The Repeating Fast Radio Burst FRB 121102 as Seen on Milliarcsecond Angular Scales
B. Marcote, Z. Paragi, J. W. T. Hessels, A. Keimpema, H. J. van Langevelde, Y. Huang, C. G. Bassa, S. Bogdanov, G. C. Bower, S. Burke-Spolaor, B. J. Butler, R. M. Campbell, S. Chatterjee, J. M. Cordes, P. Demorest, M. A. Garrett, T. Ghosh, V. M. Kaspi, C. J. Law, T. J. W. Lazio, M. A. McLaughlin, S. M. Ransom, C. J. Salter, P. Scholz, A. Seymour, A. Siemion, L. G. Spitler, S. P. Tendulkar, R. S. Wharton
DDraft version January 5, 2017
Typeset using L A TEX twocolumn style in AASTeX61
THE REPEATING FAST RADIO BURST FRB 121102 AS SEEN ON MILLIARCSECOND ANGULAR SCALES
B. Marcote, Z. Paragi, J. W. T. Hessels,
2, 3
A. Keimpema, H. J. van Langevelde,
1, 4
Y. Huang,
5, 1
C. G. Bassa, S. Bogdanov, G. C. Bower, S. Burke-Spolaor,
8, 9, 10
B. J. Butler, R. M. Campbell, S. Chatterjee, J. M. Cordes, P. Demorest, M. A. Garrett,
12, 4, 2
T. Ghosh, V. M. Kaspi, C. J. Law, T. J. W. Lazio, M. A. McLaughlin,
9, 10
S. M. Ransom, C. J. Salter, P. Scholz, A. Seymour, A. Siemion,
15, 2, 19
L. G. Spitler, S. P. Tendulkar, and R. S. Wharton Joint Institute for VLBI ERIC, Postbus 2, 7990 AA Dwingeloo, The Netherlands ASTRON, Netherlands Institute for Radio Astronomy, Postbus 2, 7990 AA Dwingeloo, The Netherlands Anton Pannekoek Institute for Astronomy, University of Amsterdam, Science Park 904, 1098 XH Amsterdam, The Netherlands Sterrewacht Leiden, Leiden University, Postbus 9513, 2300 RA Leiden, The Netherlands Department of Physics and Astronomy, Carleton College, Northfield, MN 55057, USA Columbia Astrophysics Laboratory, Columbia University, New York, NY 10027, USA Academia Sinica Institute of Astronomy and Astrophysics, 645 N. A’ohoku Place, Hilo, HI 96720, USA National Radio Astronomy Observatory, Socorro, NM 87801, USA Department of Physics and Astronomy, West Virginia University, Morgantown, WV 26506, USA Center for Gravitational Waves and Cosmology, West Virginia University, Chestnut Ridge Research Building, Morgantown, WV 26505,USA Cornell Center for Astrophysics and Planetary Science and Department of Astronomy, Cornell University, Ithaca, NY 14853, USA Jodrell Bank Centre for Astrophysics, University of Manchester, M13 9PL, UK Arecibo Observatory, HC3 Box 53995, Arecibo, PR 00612, USA Department of Physics and McGill Space Institute, McGill University, 3600 University St., Montreal, QC H3A 2T8, Canada Department of Astronomy and Radio Astronomy Lab, University of California, Berkeley, CA 94720, USA Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USA National Radio Astronomy Observatory, Charlottesville, VA 22903, USA National Research Council of Canada, Herzberg Astronomy and Astrophysics, Dominion Radio Astrophysical Observatory, P.O. Box 248,Penticton, BC V2A 6J9, Canada Radboud University, 6525 HP Nijmegen, The Netherlands Max-Planck-Institut f¨ur Radioastronomie, Auf dem H¨ugel 69, Bonn, D-53121, Germany (Received 2016 December 10; Revised 2016 December 21; Accepted 2016 December 22)
Submitted to ApJABSTRACTThe millisecond-duration radio flashes known as Fast Radio Bursts (FRBs) represent an enigmatic astrophysicalphenomenon. Recently, the sub-arcsecond localization ( ∼
100 mas precision) of FRB 121102 using the VLA has ledto its unambiguous association with persistent radio and optical counterparts, and to the identification of its hostgalaxy. However, an even more precise localization is needed in order to probe the direct physical relationship betweenthe millisecond bursts themselves and the associated persistent emission. Here we report very-long-baseline radiointerferometric observations using the European VLBI Network and the 305-m Arecibo telescope, which simultaneouslydetect both the bursts and the persistent radio emission at milliarcsecond angular scales and show that they are co-located to within a projected linear separation of (cid:46)
40 pc ( (cid:46)
12 mas angular separation, at 95% confidence). Wedetect consistent angular broadening of the bursts and persistent radio source ( ∼ Corresponding author: J. W. T. [email protected], [email protected] a r X i v : . [ a s t r o - ph . H E ] J a n Marcote et al. constrained to be (cid:46) . (cid:46) . T b (cid:38) × K. Together, these observations provide strong evidence for a direct physical link between FRB 121102and the compact persistent radio source. We argue that a burst source associated with a low-luminosity active galacticnucleus or a young neutron star energizing a supernova remnant are the two scenarios for FRB 121102 that best matchthe observed data.
