The mode switching of PSR B2020+28
Z. G. Wen, N. Wang, W. M. Yan, J. P. Yuan, Z. Y. Liu, M. Z. Chen, J. L. Chen
TThe mode switching of PSR B2020+28
Z. G. Wen • N. Wang • W. M. Yan • J. P. Yuan • Z. Y. Liu • M. Z. Chen • J. L. Chen
Abstract
This paper reports on polarimetric radia-tion properties based on the switching modes of nor-mal PSR B2020+28 by analysing the data acquiredfrom the Nanshan 25-m radio telescope at 1556 MHz.With nearly 8 hours quasi-continuous observation, thedata presented some striking and updated phenomena.The change of relative intensity between the leadingand trailing components is the predominant feature ofmode switching. The intensity ratio between the lead-ing and trailing components are measured for the in-dividual profiles averaged over 30 seconds. It is foundthat there is an excess of high ratios over the normaldistribution, which indicates that two modes exist inthe pulsar. The distribution of abnormal mode has anarrower width indicating that the abnormal mode ismore stable than the normal mode. A total of 76 modeswitching events are detected in our data. It spends89% in the normal mode and 11% in the abnormalmode. The intrinsic distributions of mode timescalesare constrained with power-law distributions. The sig-nificant difference in the index of the duration distribu-tion between normal and abnormal modes possibly in-
Z. G. WenN. WangW. M. YanJ. P. YuanZ. Y. LiuM. Z. ChenJ. L. Chen Xinjiang Astronomical Observatory (XAO), Chinese Academyof Sciences, Urumqi 830011, China. [email protected] University of Chinese Academy of Sciences, 19A Yuquan road,Beijing, 100049, China. Key laboratory of Radio Astronomy, CAS, Nanjing, 210008,China. Department of Physics & Electronic Engineering, YunchengUniversity, 044000, Yuncheng, Shanxi, China dicates that the timescale for the abnormal mode to getstable is shorter than that for the normal mode. Thefrequent switching between both modes may indicatethat the oscillations between different magnetosphericstates are rapid.
Keywords
Stars: neutron — pulsars: individual: po-larization : mode switching: PSR B2020+28
Pulsars are rapidly rotating, extremely dense neutronstars that emit radio frequency electromagnetic radia-tion from regions above their magnetic polar caps. Asa pulsar rotates, the radiation originating from theseregions sweeps through space much like a beam of alighthouse. It sweeps across our line of sight once perrotation, hence we can receive the pulse signal. Thepulsar presents a regularly spaced pulses with a repe-tition respectively stable period (hereafter P ). In thiscase, the rotation frequency ν = 1 /P . The individ-ual pulses are very weak and vary with time signif-icantly. In order to produce a stable profile, it re-quires the coherent addition of many hundreds or eventhousands of pulses together. However, some pulsarsshow two or more patterns of averaged pulse profiles,which is called mode switching (or mode changing) phe-nomenon. Since the discovery of mode switching inPSR B1237+25 (Backer 1970), this phenomenon hasbeen seen in a few dozens of pulsars. However, theintegrated profiles of these pulsars are quite complex,e.g. PSR B1237+25 has 5 components (Backer 1970;Rankin 1986), PSR B0329+54 has triple components(Rankin 1986; Chen et al. 2011). The recent obser-vations of PSRs B0723+26 (Sobey et al. 2015) andB0943+10 (Bilous et al. 2014) with LOFAR show tworelatively stable states (bright and quiet modes). The a r X i v : . [ a s t r o - ph . H E ] M a y forced Markov process was recently analysed by Cordes(2013) to model state changing pulsars.The radio emission from pulsars is highly polarized(Lyne & Smith 1968). The mean pulse profile and thepolarization measurements can yield a wealth of infor-mation, not only about the emission process itself, in-cluding the pulse emission mechanism, the beaming ofpulsar radiation and the geometry of the system, butalso about the medium through which it propagates.The integrated pulse profiles can be described withcore/double-cone geometric model (Rankin 1983). Theobserved polarization position angle (hereafter P A )variations can be approximately described by the Ro-tating Vector Model (RVM) (Radhakrishnan & Cooke1969). This model results from the idea that the radia-tion is polarized in the plane of curvature of field linesemanating from a magnetic pole on the star (Kome-saroff 1970). For a simple dipole field, the observedPA variation is then determined by the projected di-rection of the magnetic axis as the star rotates. Therapid swings often observed near the profile midpointimply that magnetic axis is nearly aligned with the ob-server’s line of sight at that profile phase. The observedPA swings are various. Yan et al. (2011) presentedpolarization profiles of 20 pulsars with smooth, con-tinuous and discontinuous PA variation. Discontinu-ities of approximately 90 ◦ are often observed (Manch-ester et al. 1975; Backer & Rankin 1980; Stinebringet al. 1984; Han et al. 2009), and these are interpretedas resulting from overlapping emission from orthogo-nally polarized emission modes (McKinnon & Stine-bring 