Fast Blue Optical Transients¶
This page summarizes FBOT detection simulations for both ZTF and Rubin LSST, using the Bazin parametric model with timescales drawn from the Ho et al. (2021) ZTF Phase I sample.
Summary of results¶
| Survey | Duration | z_max | Injections | Efficiency | Expected | Observed |
|---|---|---|---|---|---|---|
| ZTF (Ho+2021 validation) | 2.67 yr | 0.3 | 100K | 4.2% | 38.1 | 38 |
| Rubin (photometric) | 10 yr | 0.5 | 100K | 12.4% | 3,600 (360/yr) | — |
The ZTF simulation reproduces the Ho et al. (2021) sample to < 1% accuracy. Rubin is predicted to detect ~360 FBOTs per year, a ~25× increase over ZTF.
Background¶
Fast Blue Optical Transients (FBOTs) are a class of rapidly evolving transients with rise times of ~1–5 days and half-light durations of 1–12 days. They are blue at peak (g−r < −0.2 mag) and span a wide luminosity range (\(M_g\) = −16 to −21). The most luminous and exotic subclass — exemplified by AT2018cow ("the Cow") — shows featureless spectra, luminous radio/X-ray emission, and may be powered by central engine activity (accreting black hole or magnetar).
Ho et al. (2021, arXiv:2105.08811) conducted a systematic search of ZTF Phase I data (March 2018 – October 2020), identifying 38 FBOTs with well-sampled lightcurves.
Lightcurve model¶
We use the Bazin model (Bazin et al. 2009), a simple analytical rise/fall profile commonly used for fast transients:
This is a built-in Rust model evaluated via ParametricModel() — no Python callback or GPU required.
Timescale parameters¶
Drawn from distributions calibrated to the Ho et al. (2021) Table 10 sample (g-band):
| Parameter | Distribution | Ho+2021 range |
|---|---|---|
| \(t_{1/2,\mathrm{rise}}\) | \(\mathcal{N}(2.5, 1.2)\), clamp [0.3, 6.0] days | 0.5–4.6 days |
| \(t_{1/2,\mathrm{fade}}\) | \(\mathcal{N}(5.5, 2.0)\), clamp [1.0, 15.0] days | 1.5–8.0 days |
The Bazin e-folding times are related to observed half-light times by \(\tau \approx t_{1/2} / \ln 2\).
Peak magnitudes¶
From Ho+2021 Table 10, the 38 FBOTs span \(M_g\) = −16.4 to −21.2 with a median of approximately −18.7. We draw peak absolute magnitudes from \(\mathcal{N}(-18.7, 1.5)\), clamped to [−22.7, −14.7].
Selected events from Table 10:
| Event | \(M_g\) | \(t_\mathrm{rise}\) (d) | \(t_\mathrm{fade}\) (d) | Type |
|---|---|---|---|---|
| AT2018lug | −21.17 | 1.12 | 2.92 | Exotic (cow-like) |
| AT2020xnd | −21.03 | 1.6–4.8 | 2.39 | Exotic (cow-like) |
| SN2018gep | −19.84 | 3.27 | 6.00 | SN Ic-BL |
| SN2019deh | −19.73 | 4.35 | 6.33 | SN II |
| SN2018bcc | −19.82 | 3.20 | 5.87 | SN Ib |
| SN2020rsc | −16.37 | 1.62 | 1.72 | SN Ibn |
Volumetric rate¶
Ho et al. (2021) find that FBOTs represent approximately 0.1% of the local core-collapse supernova rate. With a CC SN rate of ~70,000 Gpc⁻³ yr⁻¹ (Perley et al. 2020), this gives:
This rate encompasses all FBOTs with \(1 < t_{1/2} < 12\) days and blue colors (g−r < −0.2), not just the exotic AT2018cow-like subset.
ZTF validation (Ho et al. 2021)¶
Setup¶
Reproduces the FBOT sample from Ho et al. (2021): 38 FBOTs from ZTF Phase I (March 2018 – October 2020).
from survey_sim import (
FbotPopulation, ParametricModel, DetectionCriteria,
SimulationPipeline, load_ztf_survey,
)
survey = load_ztf_survey(start="201803", end="202010", nside=64)
pop = FbotPopulation(rate=65.0, z_max=0.3, peak_abs_mag=-18.7)
model = ParametricModel() # built-in Rust Bazin model
Detection criteria¶
Ho+2021 applied strict selection criteria: fast-rising (\(\geq 1\) mag in 6.5 days), well-sampled near peak in both g and r, blue color (g−r < −0.2), and spectroscopic classification. We approximate these with:
| Criterion | Value |
|---|---|
| SNR threshold | \(\geq 5\sigma\) |
| Min detections | \(\geq 5\) |
| Min bands | \(\geq 2\) (g + r) |
| Min per band | \(\geq 2\) |
| Max timespan | 24 days |
| Min time separation | \(\geq 24\) hours |
| Fast transient | Required |
| Min rise rate | 0.15 mag/day (\(\geq 1\) mag in 6.5d) |
| Min fade rate | 0.1 mag/day |
| Pre-peak detections | \(\geq 1\) |
| Post-peak detections | \(\geq 1\) |
| Phase range | \(\geq 3\) days |
| Galactic latitude | \(\|b\| > 7°\) |
| Host brightness cut | Logistic: \(k = 3.0\), \(m_0 = 19.0\) |
det = DetectionCriteria(
snr_threshold=5.0,
snr_threshold_secondary=5.0,
min_detections=5,
min_detections_primary=5,
min_bands=2,
min_per_band=2,
max_timespan_days=24.0,
min_time_separation_hours=24.0,
require_fast_transient=True,
min_rise_rate=0.15,
min_fade_rate=0.1,
min_pre_peak_detections=1,
min_post_peak_detections=1,
min_phase_range_days=3.0,
min_galactic_lat=7.0,
spectroscopic_completeness_k=3.0,
spectroscopic_completeness_m0=19.0,
)
Results (100K injections)¶
| Quantity | Value |
|---|---|
| Comoving volume (\(z < 0.3\)) | 11.0 Gpc³ |
| ZTF sky fraction | 47% |
| Survey duration | 2.67 yr |
| Volumetric rate | 65 Gpc⁻³ yr⁻¹ |
| Total FBOTs in volume | ~900 |
| Detection efficiency | 4.2% |
| Expected detections | 38.1 |
| Ho+2021 actual | 38 |
| Agreement | < 1% |
Rubin LSST prediction¶
Setup¶
Predicts the number of FBOTs that Rubin will detect over 10 years. Uses the same population parameters as ZTF.