Keywords: radio continuum: galaxies — radiation mechanisms: non-thermal — techniques: highangular resolution — scattering
RB 121102 as Seen on Milliarcsecond Angular Scales INTRODUCTIONFast Radio Bursts (FRBs) are transient sources of un-known physical origin, which are characterized by theirshort ( ∼ ms), highly dispersed, and bright ( S peak ∼ . z ∼ .
5. However,further studies have shown that the transient sourcecontinues to vary in brightness well after the initialFRB 150418 burst detection, and can be explained bya scintillating, low-luminosity active galactic nucleus(AGN; e.g. Williams & Berger 2016; Giroletti et al.2016; Bassa et al. 2016; Johnston et al. 2016), whichleaves limited grounds to claim an unambiguous associ-ation with FRB 150418.Thus far, FRB 121102 is the only known FRB to haveshown repeated bursts with consistent dispersion mea-sure (DM) and sky localization (Spitler et al. 2014, 2016;Scholz et al. 2016). Recently, using fast-dump interfero-metric imaging with the Karl G. Jansky Very Large Ar-ray (VLA), FRB 121102 has been localized to ∼
100 mil-liarcsecond (mas) precision (Chatterjee et al. 2017). Theprecise localization of these bursts has led to associationswith both persistent radio and optical sources, and theidentification of FRB 121102’s host galaxy (Chatterjeeet al. 2017; Tendulkar et al. 2017). European VLBI Net-work (EVN) observations, confirmed by the Very LongBaseline Array (VLBA), have shown that the persistentsource is compact on milliarcsecond scales (Chatterjeeet al. 2017). Optical observations have identified a faint( m r (cid:48) = 25 . ± . . . z = 0 . ± . D L ≈
972 Mpc, and implying an angular diameterdistance of D A ≈
683 Mpc (Tendulkar et al. 2017). Thecentroids of the persistent optical and radio emission areoffset from each other by ∼ . ∼ . (cid:12) yr − (Tendulkar et al. 2017). In X-rays, XMM-Newton and
Chandra observations provide a 5- σ upper limit in the0 . L X < × erg s − (Chatterjeeet al. 2017).In the past few years, significant efforts have beenmade to detect and localize millisecond transient sig- nals using the EVN (Paragi 2016). This was made pos-sible by the recently commissioned EVN Software Cor-relator (SFXC; Keimpema et al. 2015) at the Joint In-stitute for VLBI ERIC (JIVE; Dwingeloo, the Nether-lands). Here we present joint Arecibo and EVN observa-tions of FRB 121102 which simultaneously detect boththe persistent radio source as well as four bursts fromFRB 121102, localizing both to milliarcsecond precision.In § § § § OBSERVATIONS AND DATA ANALYSISWe have observed FRB 121102 using the EVN at1.7 GHz and 5 GHz central frequencies (with a maxi-mum bandwidth of 128 MHz in both cases) in 8 observ-ing sessions that span 2016 Feb 1 to Sep 21 (Table 1).These observations included the 305-m William E. Gor-don Telescope at the Arecibo Observatory (which pro-vides raw sensitivity for high signal-to-noise burst de-tection) and the following regular EVN stations: Ef-felsberg, Hartebeesthoek, Lovell Telescope or Mk2 inJodrell Bank, Medicina, Noto, Onsala, Tianma, Toru´n,Westerbork (single dish), and Yebes. Of these antennas,Hartebeesthoek, Noto, Tianma, and Yebes only partici-pated in the single 5-GHz session.We simultaneously acquired both EVN VLBI dataproducts (buffered baseband data and real-time corre-lations) as well as wideband, high-time-resolution datafrom Arecibo as a stand-alone telescope. The Arecibosingle-dish data provide poor angular resolution ( ∼ Arecibo Single Dish Data
For the 1.7-GHz observations, Arecibo single-dish ob-servations used the Puerto-Rican Ultimate Pulsar Pro-cessing Instrument (PUPPI) in combination with theL-band Wide receiver, which provided ∼
600 MHzof usable bandwidth between 1150–1730 MHz. ThePUPPI data were coherently dedispersed to a DM =557 pc cm − , as previously done by Scholz et al. (2016).Coherent dedispersion removes the dispersive smear-ing of the burst width within each spectral channel.The time resolution of the data was 10 . µ s, and we Marcote et al.
Table 1.
