UV and Optical Signatures of Late-time Disk Instabilities in Tidal Disruption Events

arXiv:2602.15103 · astro-ph.HE · Submitted 2026-02-16 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Astrophysics Radio. Generated commentary on the latest astrophysics papers.

Vera: I'm Vera.

Jocelyn: And I'm Jocelyn, here with Subrahmanyan.

Vera: We're looking at the paper: "UV and Optical Signatures of Late-time Disk Instabilities in Tidal Disruption Events".

Jocelyn: Here's the gist of the paper. Tidal disruption events (TDEs) are unique probes of evolving accretion in supermassive black holes, and this work models their UV and optical signatures arising from late-time thermal instabilities that may explain the ubiquitous late radio emergence in TDEs Modeling Disk Instabilities The authors model the ultraviolet (UV) and optical signatures of disk instabilities expected from accretion power being reprocessed by an optically-thick outflow following super-Eddington accretion Their model predicts brief UV-bright transients lasting for days, with luminosities of 1042–1043 erg s−1 in near-UV and 1041–1042 erg s−1 in optical for a typical TDE by a 106 M⊙ black hole These transients could be detectable by near-future surveys such as ULTRASAT, Vera C.

Vera: Rubin Observatory and Argus Array, for TDEs of redshifts out to ≈ 0.1.

Jocelyn: Emission Physics and Light Curve The accretion flare model is inspired by timedependent one-zone modeling of Piro & Mockler (2025) and involves three parameters: the total mass ejected by the accretion flare Mfl, the flare’s duration tfl, and the characteristic disk radius rd during the flare.

Vera: The model calculates a light curve by splitting the wind into multiple “shells” launched at different times and calculating their radiative output.

Paper summary: Vera: So we're diving into this paper called "UV and Optical Signatures of Late-time Disk Instabilities in Tidal Disruption Events." We're looking at how accretion power gets reprocessed in these events and what that might mean for those late radio emissions.

Jocelyn: Yeah, basically the thesis is that if you have these disk instabilities happening after super-Eddington accretion, you should see specific UV and optical flares.

Subrahmanyan: The core idea here is linking a specific type of disk instability to those late radio flares that we see coming from some TDEs later on.

Vera: They model this by splitting the wind into different shells and calculating how much light each shell throws off over time, which is what they call the accretion flare model.

Jocelyn: And they found these transients should be very hot, with peak brightness of one thousand forty-two to one thousand forty-three erg per second in near-UV and one thousand forty-one to one thousand forty-two erg per second in the optical for a typical event <ref:2602.15103#pg2>.

Subrahmanyan: That level of brightness means these events should be visible by high-cadence UV surveys like ULTRASAT or Vera C Rubin Observatory.

Vera: But the paper also tests this against archival data and didn't find any strong matches to this specific disk instability model in the existing TDE samples.

Jocelyn: Because of that non-detection, they set upper limits on how often we might see these flares, putting a constraint of one to two events per year for the most constrained TDEs.

Subrahmanyan: That rate limit tells us either these flares are rarer than their model predicts or the physics causing the instability is different from what this paper assumes.

Vera: The authors themselves say they need to improve their model by using a more realistic two dimensional disk model and better modeling of how the light interacts with the outflow, which they call a greybody assumption.

Jocelyn: So to sum up these UV and optical signatures from late-time disk instabilities in TDEs are expected to be very hot transients with near-UV luminosities around one thousand forty-two to one thousand forty-three erg per second <ref:2602.15103#pg2>.

Subrahmanyan: And the big thing here is that we need better physical modeling, like that two dimensional model, to really confirm whether this is a real thing for explaining those late radio flares.

Vera: We've covered the summary and the conclusion of this paper on "UV and Optical Signatures of Late-time Disk Instabilities in Tidal Disruption Events."

Conclusion: Vera: So we're wrapping up this look at how accretion power gets reprocessed in tidal disruption events and what that might mean for late radio emission, talking about this paper "UV and Optical Signatures of Late-time Disk Instabilities in Tidal Disruption Events."

Jocelyn: Yeah, what they are really saying is that if you have these instabilities happening after the black hole has eaten a lot of matter, you should see specific UV and optical flares.

Subrahmanyan: The core idea is linking a specific type of disk instability to those late radio flares that we see coming from some TDEs later on.

Vera: They model this by splitting the wind into different shells and calculating how much light each shell throws off over time.

Jocelyn: And they found these transients should be very hot, with a peak brightness of one thousand forty-two to one thousand forty-three erg per second in the near-UV and one thousand forty-one to one thousand forty-two erg per second in the optical for a typical event.

Subrahmanyan: That level of brightness means these events should be visible by high-cadence UV surveys like ULTRASAT or Vera C Rubin Observatory.

Vera: But they also tested the real data and didn't find any strong matches to this specific disk instability model in the archival TDEs.

Jocelyn: Because of that non-detection, they set upper limits on how often we might see these flares, putting a constraint of one to two events per year for the most constrained TDEs.

Subrahmanyan: This tells us that either these flares are rarer than their model predicts or the physics causing the instability is different from what this paper assumes.

Vera: The authors themselves say they need to improve their model by using a more realistic two dimensional disk model and better modeling of how the light interacts with the outflow, which is called greybody assumption.

Jocelyn: So, to sum up, these UV and optical signatures from late-time disk instabilities in TDEs are expected to be very hot transients with near-UV luminosities around one thousand forty-two to one thousand forty-three erg per second.

