SN Requiem: Multiply Imaged SN Ia

• SN Requiem is a multiply imaged Type Ia supernova observed in a strongly lensed galaxy at z ≈ 1.95, serving as a prototype for cluster-scale time-delay cosmography.
• The distinct multiple images (2a, 2b, 2c) and predicted future images precisely map the mass profile of MACS J0138, refining lens models and time-delay estimates.
• Combining archival HST and JWST data with spectroscopy, studies of SN Requiem enhance H0 measurements and advance methodologies in gravitational lensing.

SN Requiem is a multiply imaged Type Ia supernova in the massive, red, multiply imaged galaxy MRG-M0138 at $z_s = 1.949$, strongly lensed by the galaxy cluster MACS J0138.0−2155 at $z_l = 0.336$. It was discovered retrospectively in archival 2016 Hubble Space Telescope imaging and is widely treated as the first cluster-scale photometrically classified multiply-imaged SN Ia. Its scientific importance increased substantially after the discovery of SN Encore in 2023 in the same host galaxy, making MACS J0138.0−2155 the first known system to produce more than one observed multiply-imaged supernova from the same host and establishing the field as an unusually powerful laboratory for cluster-scale time-delay cosmography (Rodney et al., 2021, Pierel et al., 2024, Bazzanini et al., 23 Jun 2026).

1. Discovery, classification, and host environment

SN Requiem, also designated AT2016jka, was identified in archival HST WFC3/IR imaging obtained on 18–19 July 2016 and had faded by the time of the 2019 REQUIEM program observations, when the transient was recognized. The source lies in MRG-M0138, described as a quiescent, strongly lensed galaxy at $z = 1.95$. The discovery analysis characterized the host as a massive, slowly quenched red galaxy with $\log_{10}(M/M_\odot) = 11.7$, specific SFR $\approx 10^{-11.3}\ {\rm yr}^{-1}$, and an exponential SFH age $\approx 1.4$ Gyr (Rodney et al., 2021).

The original classification was probabilistic and template-based. Host-based classification yielded $p({\rm Ia}) = 0.75$ from host color–magnitude via galsnid and $p({\rm Ia}) = 0.62$ from rates based on stellar mass and SFR. SN-based classification gave $p({\rm Ia}) \approx 0.95$ from color–magnitude comparison of demagnified SN images to simulated populations at $z = 1.95$, and a joint SALT2-extended analysis yielded $z_l = 0.336$0. The discovery paper therefore concluded that SN Requiem is very likely a Type Ia supernova, at $z_l = 0.336$1 confidence (Rodney et al., 2021).

In the 2016 data, three images of the supernova were observed. The discovery paper denoted them SN1–SN3; later MACS J0138 analyses standardized the nomenclature to 2a, 2b, and 2c. That renaming became important once SN Encore was incorporated into a unified system-level lensing analysis, with the Encore family labeled 1a–1e and the Requiem family labeled 2a–2e (Rodney et al., 2021, O'Donnell et al., 27 Aug 2025).

The joint occurrence of Requiem and Encore is itself statistically unusual. For the host galaxy MRG-M0138, one analysis estimated the probability of detecting two SNe Ia over a $z_l = 0.336$2 year window to be $z_l = 0.336$3, while also noting that an alternative mass-weighted rate gives a higher feasibility of $z_l = 0.336$4; the headline value used in the abstract and discussion is $z_l = 0.336$5 (Pierel et al., 2024).

2. Image configuration and lensing geometry

MACS J0138.0−2155 is a massive cluster-scale lens whose multiply imaged background constraints now include the host galaxy of both supernovae. Later JWST-era analyses describe the host itself as quintuply imaged, with images A–E distributed across the field, and the bright central region of MRG-M0138 as multiply imaged at least five times in two giant tangential arcs and a radial arc northeast of the BCG (Pierel et al., 2024, Ertl et al., 12 Mar 2025).

For SN Requiem, the three observed images 2a, 2b, and 2c were seen in 2016 HST imaging. Their positions relative to the BCG were tabulated in subsequent lensing work as 2a: $z_l = 0.336$6, 2b: $z_l = 0.336$7, and 2c: $z_l = 0.336$8, with $z_l = 0.336$9 positional uncertainty ellipses (Ertl et al., 12 Mar 2025). The 2021 discovery paper also provided absolute J2000 coordinates for the observed images and for two predicted future images, SN4 and SN5 (Rodney et al., 2021).

