High Magnification Events (HMEs)

• High Magnification Events (HMEs) are gravitational lensing occurrences where the source closely approaches the lens center or caustic, resulting in extreme magnification.
• They play a crucial role in exoplanet detection by enhancing sensitivity to central caustic perturbations and enabling the discovery of multiple planetary systems.
• Simulations and follow-up observations leverage HMEs to constrain lens parameters such as mass ratios and lens-source separations, advancing microlensing astrophysics.

Searching arXiv for recent and foundational papers on high-magnification events in microlensing and related lensing contexts. High Magnification Events (HMEs) are gravitational-lensing events in which the source approaches the lens center or a caustic closely enough that the observed magnification becomes very large. In stellar microlensing, the term usually denotes small-impact-parameter events with very high peak amplification, often with the practical threshold $A_{\max}\gtrsim 100$, corresponding to $u_0\lesssim 0.01$ in the high-magnification limit $A\simeq u^{-1}$ (Gould et al., 2012). In quasar microlensing and in lensing near galaxy-cluster critical curves, the same term is used more broadly for strong brightening or dimming episodes associated with source motion through microcaustic structure (Neira et al., 29 Jul 2025). Across these regimes, the importance of HMEs derives from the same geometric fact: they concentrate observational sensitivity into the small regions of lensing phase space where the response to source structure, companion masses, or caustic topology is strongest.

1. Definition and geometric regime

For a point lens, the magnification is

$A(u)=\frac{u^2+2}{u\sqrt{u^2+4}},$

and the high-magnification approximation is $A\approx 1/u$ when the source-lens separation $u$ is small (Choi et al., 2011). In this regime, an HME is fundamentally a small-$u_0$ event: the source trajectory passes extremely close to the lens, the peak becomes large, and the light curve is dominated by the immediate neighborhood of the Einstein ring or the central caustic. A source-crossing HME is the more restrictive case $u_0<\rho_\star$, where $\rho_\star$ is the source radius in units of $\theta_{\rm E}$; then finite-source effects reshape the peak and encode the source surface-brightness profile (Choi et al., 2011).

In planetary microlensing, the defining geometric feature of HMEs is that they preferentially probe the central caustic near the host star. That is why they are unusually sensitive to companions that perturb the lens potential close to the host, including planets near and beyond the snow line and, in special cases, multiple planets in the same system (Shin et al., 2015). Simulations of KMTNet-like observations make the same point in survey terms: high magnification biases detections toward smaller impact parameters and therefore toward central perturbations, even though the very highest magnifications can become less favorable under uniform 10-minute cadence because the anomalies near peak become too short and subtle to sample adequately (Zhu et al., 2014).

The term also has a wider lensing usage. In quasar microlensing, an HME is a pronounced episode in the microlensing light curve produced when the accretion disk moves through a complicated magnification map generated by stars in the lens galaxy; in that literature the event may be operationally defined by an amplitude threshold rather than solely by impact parameter (Neira et al., 29 Jul 2025). Near galaxy-cluster critical curves, ultra-high magnification events arise when compact sources such as giant stars, supernovae, or Pop III stars approach fold caustics closely enough for magnifications of thousands or more, subject to finite-source and microlens-induced limits (Diego, 2018).

2. HMEs as a planetary detection channel

The modern importance of HMEs in exoplanet microlensing rests on two linked properties: predictability and central-caustic sensitivity. High-magnification peaks can often be anticipated from the rising light curve, which allows telescope time to be concentrated near maximum, and the same geometry naturally probes the central perturbation region where a planetary companion is detectable (Gould et al., 2010). In a controlled sample of $u_0\lesssim 0.01$0 events monitored during 2005–2008, 13 HMEs entered the final sensitivity analysis and 6 planets were found; from that sample the measured planet frequency beyond the snow line was

$u_0\lesssim 0.01$1

at mean mass ratio $u_0\lesssim 0.01$2, with a distribution consistent with being flat in $u_0\lesssim 0.01$3 (Gould et al., 2010). The same study emphasized that the high-magnification channel is efficient: about half of all high-mag events were successfully monitored, and about half of those monitored events yielded planet detections (Gould et al., 2010).