Subrahmanyan: And the big thing here is that we need better physical modeling, like that two dimensional model, to really confirm whether this is a real thing for explaining those late radio flares.

Vera: We've covered the summary and the conclusion of this paper on "UV and Optical Signatures of Late-time Disk Instabilities in Tidal Disruption Events."

Daichi Tsuna, V. Ashley Villar, Anthony L. Piro, Samantha C. Wu

Center for Astrophysics | Harvard & Smithsonian · The NSF AI Institute for Artificial Intelligence and Fundamental Interactions · The Observatories of the Carnegie Institution for Science, Pasadena, CA 91101, USA · Center for Interdisciplinary Exploration & Research in Astrophysics (CIERA), Physics & Astronomy, Northwestern University

astro-ph.HE

Submitted: 2026-02-16

Updated: 2026-10-31

Comments: Accepted to ApJ; code for model results: https://github.com/DTsuna/TDEflares

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 85/100

The gist: Tidal disruption events (TDEs) are unique probes of evolving accretion in supermassive black holes, and this work models their UV and optical signatures arising from late-time thermal instabilities…

Key concepts

Disk Instabilities
This model simulates how accretion power is reprocessed by an optically-thick outflow after super-Eddington accretion. It predicts a flare mechanism where the wind splits into multiple 'shells' launched at different times, causing brief, bright UV and optical transients.
Accretion Flare Model
The emission physics is based on timedependent one-zone modeling. It uses three parameters—the total mass ejected ($M_{fl}$), the flare duration ($t_{fl}$), and the characteristic disk radius ($r_d$)—to calculate a light curve by summing contributions from these emitted shells over time.
UV/Optical Signatures
The model predicts specific observable signatures. For a typical $10^6 M_{\odot}$ black hole, the near-UV emission peaks at magnitudes between -17 and -19 mag for 3–4 days, while optical emission is predicted to be around (-14) to (-17) mag lasting 3–5 days.
TDE Probes
Tidal Disruption Events are unique probes of evolving accretion onto supermassive black holes. The late-time instabilities modeled here are hypothesized to explain the ubiquitous emergence of late radio emission observed in many TDEs.

Terminology

Summary

Tidal disruption events (TDEs) are unique probes of evolving accretion in supermassive black holes, and this work models their UV and optical signatures arising from late-time thermal instabilities that may explain the ubiquitous late radio emergence in TDEs

Modeling Disk Instabilities

The authors model the ultraviolet (UV) and optical signatures of disk instabilities expected from accretion power being reprocessed by an optically-thick outflow following super-Eddington accretion Their model predicts brief UV-bright transients lasting for days, with luminosities of 1042–1043 erg s−1 in near-UV and 1041–1042 erg s−1 in optical for a typical TDE by a 106 M⊙ black hole These transients could be detectable by near-future surveys such as ULTRASAT, Vera C. Rubin Observatory and Argus Array, for TDEs of redshifts out to ≈ 0. p. 0

Where the paper says it

These could be detectable by near-future surveys such as ULTRASAT, Vera C. Rubin Observatory and Argus Array, for TDEs of redshifts out to ≈ 0.

1.

Emission Physics and Light Curve

The accretion flare model is inspired by timedependent one-zone modeling of Piro & Mockler (2025) and involves three parameters: the total mass ejected by the accretion flare Mfl, the flare’s duration tfl, and the characteristic disk radius rd during the flare. The model calculates a light curve by splitting the wind into multiple “shells” launched at different times and calculating their radiative output. p. 2

Where the paper says it

Light curve Our approach below is to split the wind into multiple “shells” launched at times t˜ (0 < t < t ˜ fl) and calculate the radiative output of each shell at time t ′ from launch, to obtain the luminosity observed at time t = t˜+ t ′ . Let us consider a shell of wind material launched over a time ∆t at radius rin and time t˜, with internal energy Eint,0 ≈ 1 2 Lacc∆t (7) and with velocity vw ∼ q (Lacc/2)/(M˙ w/2) ∼ 0.

The resulting luminosity Lobs(t) is obtained by summing up the contributions from all emitted shells over time. p. 2
Where the paper says it

(13) We can now integrate over all the emitted shells to obtain the light curve as a function of t. Using the solution for Lrad(t ′ ) and substituting t ′ = t − t˜ for each wind shell launched at t˜, we sum up the contributions from all shells emitted within t and obtain Lobs(t) =Z t 0 dt˜ Lacc(t˜) 2At˜ B + t − t˜ t˜ 2 rin rin + vw(t − t˜) 2/3 ×exp ( − B3 3A " 1 + 1 B t − t˜ t˜ 3 − 1 #) , (14) The top-left panel of Figure 1 shows a light curve for our fiducial parameter set of MBH = 106M⊙, Mfl = 0.

Parameter Exploration and Detectability

The study explores parameter space by calculating a grid of light curves varying the flare mass Mfl and disk radius rd. For a fiducial choice of rd = 5rT (dashed lines) and MBH = 106 M⊙, the near-UV emission is found to peak at magnitudes between −17 and −19 mag, with durations of 3–4 days. p. 5

Where the paper says it

Figure 2 shows the range of Mpk and ∆tpk at nearUV (250 nm) for the parameter space of rd and Mfl, with MBH of 106 M⊙ and 107 M⊙ in the top and bottom panels. For our fiducial choice of rd = 5rT (dashed lines) and MBH = 106 M⊙, we find the near-UV emission to peak at magnitudes between −17 and −19 mag, with durations of 3–4 days.

