astrophysical processes (e.g., planetary companions, binary‑lens orbital motion, or blending effects) contribute to the observed repeats.

To resolve the tension, the paper recommends several avenues for future work: (i) refine the simulation inputs by adopting empirically measured distributions of source brightness ratios, orbital periods, and lens mass ratios; (ii) cross‑validate the repeat statistics with independent microlensing surveys such as KMTNet and MOA, which have different cadences and photometric depths; (iii) improve the repeat‑detection pipeline to be more sensitive to low‑amplitude or closely spaced secondary peaks, perhaps by employing machine‑learning classifiers trained on simulated binary‑source events; and (iv) conduct targeted follow‑up observations (e.g., high‑resolution imaging or spectroscopy) of the identified repeat events to directly assess the binary nature of the lens or source.

In summary, the study demonstrates that binary sources can indeed produce repeat microlensing signatures, but the predicted occurrence rate falls short of the empirical count unless unrealistically high binary fractions are assumed. This mismatch highlights a gap in our understanding of microlensing event morphology and underscores the need for more sophisticated modeling, improved detection algorithms, and multi‑survey comparisons to fully unravel the origins of repeating microlensing events.