HME-based strategies were subsequently extended toward lower planet masses. A dedicated proposal argued that moderately high magnifications, roughly $u_0\lesssim 0.01$4, are a practical sweet spot for detecting planets of $u_0\lesssim 0.01$5–$u_0\lesssim 0.01$6 with 1–2 m class telescopes, because such events are much more common than the rarest extreme peaks while still producing short, detectable perturbations near maximum (Abe et al., 2013). In that framework, planets of a few Earth masses were found to produce deviations of $u_0\lesssim 0.01$7 lasting $u_0\lesssim 0.01$8–$u_0\lesssim 0.01$9 hr in events with magnification $A\simeq u^{-1}$0 when the projected separation lies in an annular region around the Einstein ring; Earth-mass sensitivity becomes confined more narrowly to $A\simeq u^{-1}$1–$A\simeq u^{-1}$2 (Abe et al., 2013).

Survey simulations reinforced both the value and the limits of the HME channel. In a KMTNet-like experiment based on 6690 simulated microlensing events, 313 exceeded $A\simeq u^{-1}$3, 292 were confirmed as planetary events, and 16 were two-planet events, implying a planetary-event fraction of $A\simeq u^{-1}$4 and a multiple-planet fraction of $A\simeq u^{-1}$5 among planetary detections (Zhu et al., 2014). High magnification and large planet mass ratios were both favored, but events with $A\simeq u^{-1}$6 were significantly less sensitive than lower-$A\simeq u^{-1}$7 events under survey-only 10-minute cadence, which led to the conclusion that follow-up observations remain essential for extracting the full science return from very high-magnification events (Zhu et al., 2014).

The transition from survey-plus-follow-up discovery to survey-only discovery did not eliminate the special role of HMEs. OGLE-2016-BLG-0596Lb, a pure-survey HME discovered in archival data, reached a point-lens magnification of about $A\simeq u^{-1}$8 and yielded a resonant-caustic planetary solution with

$A\simeq u^{-1}$9

showing that high-magnification sensitivity can be exploited even when no real-time anomaly recognition occurs (Mróz et al., 2016).

3. Multiple planets and exclusion constraints

HMEs are unusually valuable for multiplanet searches because the source trajectory passes near the host star and therefore samples the region where the central caustic associated with each planet lies. This motivated a reanalysis of eight previously known single-planet HMEs—OGLE-2005-BLG-071, OGLE-2005-BLG-169, MOA-2007-BLG-400, MOA-2008-BLG-310, MOA-2009-BLG-319, MOA-2009-BLG-387, MOA-2010-BLG-477, and MOA-2011-BLG-293—using triple-lens models consisting of a host star, the known planet, and a putative second planet (Shin et al., 2015). The analysis held the known planet fixed and scanned the second planet over

$A(u)=\frac{u^2+2}{u\sqrt{u^2+4}},$0

with the remaining parameters optimized by MCMC (Shin et al., 2015).

The result was negative in the strict detection sense but positive in the constraint sense. Adding a second planet improved the fit in 7 of 8 events, but seven had $A(u)=\frac{u^2+2}{u\sqrt{u^2+4}},$1, and the largest apparent improvement, in MOA-2009-BLG-319, was $A(u)=\frac{u^2+2}{u\sqrt{u^2+4}},$2, from $A(u)=\frac{u^2+2}{u\sqrt{u^2+4}},$3 to $A(u)=\frac{u^2+2}{u\sqrt{u^2+4}},$4 (Shin et al., 2015). Detailed inspection showed that the light-curve peak in the binary and triple models was visually almost indistinguishable, the improvement was only at the $A(u)=\frac{u^2+2}{u\sqrt{u^2+4}},$5 mag level, and the excess was dominated by specific data sets, consistent with systematics rather than a coherent second-planet anomaly (Shin et al., 2015). The authors therefore reported no firm detection of additional planets.