The optical emission is predicted to be approximately (−14)–(−17) mag (1041–1042 erg s−1) and last for 3–5 days for a typical SMBH mass of MBH = 106 M⊙. p. 5
Where the paper says it

Figure 3 shows the same parameter grid as Figure 2 but for the optical at 600 nm. For a typical SMBH mass of MBH = 106 M⊙ and rd ≈ 5rT , the emission in the optical is ≈ (−14)–(−17) mag (1041–1042 erg s−1 ) and lasts for 3–5 days.

Observational Search Results

A search was conducted in archival TDE data using optical data from the OTTER database to find late-time flares consistent with disk instability models. The analysis did not find any strong candidates of a disk instability flare, and the non-detection led to upper limits in their event rates of 1–2 yr−1 for each of the three most constraining TDEs. p. 8

Where the paper says it

We did not find any promising candidates from late-time photometric detections. The non-detection led to upper limits in their event rates of 1–2 yr−1 , for each of the three most constraining TDEs.

Conclusion and Future Work

The study concludes that the emission from such disk instabilities is expected to be a very hot (a few – 10 ×104 K), luminous transient with luminosities of 1042–1043 erg s−1 in the near-UV. Suggestions for future work include improving disk modeling to a more realistic 1D/2D model and extending the greybody assumption by detailed radiative transfer modeling. p. 8

Where the paper says it

This could provide more robust predictions for both the timing and the mass of the flares. Second, the greybody assumption for the spectra could be improved by detailed radiative transfer modeling of the outflow, which could impact more on the predictions for optical that are far from the spectral peak.

The search can be improved by combining information of multiple filters and applying more physically informed priors on the flares for significance estimates and rate constraints. p. 8
Where the paper says it

Second, the greybody assumption for the spectra could be improved by detailed radiative transfer modeling of the outflow, which could impact more on the predictions for optical that are far from the spectral peak. Finally the search can be improved in several ways as more sensitive surveys are realized, such as combining information of multiple filters and applying more physically informed priors on the flares for significance estimates and rate constraints.

Appendix A

Table 1 summarizes the redshift, luminosity distance (DL), epoch of last ZTF detection from discovery, and the supermassive BH mass estimates for the 28 TDEs analyzed in this work. The information for these TDEs and their host-subtracted photometric data were obtained from the literature, via the OTTER database as of December 2025. p. 8

Where the paper says it

The information for these TDEs and their host-subtracted photometric data were obtained from the literature, via the OTTER database6 (Franz et al. 2025) as of December 2025.

The authors note that for AT 2022dbl there is an observation in 2018 labeled as “detection” by OTTER, which could be an image subtraction artifact. p. 9
Where the paper says it

(2025), v: Charalampopoulos et al. (2024) Note—For AT 2022dbl there is an observation in 2018 labeled as “detection” by OTTER, which could be an image subtraction artifact (Makrygianni et al.

p. 9
Where the paper says it

(2025), v: Charalampopoulos et al. (2024) Note—For AT 2022dbl there is an observation in 2018 labeled as “detection” by OTTER, which could be an image subtraction artifact (Makrygianni et al.

p. 9
Where the paper says it

(2025), v: Charalampopoulos et al. (2024) Note—For AT 2022dbl there is an observation in 2018 labeled as “detection” by OTTER, which could be an image subtraction artifact (Makrygianni et al.

References

The paper cites numerous works supporting the modeling and context of TDEs, including papers by Alexander et al. p. 8

Where the paper says it

First, the disk modeling could be improved by extending the one-zone model (e. , Lu 2022; Piro & Mockler 2025) to a more realistic 1D/2D model, and/or including the effects of delayed circularization over a timescale of months to years as suggested in recent numerical studies.

(2020), Piro & Mockler (2025), and various others detailing TDE physics. p. 0
Where the paper says it

2024, 2025). Another possibility is thermal instabilities in TDE disks occurring at late times (Lu 2022; Piro & Mockler 2025; Guo & Qiao 2025, but see also Alush & Stone 2025).

p. 0
Where the paper says it

2024, 2025). Another possibility is thermal instabilities in TDE disks occurring at late times (Lu 2022; Piro & Mockler 2025; Guo & Qiao 2025, but see also Alush & Stone 2025).

The references include foundational works like Hills 1975, Rees 1988, and Shakura & Sunyaev 1973. p. 0
Where the paper says it

In the era of future surveys, combined UV/optical and radio monitoring would be an important test to the disk instability phenomena, as well as its explanation for the late-time radio emission in TDEs. INTRODUCTION A star is tidally disrupted when it reaches too close to a supermassive black hole (BH), such that tidal forces overwhelm the star’s self-gravity (Hills 1975; Rees 1988; Phinney 1989).

p. 0
Where the paper says it

In the era of future surveys, combined UV/optical and radio monitoring would be an important test to the disk instability phenomena, as well as its explanation for the late-time radio emission in TDEs. INTRODUCTION A star is tidally disrupted when it reaches too close to a supermassive black hole (BH), such that tidal forces overwhelm the star’s self-gravity (Hills 1975; Rees 1988; Phinney 1989).

The sample of TDEs analyzed is detailed in Appendix A. p. 8
Where the paper says it

APPENDIX A. TDE SAMPLE We summarize our sample of 28 low-redshift TDEs analyzed in this work (Section 4).

p. 8
Where the paper says it

APPENDIX A. TDE SAMPLE We summarize our sample of 28 low-redshift TDEs analyzed in this work (Section 4).

