Datasets:

SouthernCrossAI
/

Arxiv

Dataset card Viewer Files Files and versions Community

Split (1)

train · 132k rows

Filename stringlengths 22 64	Paragraph stringlengths 8 5.57k
Processed_Nonlinear_Self-Duality_and_Supergravity.txt	which involves the naked prepotential V and therefore does not allow a naive generaliza- tion of the Gaillard-Zumino proof [7, 12] to superspace.
Processed_Nonlinear_Self-Duality_and_Supergravity.txt	Any system of matter superﬁelds Ψ coupled to new minimal supergravity can be treated as a super-Weyl invariant coupling of old minimal supergravity to the matter superﬁelds Ψ and L (see [31] for a review). It is clear that matter couplings in the new minimal formulation of supergravity are more restrictive as compared to the old minimal version. Here we will demonstrate how to couple the models of self-dual nonlinear electrodynamics to new minimal supergravity.
Processed_Nonlinear_Self-Duality_and_Supergravity.txt	which is (i) super-Weyl invariant and (ii) self-dual, i.e. it solves the self-duality equation (2.14). This action describes self-dual supersymmetric electrodynamics in new minimal supergravity.
Processed_Nonlinear_Self-Duality_and_Supergravity.txt	Z and this theory possesses several important symmetries: (i) super-Weyl invariance; (ii) K¨ahler invariance; (iii) U(1) duality invariance. We should now uncover the description of this theory in the framework of old minimal supergravity.
Processed_Nonlinear_Self-Duality_and_Supergravity.txt	This is the old minimal supergravity counterpart of the model (2.42). To preserve the super-Weyl gauge condition U = 0, any K¨ahler transformation (2.41) must now be ac- companied by the induced super-Weyl transformation with σ = 1 3λ(φ). As a result, one ends up with the so-called super-Weyl–K¨ahler transformations (see [29, 30, 31] for more details).
Processed_Nonlinear_Self-Duality_and_Supergravity.txt	In the above analysis of the coupling of self-dual supersymmetric electrodynamics to K¨ahler sigma models in curved superspace, it was assumed that the matter superﬁelds, φ and ¯φ, are inert under the electromagnetic duality rotations. Of some interest is a more general situation when, say, a chiral matter superﬁeld Φ and its conjugate ¯Φ do transform under duality rotations, which can now span a larger group than the one corresponding to the pure gauge ﬁeld case. Coupling to the so-called dilaton-axion supermultiplet is an important example.
Processed_Nonlinear_Self-Duality_and_Supergravity.txt	with d = −a. We see that the maximal group of duality transformations is Sp(2, R) ∼= SL(2, R). The complex variable Φ should then parametrize the homogeneous space SL(2, R)/U(1), with the vector ﬁeld ξ(Φ) in (3.2) generating the action of SL(2, R) on the coset space. The matter-free case, which was considered before, corresponds to freez- ing the superﬁeld Φ(z) to a given point of the space SL(2, R)/U(1). In such a case, the duality group, SL(2, R), reduces to U(1) – the stabilizer of the point chosen.
Processed_Nonlinear_Self-Duality_and_Supergravity.txt	L2 Λ and u is deﬁned in (2.19). Here K(Φ, ¯Φ) is the K¨ahler potential of the K¨ahler manifold d8z E−1 L K(φ, ¯φ) in (3.4) corresponds to the chiral matter SL(2, R)/U(1). The term which is inert under the duality rotations. For Φ = −i, the action (3.4) reduces to (2.42).
Processed_Nonlinear_Self-Duality_and_Supergravity.txt	Acknowledgements. We are grateful to Ian McArthur for discussions and for reading the manuscript. The work of SK is supported by the Australian Research Council, the Australian Academy of Science as well as by UWA research grants. The work of SMc is supported by the Hackett Postgraduate Scholarship.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	We present the Australia Telescope Compact Array (ATCA) observations of the SiO masers in the Galactic center in transitions of v=1, J=2–1 at 86 GHz and v=1, J=1–0 at 43 GHz. Two 86-GHz SiO masers were detected within the central parsec, and they are associated with IRS 10EE and IRS 15NE, respectively. We detected eighteen 43-GHz SiO masers within a projected separation of .2 pc from Sagittarius A* (Sgr A*), among which seven masers are newly discovered from our observations. This raises the total number of 43-GHz SiO masers within the central 4 parsecs of the GC region to 22. Simultaneous observations at 86 and 43 GHz showed that the intensity of 43-GHz SiO maser is ∼3 times higher than that of 86-GHz maser in IRS 10EE (an OH/IR star), while the integrated ﬂux of the SiO maser emission at 43 GHz is comparable with that at 86 GHz in IRS 15NE (an ordinary Mira variable). These results are consistent with previous observations of massive late-type stars in the Galaxy in which the 86-GHz SiO maser is in general weaker than the 43-GHz SiO maser in OH/IR stars, while the two transitions are comparably strong in Mira stars.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	SiO masers located in the extended circumstellar envelopes of late-type giant stars are believed to be reliable tracers of the Galactic dynamics since they can be treated as point-like objects and are not subject to non-gravitational forces such as magnetic ﬁelds or stellar wind collisions. A number of SiO maser surveys toward the Galactic center (GC) region have been carried out at 43 GHz in order to study the structure and dynamics of the stellar disk in the Inner Galaxy (Izumiura et al. 1998; Miyazaki et al. 2001; Deguchi et al. 2000, 2002; Sjouwerman et al. 2002, 2004; Imai et al. 2002). So far the most sensitive search for 43-GHz SiO masers in the central few parsec region of the GC was made with the VLA and VLBA, and the accurate position determination of the SiO masers has been used to calibrate the astrometry in the infrared images of the same region with an accuracy of ∼1 mas (Menten et al. 1997; Reid et al. 2003, 2007; Oyama et al. 2008). Moreover ﬁfteen SiO maser sources have been observed with a projected separation < 50′′ from Sgr A* at 43 GHz; lower limit of the dynamic mass enclosed within a given radius was derived from the 3-dimensional velocity of the maser star. Whereas only in IRS 9, the derived enclosed mass limit is ∼20 percent higher than the mass of known objects in the GC, i.e., the supermassive black hole (SMBH) and the luminous stars (Reid et al. 2007). In fact previous VLA observations of SiO masers were constrained by the limited velocity coverage of the receiver band. Searching for high-velocity SiO maser stars is critical for exploring the mass distribution in the GC.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	At 86 GHz, Lindqvist et al. (1991) searched for SiO maser line emission in a sample of 31 OH/IR stars close to the GC using the Nobeyama 45m telescope, but they failed to obtain distinctive detection of any 86 GHz masers. Messineo et al. (2002) detected 271 SiO masers in a sample of 441 late-type stars in the Inner Galaxy from their high-sensitivity IRAM-30m observations. Limited by the low angular resolution and sensitivity of single-dish observations, the number and spatial distribution of SiO maser sources within a few parsecs of the GC remain uncertain. Until recently, high-angular-resolution observations of 86-GHz SiO maser line emission from the GC have been made possible using the Australia Telescope Compact Array (ATCA) of the Australia Telescope National Facility (ATNF) (this paper and L.O. Sjouwerman et al., private communication).
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	In this Letter, we present the results from the ATCA observations of SiO maser lines at two transitions (v=1, J=2–1 (86 GHz) and v=1, J=1–0 (43 GHz)) within a few parsecs from Sgr A*. The ATCA observations and data reduction are described in Section 2, and the results and analysis are presented in Section 3. In Section 4 we summarize the observational results. Throughout this paper we adopt a GC distance of 8 kpc (Reid 1993) and a SMBH mass of 4 × 106M⊙ at the GC (Ghez et al. 2008; Gillessen et al. 2009).