The same analysis formalized nondetections through exclusion diagrams. For each $A(u)=\frac{u^2+2}{u\sqrt{u^2+4}},$6, the model fit was compared to the original binary-lens model, and the detection efficiency—defined as the fraction of orientation angles $A(u)=\frac{u^2+2}{u\sqrt{u^2+4}},$7 yielding significant deviations—was interpreted as the confidence level for excluding such a planet (Shin et al., 2015). The adopted detection threshold was $A(u)=\frac{u^2+2}{u\sqrt{u^2+4}},$8, and the physical mapping was

$A(u)=\frac{u^2+2}{u\sqrt{u^2+4}},$9

The strongest exclusions were for large mass ratios, favorable separations, high-quality peak coverage, weak finite-source suppression, and high $A\approx 1/u$0 (Shin et al., 2015).

At the 90% confidence level, the reported exclusion ranges were as follows (Shin et al., 2015).

Event	Jupiter 90% exclusion	Saturn/Uranus 90% exclusion
OGLE-2005-BLG-071	$A\approx 1/u$1–$A\approx 1/u$2 AU	—
OGLE-2005-BLG-169	$A\approx 1/u$3–$A\approx 1/u$4 AU	Saturn: $A\approx 1/u$5–$A\approx 1/u$6 AU
MOA-2007-BLG-400	$A\approx 1/u$7–$A\approx 1/u$8 AU	Saturn: $A\approx 1/u$9–$u$0 AU
MOA-2008-BLG-310	$u$1–$u$2 AU	Saturn: $u$3–$u$4 AU
MOA-2009-BLG-319	$u$5–$u$6 AU	Saturn: $u$7–$u$8 AU; Uranus: $u$9–$u_0$0 AU
MOA-2009-BLG-387	$u_0$1–$u_0$2 AU	Saturn: $u_0$3–$u_0$4 AU
MOA-2010-BLG-477	$u_0$5–$u_0$6 AU	Saturn: $u_0$7–$u_0$8 AU
MOA-2011-BLG-293	$u_0$9–$u_0<\rho_\star$0 AU	—

Roman predictions suggest that HMEs will become a major route to direct two-planet detections rather than only exclusion constraints. Simulations of $u_0<\rho_\star$1 synthetic high-magnification triple-lens light curves over a 72-day Roman observing window, sampled at 15-minute cadence and analyzed with a conservative second-planet threshold $u_0<\rho_\star$2, yielded an overall detection efficiency of $u_0<\rho_\star$3 (Saggese et al., 4 Dec 2025). Systems with resonant configurations were most favorable, with $u_0<\rho_\star$4 for resonant/resonant, $u_0<\rho_\star$5 for close/resonant, and $u_0<\rho_\star$6 for wide/resonant; the predicted Roman yield was approximately $u_0<\rho_\star$7 of multi-planet microlensing events, corresponding to about 64 HME triple-lens detections over the full survey (Saggese et al., 4 Dec 2025).

4. Central perturbations, degeneracies, and morphological diagnostics

The observational utility of HMEs is inseparable from the degeneracies they produce. Central perturbations near the peak can be caused by a planet or by a very close or very wide binary companion, creating the planet-binary degeneracy (Chung et al., 2012). In the canonical comparison, the planetary central caustic is a small asymmetric arrow-shaped figure whose tip points toward the planet, whereas the wide-binary central caustic is approximately symmetric and asteroid-shaped in the Chang-Refsdal limit (Chung et al., 2012). The corresponding width ratio

$u_0<\rho_\star$8

is therefore far from unity for planetary caustics near resonance but close to unity for wide binaries (Chung et al., 2012).

A useful observational discriminator exists in caustic-crossing HMEs. Using the magnification excess

$u_0<\rho_\star$9

it was shown that planetary and binary lenses produce different excess patterns inside their central caustics, and therefore different interpeak morphologies (Chung et al., 2012). Planetary caustic-crossing HMEs exhibit a smooth convex or boxy interpeak region between the two caustic-crossing peaks, whereas binary-lensing events exhibit a smooth concave interpeak region (Chung et al., 2012). The diagnostic is strongest when the source trajectory passes near the center of the planetary caustic, and it weakens under finite-source smoothing; for weak finite-source effects, the convex feature appears for $\rho_\star$0 in the discussed examples (Chung et al., 2012).