The data sources are referenced via the OTTER database (Franz et al. p. 6
Where the paper says it

For example, TDEs at distances ≲ 100 Mpc could have flares of ≲ 21 mag in optical, potentially within reach of surveys like ZTF. Motivated by this, we conduct a search of late-time flares in existing TDEs, using light curve data from the OTTER database (Franz et al.

p. 6
Where the paper says it

For example, TDEs at distances ≲ 100 Mpc could have flares of ≲ 21 mag in optical, potentially within reach of surveys like ZTF. Motivated by this, we conduct a search of late-time flares in existing TDEs, using light curve data from the OTTER database (Franz et al.

2025) and various literature sources. p. 8
Where the paper says it

The information for these TDEs and their host-subtracted photometric data were obtained from the literature, via the OTTER database6 (Franz et al. 2025) as of December 2025.

The constraints on the event rate are based on artificial injections into ZTF data. p. 6
Where the paper says it

We believe these candidates are also unlikely to be disk instability flares, as the g-band flux in the vicinity of each candidate is not enhanced like the r-band flux. We finally place rough rate constraints of these flares by artificial injections of our theoretical light curve models into ZTF data.

The paper relies on established cosmological parameters for distance calculations. p. 9
Where the paper says it

Redshift (z), luminosity distance (DL), epoch of last ZTF detection from discovery, and the supermassive BH mass estimates (obtained by host galaxy scaling relation, or light curve modeling if not available), for the 28 TDEs analyzed in this work. DL is calculated assuming a flat ΛCDM cosmology with parameters H0 = 70 km s−1 Mpc−1 , Ωm = 0.

The analysis of existing TDEs is based on OTTER data (Franz et al. p. 6
Where the paper says it

For example, TDEs at distances ≲ 100 Mpc could have flares of ≲ 21 mag in optical, potentially within reach of surveys like ZTF. Motivated by this, we conduct a search of late-time flares in existing TDEs, using light curve data from the OTTER database (Franz et al.

2025) and literature searches. p. 8
Where the paper says it

The information for these TDEs and their host-subtracted photometric data were obtained from the literature, via the OTTER database6 (Franz et al. 2025) as of December 2025.

The search methodology involves comparing flux measurements with surrounding flux measurements. p. 6
Where the paper says it

2019), we search for any short-duration excess that could be consistent with a flare on top of a longer plateau. For each TDE, we extract all host-subtracted ZTF data at > 1 yr from discovery, and compare the flux of each detection with the surrounding flux measurements in the same band at ±(3–50) days.

The conclusion regarding the marginal nature of the constraints is based on the Poisson process assumption. p. 7
Where the paper says it

bands, we set the constraint on the rate of the signal. We assume the flares occur according to a Poisson process, a model-agnostic simplification which could be potentially improved in future works by employing detailed model predictions of the flare pattern.

The paper discusses potential improvements by extending one-zone models and improving greybody assumptions. p. 8
Where the paper says it

We conclude with suggestions for potential improvements to theoretical predictions and observational searches. First, the disk modeling could be improved by extending the one-zone model (e.

The acknowledgments thank several individuals and institutions for their discussions and support. p. 8
Where the paper says it

Finally the search can be improved in several ways as more sensitive surveys are realized, such as combining information of multiple filters and applying more physically informed priors on the flares for significance estimates and rate constraints. ACKNOWLEDGMENTS We thank Kate Alexander and Noah Franz for discussions on OTTER data, Xiaoshan Huang, Brenna Mockler, David Sand and members of Edo Berger’s group for discussions, and Assaf Horesh for comments that motivated this study.

Acknowledgments

The authors thank Kate Alexander and Noah Franz for discussions on OTTER data, Xiaoshan Huang, Brenna Mockler, David Sand and members of Edo Berger’s group for discussions, and Assaf Horesh for comments that motivated this study. D. T. is supported by Harvard University through the Institute for Theory and Computation Fellowship. p. 8

Where the paper says it

ACKNOWLEDGMENTS We thank Kate Alexander and Noah Franz for discussions on OTTER data, Xiaoshan Huang, Brenna Mockler, David Sand and members of Edo Berger’s group for discussions, and Assaf Horesh for comments that motivated this study. is supported by Harvard University through the Institute for Theory and Computation Fellowship.

The Villar Astro Time Lab acknowledges support through the David and Lucile Packard Foundation, the Research Corporation for Scientific Advancement (through a Cottrell Fellowship), the National Science Foundation under AST-2433718, AST-2407922 and AST-2406110, as well as an Aramont Fellowship for Emerging Science Research. p. 8
Where the paper says it

is supported by Harvard University through the Institute for Theory and Computation Fellowship. The Villar Astro Time Lab acknowledges support through the David and Lucile Packard Foundation, the Research Corporation for Scientific Advancement (through a Cottrell Fellowship), the National Science Foundation under AST-2433718, AST-2407922 and AST-2406110, as well as an Aramont Fellowship for Emerging Science Research.