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	We carried out ATCA observations of the Galactic center at 86 and 43 GHz in ﬁve sessions (2006 June 9, 2006 August 13, 2007 October 21, 2008 June 3-4 and 2008 October 3). These observations were initially made with the purpose of investigating the variability of Sgr A* at millimeter wavelengths (Li et al. 2009). The observations on 2008 October 3 were made quasi simultaneously at 43 and 86 GHz by interchanging the receivers at the two frequencies with a duty cycle of about 20 minutes. The observations on other dates were performed solely at a single frequency. Except for the observation on 2008 June 3-4, other observations were made under good weather condition. Table 1 summarizes the observational parameters.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	Quasar 3C 279 was observed at the beginning of each observing run to calibrate the bandpass. The observations were pointed at Sgr A, allowing for detection of SiO masers located in the primary beam of the ATCA antenna, i.e., θF W HP of ∼ 36′′ and ∼ 72′′ at 86 and 43 GHz, respectively. The pointing were checked and corrected every half an hour by observing either a nearby quasar PKS 1730-130 (NRAO 530) or a strong SiO maser VX Sgr. Quasar PKS 1741-312 and SiO maser OH 2.6–0.4 were interlaced with Sgr A to track the elevation-dependent gains.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	We searched for SiO maser sources within a radius of 25′′ and 50′′ centering at Sgr A* at 86 and 43 GHz, respectively. For those SiO maser sources detected above 5σ, we derived their positions in two steps: ﬁrst, we ﬁtted their brightness distribution with a two-dimensional Gaussian using the task IMFIT in each channel; then, we made variance-weighted averaging to obtain its mean position and the uncertainty.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	Two SiO masers were detected in the 86 GHz images with peak ﬂux densities higher than 5σ. These two sources were identiﬁed to be associated with IRS 10EE and IRS 15NE, based on the consistent positions and radial velocities determined from previous observations (Reid et al. 2003, 2007). IRS 10EE appeared at four epochs (∼ 4σ on 2008 October 3) and displayed strong variability on timescale of months. IRS 15NE was only distinctively detected on 2007 October 21. It was marginally detected on 2008 June 4 (∼ 3σ) as a result of high rms noise level. The late-type giants and supergiants in the GC are likely detectable over a wide radial velocity range of ±350 km s−1 (e.g., Figer et al. 2003). However, current 86-GHz observations were constrained by the limited velocity coverage and relatively low sensitivity. The positions, peak ﬂux densities and velocity-integrated ﬂux densities of IRS 10EE and IRS 15NE are listed in Table 2. The rms noise of the integrated ﬂux density is estimated as △I = σrms△v(Nchan)1/2, where σrms is the rms noise level in individual channel maps, △v is the velocity resolution, and Nchan is the number of channels over which the ∆I is calculated (Klaassen & Wilson 2007).
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	H2O maser line emission (Lindqvist et al. 1990, 1992; Menten et al. 1997; Sjouwerman et al. 2002; Reid et al. 2003, 2007; Oyama et al. 2008; Peeples et al. 2007). In infrared (IR) bands it was classiﬁed as a long-period variable, and exhibited variability on various timescales ranging from several days to a few months (Tamura et al. 1996; Wood et al. 1998; Ott et al. 1999; Peeples et al. 2007). X-ray emission was detected within 0.′′8 of IRS 10EE (Muno et al. 2004) and a binary system was proposed to explain its relatively high X-ray luminosity as an AGB star (Peeples et al. 2007). The binary scenario of IRS 10EE was further supported by the VLBA monitoring of its proper motion (Oyama et al. 2008). Spectra of 86- and 43-GHz SiO maser lines of IRS 10EE from our observations are shown in the upper panel of Figure 1. They indicated a ﬂux density variation on timescale of months, in agreement with previous radio and IR observations. Long-term monitoring of SiO maser emission in IRS 10EE and comparison with the optical phase change would provide an important tool for understanding the pumping mechanisms in this source (Pardo et al. 2004).
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	where G is the gravitational constant, α is the angle between the radius vector to the star and the plane of the sky containing the central mass (0◦≤ α ≤90◦), and θ is the observed angular separation. The ar gets its maximum value in the case of α = 35◦. Calculations based on the observed parameters of IRS 15NE suggest that the acceleration caused by the enclosed mass is not signiﬁcant and far lower than the rate of the peak velocity variation. On the other hand, IRS 15NE seems to be a special case showing large variation of the peak velocity at the 43 GHz (Table 2). Therefore, the observed variation of the peak velocity of IRS 15NE seems unlikely to be caused by the gravitation acceleration by the dynamic center of the GC. The resolution of current observations is not high enough to explore whether the changes of the peak velocity are related to the variation in the spatial distribution of the masering regions.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	Seventeen 43-GHz SiO masers were identiﬁed on 2008 June 3 based on both radial velocity and position information. Among them eleven sources have been observed previously (Menten et al. 1997; Reid et al. 2003, 2007) and six are newly detected. One of the six new SiO masers, oﬀset (+0.′′9, −8.′′0) from Sgr A*, was associated with IRS 14NE, a known AGB star (Blum et al. 1996; Ott et al. 1999; Peeples et al. 2007). Following the nomenclature of the GC SiO masers adopted by Reid et al. (2007), the other ﬁve new SiO masers were named as SiO-18 (−77.6 km s−1), SiO-19 (−29.4 km s−1), SiO-20 (−20.3 km s−1), SiO-21 (+13.8 km s−1) and SiO-22 (+33.6 km s−1) in a sequence of ascending velocity.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	We detected fourteen 43-GHz SiO masers on 2008 October 3, among which eleven masers were in agreement with previous observations (Menten et al. 1997; Reid et al. 2003, 2007) and two masers (SiO-18 and SiO-22) were also found on 2008 June 3. The remaining one maser (VLSR = +80 km s−1) was found at (+24.′′0, +20.′′3) northeast of Sgr A* and represents a new detection. We named it as SiO-23.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	These detections at both 86 and 43 GHz are tabulated in Table 2. Together with previous observations, our new detections of SiO masers in the GC raise the total number of the 43- GHz SiO masers in the central 100′′ (or 4 pc) region to 22, increasing by ∼50 percent. Figure 2 shows composite spectra of the 43-GHz SiO masers derived from two observations in 2008. Figure 3 displays a sketch map of the 2-dimensional distribution of 43-GHz SiO masers in the inner 100′′ region of the GC. The contours in Figure 3 represent the continuum emission at 43 GHz. It is interesting to note that almost all sources with negative radial velocities have projected separations away from Sgr A* less than 25′′ (∼1 pc); while most sources with positive radial velocities are located at larger projected distances > 25′′. The origin for such a velocity distribution of the GC SiO maser stars is not clear. The IR observations of the massive stars in the GC suggest that the red giants and supergiants are belonging to a relax system and show isotropic 3-dimensional velocity distribution. Was the systematic change of the radial velocities of the SiO maser stars related to an intrinsically extraordinary property of the SiO maser stars (diﬀerent from those non-maser red giants) or just a selection eﬀect? More observations using the ATCA and EVLA with receiver bandwidth that covers the full velocity range of the GC SiO masers are necessary to obtain a complete view of the radial velocity and spatial distribution of SiO masers in the GC region and thus to solve this velocity distribution puzzle.