iPTF Survey for Cool Transients

Transient astronomy is benefiting enormously from the deployment of increasingly capable wide-field cameras in time-domain surveys such as ASAS-SN (Shappee et al. 2014), ATLAS (Tonry 2011), MASTER (Gorbovskoy et al. 2013), the Palomar Transient Factory (PTF; Law et al. 2009), and Pan-STARRS (Chambers et al. 2016). However, with the upcoming Zwicky Transient Facility (ZTF; Bellm 2014) and, later, the Large Synoptic Survey Telescope (LSST Science Collaboration et al. 2009), the torrent of transient discoveries will vastly exceed spectroscopic follow-up capabilities. While some interesting transients can be prioritized for spectroscopic follow-up on the basis of rapidly evolving light curves from high-cadence experiments, for many science cases it will only be practical to select targets based on photometric colors.

Despite the impending need for the next generation of transient surveys to utilize multiple filters concurrently, such an approach has generally been limited to transient surveys with relatively small footprints (e.g., the Pan-STARRS Medium Deep Survey—Chambers et al. 2016; the DES SN survey—Kessler et al. 2015; SNLS—Howell et al. 2005; Astier et al. 2006. SDSS II Supernova Survey—Frieman et al. 2008). Recently, Miller et al. (2017) described the first quasi-contemporaneous multi-band (g and R) survey to be undertaken with the Intermediate Palomar Transient Factory (iPTF; Kulkarni 2013).

Extragalactic transient surveys have generally avoided I band because quickly evolving transients are hot (at least initially) and the sky background is higher in I band than in bluer optical filters. Still, adding I band imaging as an additional filter to a transient survey may provide a number of benefits.

An I band filter can increase the sensitivity to and improve the selection of cool or dusty events such as luminous red novae (LRNe), intermediate luminosity red transients (ILRTs), and obscured SNe. This is illustrated by Figure 1, which shows the color evolution of representatives of various classes of transients. While most SNe have modest ($g-I\lesssim 0.5$ mag) colors at early times, a significant fraction (∼19%; Mattila et al. 2012) of local SNe are so heavily reddened that they are missed by standard optical surveys.

Zoom In Zoom Out Reset image size

LRNe and ILRTs are loosely defined observational classes of transients with peak magnitudes between novae and SNe Ia ($-10\lesssim M\lesssim -16$), red optical colors, and low inferred ejecta velocities ($\lesssim 1000\,\mathrm{km}\,{{\rm{s}}}^{-1}$). ILRTs (also referred to as SN 2008S-like transients) arise from obscured massive stars (Thompson et al. 2009). It has been suggests that ILRTs may be the result of electron-capture SNe (Botticella et al. 2009), mass transfer episodes (Kashi et al. 2010), or some other non-terminal outburst (Berger et al. 2009; Smith et al. 2009). The most recent evidence from the late-time evolution of the best-studied events suggests that they are supernovae (Adams et al. 2016). LRNe leave infrared-bright remnants and are of particular interest because they have been shown to be associated with stellar mergers or the onset of common envelope events (Tylenda et al. 2011; Ivanova et al. 2013). Although they are common (Kochanek et al. 2014), their low luminosities and red colors ($g-I\gtrsim 1$ mag) have limited the number of identified candidates to a handful of events in the Milky Way (Martini et al. 1999; Soker & Tylenda 2003; Tylenda et al. 2011, 2013) and nearby ($\lesssim 7$ Mpc) galaxies (Rich et al. 1989; Kulkarni et al. 2007; Williams et al. 2015; Smith et al. 2016; Blagorodnova et al. 2017a). More events are needed to characterize the range of LRN and ILRT properties.

Furthermore, the addition of I band survey data is important for SN Ia cosmology. SN Ia distances are best determined with at least two colors (Riess et al. 1996; Jha et al. 2007). Employing I band for one of these colors has the added benefit that the distinctive 2nd peak seen at I band and redder filters (see Figure 1) for a majority of (low redshift) SNe Ia facilitates photometric typing (Poznanski et al. 2002).

In this paper we describe a 2-month long experiment undertaken with the Palomar 48'' Telescope observing a wide area at low cadence in g and I bands. We present the 36 transients discovered, their spectroscopic classifications, rate estimates, and a description of three interesting transients discovered: a heavily reddened SN Ia, a transitional SN Ibn/IIn, and an LBV outburst. All magnitudes in this paper are in the AB system.

This experiment was undertaken as part of the extension to iPTF, which succeeded PTF (Law et al. 2009; Rau et al. 2009). iPTF utilizes the Palomar 48'' Telescope (P48) with the CFH12K camera (Rahmer et al. 2008; Law et al. 2010) comprised of 11 working CCDs, each with 2048 × 4096 pixels, $1\buildrel{\prime\prime}\over{.} 01$ pixel⁻¹, for a 7.3 deg² field of view. The experiment was allocated 38 nights (of which observing conditions permitted data collection on 17 nights) during dark and gray time between 2017 January 7 and February 25, with the remainder of the nights used to complete the iPTF Census of the Local Universe Hα survey (Cook et al. 2017) and follow up target of opportunity triggers.