This work is supported by the National Science Foundation under Cooperative Agreement PHY-2019786 (the NSF AI Institute for Artificial Intelligence and Fundamental Interactions). p. 8
Where the paper says it

The Villar Astro Time Lab acknowledges support through the David and Lucile Packard Foundation, the Research Corporation for Scientific Advancement (through a Cottrell Fellowship), the National Science Foundation under AST-2433718, AST-2407922 and AST-2406110, as well as an Aramont Fellowship for Emerging Science Research. This work is supported by the National Science Foundation under Cooperative Agreement PHY-2019786 (the NSF AI Institute for Artificial Intelligence and Fundamental Interactions).

The references section lists all cited works, including foundational physics texts and recent astrophysical literature. The appendix provides detailed information on the TDE sample used in the analysis. p. 8
Where the paper says it

TDE SAMPLE We summarize our sample of 28 low-redshift TDEs analyzed in this work (Section 4). The information for these TDEs and their host-subtracted photometric data were obtained from the literature, via the OTTER database6 (Franz et al.

p. 8
Where the paper says it

TDE SAMPLE We summarize our sample of 28 low-redshift TDEs analyzed in this work (Section 4). The information for these TDEs and their host-subtracted photometric data were obtained from the literature, via the OTTER database6 (Franz et al.

The data presentation relies on OTTER database observations (Franz et al. p. 6
Where the paper says it

For example, TDEs at distances ≲ 100 Mpc could have flares of ≲ 21 mag in optical, potentially within reach of surveys like ZTF. Motivated by this, we conduct a search of late-time flares in existing TDEs, using light curve data from the OTTER database (Franz et al.

p. 6
Where the paper says it

For example, TDEs at distances ≲ 100 Mpc could have flares of ≲ 21 mag in optical, potentially within reach of surveys like ZTF. Motivated by this, we conduct a search of late-time flares in existing TDEs, using light curve data from the OTTER database (Franz et al.

2025) and cosmological calculations. p. 0
Where the paper says it

2021; Mockler et al. 2025, for reviews).

p. 0
Where the paper says it

2021; Mockler et al. 2025, for reviews).

p. 0
Where the paper says it

2021; Mockler et al. 2025, for reviews).

The search methodology is based on comparing observed fluxes with theoretical models of flares. p. 6
Where the paper says it

Our search is hence less sensitive to flares frequently occurring with intervals ≲ 50 days, as the flares can blend into both F and f. However we believe these short intervals are unlikely at ≳ 1 yr after the TDE, based on existing disk models (Lu 2022; Piro & Mockler 2025).

p. 6
Where the paper says it

Our search is hence less sensitive to flares frequently occurring with intervals ≲ 50 days, as the flares can blend into both F and f. However we believe these short intervals are unlikely at ≳ 1 yr after the TDE, based on existing disk models (Lu 2022; Piro & Mockler 2025).

p. 6
Where the paper says it

Our search is hence less sensitive to flares frequently occurring with intervals ≲ 50 days, as the flares can blend into both F and f. However we believe these short intervals are unlikely at ≳ 1 yr after the TDE, based on existing disk models (Lu 2022; Piro & Mockler 2025).

The rate constraints are derived using artificial injections into ZTF data. p. 6
Where the paper says it

We believe these candidates are also unlikely to be disk instability flares, as the g-band flux in the vicinity of each candidate is not enhanced like the r-band flux. We finally place rough rate constraints of these flares by artificial injections of our theoretical light curve models into ZTF data.

p. 6
Where the paper says it

We believe these candidates are also unlikely to be disk instability flares, as the g-band flux in the vicinity of each candidate is not enhanced like the r-band flux. We finally place rough rate constraints of these flares by artificial injections of our theoretical light curve models into ZTF data.

p. 6
Where the paper says it

We believe these candidates are also unlikely to be disk instability flares, as the g-band flux in the vicinity of each candidate is not enhanced like the r-band flux. We finally place rough rate constraints of these flares by artificial injections of our theoretical light curve models into ZTF data.

The conclusion suggests that future work should focus on more realistic 1D/2D models and improved greybody assumptions. p. 7
Where the paper says it

We expect the search to be more sensitive in the future, with longer baselines and higher sensitivities to be realized for more TDEs. CONCLUSION We have studied the transient UV and optical emission from accretion disks formed in TDEs, under the picture suggested from recent models that they undergo latetime thermal instabilities.

p. 7
Where the paper says it

We expect the search to be more sensitive in the future, with longer baselines and higher sensitivities to be realized for more TDEs. CONCLUSION We have studied the transient UV and optical emission from accretion disks formed in TDEs, under the picture suggested from recent models that they undergo latetime thermal instabilities.

p. 7
Where the paper says it

We expect the search to be more sensitive in the future, with longer baselines and higher sensitivities to be realized for more TDEs. CONCLUSION We have studied the transient UV and optical emission from accretion disks formed in TDEs, under the picture suggested from recent models that they undergo latetime thermal instabilities.

The paper is structured with sections covering modeling, parameter study, search, and conclusion. p. 1
Where the paper says it

We present a parameter study of this in Section 3, and discuss the prospects of UV/optical surveys to test the existence of disk instabilities using these transients. In Section 4 we present our search for these transients in archival TDEs, using existing high-cadence optical data.

p. 1
Where the paper says it

We present a parameter study of this in Section 3, and discuss the prospects of UV/optical surveys to test the existence of disk instabilities using these transients. In Section 4 we present our search for these transients in archival TDEs, using existing high-cadence optical data.

p. 1
Where the paper says it

We present a parameter study of this in Section 3, and discuss the prospects of UV/optical surveys to test the existence of disk instabilities using these transients. In Section 4 we present our search for these transients in archival TDEs, using existing high-cadence optical data.

p. 1
Where the paper says it

We present a parameter study of this in Section 3, and discuss the prospects of UV/optical surveys to test the existence of disk instabilities using these transients. In Section 4 we present our search for these transients in archival TDEs, using existing high-cadence optical data.