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	Figure 1 compares the SiO maser line spectra of IRS 10EE from the quasi-simultaneous observations at 86 and 43 GHz on 2008 June 4 and 2008 October 3, and the quasi-simultaneous spectra of IRS 15NE at 86 and 43 GHz on 2008 June 4. The 43-GHz integrated ﬂux of IRS 10EE is about 3-5 times higher than that of 86 GHz in the two observations. Diﬀerent from IRS 10EE, the 43 GHz integrated ﬂux of IRS 15NE is comparable with that of 86 GHz. The diﬀerent behaviors of the SiO maser line emission in IRS 10EE (an OH/IR star) and IRS 15NE (an ordinary Mira) seem to be consistent with previous observations of massive late-type stars in the Galaxy (Nyman et al. 1986, 1993; Lindqvist et al. 1992) which showed that the 86-GHz SiO maser emission was weak in OH/IR stars compared with ordinary Mira variables. The high mass loss rate in OH/IR stars results in much turbulence to quench the 86-GHz maser excitation, as provides a possible explanation for relatively weak 86-GHz SiO maser emission in OH/IR stars. The radiative pumping should yield similar 43- and 86- GHz SiO intensities over broad ranges of SiO densities and abundances, while the collisional pumping is sensitive to the SiO column density and may give rise to diﬀerent spatial dis- tribution and diﬀerent intensities at 86 and 43 GHz (Phillips et al. 2003 and references therein). Although the low-resolution observations in the present paper are not allowed for inspecting the spatial correspondence of 43- and 86-GHz maser spots, the large discrepancy between the integrated ﬂux of IRS 10EE at the two frequencies would suggest that the ra- diative pumping might not be the dominant mechanism in this source. On the other hand, the turbulent outﬂows in IRS 10EE make it diﬃcult for the circumstellar envelope to sustain at a density of (4 − 6) × 109 cm−3, which is a scope for the 86-GHz maser emission to be eﬀectively ampliﬁed (Doel et al. 1995). For IRS 15NE, either radiative or collisional pump- ing may be responsible for the observed 43- and 86-GHz SiO maser emission. In order for the collisional mechanism to be at work for IRS 15NE, the H2 gas density in most regions of the envelope is required to be close to a critical density ∼ 5 × 109 cm−3 (Doel et al. 1995). Long-term monitoring of the maser line and infrared emissions from the GC SiO maser stars might be helpful for understanding the connection between the SiO maser variability and the IR stellar pulsation cycle.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	eighteen 43-GHz masers, of which seven are discovered for the ﬁrst time. Combined with previous observations (Reid et al. 2007), the total number of the 43-GHz SiO masers in GC region has been raised to 22. The current observations of the GC SiO masers mainly made with the ATCA and VLA are concentrated in the radial velocity range of ±100 km s−1. The updated broad-band observing facilities such as the ATCA (with the CABB) and the EVLA are excellent tools to gain a complete view of the distribution of GC SiO masers. In particular, monitoring the high-radial-velocity maser stars using the VLBI is especially important for constraining the mass distribution in the GC.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	In addition, the relative strength of SiO masers between 43 and 86 GHz from the simultaneous observations of IRS 10EE tends to favor a collisional pumping mechanism in IRS 10EE. For IRS 15NE, either radiative or collisional pumping mechanism, or both, might be at work.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	The Australia Telescope Compact Array is part of the Australia Telescope which is founded by the Commonwealth of Australia for operation as a National Facility managed by the CSIRO.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	This work was supported in part by the National Natural Science Foundation of China (grants 10625314, 10633010 and 10821302), the CAS/SAFEA International Partnership Pro- gram for Creative Research Teams and the Knowledge Innovation Program of the Chinese Academy of Sciences (Grant No. KJCX2-YW-T03), and sponsored by the Program of Shang- hai Subject Chief Scientist (06XD14024) and the National Key Basic Research Development Program of China (No. 2007CB815405, 2009CB24903). We thank Willem Baan for proof- reading and constructive comments on the manuscript. JL would like to thank helpful discussions with Lorant Sjouwerman on the ATCA observations, and Xi Chen, Yongjun Chen, Bing Jiang and Zhiyu Zhang for comments and suggestions on this work.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	Humphreys, E. M. L., Gray, M. D., Yates, J. A., Field, D., Bowen, G. H. & Diamond, P. J.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	a: channel width in unit of kHz; b: velocity resolution corresponding to the channel width ∆ν; c: velocity coverage; d: channels were binned with a width of 1 km s−1 at 86 GHz and 2 km s−1 at 43 GHz while mapping the SiO maser line emission, then the rms noise in individual channels was estimated.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	Fig. 1.— SiO maser spectra of IRS 10EE (upper panel) and IRS 15NE (lower panel) at 86 GHz (solid line) and 43 GHz (dashed line) from the ATCA observations. The velocity resolutions is 1 km s−1 at 86 GHz and 2 km s−1 at 43 GHz. The spectra were constructed at the pixel of peak brightness for SiO masers.
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	Fig. 2.— Composite spectra of 43 GHz SiO masers observed on 2008 June 3 (empty black bar) and October 3 (ﬁlled red bar).
Processed_ATCA_observations_of_SiO_masers_in_the_Galactic_ce.txt	Fig. 3.— The 7 mm ATCA continuum map of the GC region observed on 2008 June 3 with a synthesized beam of 8.′′4 × 2.′′3 (shown at the bottom left corner). The size of the region corresponds to an area of projected size 4 pc × 4 pc centered on the position of Sgr A. Late-type giant and supergiant stars with eighteen 43 GHz SiO maser (solid circle) emission and two 86 GHz SiO maser (open circle) emission are marked. The seven new detections are marked with .
Processed_ttbar_Spin_Correlations_at_D0.txt	The heaviest known elementary particle today, the top quark, has been discovered in 1995 by the CDF and D0 collaborations at the Tevatron collider at Fermilab. Its high mass and short lifetime, shorter than the timescale for hadronization, makes the top quark a special particle to study. Due to the short lifetime, the top quark’s spin information is preserved in the decay products. In this article we discuss the studies of t ¯t spin correlations at D0, testing the full chain from production to decay. In particular, we present a measurement using angular information and an analysis using a matrix-element based technique. The application of the matrix-element based technique to the t ¯t dilepton and semileponic ﬁnal state resulted in the ﬁrst evidence for non-vanishing t ¯t spin correlations.
Processed_ttbar_Spin_Correlations_at_D0.txt	c(cid:13) Copyright owned by the author(s) under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike Licence.
Processed_ttbar_Spin_Correlations_at_D0.txt	Discovered in 1995 by the CDF and D0 collaborations at the Tevatron proton antiproton (p ¯p) collider at Fermilab, the top quark [1, 2] is the heaviest known elementary particle today. The top quark mass is measured to be mt = 173.18 0.94 GeV [3]. The lifetime of the top quark is shorter than the time scale for hadronization, therefore it is the only quark that can be studied as bare quark. Due to the short lifetime, the spin information of the top quark is preserved in its decay products. While t ¯t pairs are produced unpolarized at the Tevatron, the correlation of the spin orientation of the top and the anti-top quark can be studied. By investigating the t ¯t spin correlations, we can study the full chain from production to decay and thus test the couplings in production and decay for possible new physics that would change the t ¯t spin correlation strength. In the following, two methods of measuring t ¯t spin correlations, performed by the D0 collaboration using Tevatron Run II data, are presented. The ﬁrst method explores angular distributions, while the second uses a matrix element based approach.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	ABSTRACT We present 5.5 and 9.0 GHz observations of a sample of seventeen GRB host galaxies at 0.5 < z < 1.4, using the radio continuum to explore their star formation properties in the context of the small but growing sample of galaxies with similar observations. Four sources are detected, one of those (GRB 100418A) likely due to lingering afterglow emission. We suggest that the previously-reported radio afterglow of GRB 100621A may instead be due to host galaxy ﬂux. We see no strong evidence for redshift evolution in the typical star formation rate of GRB hosts, but note that the fraction of ‘dark’ bursts with detections is higher than would be expected given constraints on the more typical long GRB population. We also determine the average radio-derived star formation rates of core collapse supernovae at comparable redshift, and show that these are still well below the limits obtained for GRB hosts, and show evidence for a rise in typical star formation rate with redshift in supernova hosts.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Long Gamma Ray Bursts (GRBs) are potentially a valu- able probe of star forming galaxies. Arising from the core collapse of a massive star (e.g. Woosley & Heger 2006), they indicate the presence of recent star formation and are typically associated with the most intensely star form- ing region of their host galaxies (Fruchter et al. 2006; Svensson et al. 2010). Optical and ultraviolet observations of those hosts galaxies have suggested that these tend to be relatively low in metallicity (e.g. Graham & Fruchter 2013) and mass (e.g. Castro Cer´on et al. 2010), with modest star formation rates (e.g. Christensen, Hjorth, & Gorosabel 2004; Savaglio, Glazebrook, & LeBorgne 2009). Thus they allow the selection and study of a galaxy population that may represent the low mass tail of the ultraviolet-selected, high redshift Lyman break galaxy luminosity function, and could plausibly dominate the ionising photon density of the Universe at early times (Chen et al. 2009; Tanvir et al. 2012).