That diagnostic is powerful but not universal. A distinct ambiguity was identified for flat or blunt-topped peaks in HMEs, where the source does not execute an obvious caustic crossing but instead samples a negative perturbation region near the central caustic (Choi et al., 2012). In the planetary case, the source passes through the negative region behind the back end of the arrowhead caustic with $\rho_\star$1; in the binary case, it passes through the negative region between two cusps of an astroid-shaped caustic with $\rho_\star$2 (Choi et al., 2012). Two 2011 HMEs illustrated the effect. For OGLE-2011-BLG-0526, the difference between planetary and binary solutions was only $\rho_\star$3 in $\rho_\star$4, while for OGLE-2011-BLG-0950/MOA-2011-BLG-336 the binary model was formally excluded by $\rho_\star$5, yet systematics at the sub-1% level prevented a secure planet claim (Choi et al., 2012).

Post-event high-resolution imaging later resolved the OGLE-2011-BLG-0950 ambiguity. Keck observations in 2019 and 2021 measured a heliocentric lens-source relative proper motion of $\rho_\star$6, inconsistent with the planetary prediction and consistent with the close stellar-binary model (Terry et al., 2022). The preferred physical solution yielded $\rho_\star$7, $\rho_\star$8, $\rho_\star$9 kpc, and $\theta_{\rm E}$0 AU (Terry et al., 2022). This established the central-caustic cusp-approach degeneracy as a practical concern for exoplanet demographics rather than only a formal modeling issue.

Binary-lens HMEs also exhibit their own internal ambiguity. In a sample of eight binary microlensing events discovered through the HME channel during 2007–2010, the perturbations were all confined near the peak and were readily distinguished from ordinary planetary central perturbations, but the close/wide binary degeneracy could not be resolved at $\theta_{\rm E}$1 for three events (Shin et al., 2011). The severity of that degeneracy was found to increase as the binary separation and mass ratio deviated from resonant caustic values (Shin et al., 2011).

5. Finite-source physics, limb darkening, parallax, and chromatic structure

Some of the most information-rich HMEs are not planetary anomalies at all but single-lens transits in which the lens passes over the source star. In a sample of nine such events—OGLE-2004-BLG-254, MOA-2007-BLG-176, MOA-2007-BLG-233/OGLE-2007-BLG-302, MOA-2009-BLG-174, MOA-2010-BLG-436, MOA-2011-BLG-093, MOA-2011-BLG-274, OGLE-2011-BLG-0990/MOA-2011-BLG-300, and OGLE-2011-BLG-1101/MOA-2011-BLG-325—the finite-source perturbation at peak yielded limb-darkening measurements for all 9 events and Einstein-radius measurements for 7 (Choi et al., 2011). With

$\theta_{\rm E}$2

the source-crossing geometry converts light-curve morphology into lens characterization (Choi et al., 2011). Five events had $\theta_{\rm E}$3 mas, suggesting very low-mass stars or brown dwarfs, and MOA-2011-BLG-274, with $\theta_{\rm E}$4 mas and $\theta_{\rm E}$5 days, was identified as a possible free-floating planet candidate (Choi et al., 2011).

The same finite-source information can be inverted more flexibly than by assuming a fixed analytic limb-darkening law. A finite-element treatment of the magnification–limb-darkening integral equation was applied to seven single-filter light curves from six high-cadence transit HMEs and recovered intensity profiles consistent with linear limb darkening in five cases and square-root profiles in two (Golchin et al., 2019). The method was formulated as a Fredholm integral equation of the first kind, regularized by penalizing curvature, and its principal advantage was that it imposed no explicit profile shape beyond flatness near the stellar center (Golchin et al., 2019).

High magnification also simplifies space-based microlens parallax. For $\theta_{\rm E}$6 events, accurate microlens parallaxes can be obtained from three or fewer photometric measurements from a small telescope on a satellite in solar orbit at $\theta_{\rm E}$7 AU from Earth, requiring 1–2 orders of magnitude fewer observing resources than standard space-based parallaxes (Gould et al., 2012). When a finite-source measurement gives $\theta_{\rm E}$8, the combination with $\theta_{\rm E}$9 yields

$u_0\lesssim 0.01$00

so lens mass and distance follow directly (Gould et al., 2012). The method is especially well matched to HMEs because they are both highly planet-sensitive and unusually likely to show finite-source effects near the peak (Gould et al., 2012).