The final summary highlights the expected UV peak magnitudes for ULTRASAT detection. p. 5
Where the paper says it

For our fiducial choice of rd = 5rT (dashed lines) and MBH = 106 M⊙, we find the near-UV emission to peak at magnitudes between −17 and −19 mag, with durations of 3–4 days. ULTRASAT, with a singlevisit sensitivity of 22.

p. 5
Where the paper says it

For our fiducial choice of rd = 5rT (dashed lines) and MBH = 106 M⊙, we find the near-UV emission to peak at magnitudes between −17 and −19 mag, with durations of 3–4 days. ULTRASAT, with a singlevisit sensitivity of 22.

p. 5
Where the paper says it

For our fiducial choice of rd = 5rT (dashed lines) and MBH = 106 M⊙, we find the near-UV emission to peak at magnitudes between −17 and −19 mag, with durations of 3–4 days. ULTRASAT, with a singlevisit sensitivity of 22.

p. 5
Where the paper says it

For our fiducial choice of rd = 5rT (dashed lines) and MBH = 106 M⊙, we find the near-UV emission to peak at magnitudes between −17 and −19 mag, with durations of 3–4 days. ULTRASAT, with a singlevisit sensitivity of 22.

The paper's arXiv identifier is 2602.15103 [astro-ph. p. 0
Where the paper says it

arXiv:2602. 15103v1 [astro-ph.

p. 0
Where the paper says it

arXiv:2602. 15103v1 [astro-ph.

HE] 16 Feb 2026. p. 0
Where the paper says it

15103v1 [astro-ph. HE] 16 Feb 2026

p. 0
Where the paper says it

15103v1 [astro-ph. HE] 16 Feb 2026

The references section lists all cited works, including foundational physics texts and recent astrophysical literature. The paper cites numerous works supporting the modeling and context of TDEs, including papers by Alexander et al. (2020), Piro & Mockler (2025), and various others detailing TDE physics. The references include foundational works like Hills 1975, Rees 1988, and Shakura & Sunyaev 1973. The sample of TDEs analyzed is detailed in Appendix A. The data sources are referenced via the OTTER database (Franz et al. 2025) and cosmological calculations. The search methodology is based on comparing observed fluxes with theoretical models of flares. The rate constraints are derived using artificial injections into ZTF data. The conclusion suggests that future work should focus on more realistic 1D/2D models and improved greybody assumptions. The paper is structured with sections covering modeling, parameter study, search, and conclusion. The final summary highlights the expected UV peak magnitudes for ULTRASAT detection. The paper's arXiv identifier is 2602.15103 [astro-ph.HE] 16 Feb 2026.

Table 1

Table 1 summarizes the redshift (z), luminosity distance (DL), epoch of last ZTF detection from discovery, and the supermassive BH mass estimates for the 28 TDEs analyzed in this work. DL is calculated assuming a flat ΛCDM cosmology with parameters H0 = 70 km s−1 Mpc−1,omegam = 0. p. 9

Where the paper says it

Redshift (z), luminosity distance (DL), epoch of last ZTF detection from discovery, and the supermassive BH mass estimates (obtained by host galaxy scaling relation, or light curve modeling if not available), for the 28 TDEs analyzed in this work. DL is calculated assuming a flat ΛCDM cosmology with parameters H0 = 70 km s−1 Mpc−1 , Ωm = 0.

3 and omegaΛ = 0.7. The authors note that for AT 2022dbl there is an observation in 2018 labeled as “detection” by OTTER, which could be an image subtraction artifact.

Figure 7

Figure 7 shows the host-corrected ZTF light curves of 28 TDEs selected from OTTER. Dashed lines show 1 year from discovery, and a search is conducted for flares beyond this epoch. p. 6

Where the paper says it

1, (ii) host correction applied, and (iii) with at least one ZTF detection (flagged by > 3σ confidence in OTTER) at > 1 year from discovery. We show the redshift and the epoch of the last ZTF detection (from discovery) in Figure 4 (for details of the events see Appendix A).

Grey shaded regions show the flare magnitude for an expected absolute magnitude of (-14) – (-17) mag. p. 8
Where the paper says it

The near-UV magnitude peaks between −17 and −19 mag, which is a promising target for ULTRASAT for TDEs out to redshift ≈ 0. The optical emission is expected to be an order of magnitude fainter, but may still be well constraining in the era of future high-cadence surveys like LSST and Argus Array.

The figure illustrates the comparison between observed data and expected flare magnitudes. p. 4
Where the paper says it

In the bottom-right panel of Figure 1 we show predictions for the AB magnitude in the near-UV (250 nm) and optical (600 nm), assuming the transient has greybody spectra with temperature Tobs and luminosity Lobs. Due to the high temperature of ∼ 105 K at early times, the AB magnitudes peak closer to the end of the flare.

Figure 6

Figure 6 shows the 90% confidence upper limits on the event rate as a function of the flare mass Mfl, for the three TDEs, AT 2019qiz, AT 2019azh, and AT 2021ehb. The constraints found are of ≲ 1–2 yr−1 for each of the three TDEs. p. 6

Where the paper says it

Redshift and epoch of last ZTF detection (relative to discovery), for the 28 TDE samples analyzed in this work. Stars show three TDEs used for constraints on the flare rate, AT 2019qiz, AT 2019azh, and AT 2021ehb (see main text).