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	radio luminosities, corresponding to star formation rates of 1.5-15 M⊙ yr−1 (Condon, Cotton, & Broderick 2002). More massive galaxies formed their stars at earlier times, typi- cally at z > 1, and often at a higher speciﬁc star forma- tion rate than is seen in the local Universe (see, for exam- ple, Cowie et al. 2004; Hopkins et al. 2010, and references therein). Thus the peak in the volume-averaged cosmic star formation rate, in AGN activity (indicative of co-evolution of spheroids) and in the number density of obscured star- bursts such as submillimeter galaxies (Chapman et al. 2005) all occur at z > 1, and the mass and luminosity of a typical starburst decrease sharply towards z = 0.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	However, there are indications that the host galaxies of GRBs do not necessarily trace this evolving, typical star forming population, but rather that the occurence of a GRB may be likely in galaxies with certain physical properties re- gardless of redshift (see Levesque 2014, for a recent review). Extensive work in the optical and near-infrared has now pro- duced a reasonably clear picture of the stellar populations in GRB hosts, in those examples suﬃciently close to carry out a detailed analysis (e.g. Savaglio, Glazebrook, & LeBorgne 2009; Svensson et al. 2010) and in samples designed to have well understood completeness and selection eﬀects (e.g. the ‘TOUGH’ sample, Hjorth et al. 2012; Micha lowski et al. 2012).
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	limited to damped Lyman-α system measurements at z > 2. Salvaterra et al. (2012) did ﬁnd evidence for a strong evolu- tion in either the luminosity or density distribution of GRB hosts and attributed this to a metallicity dependence in the GRB progenitor population.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	While Svensson et al. (2010) did not consider redshift evolution, they compared GRB host locations to those of the core-collapse supernova population (which should also trace star formation) and found that the former occupy galax- ies with higher optical surface brightness but less mass and smaller radii, as also suggested by Fruchter et al. (2006) and others. Work on supernovae themselves also support this. GRBs have been observationally associated with type Ic su- pernovae (Galama et al. 1998) but not with other types, and may well trace only a part of the Ic population, with addi- tional constraints (e.g. on the progenitor rotation) neces- sary to generate a GRB. In local supernova samples, SNe Ic occur in more metal rich, exceptionally star forming galax- ies, compared to the hosts of other core collapse supernovae (Kelly & Kirshner 2012), strengthening the case for GRBs picking out some subset of the full star forming galaxy pop- ulation.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	However this conclusion is by no means undisputed. As Coward et al. (2013) point out, a large number of ob- servational selection eﬀects as well as astrophysical biases must be taken into account. Initial suggestions that the GRB host population may show a sharp metallicity cut- oﬀ (e.g. Salvaterra et al. 2012; Kocevski, West, & Modjaz 2009) have been moderated by the detection of higher metal- licity systems (e.g. Levesque et al. 2010a; Kr¨uhler et al. 2012) with a more gradual metallicity-dependent fall oﬀ in GRB rate now suggested (e.g. Kocevski & West 2011; Robertson & Ellis 2012, and references therein). While some recent studies conclude that GRBs show a pronounced aver- sion to high metallicity (e.g. Graham & Fruchter 2013), there are also those who believe no such dependence ex- ists. Studies of the GRB host galaxy mass function and long GRB burst rate, based on a large sample observed by the GROND team, have suggested that no obvious biases in the host galaxy properties with metallicity, mass or redshift (Elliott et al. 2012) can be inferred, leaving this still very much an open question.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	So, if the degree to which the properties of GRB hosts track those of the typical star forming galaxy at a given red- shift is uncertain, converting a gamma ray burst rate to a volume averaged star formation density (and hence to the ionizing photon density vital for, amongst other things, hy- drogen and helium reionisation) is fraught with diﬃculties and assumptions. Not only is the detection of bursts, their follow-up and redshift conﬁrmation, rather an erratic pro- cess, with a diﬃcult to establish completion function (see Fynbo et al. 2009; Hjorth et al. 2012; Kr¨uhler et al. 2012), but the derived star formation rates rely on prescriptions relating the observed death of a single massive star in each host galaxy to the stellar birth rate (e.g. Robertson & Ellis 2012).
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	GRBs could arise in these dusty and relatively massive sys- tems (Berger et al. 2003; Priddey et al. 2006; Tanvir et al. 2004). The deﬁnition of a population of ‘Dark GRBs’ with low optical emission relative to their X-ray ﬂux (Jakobsson et al. 2004) again strengthened earlier sugges- tions that dust extinction may play a role that must be understood in the interpretation of these sources (e.g. see discussion in Groot et al. 1998; Djorgovski et al. 2001; Rol et al. 2005; Cenko et al. 2009; Greiner et al. 2011, and references therein). Including this ‘dark’ population leads to determining a much higher average dust content in the GRB host population than previously believed, but is still insuﬃcient to explain the early submillimeter re- sults (Perley et al. 2013). Studies of the lowest redshift bursts (Stanway, Davies, & Levan 2010) and other more recent work (e.g. Hunt et al. 2014; Perley & Perley 2013; Micha lowski et al. 2012; Wang, Chen, & Huang 2012) has suggested that the typical GRB host galaxy does not in fact host large quantities of extinguished star formation.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	As a result, samples of GRBs with measured ra- dio ﬂuxes remain small, with early work focused on de- tailed studies of individual targets (e.g. Kohno et al. 2005; Micha lowski et al. 2009; Stanway et al. 2011). Recent work by Micha lowski et al. (2012) represented a substantial im- provement in the sample of GRB hosts observed in the radio, presenting observations for 22 z < 1 sources, including those derived from the ‘TOUGH’ sample (Hjorth et al. 2012) and sources compiled from the literature. Of these, three were detected. While the constraints placed on individual sources varied signiﬁcantly from object to object, a stacking analysis based on the median redshift of the sample suggested that the typical undetected source may be relatively low in star formation rate (SFR<15 M⊙ yr−1) and have dust extinction (A2800µm < 6.7 mag).