The experiment monitored ∼2000 square degrees with fields chosen to avoid the Galactic and ecliptic planes and virtually no bias toward nearby galaxies. When observed, fields were imaged twice within the same night, once in g band and once in I band, with $\simeq 1\,\mathrm{hr}$ between the pair of observations. This approach was chosen to provide g − I color information while rejecting asteroids (by requiring two co-spatial detections) and minimizing filter changes. The observation pairs were repeated with a cadence varying between one night and two weeks (depending on weather and the aforementioned scheduling considerations; see Figure 2). The schedule was manually adjusted each night to maintain similar cadence for all fields targeted by the experiment (fields with longer elapsed time since a previous observation were given the higher priority). The particular set of fields scheduled for a given night was chosen to minimize slew and airmass, but also anticipated the schedule for the following nights. The effective areal coverage of the survey is shown in Figure 3.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The survey data were processed independently with the IPAC (Masci et al. 2017) and NERSC realtime image subtraction (Cao et al. 2016) pipelines. Machine learning software, known as Real-Bogus, was used to separate real transients from image subtraction artifacts (Bloom et al. 2012; Brink et al. 2013; Rebbapragada et al. 2015). Sources with two detections within 24 hr (essentially requiring detections in both g and I bands¹⁵ ) predicted by Real-Bogus to likely be real transients were then examined by human scanners. Sources with positions consistent with a known AGN/QSO or stellar source (based on SDSS classification and past history of variability within the PTF data set) were discarded. The median blank-sky 5σ point source limiting magnitudes are 20.5 and 20.1 mag in g and I bands respectively (Figure 4). However, the cumulative distributions of the discovery magnitudes of the transients (Figure 5) shows that the survey completeness to transient sources begins to degrade significantly for magnitudes fainter than ∼19.5. The volume probed by a magnitude-limited survey is proportional to ${10}^{0.6m}$, where m is the limiting magnitude. Thus for a non-cosmological extragalactic survey (where the target density is independent of distance and K corrections are negligible) the cumulative number of detected sources should be proportional to ${10}^{0.6m}$ and a cumulative distribution starting to fall below $\propto {10}^{0.6m}$ indicates diminishing completeness.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We spectroscopically classified all transients with $| g-I| \gt 0.5$ (at peak brightness),¹⁶ that occurred in a known nearby (<200 Mpc) galaxy, or were brighter than 19.5 mag in either filter. Spectroscopic observations were performed with the Spectral Energy Distribution Machine (SEDM; Blagorodnova et al. 2017b) on the Palomar 60 inch robotic telescope (P60), the Double Beam Spectrograph (DBSP; Oke & Gunn 1982) on the 200 inch Hale telescope at Palomar Observatory (P200), the Low Resolution Imaging Spectrograph (LRIS; Oke et al. 1995) on the 10 m Keck I telescope, DEep Imaging Multi-Object Spectrograph (DEIMOS) (Faber et al. 2003) on the Keck II telescope, the DeVeny Spectrograph (Bida et al. 2014) on the Discovery Channel Telescope (DCT), the Double Imaging Spectrometer on the Astrophysics Research Consortium 3.5 m Telescope, the Andalucia Faint Object Spectrograph and Camera (ALFOSC) on the Nordic Optical Telescope (NOT), the Gemini Multi-Object Spectrograph on the Gemini North Telescope, and the Device Optimized for the LOw RESolution (DOLORES) on the Telescopio Nazionale Galileo (TNG). Spectral classification was done with SNID (Blondin & Tonry 2007) and Superfit v.3.5 (Howell et al. 2005).

For select transients we supplemented the survey light curves from the P48 with gri photometry from the Rainbow camera (RC) on the SEDM and the GRB Cam (Cenko et al. 2006) on the P60. The RC photometry was extracted using the automatic reference-subtraction photometry pipeline, FPipe (Fremling et al. 2016), which utilizes reference images of the host galaxy from the Sloan Digital Sky Survey to subtract the background light. Additional photometry was obtained for iPTF17lf using the Reionization And Transients Infra-Red (RATIR; Fox et al. 2012), camera mounted on the robotic 1.5 m Harold Johnson telescope of the Observatorio Astronómico Nacional. For the RATIR data, the SN and the host galaxy were fitted simultaneously by modeling the two using the point-spread function of the image and a Sérsic profile (Sérsic 1963). We obtained two epochs of BVgri images for iPTF17bkj from Las Cumbres Observatory's 1 m telescope in Texas (Brown et al. 2013). Using lcogtsnpipe (Valenti et al. 2016), we measured PSF photometry and calibrated it to the APASS 9 (Henden et al. 2016) and SDSS 13 (SDSS Collaboration et al. 2016) catalogs for BV and gri, respectively. Supplemental photometry for iPTF17be was obtained with LRIS and with the NIRC2 near-IR camera on the Keck II telescope.