At still higher photometric precision, HMEs become probes of chromatic source structure. Simulations of dual-color monitoring with the Danish 1.54 m telescope examined limb darkening, close-in giant planets, stellar spots, and gravity darkening in HMEs and caustic-crossing events (Sajadian et al., 2021). The study concluded that the likelihood of detecting these phenomena per object where they are present is $u_0\lesssim 0.01$01 during high magnification and $u_0\lesssim 0.01$02 during caustic crossings, and it identified HMEs as the regime in which color variations from finite-source structure become most measurable (Sajadian et al., 2021).

6. Broader lensing uses: quasar HMEs, ultra-high magnification, and forecasting

Outside Galactic stellar microlensing, HME has become a standard term in quasar microlensing for the sharp, information-rich variability episodes generated when a lensed quasar image moves through a magnification map created by stars in the lens galaxy (Neira et al., 29 Jul 2025). In one recent operational definition, an HME is any brightening or dimming event whose amplitude exceeds $u_0\lesssim 0.01$03 mag in the LSST $u_0\lesssim 0.01$04-band (Neira et al., 29 Jul 2025). For observable systems with minimum image separation at least $u_0\lesssim 0.01$05 and second-dimmest image brighter than $u_0\lesssim 0.01$06, the predicted rate is

$u_0\lesssim 0.01$07

in either the northern or southern sky (Neira et al., 29 Jul 2025). Saddle images are about four times more likely to host HMEs than minima, yet only about 10% of saddle-image HMEs are caustic crossings compared with about 50% for minima, so a large fraction of quasar HMEs are not literal caustic-crossing events (Neira et al., 29 Jul 2025).

Because LSST cadence is not ideal for sampling the inner accretion disk during these episodes, there is strong interest in forecasting HMEs early enough to trigger follow-up. A recurrent-neural-network study trained on 282,100 simulated multiband quasar microlensing light curves reported that LSST-like observations should permit prediction of about 55% of HME peaks with a false-positive rate of around 20% relative to the total number of HMEs (Fagin et al., 2024). The performance improved substantially with more bands and with the removal of seasonal gaps, confirming that the scientific value of quasar HMEs depends not only on their intrinsic rate but also on pre-peak alerting (Fagin et al., 2024).

In strong lensing by galaxy clusters, HMEs denote the extreme tail of magnification near macro-caustics. In the smooth fold-caustic limit, the cumulative probability obeys $u_0\lesssim 0.01$08 and the differential distribution obeys $u_0\lesssim 0.01$09 (Diego, 2018). Finite source size would in principle limit the maximum magnification roughly as $u_0\lesssim 0.01$10, but the more important practical bound comes from microlenses near the critical curve, which corrugate the macrocaustic into a dense microcaustic network and produce a natural ceiling of order $u_0\lesssim 0.01$11 for many configurations (Diego, 2018). For Pop III-star caustic crossings, a practical upper limit of $u_0\lesssim 0.01$12 was adopted in realistic high-redshift environments (Diego, 2018).

An analytic treatment of the same regime framed the high-magnification tail in terms of the number of independent micro-critical curves. With

$u_0\lesssim 0.01$13

the point-source tail normalization reduces to the compact scaling

$u_0\lesssim 0.01$14

where the linear regime $u_0\lesssim 0.01$15 corresponds to mostly isolated micro-critical curves and the nonlinear regime $u_0\lesssim 0.01$16 to exponentially suppressed independence because of overlap (Kawai et al., 2024). That model reproduced the dependence of the ultra-high-magnification PDF on microlens mass fraction and background magnification, and it yielded Icarus-like event-rate estimates consistent with HST observations (Kawai et al., 2024).

Across these disparate applications, the unifying feature of HMEs is not a single numerical threshold but a shared lensing geometry: the source approaches a singular or near-singular magnification structure so closely that otherwise inaccessible physical information becomes observable. In planetary microlensing, HMEs define a controlled and efficient discovery channel for cold planets, multiplanet systems, and finite-source astrophysics. In quasar microlensing and near cluster critical curves, they isolate the rare epochs in which the source structure and the microlens population are most strongly imprinted on the light curve.