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Recent improvements in correlator bandwidth, and hence sensitivity, at the major radio observatories has opened up the potential to change this, with observations reaching star formation rates within a factor of a few of the optical and ultraviolet derived values for GRB hosts now achievable with a reasonable investment of time. In this paper we extend the radio analysis of long GRB hosts to higher redshifts, and increase the number of observed sources, aiming to characterise their star formation with- out assumptions regarding their redshift evolution. We use the radio continuum to explore the star formation rates of the GRB host galaxies in our sample, in the context of the small but growing number of galaxies with similar observa- tions. We present 5.5 and 9.0 GHz observations of seventeen GRB host galaxies at z = 0.5 − 1.4. Of these, radio observa- tions of twelve sources are reported here for the ﬁrst time. Of the remaining sources, four have previously been observed at ∼1.4 GHz (Micha lowski et al. 2012). We report 5.5 GHz ﬂux constraints and signiﬁcantly improve the observed star formation rate limits in two cases, and present a tentative (2.3 σ) detection at 5.5 GHz of a third source. For the ﬁnal source we report a ﬂux limit consistent with the detection reported by Perley & Perley (2013).
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	the conversion from radio ﬂux to star formation rate. In sec- tion 5 we discuss our results in the context of previous work in this ﬁeld, before presenting our conclusions in section 6. Throughout, optical magnitudes are presented in the AB system. Host galaxies are typically referred to by the name of the associated GRB, but measurements are those of the host galaxy unless otherwise speciﬁed. Where necessary, we use a standard ΛCDM cosmology with H0 =70 km s−1 Mpc−1, ΩM = 0.3 and ΩΛ = 0.7.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Gamma ray bursts were selected from the data compila- tion table1 recording burst triggers from the Swift telescope (Gehrels et al. 2004). The primary criterion for sample se- lection was the existence of a known redshift for the GRB or its host galaxy, with suﬃcient precision to accurately derive rest-frame properties, and suﬃciently low to allow useful ra- dio constraints.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	As mentioned below and discussed in section 5.3, one initial goal of our ATCA programme (see section 2.2) was to evaluate whether ‘dark bursts’ diﬀered in their radio prop- erties from the wider population. As a result four of the sources in this study were identiﬁed as dark bursts (lying at z = 0.5 − 0.8), and other long GRBs at similar redshifts were preferentially selected for our ATCA observations so as to allow more direct comparison of their properties. The targets for our VLA observations (section 2.3) were chosen to lie at slightly higher redshifts, ideally at z = 1.0 − 1.4, to extend this work to earlier times and take best advantage of the higher sensitivity of the VLA.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Southern sources were prioritized so as to maximize cov- erage of the uv-plane in interferometric radio telescopes. If multiple targets satisﬁed these selection criteria and were accessible to the telescope during the observing programme, then the sources observed were selected at random. As a re- sult, and given the patchy spectroscopic follow-up of bursts (particularly in the early years of the Swift mission), the sample is unlikely to be complete, but nonetheless samples the known population. All sources all satisfy the T90 > 2.0 s criterion for a ‘long’ duration gamma-ray burst (where T90 is the interval within which 90% of the integrated counts from the burst are detected Kouveliotou et al. 1993).
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Our primary source for redshift information was the compilation provided in the Swift burst lookup table2. This table is hosted by Swift and seeks to compile data on these triggers and their follow up from GCN announcement and other sources, although its data is occassionally superceded by more detailed analyses. For two sources in our sample, GRB 060814 and GRB 071003, two highly discrepant red- shifts appear in the literature.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	GRB 060814 was selected for follow-up based on the z = 0.84 interpretation given in the Swift burst table, derived from Keck spectroscopy (Thoene, Perley, & Bloom 2007) at early times and appearing in several catalog pa- pers (e.g. Kann et al. 2010). However later observations have suggested that the faint host galaxy was misidentiﬁed and a second redshift z = 1.92 was reported, from line detec- tions in a star forming galaxy determined to be closer to the burst location (Salvaterra et al. 2012; Kr¨uhler et al. 2012). Throughout, we give results for the z = 1.92 hypothesis based on the higher probability that this is the host galaxy redshift. We note that the measured ﬂux limit would corre- spond to a substantially lower star formation rate if z = 0.84 is used instead (as ﬁgure 2 illustrates). If this target were detected, it would also be subject to some uncertainty as to whether the ﬂux arose from the GRB host, or the interven- ing z = 0.84 galaxy. Due to the likely high redshift (which renders comparison with the rest of our sample problem- atic) and degree of ambiguity, this source is omited from statistical analyses in section 5.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Radio continuum observations of 13 GRB host galaxies were undertaken in 2011 April 15-19 using the Australia Tele- scope Compact Array (ATCA)3. The targets for this pro- gramme were selected as Swift-detected long gamma-ray bursts lying at z ∼ 0.5 − 0.8, based on published obser- vations. Four targets were included which satisfy the condi- tions for being deﬁned as ‘Dark GRBs’ (i.e. optically faint, see section 5.3).
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Data were taken simultaneously at 5.5 and 9.0 GHz, with a correlator bandwidth of 2 GHz centered on each fre- quency. The telescope was in its most elongated 6A conﬁgu- ration, with maximum and minimum baselines of 5.938 and 0.337 km respectively, aligned East-West, and earth rota- tion synthesis was used to improve coverage of the uv-plane. Data were taken across the full range of possible hour angles, with a total on-source integration time per target of ∼140 minutes. A bright, compact source close on the sky to each target was used for phase calibration, and absolute ﬂux and bandpass calibration were determined through observations of PKS 1934-638 (the standard calibrator for ATCA).
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	close to the centre of the 8.5 arcmin primary beam. Given their redshifts and the typical identiﬁcation of GRB hosts as relatively low mass galaxies (e.g. Svensson et al. 2010; Savaglio, Glazebrook, & LeBorgne 2009) our GRB host tar- gets were expected to be compact or point sources in these observations. In each image, multiple faint sources were de- tected, and the images were ‘cleaned’ using standard pre- scriptions.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	The resulting 5.5 GHz synthesis images were inspected for any ﬂux excess in close proximity to the burst or its host galaxy (where known). The typical uncertainty in the en- hanced Swift XRT position for the burst (Evans et al. 2009) is 1.4 arcseconds, with a few cases where the constraints are poorer (∼3 arcsec). In four ﬁelds, a ﬂux excess exceed- ing the typical image noise by a factor of two were observed within the XRT error circle, or coincident with a known host galaxy. Either incomplete sampling of the uv-plane or the presence of a bright object close to the telescope beam can leads to ﬂux being scattered across a synthesised image in a correlated noise pattern. We inspected each target to ensure that there was no sign of large scale pattern noise that might explain a ﬂux excess at the target location.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	In order to rigourously determine the 5.5 GHz ﬂux (or limit) for each target and its associated uncertainty, we made use of the miriad task ‘imﬁt’, specifying that the software attempt to ﬁt a point source, matching the dimensions of the synthesised beam. In order to allow the most conservative limit on the host galaxy ﬂux (i.e. the maximum value per- mitted by the data), we permitted the algorithm to search around the XRT position by up to twice the width of the elongated synthesised beam in Right Ascension and a beam width in declination, suﬃcient to encompass both the XRT error circle and any plausible uncertainty associated with bandwidth smearing or the telescope pointing algorithm. This typically resulted in a search box of ∼ 10 × 3′′, and a measured oﬀset of peak ﬂux from the provided coordinates of < 2 arcseconds.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	For non-detections, we report the maximum permitted point source ﬂux and its associated error (scaled by the soft- ware to account for correlated noise), based on this pro- cedure in table 1. These represent conservative limits. The scaled errors, which account for the diﬃculty in determining a ﬂux excess above the structured background noise exceed the image root-mean-square noise (often quoted as a ﬂux uncertainty) by a factor > 5.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	For the four sources with an identiﬁed ﬂux excess within the search region, we explored the possibility of ﬁtting an ex- tended source, resulting in a poorer ﬁt to the data in all four cases. In each case, ‘imﬁt’ returned a position consistent with the peak ﬂux identiﬁed by visual inspection of the images. For consistency, and based on the large synthesised beam of the telescope at these frequencies, we report the point source ﬂux for these objects together with the reported estimate of uncertainty in table 1, and present 5.5 GHz radio maps of these objects in ﬁgure 1.