We identified 36 extragalactic transients during the experiment (see Table 1) and secured spectroscopic classifications for 34 of these: 23 SNe Ia, 5 SNe II, 2 SNe Ic, 1 SN Ib, 1 SN Ibn/IIn, 1 cataclysmic variable (CV), and 1 likely luminous blue variable (LBV) outburst. The sample is spectroscopically complete for the 31 of these transients brighter than 19.5 mag in either filter (20 SN Ia, 5 SN II, 2 SN Ic, 1 SN Ib, 1 SN Ibn/IIn, 1 CV, and 1 LBV outburst). The results of the spectroscopic follow-up are summarized in Table 2. The sample is also spectroscopically complete for transients with $| g-I| \gt 0.5$ mag or coincident with a known nearby (<200 Mpc) galaxy, but these criteria did not select any transients not already included in the magnitude-limited sample. We also note that the host galaxies of 4 transients (not including the CV) with redshifts corresponding to $\lt 200$ Mpc did not have redshifts listed in the NASA Extragalactic Database.¹⁷ Though no LRNe were detected the color distribution of the detected transients confirms that a red color selection will filter out most other transients.¹⁸ All classification spectra have been made publicly available on the Transient Name Server¹⁹ and the Open Supernova Catalog.²⁰

Table 1.  Catalog of Transients

Name	R.A.	Decl.	Classification	Spectrum	Telescope/	Redshift	Peak	MJD	g − I
				MJD	Instrument		g Mag	at peak	at peak
iPTF16iyy(1)	06:56:34.60	+46:53:38.7	SN II(2)	⋯	⋯	0.03	18.42 ± 0.12	57763.8	0.19 ± 0.22
iPTF16jdl(2)	12:27:39.98	+40:36:52.7	SN Ia(4)	⋯	⋯	0.02	18.58 ± 0.07	57785.1	⋯
iPTF16jhb(1)	04:00:25.51	+20:19:36.2	SN Ia	57809.4	P200/DBSP	0.03	17.40 ± 0.06	57786.7	−0.47 ± 0.14
iPTF17be(3)	08:13:13.38	+45:59.28.9	LBV(7)	57765.3	APO/DIS	0.00	18.09 ± 0.12	57771.9	0.43 ± 0.16
iPTF17jv	08:15:04.50	+57:10:50.0	SN Ia	57779.5	Keck I/LRIS	0.08	19.47 ± 0.08	57770.9	$\lt -0.44$
iPTF17le(1)	11:10:27.22	+28:41:42.2	SN Ia(8)	⋯	⋯	0.03	16.92 ± 0.07	57772.0	0.20 ± 0.18
iPTF17lf(4)	03:12:33.59	+39:19:14.2	SN Ia(4)	57774.2	Gemini-N/GMOS	0.01	19.42 ± 0.08	57786.6	1.77 ± 0.13
iPTF17lj(5)	07:39:01.54	+50:46:05.4	SN Ia	57780.3	P60/SEDM	0.03	17.24 ± 0.05	57782.9	−0.66 ± 0.12
iPTF17ln(2)	11:38:33.71	+25:23:49.7	SN Ia(8)	⋯	⋯	0.03	16.45 ± 0.04	57781.0	−0.49 ± 0.11
iPTF17nh(2)	07:10:13.54	+27:12:10.5	SN Ia(4)	⋯	⋯	0.06	18.74 ± 0.05	57785.8	−0.61 ± 0.20
iPTF17pq	08:03:41.63	+09:50:19.5	SN II	57785.3	DCT/DeVeney	0.09	19.40 ± 0.07	57780.8	−0.23 ± 0.13
iPTF17wj	07:24:08.95	+39:51:43.6	SN Ib	57785.3	P200/DBSP	0.03	19.15 ± 0.09	57782.8	0.15 ± 0.15
iPTF17wk	07:08:07.02	+49:16:29.2	⋯	⋯	⋯	⋯	20.71 ± 0.17	57782.8	0.43 ± 0.24
iPTF17wm	06:30:52.76	+73:27:31.4	SN Ia	57789.3	P200/DBSP	0.08	18.96 ± 0.07	57782.8	−0.22 ± 0.14
iPTF17wo	01:50:08.81	+39:28:52.3	⋯	⋯	⋯	⋯	19.64 ± 0.16	57782.6	−0.06 ± 0.21
iPTF17wp	01:23:44.06	+14:46:32.1	SN Ia	57786.1	P200/DBSP	0.09	18.95 ± 0.12	57785.6	−0.26 ± 0.19
iPTF17wr	05:02:44.92	+75:01:44.6	SN Ia	57787.3	P200/DBSP	0.07	18.50 ± 0.07	57789.7	−0.45 ± 0.13