Properties of the persistent radio source and detected FRB 121102 bursts from the Arecibo+EVN ob-servations. All positions are referred to the 5-GHz detection of the persistent source (RP026C epoch): α J2000 =5 h m . s , δ J2000 = 33 ◦ (cid:48) . (cid:48)(cid:48) . The observations conducted on 2016 Feb 1 (RP024A) and 2016 Sep 19(RP026A) did not produce useful data, and are not included here (see main text). The arrival times of the burstsare UTC topocentric at Arecibo at the top of the observing band (1690.49 MHz). All these bursts had gate widthsof 2–3 ms, and the quoted flux densities are averages over these time windows. We note that the larger error on theflux density of Burst ξ = F/ √ w (fluence divided by the square-root of the burst width).Session Epoch ν ∆ α ∆ δ S ν ξ (YYYY-MM-DD) (GHz) (mas) (mas) ( µ Jy) (Jy ms / )RP024B 2016-02-10 1 . . ± − ± ±
20 —RP024C 2016-02-11 1 . − ± − ± ±
14 —RP024D 2016-05-24 1 . ± − ± ±
40 —RP024E 2016-05-25 1 . ± ± ±
40 —RP026B 2016-09-20 1 . . ± . − . ± . ±
11 —RP026C 2016-09-21 5 . . ± . . ± . ±
14 —(YYYY-MM-DD hh:mm:ss.sss) (Jy)Burst . − ± − . ± . . ± . ∼ . . − . ± . . ± . . ± . ∼ . − ± . ± . ± . ∼ . . ± ± . ± . ∼ . . − ± . ± . recorded full Stokes parameters. At 5 GHz, the Arecibosingle-dish observations were recorded with the MockSpectrometers in combination with the C-band receiver,which together provided spectral coverage from 4484–5554 MHz. The Mock data were recorded in 7 par-tially overlapping subbands of 172 MHz, with 5.376-MHz channels and 65 . µ s time resolution. In addi-tion to the PUPPI and Mock data, the autocorrelationsof the Arecibo data from the VLBI recording were alsoavailable (these are restricted to only 64 MHz of band-width, see below).The Arecibo single-dish data were analyzed usingtools from the PRESTO suite of pulsar software (Ran-som 2001), and searched for bursts using standard pro-cedures (e.g., Scholz et al. 2016). The data were firstsubbanded to 8 × lower time and frequency resolutionand were then dedispersed using prepsubband to trialDMs between 487–627 pc cm − in order to search forpulses that peak in signal-to-noise ratio (S/N) at the Available at https://github.com/scottransom/presto expected DM of FRB 121102. This is required toseparate astrophysical bursts from radio frequency in-terference (RFI). For each candidate burst found us-ing single pulse search.py (and grouping commonevents across DM trials), the astrophysical nature wasconfirmed by producing a frequency versus time diagramto show that the signal is (relatively) broadband com-pared to the narrow-band RFI signals that can some-times masquerade as dispersed pulses.2.2.
Arecibo+EVN Interferometric Data
EVN data were acquired in real time using the e-EVNsetup, in which the data are transferred to the centralprocessing center at JIVE via high-speed fibre networksand correlated using the SFXC software correlator. Thehigh data rate of VLBI observations requires visibilitiesto be typically averaged to 2-s intervals during corre-lation, which is sufficient to study persistent compactsources near the correlation phase center, like the per-sistent radio counterpart to FRB 121102. However, wealso buffered the baseband EVN data to produce high-time-resolution correlations afterwards for specific times
RB 121102 as Seen on Milliarcsecond Angular Scales . ◦ away from FRB 121102). In the first five sessions(conducted in Feb and May) we scheduled phase ref-erencing cycles of 8 min on the target and 1 min onthe phase calibrator. Whereas this setup maximizedthe on-source time for burst searches, it provided lessaccurate astrometry due to poorer phase solutions. Thepulsar B0525+21 was also observed in one of these ses-sions following the same strategy (phase referenced usingJ0521+2112), in order to perform an empirical analysisof the derived astrometry in interferometric single-burstimaging. In the following three sessions in September,however, we conducted 5-min cycles with 3.5 min on thetarget and 1.5 min on the phase calibrator, improvingthe phase referencing, and hence providing more accu-rate astrometry. Two sessions failed to produce usefulcalibrated data on the faint target, and are not listedin Table 1. The first session (2016 Feb 1) was usedto explore different calibration approaches, whereas the2016 Sep 19 session was unusable because the largestEVN stations were unavailable and the data could notbe properly calibrated without them. An extragalactic ∼ α J2000 = 5 h m . s , δ J2000 =33 ◦ (cid:48) . (cid:48)(cid:48) ). This source has been used to acquirerelative astrometry of FRB 121102 during all the ses-sions and to provide a proper motion constraint.The 2-s integrated data were calibrated using stan-dard VLBI procedures within AIPS and ParselTongue(Kettenis et al. 2006), including a priori amplitude cal-ibration using system temperatures and gain curves foreach antenna, antenna-based delay correction and band-pass calibration. The phases were corrected by fringe-fitting the calibrators. The phase calibrator J0529+3209was then imaged and self-calibrated using the CaltechDifmap package (Shepherd et al. 1994). These correc-tions were interpolated and applied to FRB 121102,which was finally imaged in Difmap.The arrival times of the bursts were first identifiedusing Arecibo single-dish data, and then slightly re-fined for application to the EVN data. First using co-herently dedispersed Arecibo auto-correlations from theEVN data, we performed a so-called gate search by cre-ating a large number of short integrations inside a 50-ms window around the nominal Arecibo single-dish ar- The Astronomical Image Processing System, AIPS, is a soft-ware package produced and maintained by the National RadioAstronomy Observatory (NRAO). rival times. A pulse profile was then created for eachof the bursts by plotting the total power in the cross-correlations as a function of time. We then used thispulse profile to determine the exact time window forwhich the correlation function was accumulated, i.e. the‘gate’. We de-dispersed and correlated the EVN data toproduce visibilities for windows covering only the timesof detected bursts. We applied the previously describedcalibration to the single-pulse data and imaged them.The final images were produced using a Briggs robustweighting of zero (Briggs 1995) as it produced the mostconsistent results (balance between the longest baselinesto Arecibo and the shorter, intra-European baselines).Images with natural or uniform weighting did not pro-duce satisfactory results due to the sparse uv -coverage.The flux densities and positions for all datasets weremeasured using Difmap and CASA by fitting a circu-lar Gaussian component to the detected source in the uv -plane. RESULTS3.1.