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Of these objects, two (GRBs 100621A and 100418A) are well deﬁned, clear radio detections coincident with the GRB X-ray location. A third (GRB 050223) shows a signiﬁcant (3 σ) point source coincident with the host galaxy identiﬁed by (Pellizza et al. 2006). The fourth source (GRB 060729) presents an excess of 2.3 σ over the background noise, coin- cident with the burst location.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	While marginal as a claimed detection, this ﬂux excess remains when the data is sub-divided, or a diﬀerent weight- ing is used for image reconstruction. In addition to being co- incident with the X-ray location, the ﬂux excess is stronger than any of the likely noise features in its vicinity on the im- age, and is not straightforwardly attributable to the beam pattern extending from any other object. The probability of a noise ﬂuctuation of this strength occuring in our search region by chance is 3%. In our sample of 17 sources (combin- ing ATCA targets with the VLA observations discussed in section 2.3) we might expect 0.5 such ﬂuctuations (assuming gaussian statistics). We note that no targets in our sample show a negative ﬂux of comparable signiﬁcance. Thus, while it remains possible that this radio ﬂux is not attributable to the target, in the analysis that follows, we treat the mea- sured 5.5 GHz ﬂux as a radio detection, with the caveat that it might also be treated as a robust upper limit.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Observations at 9.0 GHz are more sensitive to atmo- spheric conditions and yield less sensitive constraints on star formation rate for all but the most peculiar of radio spectral slopes. We nonetheless inspect our 9.0 GHz for targets co- incident with the GRB locations, using the same procedure described above. Two of the 5.5 GHz-detected sources were also detected at 9.0 GHz. The remaining sources showed no evidence for a 9.0 GHz detection.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	We note that the constraints obtained above may be weakened if the sources are extended relative to the synthesised beam. The median eﬀective radius of GRB host galaxies is ∼ 1.7 kpc (Wainwright, Berger, & Penprase 2007). This corresponds to a projected size of just 0.3′′ at z = 0.5 and 0.2′′ at z = 1.0. Of 47 GRB hosts in the Wainwright, Berger, & Penprase (2007) sample, only one (GRB 011121 at z = 0.34) would be comparable to a 1.5′′ beam size in our imaging. While it is possible that a more extended host may exist within our relatively small sample, such studies of archival space-based imaging suggest that this is unlikely.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	The survey was extended to higher redshifts (z = 1.0 − 1.6) through the observation of four additional sources at the Karl G. Jansky Very Large Array (VLA). Observations were obtained in service mode in June-July 2012 and were associ- ated with programme 12A-279 (PI: Stanway). A total band- width of 2 GHz was centred at 5.5 GHz, with data collected in 1024 channels, each 2 MHz in width. Observations were performed in B conﬁguration.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Three targets - GRBs 071003, 080413B and 091208B - were each observed for ∼80 minutes on source, and a fourth target - GRB 100901A - for ∼160 minutes. As in the case of the ATCA observations above, a nearby point source was used for phase calibration on each target and absolute ﬂux calibration was provided by the standard calibrator 3C48. Data were ﬂagged and reduced using standard CASA tasks, producing a multi-frequency synthesis image for each target, using natural weighting. Both the improved instantaneous uv-plane coverage and the northern location of the VLA relative to the ATCA led to a more compact synthesised beam in these observations, as shown in table 1.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Table 1. Results from the radio observations at 5.5 and 9.0 GHz, taken at the ATCA and VLA. Objects are ordered by redshift, with the exception of GRB 060814 (see section 3). Fluxes and 1 σ uncertainties are given in µJy/beam. The beam size is given at 5.5 GHz and is half this size at 9.0 GHz. For measurements with an associated S/N> 2, the signal to noise is shown in the fourth column. The penultimate column gives the inferred star formation rate in solar masses per year, assuming the radio ﬂux is driven by star formation, as described in section 4, with 2 σ limits where appropriate. The detection of GRB 100418A is not interpreted as star formation (see section 3). If GRB 060814 is instead placed at z = 0.84 (see section 2.1), the limit instead corresponds to ∼ 107 M⊙ yr−1. For sources detected at 5.5 GHz, the ﬁnal column gives the observed-frame 9.0 GHz ﬂux in µJy. No other sources are detected at 9.0 GHz. The ﬁnal four sources were observed at the VLA, and 9.0 GHz data was not taken. Locations are given for the GRB afterglow and taken from the Swift XRT enhanced positions catalogue.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	we recover the known NVSS galaxy J200658+110024 at a separation of 7.3 arcmin from our target. As before, we per- mit a small search region, allowing for the X-ray position uncertainty, any pointing uncertainty or oﬀsets between the GRB or its host. None of the four targeted sources were indi- vidually detected. Given the non-detection, we measure the ﬂux within the GRB error circle, allowing for small (< 2′′) shifts in the centroid if these maximise the ﬂux, thus achiev- ing the most conservative constraints. Our resulting sensitive ﬂux limits as shown in table 1 and ﬁgure 2.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	imag- ing and the literature. It has been established that GRB 100621A was unusually bright. Greiner et al. (2013) anal- ysed the multi-wavelength afterglow properties, including observations taken with ATCA in Jun-Jul 2010 (4-27 days after burst) at the same frequencies as those observed here. In table 2 we report the ﬂuxes measured by Greiner et al. (2013) and compare with those measured in our observa- tions. Greiner et al. suggested that their 2010 Jul 17 obser- vations (measured at 17 days post-burst in the rest frame) might suggest a rapid fall oﬀ of the afterglow, or that the early time data might be explained by scintillation.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	The radio afterglow of long GRBs can be observed across an extended period of time. Unlike the optical af- terglow, which typically fades on a time scale of hours, ra- dio afterglows extend across days or months. The timescales for this afterglow depend on the observed frequency, since high frequency ﬂux peaks earlier than low frequency ﬂux. Chandra & Frail (2012) compiled data on 95 GRBs for which there were radio detections spanning from minutes to ∼500 days following the Swift gamma-ray trigger. In 63 cases there was suﬃcient data to determine a peak ﬂux epoch for at least one radio frequency. Only two sources have radio afterglows peaking at >50 days at ∼5 GHz or higher, with the vast majority peaking at <10 days. Typical sources show a fall in ﬂux of more than an order of magnitude between the ﬁrst few days and a rest frame epoch of 200 days post- burst, with the ﬂux evolution at 8.5 GHz being consistent with f (t) ∝ t−1 for times after peak ﬂux.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Our late time observations of GRB 100621A are entirely consistent with those measured almost 200 days earlier in the galaxy rest frame. The 2010 July observation remains discrepant but we note that this was also the noisiest obser- vation, and given the large errors, the results are still con- sistent within 2 σ of our measured value. Since the initial claim of a fading radio afterglow hinged entirely on this rel- atively noisy datapoint, our late time observation suggests that such an interpretation should be reconsidered.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Figure 1. 5.5 GHz radio maps of the four detections identiﬁed in table 1. We note again that the detection of the host of GRB 060729 is very tentative, at 2.3σ. Dark colours indicate higher ﬂuxes, and contours are indicated at ±2, 3 and 4 times the image noise level. The yellow cross and circle mark the Swift XRT location identiﬁed for the afterglow and its associated error circle. In the case of GRB 050223, the second cross indicates the position of the host galaxy identiﬁed by Pellizza et al. (2006), with which the radio detection appears to be associated. Axes are labelled in arcsecond oﬀsets from the XRT location, with North up. Note that the synthesised beam for GRB 100418A is very extended in declination, as shown in table 1.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	lows, while noting that it is inevitably subject to debate. Combining all available ATCA data in a single image (and thereby improving uv-plane coverage and sensitivity), we determine a best 5.5 GHz ﬂux for the host of GRB 100621A of 142±19 µJy, corresponding to a star formation rate of 119±16 M⊙ yr−1 (well within the range of star formation rates seen for the GRB population, see next section) - con- sistent with each of the reported individual measurements to within their errors. The source appears to be a point source at the resolution of the 3.2 × 1.5′′ synthesised beam.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Table 2. ATCA observations of GRB 100621A including those presented here for the ﬁrst time, and those reported by Greiner et al. (2013). Fluxes and 1 σ ﬂux errors (measured for a point source at the burst location) are given in µJy/beam. Time after the Swift GRB trigger is given in days, calculated in the source rest frame (observed frame / 1.5).