iPTF17wu	05:46:21.12	+65:05:04.3	SN Ia	57789.3	P200/DBSP	0.12	19.96 ± 0.11	57771.8	−0.02 ± 0.23
iPTF17xe(5)	12:34:17.41	+44:18:08.9	SN Ia	57785.5	P200/DBSP	0.14	19.24 ± 0.10	57787.0	−0.19 ± 0.17
iPTF17xf	10:45:39.12	+54:07:46.3	SN Ic	57785.3	P200/DBSP	0.06	19.99 ± 0.08	57782.9	0.64 ± 0.16
iPTF17yw	11:47:35.37	+52:33:01.1	SN Ia	57785.5	P200/DBSP	0.15	20.33 ± 0.17	57783.0	−0.06 ± 0.28
iPTF17zb(1)	08:02:49.31	+56:37:38.8	SN Ia	57807.4	P200/DBSP	0.07	18.46 ± 0.09	57785.9	−0.46 ± 0.16
iPTF17zp	04:26:45.85	+74:04:11.5	SN Ia	57786.2	P200/DBSP	0.08	19.10 ± 0.08	57789.6	0.29 ± 0.14
iPTF17aak	08:30:54.62	+11:40:58.6	SN Ia	57786.4	P200/DBSP	0.13	19.18 ± 0.07	57789.8	−0.51 ± 0.14
iPTF17aal	07:43:02.52	+11:38:20.4	SN II	57786.4	P200/DBSP	0.07	18.77 ± 0.07	57789.8	−0.36 ± 0.11
iPTF17aam	07:38:40.01	+09:22:03.7	SN Ia	57793.7	DCT/DeVeney	0.15	19.87 ± 0.06	57785.8	−0.36 ± 0.16
iPTF17aan	07:39:26.00	+13:55:09.7	CV	57786.3	P200/DBSP	0.00	19.17 ± 0.08	57785.8	−0.71 ± 0.16
iPTF17atf	07:58:33.37	+11:55:22.0	SN Ia	57800.3	P60/SEDM	0.04	18.88 ± 0.06	57800.8	0.27 ± 0.10
iPTF17atg	07:39:32.53	+18:24:46.9	SN Ia	57800.3	P60/SEDM	0.08	18.98 ± 0.08	57800.7	−0.46 ± 0.18
iPTF17aua	08:59:38.73	+56:52:44.2	SN Ia	57800.5	P60/SEDM	0.03	17.47 ± 0.06	57799.9	−0.37 ± 0.13
iPTF17aub	06:40:24.70	+64:33:02.7	SN II	57800.2	P60/SEDM	0.01	17.47 ± 0.05	57799.8	−0.02 ± 0.14
iPTF17avb	08:01:15.98	+11:01:58.6	SN II	57801.2	P60/SEDM	0.10	19.53 ± 0.06	57800.8	0.44 ± 0.10
iPTF17avc	08:36:47.39	+45:11:35.7	SN Ia	57828.3	P200/DBSP	0.08	18.92 ± 0.12	57800.9	−0.34 ± 0.21
iPTF17axg(1)	12:36:30.75	+51:38:40.0	SN Ic-BL	57809.4	P200/DBSP	0.06	18.51 ± 0.11	57800.0	0.33 ± 0.16
iPTF17bgb	13:23:06.73	+52:05:57.2	SN Ia	57809.4	P200/DBSP	0.10	18.89 ± 0.06	57809.0	−0.67 ± 0.13
iPTF17bkj(6)	11:23:27.43	+53:40:45.4	SN Ibn/IIn	57809.4	P200/DBSP	0.03	19.12 ± 0.08	57809.0	0.14 ± 0.14

Note. Names are cross-identified on the Transient Name Server (TNS) as iPTFxx = AT (or SN) 20xx. Transient discoveries and classifications first reported to TNS by other groups are indicated by footnotes in the "Name" and "Classification" columns respectively as follows: (1) ATLAS, (2) ASAS-SN, (3) LOSS, (4) TNTS, (5) Pan-STARRS, (6) PIKA, (7) LCOGT-KP, (8) PESSTO.

Download table as:  ASCIITypeset image

Table 2.  Summary of Transients

Classification	All	$g\lt 19.5$	g or $I\lt 19.5$	$g-I\gt 0.5$	$g-I\lt -0.5$	Known Galaxy	$z\lt 0.045$
	at Peak	at Peak	at Peak	at Peak	Peak	$\lt 200$ Mpc
SN Ia	23	20	20	1	4	5	8
SN II	5	4	5	0	0	1	2
SN Ib	1	1	1	0	0	1	1
SN Ic	2	1	2	1	0	0	0
SN Ibn/IIn	1	1	1	0	0	1	1
LBV	1	1	1	0	0	1	1
CV	1	1	1	0	1	0	1
Unclassified	2	0	0	0	0	0	⋯
Total	36	29	31	2	5	9	14

Download table as:  ASCIITypeset image

In the following subsections we highlight three of these transients that are of particular interest: iPTF17lf (a highly reddened SN Ia), iPTF17bkj (a rare transititional SN Ibn/IIn), and iPTF17be (likely an LBV outburst).