Burst Detections
The EVN observations detect the compact and per-sistent source found by Chatterjee et al. (2017) with asynthesized beam size (FWHM) of ≈
21 mas × ≈ × ≈ − ◦ in both cases.On 2016 Sep 20, we detected four individual burstsin the Arecibo single-dish PUPPI data that overlapwith EVN data acquisition (Table 1). No bursts weredetected in the Arecibo PUPPI (1.7 GHz) or Mock(5 GHz) data from other sessions in which there aresimultaneous EVN observations that can be used forimaging the bursts. We formed images from the cali-brated visibility data for each burst, and measured theirpositions with respect to the persistent radio source.Figure 1 shows these positions together with the persis-tent source at 1.7 and 5.0 GHz. The nominal positionsmeasured for the four bursts are spread (cid:46)
15 mas aroundthe position of the persistent source, and we discuss thisscatter in § ξ = F/ √ w (flu-ence divided by the square-root of the burst width; e.g.,Cordes & McLaughlin 2003; Trott et al. 2013). Whenmatched filtering is done to detect a pulse (as we have The Common Astronomy Software Applications, CASA, issoftware produced and maintained by the NRAO.
Marcote et al. done, starting with the single-dish PUPPI data), thenthe S/N of the detection statistic, i.e. the output ofthe correlation, is proportional to ξ . Localization of thesource in an image (whether in the image or in the uv domain) will tend to have the same scaling if the uv dataare calculated with a tight gate (time window) aroundthe pulse so that it also scales as w . Using only flu-ence as a detection statistic is not appropriate becausea high-fluence, very wide burst can still be buried in thenoise, whereas a narrower burst with equivalent fluenceis more easily discriminated from noise. Burst ξ that is also afactor of > ∼ ∼ σ level. We thus findno convincing evidence that there is a significant off-set between the source of the bursts and the persistentsource. Since Burst ξ , is signif-icantly larger than for any of the other three bursts, itsapparent position is least affected by noise in the imageplane, as we explain in the following section, § Astrometric Accuracy
The astrometric accuracy of full-track (horizon-to-horizon observations) EVN phase-referencing is usuallylimited by systematic errors due to the poorly modeledtroposphere, ionosphere and other factors. These errorsare less than a milliarcsecond in ideal cases (Pradel et al.2006), but in practice they can be a few milliarcseconds.Given the short duration of the bursts (a few millisec-onds), our interferometric EVN data only contain a lim-ited number of visibilities for each burst, which resultsin a limited uv -coverage and thus very strong, nearlyequal-power sidelobes in the image plane (see Figure 3,bottom panel). In this case we are no longer limited onlyby the low-level systematics described above. The errorsin the visibilities, either systematic or due to thermalnoise, may lead to large and non-Gaussian uncertain-ties in the position, especially for low S/N, because theresponse function has many sidelobes. It is not straight-forward to derive the astrometric errors for data withjust a few-milliseconds integration. Therefore, we con-ducted the following procedure to verify the validity ofthe observed positions and to estimate the errors.First, we independently estimated the approximateposition of the strongest burst by fringe-fitting the burst h m . s . s . s . s α (J2000) + ◦ . . . . . δ ( J )
40 pc
Figure 1.
EVN image of the persistent source at 1.7 GHz(white contours) together with the localization of thestrongest burst (red cross), the other three observed bursts(gray crosses), and the position obtained after averaging allfour bursts detected on 2016 Sep 20 (black cross). Contoursstart at a 2- σ noise level of 10 µ Jy and increase by factors of2 / . Dashed contours represent negative levels. The colorscale shows the image at 5.0 GHz from 2016 Sep 21. Thesynthesized beam at 5.0 GHz is represented by the gray el-lipse at the bottom left of the figure and for 1.7 GHz at thebottom right. The lengths of the crosses represent the 1- σ uncertainty in each direction. Crosses for each individualburst reflect only the statistical errors derived from their S/Nand the beam size. The size of the cross for the mean po-sition is determined from the spread of the individual burstlocations, weighted by ξ (see text), and is consistent with thecentroid of the persistent source to within < σ . data and using only the residual delays (delay mapping;Huang et al. 2017, in prep.). With this method wehave obtained an approximate position of α J2000 =5 h m . s ( +0 . − . ) , δ J2000 = 33 ◦ (cid:48) . (cid:48)(cid:48) ( +0 . − . ),where the quoted errors are at the 3- σ level. Thismethod provides additional confidence that the image-plane detection of the bursts is genuine, since the posi-tions obtained with the two methods are consistent atthe 3- σ level.Next, we carried out an empirical analysis of single-burst EVN astrometry by imaging 406 pulses recordedfrom the pulsar B0525+21, which was used as a testsource in the 2016 Feb 11 session. PSR B0525+21 hastypical pulse widths of roughly 200 ms and peak fluxdensities of ∼ ξ ∼ . / ,compared to the range ξ ∼ . / measuredfor the 4 detected FRB 121102 bursts. Figure 4 shows RB 121102 as Seen on Milliarcsecond Angular Scales .