Figure 5

Figure 5 shows the two highest-significance candidates found in our analysis, from AT2019azh and AT2021nwa. Left panels show the full light curve, while the right panels show that at ±100 days around the candidate. p. 7

Where the paper says it

The two highest-significance candidates found in our analysis, from AT2019azh and AT2021nwa. Left panels show the full light curve, while the right panels show that at ±100 days around the candidate.

Figure 4

Figure 4 shows redshift and epoch of last ZTF detection (relative to discovery), for the 28 TDE samples analyzed in this work. Stars show three TDEs used for constraints on the flare rate, AT 2019qiz, AT 2019azh, and AT 2021ehb. p. 6

Where the paper says it

Redshift and epoch of last ZTF detection (relative to discovery), for the 28 TDE samples analyzed in this work. Stars show three TDEs used for constraints on the flare rate, AT 2019qiz, AT 2019azh, and AT 2021ehb (see main text).

The figure illustrates the temporal evolution of detection epochs. p. 5
Where the paper says it

Such emission would be good targets in the Rubin era, for sources also out to z ≈ 0. A confident detection of the flare may require detection over multiple epochs, which is generally challenging for the predicted short durations of the flares.

Figure 3

Figure 3 shows the same parameter grid as Figure 2 but for the optical at 600 nm. For a typical SMBH mass of MBH = 106 M⊙ and rd ≈ 5rT, the emission in the optical is ≈ (−14)–(−17) mag (1041–1042 erg s−1) and lasts for 3–5 days. p. 5

Where the paper says it

Figure 3 shows the same parameter grid as Figure 2 but for the optical at 600 nm. For a typical SMBH mass of MBH = 106 M⊙ and rd ≈ 5rT , the emission in the optical is ≈ (−14)–(−17) mag (1041–1042 erg s−1 ) and lasts for 3–5 days.

Figure 2

Figure 2 shows the peak magnitude and duration within 1 mag from peak in the near-UV (250 nm). For our fiducial choice of rd = 5rT (dashed lines) and MBH = 106 M⊙, we find the near-UV emission to peak at magnitudes between −17 and −19 mag, with durations of 3–4 days. p. 5

Where the paper says it

Figure 2 shows the range of Mpk and ∆tpk at nearUV (250 nm) for the parameter space of rd and Mfl, with MBH of 106 M⊙ and 107 M⊙ in the top and bottom panels. For our fiducial choice of rd = 5rT (dashed lines) and MBH = 106 M⊙, we find the near-UV emission to peak at magnitudes between −17 and −19 mag, with durations of 3–4 days.

Figure 7 (Continued)

The top panels show the bolometric luminosity and emission temperature. The bottom right panel shows the AB magnitudes in the near-UV (250 nm) and optical (600 nm) under a greybody approximation. p. 3

Where the paper says it

02 M⊙, and tfl = 2 days. Top panels show the bolometric luminosity and emission temperature, and the bottom right panel shows the AB magnitudes in the near-UV and optical under a greybody approximation.

The bottom left panel shows the evolution of various radii defined in Section 2. p. 3
Where the paper says it

Top panels show the bolometric luminosity and emission temperature, and the bottom right panel shows the AB magnitudes in the near-UV and optical under a greybody approximation. Bottom left panel shows the evolution of various radii defined in Section 2.

2. The temperature predictions are shown in the top-right panel of Figure 7. p. 3
Where the paper says it

We adopt the temperature at rtr of Tobs ≈ Lobs 4πr2 trvwa 1/4 (rc < rtr), (25) and update rc = rtr. The temperature predictions solved this way is shown in the top-right panel of Figure 1.

Figure 1

Figure 1 shows a light curve for our fiducial parameter set of MBH = 106M⊙, Mfl = 0.02 M⊙, rd = 300Rg, and tfl = 2 days. p. 2

Where the paper says it

Using the solution for Lrad(t ′ ) and substituting t ′ = t − t˜ for each wind shell launched at t˜, we sum up the contributions from all shells emitted within t and obtain Lobs(t) =Z t 0 dt˜ Lacc(t˜) 2At˜ B + t − t˜ t˜ 2 rin rin + vw(t − t˜) 2/3 ×exp ( − B3 3A " 1 + 1 B t − t˜ t˜ 3 − 1 #) , (14) The top-left panel of Figure 1 shows a light curve for our fiducial parameter set of MBH = 106M⊙, Mfl = 0. 02 M⊙, rd = 300Rg, and tfl = 2 days.

After a brief decline during t≪ rin/vw ∼ 0. p. 2
Where the paper says it

02 M⊙, rd = 300Rg, and tfl = 2 days. After a brief decline during t ≪ rin/vw ∼ 0.

3 days as the diffusion time evolves (Piro & Lu 2020), the light curve settles to a plateau-like morphology. p. 2
Where the paper says it

After a brief decline during t ≪ rin/vw ∼ 0. 3 days as the diffusion time evolves (Piro & Lu 2020), the light curve settles to a plateau-like morphology.

Once the outflow shuts off at t = tfl, the light curve drops sharply over a timescale of a few days, which is set by the diffusion timescale in the wind. p. 2
Where the paper says it

The observed luminosity at this phase is a factor of several lower than the input luminosity Lacc due to the adiabatic loss. Once the outflow shuts off at t = tfl, the light curve drops sharply over a timescale of a few days, which is set by the diffusion timescale in the wind.