4.1. iPTF17lf: A Highly Reddened SN Ia

iPTF17lf (SN 2017lf) is a highly reddened SN Ia in NGC 1233 at α = 3^h 12'3359 δ = +39°19'142 (J2000), discovered by the survey on 2017 January 18 (and independently by the Tsinghua-NAOC Transient Survey on January 23; Rui et al. 2017). Figure 6 shows the location of the transient. The spectral evolution of iPTF17lf is shown in Figure 7.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Taking advantage of the standard candle property of SNe Ia, we fit a model for the spectral energy distribution of iPTF17lf based on the normal and unreddened SN 2011fe as described in Amanullah et al. (2015), together with the extinction law from Cardelli et al. (1989), to the data of iPTF17lf presented here. The free lightcurve parameters are the time of maximum, t₀, and the peak brightness, m_B, in the rest-frame B band. We also fit the lightcurve stretch, s, of the temporal evolution of the lightcurve model, which is used in combination with m_B when SNe Ia are used to measure cosmological distances (see e.g., Goobar & Leibundgut 2011, for a review). For the host galaxy extinction we fit the color excess, $E(B-V)$, and the total-to-selective extinction, ${R}_{V}={A}_{V}/E(B-V)$, in the V band. Note that we only fit for the extinction law in the host, while we adopt the Galactic average value of R_V = 3.1 and $E(B-V)=0.134$ mag, based on the Schlafly & Finkbeiner (2011) recalibration of Schlegel et al. (1998), for the foreground reddening in the Milky Way.

The best fit is shown together with the data in Figure 8 where the fitted parameter values were ${t}_{0}=57777.8\pm 0.5$ days, ${m}_{B}=20.34\pm 0.27$ mag and $s=1.01\pm 0.06$. For the host extinction we obtain $E(B-V)=1.9\pm 0.1$ mag and ${R}_{V}\,=1.9\pm 0.3$, where intrinsic SN Ia uncertainties have been taken into account.

Zoom In Zoom Out Reset image size

iPTF17lf is another SN Ia with high reddening and low R_V (see e.g., Burns et al. 2011; Amanullah et al. 2015). Although lower R_V values for highly reddened SNe is consistent with expectations for SNe discovered by magnitude limited surveys if extinction laws are drawn from an R_V distribution, an alternative interpretation is that such SNe Ia arise from progenitors surrounded by circumstellar dust (Wang 2005; Goobar 2008). This alternative interpretation has been further motivated by the fact that the average color correction of SNe Ia in cosmological samples is significantly lower than what is observed in the Milky Way (see e.g., Betoule et al. 2014). If the observed R_V is lower due to scattering of the SN light by circumstellar dust, the SN should appear to have to time-dependent reddening (e.g., Amanullah & Goobar 2011; Brown et al. 2015).

However, this is not seen for iPTF17lf. Although its lightcurve is sparsely sampled, it is consistent with a constant reddening law over ∼35 days. The same was seen for the highly reddened SN 2014J that was consistent with a low value of R_V = 1.4 (Amanullah et al. 2014). Not only did constant reddening imply a dust distance of $\gtrsim 30$ pc (Bulla et al. 2018), polarimetry measurements of SN 2014J also showed the extinction to mainly be associated with the interstellar medium in the host galaxy (Patat et al. 2015). Following the methodology of (Bulla et al. 2018), the constant reddening of iPTF17lf implies a dust distance of >14 pc.

4.2. iPTF17bkj (SN 2017hw): A Transitional SN Ibn/IIn

iPTF17bkj is a transitional SN Ibn/IIn in UGC 6400 at α = 11^h 23'2743, δ = +53°40'454 (J2000), discovered by the survey on 2017 February 24. An independent discovery was also reported by the Slovenian Comet and Asteroid Search Program (PIKA) on 2017 February 25. UGC 6400 is a star-forming galaxy at a luminosity distance of 117 Mpc (distance modulus 35.35 mag using NED). Figure 9 shows the location of the SN in the host galaxy.

Zoom In Zoom Out Reset image size

SNe IIn are commonly accepted as supernovae interacting with their hydrogen-rich circumstellar medium (CSM). The narrow emission lines in their spectra are produced by the recombination in the surrounding shell, which is moving at lower speed than the ejecta. SNe Ibn are far less common. Similar to SNe Ib/c, they lack hydrogen features in their spectra, but they do show narrow He lines, which are believed to originate in a helium-rich CSM (Pastorello et al. 2007). Few supernovae have been detected sharing the characteristics of both SNe IIn and SNe Ibn. These are the so-called transitional SNe IIn/Ibn, presenting both narrow He and H lines, with low expansion velocities (∼2000 km s⁻¹). In the literature, we could identify three objects with similar spectral characteristics: SN 2005la (Pastorello et al. 2008), SN 2010al and SN 2011hw (Smith et al. 2012; Pastorello et al. 2015), which we include as our comparison sample.