74 0 .
76 0 .
78 0 . ∆ t (s)1 . . . ν ( GH z ) − − ∆ t (ms)024681012 S . ( J y ) Ar-Ef − − ∆ t (ms)Ar-Mc − − ∆ t (ms)Ef-O8 Figure 2.
Top: Dynamic spectrum of the strongest burstdetected on 2016 Sep 20 (Burst a priori calibration. The measured peak bright-nesses are 11.9, 10.7, and 10.9 Jy, respectively, where theerror is typically 10 −
20% for a priori calibration. The rmson each baseline is 12, 80, and 300 mJy, respectively. the obtained positions for the different PSR B0525+21pulses along with the synthesized beam FWHM for com-parison. This demonstrates that the positional accuracyof the bursts increases for larger ξ . It shows that pulseswith ξ (cid:38) / are typically offset by less thanthe beam FWHM, whereas for ξ ∼ . / thescatter can be closer to 10 mas. This matches well withwhat we have observed in the four detected FRB 121102bursts (Table 1, Figure 1).While Burst ∼
10 masaround the average position. In Figure 1 we show theaverage position from the four observed bursts, weightedby their detection statistic ξ (Table 1). This average S . ( J y ) u v -distance (M λ ) − − θ ( ◦ ) − − ∆ α (mas)40200 − − ∆ δ ( m a s ) − − ∆ α (mas) − S . ( J y ) Figure 3.
Top: Amplitudes and phases of the obtainedvisibilities for the strongest burst observed on 2016 Sep 20(Burst uv -distance. Bot-tom: Dirty (left) and cleaned (right) image for the sameburst. The cleaned image has been obtained by fitting the uv -data with a circular Gaussian component. The synthe-sized beam is shown by the gray ellipse at the bottom rightof the figure. The coordinates are relative to the position ofthe persistent source obtained in the same epoch. position (separated ∼ σ .We therefore claim no significant positional offset be-tween the persistent radio source and the source of theFRB 121102 bursts.Finally, we place limits on the angular separation be-tween the source of the bursts and the persistent radiosource by sampling from Gaussian distributions withcenters and widths given by the source positions anduncertainties listed in Table 1 and deriving a numeri-cal distribution of offsets. Using the average burst po-sition compared to that of the persistent source, thisresults in a separation of (cid:46)
12 mas ( (cid:46)
40 pc) at the95% confidence level (or (cid:46)
50 pc at 99.5% confidencelevel). Although the positional uncertainties on individ-ual bursts are likely underestimated and non-Gaussian,as discussed previously, the effect of this should be mit-igated somewhat by using the average burst position,which includes an uncertainty determined by the scatter
Marcote et al.
10 0 − ∆ α (mas) − ∆ δ ( m a s ) F l u e n ce w i d t h − / ( J y m s / ) Figure 4.
Pulse localizations from the pulsar B0525+21 ob-served on 2016 Feb 11 at 1.7 GHz. A total of 406 pulses wereimaged. Systematic uncertainties inversely proportional tothe detection statistic ξ are observed. We note that onlypulses with ξ (cid:38) / are robustly localized to withinthe FWHM, and for pulses with ξ ∼ . / thescatter can be closer to 10 mas. in the separate burst detections, as also seen in Figure 4for B0525+21. We note that nearly identical separationlimits are obtained if we consider instead the position ofonly the strongest burst, Burst Measured Properties
Fitting the uv -plane data with a circular Gaussiancomponent shows that both the bursts and persistentradio source appear to be slightly extended. We mea-sure a source size of ∼ ± uv -plane. In the persistent sourcewe measure a similar value of 2–4 mas at 1.7 GHz inall sessions, whereas at 5.0 GHz we measure an angu-lar size of ∼ . . ∼ MHz) in the intensity, whichin principle could be due to scintillation or self noise.This will be investigated in more detail in a forthcomingpaper. S . ( µ J y ) S . ( µ J y ) Figure 5.
Top: Light curve of the persistent source at1.7 GHz during all the EVN epochs. The horizontal linerepresents the average flux density value and its 1- σ stan-dard deviation. Bottom: Light curve of the source withinthe 2016 Sep 20 epoch (last epoch in the top figure). Thevertical red lines represent the arrival times of the four de-tected bursts. We do not detect brightening of the persistentsource on these timescales after the bursts. The different epochs at which the persistent radiosource was observed allow us to obtain the light-curveof the compact source. Figure 5 shows the flux densi-ties measured for the five sessions at 1.7 GHz, whichare compatible with an average flux density of S . =177 ± µ Jy. The only session at 5.0 GHz shows a com-pact source with a flux density of S = 123 ± µ Jy.Assuming that the source exhibited a similar flux den-sity at 1.7 GHz compared with the day before, we in-fer a two-point spectral index α = − . ± .