from survey_sim import (
SurveyStore, FbotPopulation, ParametricModel,
DetectionCriteria, SimulationPipeline,
)
survey = SurveyStore.from_rubin("baseline_v5.1.1_10yrs.db", nside=64)
pop = FbotPopulation(rate=65.0, z_max=0.5, peak_abs_mag=-18.7)
model = ParametricModel()
Detection criteria¶
Relaxed relative to ZTF — Rubin will identify fast transients photometrically from multi-band difference imaging without requiring spectroscopic classification. A host brightness cut at \(m_0 = 22.5\) (2 mag brighter than the 5σ limit) models the requirement that the transient must outshine its host.
| Criterion | Value |
|---|---|
| SNR threshold | \(\geq 5\sigma\) |
| Min detections | \(\geq 3\) |
| Min bands | \(\geq 2\) |
| Min per band | \(\geq 2\) |
| Max timespan | 24 days |
| Min time separation | \(\geq 24\) hours |
| Fast transient | Required |
| Min rise rate | 0.15 mag/day |
| Min fade rate | 0.1 mag/day |
| Galactic latitude | \(\|b\| > 15°\) |
| Host brightness cut | Logistic: \(k = 5.0\), \(m_0 = 22.5\) |
det = DetectionCriteria(
snr_threshold=5.0,
snr_threshold_secondary=5.0,
min_detections=3,
min_detections_primary=3,
min_bands=2,
min_per_band=2,
max_timespan_days=24.0,
min_time_separation_hours=24.0,
require_fast_transient=True,
min_rise_rate=0.15,
min_fade_rate=0.1,
min_galactic_lat=15.0,
spectroscopic_completeness_k=5.0,
spectroscopic_completeness_m0=22.5,
)
Results (100K injections)¶
| Quantity | Value |
|---|---|
| Comoving volume (\(z < 0.5\)) | 102 Gpc³ |
| Rubin sky fraction | 44% |
| Survey duration | 10 yr |
| Volumetric rate | 65 Gpc⁻³ yr⁻¹ |
| Total FBOTs in volume | ~29,200 |
| Detection efficiency | 12.4% |
| Expected detections | ~3,600 in 10yr |
| Expected per year | ~360 |
Why does Rubin find so many more?¶
Rubin detects ~25× more FBOTs per year than ZTF (360 vs 14). The improvement comes from:
- Depth: Rubin's 5σ limit (~24.5) is ~4 mag deeper than ZTF (~20.5). With a host brightness cut at 22.5 vs 19.0, Rubin can identify FBOTs 3.5 mag fainter.
- Volume: \(z_\mathrm{max} = 0.5\) vs 0.3 → ~9× larger effective volume.
- Higher efficiency: 12.4% vs 4.2%, because Rubin's multi-band coverage and deeper imaging make it easier to satisfy the fast-transient detection criteria.
Cadence challenge
FBOTs evolve on timescales of days, but Rubin's WFD revisits each field every ~3 days. Some FBOTs may peak and fade between visits. The 12.4% efficiency already accounts for this cadence limitation — events that fall entirely between visits are naturally rejected by the detection criteria.
FBOT subclasses¶
Ho et al. (2021) classified their 38 FBOTs spectroscopically:
| Classification | Count | Fraction |
|---|---|---|
| Type II/IIb/Ib (H- or He-rich) | 11 | 29% |
| Type IIn/Ibn (interacting) | 6 | 16% |
| Type Ic/Ic-BL (stripped) | 2 | 5% |
| Exotic (AT2018cow-like) | 2 | 5% |
| Unclassified / no spectrum | 17 | 45% |
The exotic AT2018cow-like events (AT2018lug, AT2020xnd) are the rarest and most luminous (\(M_g < -21\)), with featureless spectra and luminous radio/X-ray emission. They represent a small fraction of the broader FBOT population, which is dominated by rapidly evolving core-collapse supernovae.
Scripts¶
| Script | Description |
|---|---|
python/scripts/run_fbot_ztf.py |
ZTF Ho+2021 validation (100K injections) |
python/scripts/run_fbot_rubin.py |
Rubin 10-year FBOT prediction (100K injections) |
References¶
- Ho et al. (2021), arXiv:2105.08811 — ZTF FBOT sample (38 events, rate measurement)
- Bazin et al. (2009), A&A 499 653 — Bazin lightcurve model
- Perley et al. (2020), ApJ 904 35 — ZTF Bright Transient Survey, CC SN rate