iPTF17bkj was detected close to its peak light, at ${M}_{g}=-16.3\pm 0.2$ mag and ${M}_{I}=-16.4\pm 0.1$ mag (correcting for Galactic extinction of A_V = 0.043 mag)— ∼3 magnitudes fainter than the average peak magnitude for the SN Ibn population (Hosseinzadeh et al. 2017a). Dust extinction cannot be estimated from the Na i 5890 Å, 5896 Å doublet because of He ii emission, but the color of iPTF17bkj ($B-V\sim 0.3$ mag at +15–20 days) is slightly bluer than the sample of unobscured Ibn (at similar phase) discussed in Pastorello et al. (2015), suggesting that the fainter peak magnitude of iPTF17bkj is not due to host or circumstellar dust extinction.

Comparing the iPTF17bkj lightcurve (see Figure 10) with an average SN Ibn template lightcurve shows not only that the peak differs, but also that the decay timescales are also longer, which implies a larger circumstellar mass (Karamehmetoglu et al. 2017). The light curve most closely resembles a subluminous SN Ibc. Individual lightcurves for the other three transitional objects seems to point to a large diversity in the distribution of the H and He CSM around these stars. One possibility, discussed by Smith et al. (2012), is that the progenitors of such explosions are LBVs transitioning to Wolf–Rayet stars.

Zoom In Zoom Out Reset image size

The spectroscopic evolution of iPTF17bkj is similar to that of other transitional SN IIn/Ibn objects. Figure 11 shows that, at early stages (<15 days), the objects show narrow Hα, Hβ and He i lines with Lorentzian profiles caused by electron scattering in the outer layers of the surrounding material. In the case of iPTF17bkj, we notice a narrow absorption component in both H and He i lines, with a blueshifted velocity of ∼400 km s⁻¹. At later times (>15 days), the lines develop into a wider, P-Cygni profile, with a minimum around 2000 km s⁻¹ (see Figure 12). This emission is associated with the expanding photosphere of the SN, rather than diffusion of photons through the dense CSM. Already at +22 days, we also notice the appearance of broad Ca ii lines, which were not detected in the spectrum of SN 2010al taken at a similar epoch. The log of spectroscopic observations is available in Table 3.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 3.  Log of Spectroscopic Observations of iPTF17lf, iPTF17bkj, and iPTF17be

Object	MJD	Slita	Telescope+Instrument	Grism/Grating	Dispersion	Exposure
	(days)	(arcseconds)			(Å/pix)	(s)
17lf	57774.2	−1	Gemini+GMOS	⋯	0.9	900
17lf	57780.2	2.5a	P60+SEDM	IFU		2700
17lf	57816.1	1.5	P200+DBSP	600/4000	1.5	300
17lf	57828.1	1.0	P200+DBSP	600/4000	1.5	1200
17bkj	57809.4	1.5	P200+DBSP	600/4000	1.5	900
17bkj	57811.4	1.0	Keck I+LRIS	400/3400 + 400/8500	2.0	420
17bkj	57822.0	1.5	NOT+ALFOSC	gr4	3.5	4600
17bkj	57828.3	1.5	P200+DBSP	600/4000	1.5	1800
17bkj	57848.0	1.0	TNG/DOLORES	LR-B		3000
17be	57761.4	2.4a	P60+SEDM	IFU		2100
17be	57765.3	1.5	APO+DIS	B400	1.8	900
17be	57779.5	1.5	Keck I+LRIS	600/4000	1.3	300
17be	57895.2	1.0	Keck II+DEIMOS	600/6000	1.2	2100

Note.

^aFor the IFU, the extraction radius (in arcseconds) is indicated.

Download table as:  ASCIITypeset image

4.3. iPTF17be (AT2017be): A Likely LBV Outburst

iPTF17be was discovered independently by LOSS (Stephens et al. 2017) and our survey on 2017 January 07 at α = 08^h 13'1338, δ = +45°59'289 (J2000). The discovery image is shown in Figure 13. The host galaxy, NGC 2537, is located at a distance of 6.19 ± 0.47 Mpc (distance modulus of μ = 28.96 ± 0.16 mag) as obtained from the redshift value reported by Garrido et al. (2004). In our analysis, we adopt a Galactic extinction of $E(B-V)=0.047$ mag from Schlafly & Finkbeiner (2011).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The transient was discovered a week before reaching maximum light, as shown by the lightcurve in Figure 14. At peak, its absolute magnitude was ${M}_{I}=-11.43\pm 0.05$ and ${M}_{g}=-10.95\pm 0.14$ mag, well fainter than supernovae. Our measurements show that the lightcurve faded quickly in the u band (∼1 mag in 14 days). However, slower decline was registered in redder filters (∼1 mag in 28, 52, and 54 days for g, r, and I bands, respectively). The color at peak was relatively red, $g-I\sim 0.4$ mag, as compared to the g − I evolution of most supernovae (see Figure 1). Overall, the evolution of the g − I color through ∼100 days post-explosion is much more modest than typically seen in LRNe or SN 2008S-like transients, which show g − I color larger than 1 mag.