The existence of at least one future image is a robust conclusion across later models. A future image 2d is consistently predicted near the cluster core or radial-arc region. A fifth image 2e is more model dependent: several analyses predict a demagnified central or near-central image, but its occurrence rate and parity vary with the adopted mass model. One multi-plane “ultimate” model placed 2d at $z = 1.95$0 with $z = 1.95$1 and 2e at $z = 1.95$2 with $z = 1.95$3, noting that 2e appears in only $z = 1.95$4 of samples. A joint lensing+kinematics model instead described both 2d and 2e as future images associated with the radial critical curve, with 2d expected near the central arc and 2e in the northeast radial arc region (Ertl et al., 12 Mar 2025, O'Donnell et al., 27 Aug 2025).

This model dependence is not a contradiction about the existence of the next observable image; rather, it is concentrated in the detailed topology of the inner potential, including the central-image sector. A plausible implication is that Requiem is especially sensitive to the BCG mass profile, radial critical-curve placement, and inner-halo degeneracies.

3. Strong-lensing data and mass-model development

The lensing description of SN Requiem has evolved with the quality of the MACS J0138 data set. The current “gold” strong-lensing sample consists of 23 spectroscopically confirmed multiple images from 8 distinct background sources spanning $z = 1.95$5, identified using HST and JWST imaging and VLT/MUSE spectroscopy. The cluster member catalog includes 84 galaxies, of which 50 are spectroscopically confirmed, and stellar velocity dispersions were measured for 14 early-type members to calibrate Faber–Jackson-based scaling relations. Two line-of-sight galaxies, a foreground object at $z = 1.95$6 and a background object at $z = 1.95$7, are treated in a multi-plane framework, and three jellyfish galaxies are included as individually optimized perturbers (Bazzanini et al., 23 Jun 2026).

The 2021 discovery paper modeled the lens with Lenstool using 37 dPIE potentials: one main cluster halo, the BCG, 32 cluster-member halos, and three local perturbers. Its preferred model E achieved an image-plane RMS discrepancy of $z = 1.95$8 and already predicted a long-delayed fourth image near the cluster core (Rodney et al., 2021). Subsequent analyses expanded both the data and the model space. A 2025 study constructed seven different mass models, explored both multi-plane and approximate single-plane descriptions, and found that six of the seven models fit the observed image positions well, while the model with a Navarro–Frenk–White cluster DM profile had an image-position $z = 1.95$9 value four times higher (Ertl et al., 12 Mar 2025). Another 2025 analysis using Lenstool reported a reference best-fit model with an overall RMS offset of $\log_{10}(M/M_\odot) = 11.7$0 for all 23 images and a mean precision of only $\log_{10}(M/M_\odot) = 11.7$1 for the multiple image positions of the SNe and their host galaxy (Acebron et al., 12 Mar 2025).

A blind intercomparison then assembled seven independent cluster mass models using six software packages: glafic, GLEE, Lenstool I, Lenstool II, Zitrin-analytic, 0.8/MrMARTIAN, and WSLAP+. Four of the seven baseline models achieved $\log_{10}(M/M_\odot) = 11.7$2, and two achieved $\log_{10}(M/M_\odot) = 11.7$3; across the lower-$\log_{10}(M/M_\odot) = 11.7$4 models, predictions of positions, magnifications, and time delays for both Requiem and Encore were in good mutual agreement (Suyu et al., 15 Sep 2025). The latest dedicated Requiem+Encore analysis used Gravity.jl, a Bayesian parametric lens-modeling framework in which the cluster-scale halo is modeled as a Non-singular Isothermal Ellipsoid, cluster members as dPIE subhalos scaled with luminosity, plus individually modeled perturbers, multi-plane FG/BG galaxies, and an external shear term. Posterior sampling employed non-reversible parallel tempering through Pigeons.jl and emcee. Its fixed-cosmology reference model reproduced image positions with a reduced chi-square of about $\log_{10}(M/M_\odot) = 11.7$5 and an image-plane RMS residual $\log_{10}(M/M_\odot) = 11.7$6 (Bazzanini et al., 23 Jun 2026).

Independent cross-checks on the mass scale support these lens models. One enhanced strong-lensing analysis measured a projected total mass of $\log_{10}(M/M_\odot) = 11.7$7, consistent with the Chandra hydrostatic value $\log_{10}(M/M_\odot) = 11.7$8 (Acebron et al., 12 Mar 2025). A separate joint strong-lensing and stellar-kinematics study reported that the kinematics strongly constrain the enclosed mass within $\log_{10}(M/M_\odot) = 11.7$9 kpc and help break the mass-sheet degeneracy that affects pure-lensing models (O'Donnell et al., 27 Aug 2025).