24, where S ν ∝ ν α , for the source. Table 1 summarizes the ob-tained results. DISCUSSIONChatterjee et al. (2017) have shown that the persis-tent radio source is associated with an optical coun-terpart, which Tendulkar et al. (2017) show is a low-metallicity, star forming dwarf galaxy at a redshift of z = 0 . ± . RB 121102 as Seen on Milliarcsecond Angular Scales D L ≈
972 Mpcand D A ≈
683 Mpc, respectively, determined by Ten-dulkar et al. (2017). We show that the VLBI data aloneprovide further support to the extragalactic origin ofboth radio sources. Furthermore, we argue that thebursts and the persistent radio source must be phys-ically related because of their close proximity to eachother. We assume such a direct physical link in the fol-lowing discussion.4.1.
Persistent Source and Burst Properties
The results from all the EVN observations conductedat 1.7 GHz show a compact source with a persistentemission of ∼ µ Jy, which is consistent with theflux densities inferred at ∼ × larger angular scaleswith the VLA (Chatterjee et al. 2017). No significant,short-term changes in the flux density are observed afterthe arrival of the bursts or otherwise (Figure 5). Theaverage flux density of the persistent source implies aradio luminosity of L . ≈ × erg s − . The sin-gle measured flux density at 5.0 GHz corresponds to asimilar luminosity of L . ≈ × erg s − ( νL ν , witha bandwidth of 128 MHz at both frequencies). Addi-tionally, the 5.0-GHz data allow us to set a constrainton the brightness temperature of the persistent sourceof T b (cid:38) × K. Considering the measured radioluminosities and the current 5- σ X-ray upper limit inthe 0 . × − erg s − cm − (Chat-terjee et al. 2017, which implies L X < × erg s − )we infer a ratio between the 5.0-GHz radio and X-ray luminosities of log R X > − .
4, where R X = νL ν (5 GHz) /L X (2 −
10 keV) as defined by Terashima& Wilson (2003). The strongest observed burst exhibitsa luminosity of ∼ × erg s − at 1.7 GHz in the2-ms integrated data. These values imply an energyof ∼ (∆Ω / π ) erg, where ∆Ω is the emission solidangle.With the EVN sessions spanning a period of ap-proximately 7 months, we derived a constraint on theproper motion of the persistent source of − . < µ α < . − , and − . < µ δ < . − at a 3- σ confidence level. These values have been obtained afterremoving the offsets measured in the in-beam calibratorsource (VLA2, see § (cid:38) σ con-fidence level, setting a distance for the persistent source (cid:38) . (cid:46) . . × AU, given the distance of the source).The angular size measured for the source at 1.7 and5.0 GHz ( ∼ ∼ . ∼ ∝ ν − and thus we would expect a size of ∼ . ∼ (cid:46) µ as ν − / , the limit implied by syn-chrotron self-absorption for a frequency of optical depthunity ν ∼ T b (cid:46) K by synchrotron self-Compton radiation, the sizeis (cid:38) µ as ν − . These angles are too small to resolvewith VLBI but could be probed with interstellar scintil-lations.We have constrained the projected separation betweenthe source of the bursts and the persistent radio sourceto be (cid:46)
40 pc. Such a close proximity strongly suggeststhat there is a direct physical link between the burstsand the persistent source, as we now discuss in moredetail. 4.2.
Possible Origins of FRB 121102
The data presented here, in addition to the results pre-sented by Chatterjee et al. (2017) and Tendulkar et al.(2017), allow us to constrain the possible physical sce-0
Marcote et al. narios for the origin of FRB 121102. While the fact thatthe bursts are located within (cid:46)
40 pc of the persistentradio source strongly suggests a direct physical link, thepersistent radio source and the source of the FRB 121102bursts don’t necessarily have to be the same object. Weprimarily consider two classes of models that could ex-plain FRB 121102 and its multiwavelength counterparts:a highly energetic, extragalactic neutron star in a youngsupernova remnant (SNR) or an active galactic nucleus(AGN; or analogously a black hole related system witha jet). 4.2.1.