Zoom In Zoom Out Reset image size

The location of iPTF17bkj was observed by iPTF in previous seasons and there is also archival Spitzer and HST/NICMOS pre-outburst imaging of the host (see Figure 15). We do not detect a point-source at the location of iPTF17bkj in any of these data. The limiting magnitudes are provided in Table 4 (the upper limits from the iPTF data are set from the detection of an extended source). Unfortunately these limits are not deep enough to set meaningful constraints on the progenitor.

Zoom In Zoom Out Reset image size

Table 4.  iPTF17be Progenitor Photometry

Filter	Magnitude	$\nu {L}_{\nu }/{L}_{\odot }$	Telescope	Date
g	>18.2	$\lt 1.6\times {10}^{6}$	P48	2011 Apr 04–2014 Jan 03
r	>18.2	$\lt 1.3\times {10}^{6}$	P48	2012 Nov 21–2012 Nov 25
I	>18.0	$\lt 1.2\times {10}^{6}$	P48	2016 Dec 14–2016 Dec 20
F160W	>19.76	$\lt 3.1\times {10}^{4}$	NIC3/HST	2002 Oct 14
3.6 μm	>17.38	$\lt 3.1\times {10}^{4}$	IRAC/SST	2005 Apr 16a
4.5 μm	>17.26	$\lt 1.8\times {10}^{4}$	IRAC/SST	2005 Apr 16a
5.8 μm	>15.37	$\lt 5.1\times {10}^{4}$	IRAC/SST	2005 Apr 16a
8.0 μm	>13.50	$\lt 1.2\times {10}^{5}$	IRAC/SST	2005 Jul 01a

Note.

^aMean Date of multi-epoch coadded image.

Download table as:  ASCIITypeset image

The spectroscopic evolution of iPTF17be is shown in Figure 16, along with some comparison spectra. The first classification spectrum for this event was obtained on 2017 January 08 (Hosseinzadeh et al. 2017b). Despite its low signal-to-noise ratio, the classification report describes Hα emission with a FWHM ∼ 950 km s⁻¹ in a spectrum taken 2017 January 10, and uses this and the low (∼−11 mag) peak absolute magnitude to conclude that the event was likely an LBV outburst. Our spectrum taken near peak light (2017 January 12) also shows strong Hα emission with a width consistent with that reported by Hosseinzadeh et al. (2017b), the forbidden Ca ii doublet in emission, and the Ca ii triplet ∼8500 Å in absorption. By the spectrum from 26 January 2017 (+10 days) the triplet shifted to being in emission with broad line profiles. In the final spectrum taken 2017 June 14, the forbidden Ca ii doublet is no longer visible and the Hα emission is double-peaked. The narrow Hα and Ca triplet emission are common not only to LBV outbursts, but also to stellar mergers (Rushton et al. 2005; Williams et al. 2015) and SN 2008S-like transients (Bond et al. 2009; Smith et al. 2009). However, in contrast to iPTF17be, in well-studied stellar mergers the Hα emission disappears and strong TiO absorption bands emerge within the first 100 days. Since the photometric evolution of iPTF17be more closely resembles that of LBV outbursts than of SN 2008S-like events we agree with the initial assessment of Hosseinzadeh et al. (2017b) that this event is likely an LBV outburst.

We now consider the implications of the results of this experiment for future surveys. We can make constraints on the rates of various transient types using the effective survey coverage for the typical duration of each transient type (see Figure 3) and a limiting magnitude of 19.5 (the threshold for spectroscopic completeness). We first check that our inferred rates of core-collapse and SNe Ia are consistent with values reported in the literature. Scaling for a core-collapse SN rate of ${10}^{-4}\,{\mathrm{Mpc}}^{-3}\,{\mathrm{yr}}^{-1}$ (Horiuchi et al. 2011) and a characteristic absolute magnitude $\lt -16.5$ (Li et al. 2011a) for 30 days (corresponding to an effective survey coverage of 3731 sq-deg months) we would expect to discover 6.3 core-collapse SNe $\lt 19.5$ mag, consistent (within Poisson statistics) with the nine we did find. Adopting a SN Ia rate of 34% of the core-collapse SN rate (Li et al. 2011a) and adopting a characteristic absolute magnitude for SNe Ia of $\lt -18$ for 10 days (corresponding to an effective survey coverage of 2170 sq-deg months), we would expect 19.8 SN Ia brighter than 19.5 mag, again consistent with the 20 observed.