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	dominated by emission from the GRB host galaxy and so cannot be interpreted as arising from star formation.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Figure 2. The measured ﬂuxes (or 2 σ limits thereon) of the samples discussed in this paper. The four points at z > 1 are derived from VLA data, the remaining points from ATCA data. The ﬁve points at z < 0.5 are those presented in Stanway et al (2010) and are shown to provide a point of comparison. The ﬁve observations targeting the location of ‘dark’ bursts are indicated by asterisks (two detections, three limits). The contours indicate inferred radio-continuum-derived star formation rates in solar masses per year, as discussed in section 4. We note that the highest ﬂux detection, that of GRB 100418A, should not be interpreted in terms of star formation as discussed in section 3. GRB 060814, at z = 1.92, is omitted from this ﬁgure for clarity.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	dominated by the synchrotron component, which is emit- ted by electrons accelerated in supernovae and their rem- nants. Given that these events occur at the end of the life of massive stars, the establishment of radio continuum emis- sion shows a delay relative to the ultraviolet emission from young starbursts, but is directly related to star formation in older populations with ongoing star formation.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	ferred star formation rate from ﬁtting to the UV-optical pro- vides information primarily about the often blue, relatively young star forming population dominating the emission at these wavelengths, and does not provide any useful con- straint on the fraction of star formation taking place in ob- scured regions. A relevant example is that of GRB 100621A, which has a measured obscuration from the burst itself of Av = 3.8, while a ﬁt to the SED of the host yields just AV = 0.6 (Kr¨uhler et al. 2011). The star formation associ- ated with the burst clearly occured in, or behind, a heavily obscured region, and the optical data are not representative of this extinguished population.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Figure 3. ATCA and VLA observations of GRB 100418A placing the observation reported here in the context of the radio afterglow light-curve reported by Moin et al. (2013). Fluxes and 1 σ ﬂux er- rors (measured for a point source at the burst location) are given in µJy/beam. Observations at ∼5.5 GHz are shown in black (as- terisks: 5.5 GHz from ATCA, crosses: 4.9 GHz from VLA). Data at higher frequency (∼ 7 − 9 GHz) is shown in grey. The 5.5 GHz ﬂux reported here is shown in red. Time after the Swift GRB trigger is given in the source rest frame.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Following Berger et al., we set α = −0.6, appropri- ate for faint radio sources. Using a steeper spectral slope, α = −0.75, results in star formation rate estimates 25% higher but does not signiﬁcantly aﬀect our conclusions. We ﬁx the dust temperature and emissivity index at Td = 58 K and β = 1.35 respectively (again for comparison with previ- ous studies). The derived star formation rates (in M⊙ yr−1), are relatively insensitive to these dust parameters for obser- vations at ∼1-10 GHz.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	We over-plot contours on ﬁgure 2 to show the star for- mation rates equivalent to the given 5.5 GHz ﬂux as a func- tion of redshift. The last column in table 1 gives the inferred star formation rate (or limit thereon) in solar masses per year for each object, based on our 5.5 GHz observations, as- suming that the observed radio ﬂux is attributable to star formation. The measurement of GRB 100418A is not inter- preted as star formation, as discussed in section 3.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Given the very diﬀerent beam shapes and sensitivities of the ATCA data, we do not attempt to stack undetected sources. However we do stack our three undetected sources with VLA data at z ∼ 1.1. No source is detected in our stack, which has an rms ﬂux of 1.9 µJy, limiting the mean star formation rate of these three galaxies to < 16M⊙ yr−1 (2 σ).
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Table 3. WISE limits on observed 22µm (W 4) ﬂux and inferred star formation rate for the four sources with possible radio detec- tions. 2σ limits are given in each case.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	in the long wavelength WISE bands is also excited by star formation and can be used as a star formation rate indica- tor, although this relies on calibrations that have been deter- mined locally, and will vary as a function of redshift. At the redshift of our four detected sources, the 22 µm is probing the rest frame region around 14 µm - and so is sensitive to hot dust and PAH emission. While strong ﬂux in this region is indicative of a strong photon source - often star forma- tion - the conversion is dependent on the physical properties of the star formation region and is thus poorly calibrated. WISE is also a relatively shallow survey, suﬃcient to identify very luminous sources, but not typical star forming galaxies at z > 0.5. In table 3, we give ﬂux limits and equivalent star formation rate estimates for the four sources with proposed radio detections in this work. We apply the star formation rate conversion factor calculated from z = 0 − 0.3 galaxies (Lee, Hwang, & Ko 2013) for luminosities measured in the broad W 3 ﬁlter, which spans 11-16µm (observed) and thus provides a better calibration for our distant sources than the low redshift W 4 band. As can be seen, the limits obtained by WISE are consistent with our inferred radio star forma- tion rates for these sources, and also consistent with none of the larger sample being heavily obscured, submillimeter galaxy-like intense starbursts.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	suggest that this indicates that further investigation of this source, or a revision to the existing spectral energy distri- bution ﬁts, might be appropriate.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	In ﬁgure 4 we consider the GRB host star formation rate measurements and limits derived from radio emis- sion, as a function of In addition to the data reported in this paper and our previous work (Stanway, Davies, & Levan 2010), we also plot the radio- derived star formation rates reported and compiled by Micha lowski et al. (2012). To enable direct comparison, we adjust the star formation rates of Micha lowski et al. (2012) to account for the steeper radio spectral slope assumed in that paper, and to report 2 σ limits where a source is unde- tected.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	In total, ﬁgure 4 includes data on 40 long gamma ray burst host galaxies. Of these, a mere six have secure detec- tions (excluding the afterglow of GRB 100418A), and two of those lie at z < 0.15 (i.e. 2 Gyr lookback time), making clear the diﬃculty of observing these low mass, moderately star forming galaxies over substantial intervals of cosmic time.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Five sources have only modest constraints (a 2 σ limit on star formation rate >200 M⊙ yr−1) and a sixth source is detected with a very high star formation rate (∼400 M⊙ yr−1). This source, GRB 021211 is a pre-Swift burst, ar- guably with diﬀerent selection criteria and characteristics to the bulk of the sample. We note that the high ﬂux den- sity for this source (S1.4GHz = 330 ± 31 µJy) reported by Micha lowski et al. (2012) is inconsistent with that reported by (Hatsukade et al. 2012), who constrain the star formation rate to . 60 M⊙ yr−1 (2 σ, with our assumptions). Further investigation of this source is clearly needed.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	In summary, no more than 5 of the 40 (12 ± 6%) GRB hosts with radio constraints, have star formation rates ex- ceeding 200 M⊙ yr−1. This conﬁrms, and indeed strength- ens, the ﬁndings of Micha lowski et al. (2012) who analysed data on about half of this composite sample, and is some- what lower than suggested by early, pre-Swift and higher redshift work by Berger et al. (2003, who suggested a frac- tion of 20%) and Tanvir et al. (2004, who detected 3 of 21 sources). The reasons for this remain unclear, although small number statistics and the selection criteria for follow-up tar- gets no doubt contribute to the discrepency.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	whether this anomaly resolves with added data, or remains statistically signiﬁcant. If there is, in fact, a redshift depen- dence in the typical star formation rate of GRB host galax- ies, then it is possible that z ∼ 0.5 may represent a sweet spot in current telescope sensitivity relative to the typical star formation rate in GRB host galaxies. At lower redshifts, GRBs may be occuring in galaxies with lower typical star formation rates, while at higher redshift, the current radio limits are too weak to reliably probe this regime.