Young neutron star and nebula
As previously shown by Spitler et al. (2016), the re-peatability of FRB 121102 rules out an origin in a cat-aclysmic event that destroyed the progenitor source,e.g. the collapse of a supramassive neutron star (Fal-cke & Rezzolla 2014). The repetition and energetics ofthe bursts from FRB 121102 have been used to arguethat it comes from a young neutron star or magnetar(Cordes & Wasserman 2016; Lyutikov et al. 2016; Popov& Pshirkov 2016). At birth, the rapid spin of such (po-tentially highly magnetized) objects can power a lumi-nous nebula from the region evacuated by its SNR.The measured luminosity for the persistent radiosource cannot be explained by a single SNR or a pulsarwind nebula similar to those discovered thus far in ourGalaxy. A direct comparison with one of the brightestSNRs known, Cas A (300 yr old; Baars et al. 1977; Reedet al. 1995), shows that we would expect an emissionwhich is ∼ D L ≈
972 Mpc. In the case of the CrabNebula, the expected flux density would be even fainter( ∼ . T b consistent with the persistent radio source. However,neither the SFR of ∼ (cid:12) yr − nor the size ofthe region of 250–360 pc of Arp 220 (Anantharamaiahet al. 2000) agrees with the properties of the persistentsource associated with FRB 121102 (0 . (cid:12) yr − and (cid:46) . < Active galactic nucleus / accreting black hole
Models have been proposed in which the bursts aredue to strong plasma turbulence excited by the relativis-tic jet of an AGN (Romero et al. 2016) or due to syn-chrotron maser activity from an AGN (Ghisellini 2016).It is also conceivable to have an extremely young andenergetic pulsar and/or magnetar near to an AGN (Pen& Connor 2015; Cordes & Wasserman 2016) – eitherinteracting or not.The persistent radio source is offset by ∼ . (cid:38) × M (cid:12) . This value would be hard to reconcilewith the fact that the stellar mass of the host galaxyis likely at least an order of magnitude less than that,and its optical spectrum shows no signatures of AGNactivity (Tendulkar et al. 2017).Alternatively, we could be witnessing a radio-loud,but otherwise low-luminosity AGN powered by a muchless massive black hole that accretes at a very low rate.This population is poorly known, but EVN observationsof the brightest low-luminosity AGNs (LLAGNs) in asample of Fundamental Plane outliers (i.e. radio-loud,with R X ∼ −
2) show that some of these have extended
RB 121102 as Seen on Milliarcsecond Angular Scales R X remains a mystery(Paragi et al. 2012). We note that there are other recentexamples of LLAGNs identified based on their VLBIproperties coupled with low-levels of X-ray emission andno signs of nuclear activity from the optical emissionlines (Park et al. 2016).Other possible associations, like a single X-ray binary(such as Cyg X-3; Merloni et al. 2003; Reines et al. 2011)or an ultraluminous X-ray nebula (such as S 26 and/orIC 342 X-1; Soria et al. 2010; Cseh et al. 2012), do notfit to the measured flux density of the persistent radioemission and/or the observed size by several orders ofmagnitude. CONCLUSIONSThe bursts of FRB 121102 have recently been associ-ated with a persistent and compact radio source (Chat-terjee et al. 2017) and a low-metallicity star formingdwarf galaxy at a redshift of z = 0 . ± . (cid:46)
12 mas ( (cid:46)
40 pc given the distance to thehost galaxy). This tight constraint – roughly an or-der of magnitude more precise localization compared tothat achieved with the VLA in Chatterjee et al. (2017) –strongly suggests a direct physical link, though the per-sistent radio source and the source of the FRB 121102bursts don’t necessarily have to be the same object. Al-though the origin of FRBs remains unknown, the datapresented here are consistent in many respects with ei-ther an interpretation in terms of a low-luminosity AGNor a young SNR powered by a highly energetic neutronstar/magnetar.We thank the directors and staff of all the EVN tele-scopes for making this series of target of opportunityobservations possible. We thank the entire staff of theArecibo Observatory, and in particular A. Venkatara-man, H. Hernandez, P. Perillat and J. Schmelz, for theircontinued support and dedication to enabling observa-tions like those presented here. We thank B. Stappersand M. Mickaliger for their support with simultaneouspulsar recording using the Lovell Telescope. We thankE. Adams, K. Kashiyama, N. Maddox, and E. Quataertfor useful discussions on plausible scenarios as well asO. Wucknitz and A. Deller for reviewing a draft ofthe paper. We thank F. Camilo for access to comput-ing resources. The Arecibo Observatory is operated by SRI International under a cooperative agreement withthe National Science Foundation (AST-1100968), andin alliance with Ana G. M´endez-Universidad Metropoli-tana, and the Universities Space Research Association.The European VLBI Network is a joint facility of inde-pendent European, African, Asian, and North Ameri-can radio astronomy institutes. Scientific results fromdata presented in this publication are derived from thefollowing EVN project codes: RP024 and RP026 (PIJ. Hessels). This research made use of Astropy, acommunity-developed core Python package for Astron-omy (Astropy Collaboration et al. 2013) and APLpy,an open-source plotting package for Python hosted at http://aplpy.github.com . B.M. acknowledges sup-port by the Spanish Ministerio de Econom´ıa y Com-petitividad (MINECO) under grants AYA2013-47447-C3-1-P, AYA2016-76012-C3-1-P, and MDM-2014-0369of ICCUB (Unidad de Excelencia ‘Mar´ıa de Maeztu’).J.W.T.H. is an NWO Vidi Fellow. J.W.T.H. and C.G.B.gratefully acknowledge funding from the European Re-search Council under the European Union’s SeventhFramework Programme (FP/2007-2013) / ERC GrantAgreement no. 337062 (DRAGNET). Y.H. would like toacknowledge the support of the ASTRON/JIVE Inter-national Summer Student Programme. ASTRON is aninstitute of the Netherlands Organisation for ScientificResearch (NWO). The Joint Institute for VLBI ERIC,is a European entity established by six countries andfunded by ten agencies to support the use of the Eu-ropean VLBI Network. S.C., J.M.C., P.D., M.A.M.,and S.M.R. are partially supported by the NANOGravPhysics Frontiers Center (NSF award 1430284). Workat Cornell (J.M.C., S.C.) was supported by NSF grantsAST-1104617 and AST-1008213. M.A.M. is supportedby NSF award
Marcote et al.