4. Time delays and the changing forecast for the next image

The defining observational feature of SN Requiem is its long expected reappearance delay. In the discovery paper, the three observed images had relative time delays $\approx 10^{-11.3}\ {\rm yr}^{-1}$0 days, while the predicted fourth image had a lens-model delay of $\approx 10^{-11.3}\ {\rm yr}^{-1}$1 days relative to SN1, implying a peak in $\approx 10^{-11.3}\ {\rm yr}^{-1}$2. A fifth, much fainter image was predicted for $\approx 10^{-11.3}\ {\rm yr}^{-1}$3 (Rodney et al., 2021).

That forecast shifted markedly as the MACS J0138 model incorporated JWST imaging, deeper spectroscopy, and more multiply imaged systems. Later analyses revised the long delay from an approximately two-decade baseline to a roughly decade-scale baseline. In parallel, short-delay inferences among the already observed images were refined. Using color-phase fitting with SALT2 templates via SNTD and a single HST epoch, one analysis reported $\approx 10^{-11.3}\ {\rm yr}^{-1}$4 days and $\approx 10^{-11.3}\ {\rm yr}^{-1}$5 days, with definitions relative to 2b. In the same system, lower-$\approx 10^{-11.3}\ {\rm yr}^{-1}$6 blind lens models predicted short delays of $\approx 10^{-11.3}\ {\rm yr}^{-1}$7 days and $\approx 10^{-11.3}\ {\rm yr}^{-1}$8 to $\approx 10^{-11.3}\ {\rm yr}^{-1}$9 days, explicitly noting that the new models match $\approx 1.4$0 within $\approx 1.4$1 but predict $\approx 1.4$2 about a factor of $\approx 1.4$3 shorter than the Rodney et al. color-curve estimate (Bazzanini et al., 23 Jun 2026, Suyu et al., 15 Sep 2025).

Analysis	Forecast for the next Requiem image	Notes
Discovery model (Rodney et al., 2021)	SN4 in $\approx 1.4$4; delay $\approx 1.4$5 days relative to SN1	Also predicted a much fainter SN5 in $\approx 1.4$6
Blind seven-model comparison (Suyu et al., 15 Sep 2025)	$\approx 1.4$7 to $\approx 1.4$8 days relative to 2a	For $\approx 1.4$9, reappearance in approximately April–December 2026; for $p({\rm Ia}) = 0.75$0, in approximately March–November 2027
Gravity.jl (Bazzanini et al., 23 Jun 2026)	$p({\rm Ia}) = 0.75$1 days	Expected reappearance in $p({\rm Ia}) = 0.75$2 February–July 2027; predicted $p({\rm Ia}) = 0.75$3
Joint lensing+kinematics (O'Donnell et al., 27 Aug 2025)	$p({\rm Ia}) = 0.75$4 days relative to 2b	Median calendar prediction November 2027; $p({\rm Ia}) = 0.75$5 days, median April 2028

These revisions are directly tied to model improvements rather than to any change in the physical event. The discovery paper explicitly anticipated that the classification and predicted reappearance time could be improved with further lens modelling and a comprehensive analysis of systematic uncertainties (Rodney et al., 2021). The later literature is therefore best understood as an iterative tightening of the Fermat-potential estimate rather than a sequence of incompatible claims.

5. Time-delay cosmography and the measurement of $p({\rm Ia}) = 0.75$6

SN Requiem is part of a cluster-scale time-delay cosmography program grounded in the standard thin-lens relations. The lens equation is

$p({\rm Ia}) = 0.75$7

the Fermat potential is

$p({\rm Ia}) = 0.75$8

and the time delay between images $p({\rm Ia}) = 0.75$9 and $p({\rm Ia}) = 0.62$0 is

$p({\rm Ia}) = 0.62$1

For fixed $p({\rm Ia}) = 0.62$2 and $p({\rm Ia}) = 0.62$3, $p({\rm Ia}) = 0.62$4, so measured delays combined with a well-constrained lens potential directly constrain $p({\rm Ia}) = 0.62$5 (Bazzanini et al., 23 Jun 2026).

The MACS J0138 system now supports two related cosmographic strategies. The first used the measured short delay of SN Encore, $p({\rm Ia}) = 0.62$6 days, together with the blind seven-model ensemble, yielding

$p({\rm Ia}) = 0.62$7

The uncertainty was explicitly described as dominated by the time-delay measurement rather than by inter-model scatter (Suyu et al., 15 Sep 2025, Pierel et al., 15 Sep 2025).