The observed luminosity at this phase is a factor of several lower than the input luminosity Lacc due to the adiabatic loss. p. 2
Where the paper says it

The plateau is partly due to our assumed top-hat time-dependence of M˙ and Lacc. The observed luminosity at this phase is a factor of several lower than the input luminosity Lacc due to the adiabatic loss.

Section 5

The authors conclude with suggestions for potential improvements to theoretical predictions and observational searches. First, the disk modeling could be improved by extending the one-zone model (e. p. 8

Where the paper says it

We conclude with suggestions for potential improvements to theoretical predictions and observational searches. First, the disk modeling could be improved by extending the one-zone model (e.

g., Lu 2022; Piro & Mockler 2025) to a more realistic 1D/2D model, and/or including the effects of delayed circularization over a timescale of months to years as suggested in recent numerical studies. p. 8
Where the paper says it

First, the disk modeling could be improved by extending the one-zone model (e. , Lu 2022; Piro & Mockler 2025) to a more realistic 1D/2D model, and/or including the effects of delayed circularization over a timescale of months to years as suggested in recent numerical studies.

Second, the greybody assumption for the spectra could be improved by detailed radiative transfer modeling of the outflow. p. 8
Where the paper says it

This could provide more robust predictions for both the timing and the mass of the flares. Second, the greybody assumption for the spectra could be improved by detailed radiative transfer modeling of the outflow, which could impact more on the predictions for optical that are far from the spectral peak.

Finally, the search can be improved in several ways as more sensitive surveys are realized, such as combining information of multiple filters and applying more physically informed priors on the flares for significance estimates and rate constraints. p. 8
Where the paper says it

Second, the greybody assumption for the spectra could be improved by detailed radiative transfer modeling of the outflow, which could impact more on the predictions for optical that are far from the spectral peak. Finally the search can be improved in several ways as more sensitive surveys are realized, such as combining information of multiple filters and applying more physically informed priors on the flares for significance estimates and rate constraints.

The paper is structured with sections covering modeling, parameter study, search, and conclusion. The final summary highlights the expected UV peak magnitudes for ULTRASAT detection.

Appendix A (Continued)

The authors summarize their sample of 28 low-redshift TDEs analyzed in this work. Table 1 shows the redshift, last ZTF detection from discovery, published mass estimates of the supermassive BH, and references for the photometric data and the quoted BH mass. p. 8

Where the paper says it

2025) as of December 2025. Table 1 shows the redshift, last ZTF detection from discovery, published mass estimates of the supermassive BH, and references for the photometric data and the quoted BH mass.

The authors note that for AT 2022dbl there is an observation in 2018 labeled as “detection” by OTTER, which could be an image subtraction artifact. The references section lists all cited works, including foundational physics texts and recent astrophysical literature. The appendix provides detailed information on the TDE sample used in the analysis. The data presentation relies on OTTER database observations (Franz et al. 2025) and cosmological calculations. The search methodology is based on comparing observed fluxes with theoretical models of flares. The rate constraints are derived using artificial injections into ZTF data. The conclusion suggests that future work should focus on more realistic 1D/2D models and improved greybody assumptions. The paper is structured with sections covering modeling, parameter study, search, and conclusion. The final summary highlights the expected UV peak magnitudes for ULTRASAT detection.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved transient detection capabilities: The AI system can be trained to search for UV bright transients lasting for days with specific peak luminosities of 1042–1043 erg s-1 in near-UV and 1041–1042 erg s-1 in optical, allowing it to detect flares from TDEs out to redshift z ≈ 0.1 using near-future surveys like ULTRASAT or Vera C. Rubin Observatory's LSST.

  2. Enhanced rate constraint modeling: The system can perform artificial injections of flare models into ZTF data to place upper limits on the flare event rates, constraining them to 1–2 yr-1 for each of the three most constraining TDEs based on optical data at late times.

  3. Multi-wavelength correlation analysis: The AI can correlate UV/optical light curves with radio monitoring data to test the disk instability scenario, as radio measurements can infer the radius evolution and hence the launch time of the super-Eddington outflow.

Abstract

Tidal disruption events (TDEs) are unique probes of evolving accretion in supermassive black holes. Recent models of TDE disks show that they undergo brief thermal instabilities with temporal super-Eddington accretion at late times, which has been suggested as a possibility to explain the ubiquitous late radio emergence in TDEs. We model the ultraviolet (UV) and optical signatures of such disk instabilities, expected from the accretion power being reprocessed by the optically-thick outflow following super-Eddington accretion. Our model predicts brief UV-bright transients lasting for days, with luminosities of 10 42 - 10 43 erg s-1 in near-UV and 10 41 - 10 42 erg s-1 in optical for a typical TDE by a 10 6 M black hole. These could be detectable by near-future surveys such as ULTRASAT, Vera C. Rubin Observatory and Argus Array, for TDEs of redshifts out to about 0.1. We further conduct a search for these transients in existing nearby TDEs using data from the Zwicky Transient Facility, placing upper limits on the flare rate for each TDE of 1 - 2 yr-1 dependent on the outflow mass. In the era of future surveys, combined UV/optical and radio monitoring would be an important test to the disk instability phenomena, as well as its explanation for the late-time radio emission in TDEs.

Sources

Related papers