Facility:
EVN, Arecibo Observatory
Software:
AIPS, Difmap, ParselTongue, CASA, As-tropy, APLpy, PRESTOREFERENCES
Anantharamaiah, K. R., Viallefond, F., Mohan, N. R.,Goss, W. M., & Zhao, J. H. 2000, ApJ, 537, 613Astropy Collaboration, Robitaille, T. P., Tollerud, E. J.,et al. 2013, A&A, 558, A33Baars, J. W. M., Genzel, R., Pauliny-Toth, I. I. K., &Witzel, A. 1977, A&A, 61, 99Barth, A. J., Bentz, M. C., Greene, J. E., & Ho, L. C. 2008,ApJL, 683, L119Bassa, C. G., Beswick, R., Tingay, S. J., et al. 2016,MNRAS, 463, L36Briggs, D. S. 1995, in Bulletin of the AmericanAstronomical Society, Vol. 27, American AstronomicalSociety Meeting Abstracts, 1444Chatterjee, S., Law, C. J., Wharton, R. S., et al. 2017, inpressCordes, J. M., & Lazio, T. J. W. 2002, ArXiv Astrophysicse-prints, astro-ph/0207156Cordes, J. M., & McLaughlin, M. A. 2003, ApJ, 596, 1142Cordes, J. M., & Wasserman, I. 2016, MNRAS, 457, 232Cseh, D., Corbel, S., Kaaret, P., et al. 2012, ApJ, 749, 17Falcke, H., K¨ording, E., & Markoff, S. 2004, A&A, 414, 895Falcke, H., & Rezzolla, L. 2014, A&A, 562, A137Ghisellini, G. 2016, ArXiv e-prints, arXiv:1609.04815Giroletti, M., Marcote, B., Garrett, M. A., et al. 2016,A&A, 593, L16Harris, D. E., Zeissig, G. A., & Lovelace, R. V. 1970, A&A,8, 98Huang, Y., Keimpema, A., Wen, Z., et al. 2017, in prepJohnston, S., Keane, E. F., Bhandari, S., et al. 2016, ArXive-prints, arXiv:1610.09043Keane, E. F., Johnston, S., Bhandari, S., et al. 2016,Nature, 530, 453Keimpema, A., Kettenis, M. M., Pogrebenko, S. V., et al.2015, Experimental Astronomy, 39, 259Kettenis, M., van Langevelde, H. J., Reynolds, C., &Cotton, B. 2006, in Astronomical Society of the PacificConference Series, Vol. 351, Astronomical Data AnalysisSoftware and Systems XV, ed. C. Gabriel, C. Arviset,D. Ponz, & S. Enrique, 497K¨ording, E., Falcke, H., & Corbel, S. 2006, A&A, 456, 439Kulkarni, S. R., Ofek, E. O., & Neill, J. D. 2015, ArXive-prints, arXiv:1511.09137Lorimer, D. R., Bailes, M., McLaughlin, M. A., Narkevic,D. J., & Crawford, F. 2007, Science, 318, 777 Lunnan, R., Chornock, R., Berger, E., et al. 2014, ApJ,787, 138Lyutikov, M., Burzawa, L., & Popov, S. B. 2016, MNRAS,462, 941Merloni, A., Heinz, S., & di Matteo, T. 2003, MNRAS, 345,1057Miller-Jones, J. C. A., Wrobel, J. M., Sivakoff, G. R., et al.2012, ApJL, 755, L1Murase, K., Kashiyama, K., & M´esz´aros, P. 2016, MNRAS,461, 1498Natarajan, I., Paragi, Z., Zwart, J., et al. 2016, ArXive-prints, arXiv:1610.03773Paragi, Z. 2016, ArXiv e-prints, arXiv:1612.00508Paragi, Z., Shen, Z. Q., de Gasperin, F., et al. 2012, inProceedings of the 11th European VLBI NetworkSymposium & Users Meeting. 9–12 October, 2012.Bordeaux (France), 8Park, S., Yang, J., Oonk, J. B. R., & Paragi, Z. 2016,ArXiv e-prints, arXiv:1611.05986Pen, U.-L., & Connor, L. 2015, ApJ, 807, 179Perley, D. A., Quimby, R. M., Yan, L., et al. 2016, ApJ,830, 13Petroff, E., Barr, E. D., Jameson, A., et al. 2016, PASA, 33,e045Piro, A. L. 2016, ApJL, 824, L32Plotkin, R. M., Markoff, S., Kelly, B. C., K¨ording, E., &Anderson, S. F. 2012, MNRAS, 419, 267Popov, S. B., & Pshirkov, M. S. 2016, MNRAS, 462, L16Pradel, N., Charlot, P., & Lestrade, J.-F. 2006, A&A, 452,1099Ransom, S. M. 2001, PhD thesis, Harvard UniversityReed, J. E., Hester, J. J., Fabian, A. C., & Winkler, P. F.1995, ApJ, 440, 706Reines, A. E., Sivakoff, G. R., Johnson, K. E., & Brogan,C. L. 2011, Nature, 470, 66Romero, G. E., del Valle, M. V., & Vieyro, F. L. 2016,PhRvD, 93, 023001Scholz, P., Spitler, L. G., Hessels, J. W. T., et al. 2016,ApJ, 883, 177Shepherd, M. C., Pearson, T. J., & Taylor, G. B. 1994, inBAAS, Vol. 26, Bulletin of the American AstronomicalSociety, 987–989Soria, R., Pakull, M. W., Broderick, J. W., Corbel, S., &Motch, C. 2010, MNRAS, 409, 541
RB 121102 as Seen on Milliarcsecond Angular Scales13