The second incorporated both SN Encore and SN Requiem into a joint Gravity.jl inference. Allowing $p({\rm Ia}) = 0.62$8 to vary with a uniform prior $p({\rm Ia}) = 0.62$9 and using the measured time delays of both SNe in the likelihood yielded

$p({\rm Ia}) \approx 0.95$0

This value was reported as consistent with the result from the independent seven-model combination, $p({\rm Ia}) \approx 0.95$1, and the current error budget was described as dominated by the $p({\rm Ia}) \approx 0.95$2 relative uncertainties of the measured SN time delays (Bazzanini et al., 23 Jun 2026).

The significance of Requiem in this program is its exceptionally long delayed image. The future 2d image offers a temporal baseline of roughly eleven years relative to 2a in the Gravity.jl analysis, or approximately twelve years from the 2016 images to the next appearance in the joint lensing+kinematics model. This suggests that once 2d is observed with high-S/N monitoring, the Requiem delay could become the most precise timing observable in the system. Forecasts in the literature range from “$p({\rm Ia}) \approx 0.95$3–$p({\rm Ia}) \approx 0.95$4” for high-precision $p({\rm Ia}) \approx 0.95$5 constraints to “$p({\rm Ia}) \approx 0.95$6–$p({\rm Ia}) \approx 0.95$7” for the combined system, provided that lens-model systematics are controlled (Bazzanini et al., 23 Jun 2026, Suyu et al., 15 Sep 2025, Pierel et al., 15 Sep 2025).

6. Systematics, open issues, and observational outlook

The principal systematic uncertainties in SN Requiem cosmography are now well defined. The cluster-scale literature identifies the mass-sheet degeneracy and mass-slope assumptions; substructure and galaxy-halo scaling relations; external convergence and line-of-sight structures, including multi-plane effects; and the parametric flexibility of the cluster’s total mass profile as the main lens-model systematics (Bazzanini et al., 23 Jun 2026). The software-comparison study adds a methodological control by showing that low-$p({\rm Ia}) \approx 0.95$8 models built with distinct codes are in good agreement on Requiem’s positions, magnifications, and long delays (Suyu et al., 15 Sep 2025).

Supernova-specific systematics enter differently. Type Ia SN microlensing primarily affects magnification and light-curve fits, but color-based phase methods mitigate chromatic biases. In the Encore analyses, chromatic microlensing and millilensing by dark matter substructure were explicitly propagated; millilensing was noted to affect magnification but not the time delay itself. This matters for Requiem because the standardized brightness of a SN Ia can in principle help limit the mass-sheet degeneracy, but only if microlensing and millilensing are not extreme and if template imaging is available (Bazzanini et al., 23 Jun 2026, Pierel et al., 15 Sep 2025).

Several issues remain model dependent. The presence and detectability of the fifth image 2e are uncertain. One multi-plane lens model found 2e in only $p({\rm Ia}) \approx 0.95$9 of ultimate-model samples (Ertl et al., 12 Mar 2025), whereas the joint lensing+kinematics model predicted a median calendar date of April 2028 and described it as challenging but still feasible with deep exposures or high-cadence difference imaging because of demagnification and proximity to the BCG light (O'Donnell et al., 27 Aug 2025). A plausible implication is that 2e is better regarded as a discriminant among inner-mass models than as the primary cosmographic target.

Monitoring strategies in the literature reflect the forecast spread. The blind seven-model analysis recommended vigilant monitoring in late 2025 through 2027, with emphasis on the April–December 2026 and March–November 2027 windows associated with different $z = 1.95$0 values (Suyu et al., 15 Sep 2025). The joint lensing+kinematics analysis recommended monthly cadence from mid-2026 through end-2028, with higher cadence during the most probable six-month windows around the medians, using JWST/NIRCam F150W–F200W and HST/WFC3-IR F160W; it also stressed PSF modeling, subtraction, and template differencing because of the image proximity to the BCG (O'Donnell et al., 27 Aug 2025). The Gravity.jl study further argued that continued HST/JWST monitoring, deeper spectroscopy of cluster members and line-of-sight galaxies, and incorporation of extended surface-brightness information from the lensed host would tighten constraints on the local lensing distortion field near the SN images and on the inferred $z = 1.95$1 (Bazzanini et al., 23 Jun 2026).

Within this framework, SN Requiem has moved from being a singular archival discovery to a central calibrator for cluster-scale time-delay cosmography. The event’s historical three-image appearance established the feasibility of decade-scale SN delays in a cluster lens, while the predicted return of image 2d has become the immediate experimental test of the current mass models. The present consensus is not that all model details are settled, but that MACS J0138 is now sufficiently data-rich that the next appearance of SN Requiem can materially sharpen the cosmographic utility of the system.