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	We note that other than this anomaly, there is at best very modest evidence for redshift evolution in the radio de- rived star formation rates in individual GRB host galaxies. The measurement sample is dominated by upper limits at all redshifts, but these are less constraining with increas- ing redshift. Nonetheless the upper limits on typical sources at lookback times of 6-10 Gyrs (roughly corresponding to z = 0.65 − 1.85) constrain the typical GRB host radio- derived star formation rate to be no more than an order of magnitude higher than that observed at z < 0.15. This is comparable to the expected change in the typical star formation rate for galaxies of a given mass over the same interval due to the eﬀects of downsizing (see, for example, Speagle et al. 2014, and references therein), and so while we limit the redshift evolution of GRB hosts, the current sample is insuﬃcient to conclude that their evolution is signiﬁcantly diﬀerent to that of typical star forming galaxies over cosmic time.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Long GRBs are believed to be generated by the breakout of relativistic jets during the collapse of of a massive star at the end of its life (Woosley & Heger 2006). They have been observed to be coincident with Type Ic supernovae, and may be associated with core collapse supernovae (CCSN) more generally (see Fruchter et al. 2006, and references therein). However, as discussed in section 1, Fruchter et al. (2006) found that GRB hosts are typically less optically lu- minous than those of CCSNe, and also that the transient is more closely associated with the peak of the light dis- tribution in the host, based on optical photometry with the Hubble Space Telescope. They suggested that this might well represent a diﬀerence in the metallicity of the progenitor population, with GRBs biased towards the most massive and lowest metallicity stellar populations. This conclusion was broadly supported by Svensson et al. (2010), who em- ployed template ﬁtting of the spectral energy distribution of the host galaxies to make the same comparison. However Svensson et al. (2010) concluded that the diﬀerence between the two host populations was less pronounced in the blue optical bands which are dominated by star formation. This suggests the intriguing possibility that in the star formation- dominated radio continuum, the two populations may again appear similar.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Figure 4. The radio-inferred star formation rates (and limits) for GRB host galaxies with radio observations, shown as a function of lookback time. Large symbols indicate 5.5 GHz measurements taken as part of this programme from Stanway, Davies, & Levan (2010) at z < 0.5 (lookback <5 Gyr), and from the observations presented in this paper from ATCA (0.5 < z < 1.0, 5-7.5 Gyr) and the VLA (z > 1, >7.5 Gyr). Small symbols indicate the measurements reported or compiled by Micha lowski et al. (2012) - triangles mark measurements at < 5 GHz and diamonds those at higher frequency. Note that several sources have measurements at multiple frequencies, so not every point is independent. All limits are at the 2 σ level. Large green diamonds indicate the average rates inferred for the hosts of core collapse SNe in the GOODS ﬁelds as described in section 5.2, with horizontal bars showing the limits of each redshift bin. GRB 060814 lies at z = 1.92 and is omitted from this ﬁgure.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	We divide the sample of CCSN in Svensson et al. (2010) into four redshift bins, each with ∆z = 0.2, and combine the objects in each separately. The resulting stacked images are shown in ﬁgure 5 and their measured properties given in table 4. Radio properties of individual GOODS supernova hosts are given in the appendix and not discussed further here. In three redshift bins the average CCSN host galaxy is well detected. The z ∼ 0.85 sample has no clear detection, and the measured ﬂux is treated as an upper limit.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Figure 5. The stacked radio ﬂux at 1.4 GHz for core collapse SN hosts in the GOODS ﬁelds, as described in section 5.2. The average CCSN host is well detected in three redshift bins. The third bin, centred at z ∼ 0.85 does not show any evidence for a detection and is treated as an upper limit. The bright oﬀ-centre source in the z ∼ 0.65 stack is the result of bright neighbours in two individual images biasing the mean.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	magnitude less luminous than GRBs at comparable redshift (Kamble et al. 2014). Given the typical peak radio luminos- ity of 2 × 1027 ergs s−1 Hz−1 and a time decay of index ∼ −1 (Weiler et al. 2002), we would expect radio supernovae to have dropped well below our stacked detection limit within a year in the observed frame, at all redshifts considered here. Assuming then that the host galaxy ﬂux dominates over any supernova contribution, we convert the ﬂuxes to star for- mation rates, using the same prescription given earlier to account for the diﬀerent observed-frame frequency of the observations (1.4 GHz rather than 5.5 GHz).
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	Table 4. The number of sources and measured radio properties of the CCSN host galaxy stacks discussed in section 5.2.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	collapse supernovae host sample are consistent with an in- crease in the typical radio-derived star formation rate in these galaxies with increasing redshift.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	This behaviour appears consistent with the expected picture of cosmic ‘downsizing’ (Cowie et al. 1996). Not only does the volume-averaged star formation density of the Uni- verse increase by more than an order of magnitude with increasing lookback time from z = 0 to z = 1 but the size and star formation rate of galaxies in the star-forming pop- ulation also increases. Thus, if core-collapse supernova rate is directly proportional to star formation rate (i.e. the larger core collapse population is an unbiased tracer, as GRBs are not believed to be), the majority of events would be expected to be seen in more intense starbursts at z ∼ 1 than typically exist in the local Universe. As a result the detection of a single core-collapse supernova implies the existence of more young stars if it occurs at z ∼ 1 than if it occurs more locally, and any attempt to reconstruct the cosmic star formation history would require an adjustment for this evolution in the host galaxy population.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	If GRBs are indeed still more biased towards low metal- licity environments, as theoretically and observationally sug- gested (e.g. Woosley & Heger 2006; Levesque et al. 2010a; Graham & Fruchter 2013), but not universally accepted (e.g. Levesque et al. 2010b), then this eﬀect would likely be stronger, with lower star formation rates needed, on average, to produce a burst at high redshift than in the local Uni- verse, since a fraction of the higher metallicity star formation is not represented in the GRB rate. Further observations are clearly required to determine whether GRB hosts follow a similar pattern to that seen in the small sample of CCSN presented here, which could plausibly bias attempts to in- terpret the burst rate at a given redshift as a proxy for the volume averaged cosmic star formation rate. We note that the increase in radio-derived star formation rate between our lowest redshift bin and the highest is a factor of 7.7+3.5 −2.5. If the typical rates seen in GRBs in the local Universe increase by a similar factor, it would be doubtful whether our radio observations are suﬃciently sensitive to detect them.
Processed_Radio_Observations_of_GRB_Host_Galaxies.txt	In a pessimistic scenario, where GRB hosts are no more radio luminous than those of the larger CCSN population at high redshift, securing radio detections at the level of the core collapse supernovae hosts would require an improve- ment in sensitivity by a factor of 10 over the current obser- vations. While using a larger array (for example the VLA rather than ATCA) may allow progress to be made in rea- sonable integration times at z ∼ 0.5, the hosts at z ∼ 1 will need integration times approaching 8-10 hours per source to reach star formation rates of ∼10 M⊙ yr−1, making this an expensive study to undertake.