We can now make a constraint on the rate of LRNe based on the lack of discoveries during the survey. The number, N, expected in a flux limited survey, is

$$\begin{eqnarray}&&N\propto \int \displaystyle \frac{{dN}}{{dL}}{L}^{3/2}{dL},\end{eqnarray} \tag{ 1 }$$

since the volume probed goes as ${L}^{3/2}$. For our survey the expected number depends on the luminosity function for high luminosities ($M\lt -14$), but constraints on the luminosity function have only been estimated using the more common, lower-luminosity events discovered within the Galaxy, where Kochanek et al. (2014) found the Galactic rate of LRNe to be ${dL}/{dN}\propto {L}^{-1.4\pm 0.3}$ with ∼0.1–1 yr⁻¹ brighter than ${M}_{I}\,=-4$ mag and that the peak luminosities are roughly 2000–4000 times brighter than the main-sequence progenitor luminosities.

Clearly this luminosity function must steepen for very high luminosities (the expected rate in a flux limited survey is only finite for luminosity functions steeper than ${dN}/{dL}\propto {L}^{-2.5})$, otherwise transient surveys would routinely detect distant, luminous LRNe. Instead, only a handful have been discovered, with only three beyond the local group. Although stellar mergers are expected to eject mass primarily at velocities no more than a few times the escape velocity (making them distinguishable from typical SNe regardless of their luminosity), it has been conjectured that some stellar mergers may become dust-obscured prior to their peak luminosity (Metzger & Pejcha 2017) and would be discovered as infrared transients (Kasliwal et al. 2017) rather than by an optical survey as LRNe. If the peak luminosities of LRNe are a few thousand times the progenitor main-sequence luminosity the most luminous LRNe would be roughly −16 to −18 mag.

Even the non-detection of any LRNe during just this two-month experiment is in tension with an extrapolation of the stellar merger luminosity function inferred from Galactic events to very high luminosities (${M}_{I}\lesssim -16$ mag), but it is still consistent with the peak luminosities of LRNe being a few thousand times the progenitor main-sequence luminosity. We calculate the expected number of LRNe using Equation (1) normalized by the Galactic rate, assuming 10⁻² Milky Way-like galaxies per Mpc⁻³, and scaled for a survey limiting magnitude of ${m}_{I}=19.5$ and an effective coverage 2170 square-degree months (the survey coverage for transients with a 10 day duration). The top panel of Figure 17 shows the cumulative number of LRNe as a function of peak luminosity that should have been found by this experiment based on this extrapolation and the bottom panel shows the corresponding constraint on the rate of LRNe.

Zoom In Zoom Out Reset image size

Deep all-sky time-domain photometric surveys with color information, such as the public 3π, 3 days cadence g and r band survey of the upcoming ZTF and, later, the LSST, will be capable of making much more meaningful constraints on the high-end luminosity function of LRNe. The highest luminosity stellar merger candidate known peaked at −14 mag (Smith et al. 2016). The ZTF survey will have the sensitivity, within its first year, to determine whether the LRN luminosity function steepens before or after this luminosity.

This g + I band survey is a pathfinder for future transient surveys. It was the first I band transient search with PTF/iPTF and its successful implementation laid the groundwork for including an I band component to ZTF, which will improve the photometric classification and selection of cool and dusty transients as well as SNe Ia.

We thank Mislav Balokovic, Kevin Burdge, Kishalay De, George Djorgovski, Andrew Drake, Marianne Heida, Nikita Kamraj, and Harish Vedantham for taking observations used in this paper. We thank the anonymous referee and Chris Kochanek for helpful comments.

A.A.M. is supported by a grant from the Large Synoptic Survey Telescope Corporation, which funds the Data Science Fellowship Program. J.S and F.T. gratefully acknowledge the support from the Knut and Alice Wallenberg Foundation. R.A., M.B., A.G., and R.F. acknowledge support from the Swedish Research Council and the Swedish Space Board.

The Intermediate Palomar Transient Factory project is a scientific collaboration among the California Institute of Technology, Los Alamos National Laboratory, the University of Wisconsin, Milwaukee, the Oskar Klein Center, the Weizmann Institute of Science, the TANGO Program of the University System of Taiwan, and the Kavli Institute for the Physics and Mathematics of the Universe. Part of this research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. Some of the data presented herein were obtained at the W.M. Keck Observatory, which is operated as a scientific partnership among the California Institute of Technology, the University of California and the National Aeronautics and Space Administration. The Observatory was made possible by the generous financial support of the W.M. Keck Foundation. These results made use of the Discovery Channel Telescope at Lowell Observatory. Lowell is a private, non-profit institution dedicated to astrophysical research and public appreciation of astronomy and operates the DCT in partnership with Boston University, the University of Maryland, the University of Toledo, Northern Arizona University and Yale University. The upgrade of the DeVeny optical spectrograph has been funded by a generous grant from John and Ginger Giovale. This work is partly based on observations made with the Nordic Optical Telescope, operated by the Nordic Optical Telescope Scientific Association at the Observatorio del Roque de los Muchachos, La Palma, Spain, of the Instituto de Astrofisica de Canarias. This work is partly based on observations made with DOLoRes@TNG.