1998; Gangadhara 1997). Such orthogonal po-larization modes may have resulted from the radiationemitted by positrons and electrons while moving alongthe curved magnetic field lines (Gangadhara 1997), orgenerated as the wave propagates through the pulsarmagnetosphere (Petrova 2001). The circular polariza-tion is usually relatively weaker than the linear polar-ization. It is most often associated with the central orcore component of the profile, often with a sense rever-sal near the profile mid-point (Rankin 1983).PSR B2020+28 is a relatively strong pulsar with fluxdensity of 38 mJy at 1400 MHz. The profile of thispulsar is relatively simple, having only two resolvedcomponents separated by 0.027 in pulse phase. Theperiod P is 0.3434 seconds, and its first derivative is1 . × − s s − . Correspondingly, its characteristicage ( τ = 2 . × years) is no more than the pulsarmedian. It was discovered with the east-west arm ofthe Bologna Cross telescope at 408 MHz (Bonsignori-Facondi et al. 1973). Polarimetric observations wereobtained with the 76-m Lovell telescope at Jodrell Bankat radio frequencies centred around 230, 400, 600, 920, 1400, 1600 MHz, which gives the W , W , W e , L , | V | , V and polarization profiles respectively (Gould &Lyne 1998). There exist two orthogonal modes of polar-ization that have position angles separated by 90 ◦ andopposite senses of circular polarization (Cordes et al.1978; Stinebring et al. 1984). This pulsar also showssignificant pattern change and is complicated by inter-stellar scintillation (Wang et al. 2001). However, theprofiles of normal and abnormal modes have not beenpublished so far. The radiation of various propertiesfrom pulsars are explained most naturally by a simplepicture in which the observed radiation is produced bythe acceleration of charged particles streaming outwardalong open filed lines above the poles of an essentiallydipolar magnetic field.In this paper we show a detailed investigation of theemission behavior of PSR B2020+28. In section 2, wedescribes the observing system and our observations.Section 3 presents data analysis and results. The im-plications of the results and conclusions are discussedin section 4. The observations of PSR B2020+28 were carried out us-ing the Nanshan 25-m radio telescope on 2012 June 12.A dual-channel cryogenic receiver was used to receiveorthogonal circular polarizations at a centre radio fre-quency of 1556 MHz, and with total bandwidth of 512MHz. The system temperature, T sys = T rec + T sky + T spi (in which T rec , T spi and T sky are the receiver, spilloverand the sky noise temperature, respectively), is ap-proximately 32 K. An ortho-mode transducer (OMT)was used to resolve the electromagnetic wave into left-handed circular ( L ) and right-handed circular ( R ) ba-sis modes. To determine the relative gain of the twopolarization channels and the phase between them, acalibration signal is injected at an angle of 45 ◦ to thefeed probes. The two independent polarizations werethen amplified and down-converted to an intermedi-ate frequency in the range of 0 −
512 MHz with a lo-cal oscillator at 1300 MHz. The band-limited signalswere fed to an updated back-end signal processing sys-tem, the third-generation Digital Filterbank System(DFB3), since 2010. After conversions from analoguevoltages to digital signals at the Nyquist rate with 9-bitsampling, the DFB3 uses field-programmable gate ar-ray (FPGA) processors to produce a maximum of 8192polyphase filterbank frequency channels and averagedpulse profiles. In order to obtain enough signal-to-noiseratio (S/N), about 90 pulse periods ( 30 s) were aver-aged and a time resolution of 512 bins per pulse pe-riod was used in our observations. The data lasting nearly 8 hours from 1024 frequency channels were thenrecorded for off-line processing, which contain 75400single pulses.The data were calibrated carefully. The averagedpulses were obtained by de-dispersing the data at a dis-persion measure (DM) of 24 . − . Four Stokesparameters were recorded and have been corrected fordispersion, interstellar Faraday rotation and various in-strumental polarization effects following the method de-scribed by Yan et al. (2011). The data are analysed of-fline using the PSRCHIVE package (Hotan et al. 2004)and corrected for parallactic angle and the orientationof the feed. The four Stokes parameters are accessiblein function (1), IQUV = | L | + | R | Re ( L ∗ R )2 Im ( L ∗ m ) | R | − | L | (1)where * indicates a complex conjugate (Lorimer &Kramer 2005). The structure and brightness of individual pulses areobserved to vary significantly, but the average of manyhundreds of individual pulses is usually stable, leadingto a characteristic profile that is often unique to pulsar.However, some pulsars exhibits two or more discreteand well-defined pulse profile morphologies, and theyswitches abruptly, which is known as mode-switching.The change of relative intensity between the differentcomponents is the predominant feature of mode switch-ing (Chen et al. 2011).At 1.5 GHz, the average pulse profile of PSRB2020+28 is relatively simple, having double well re-solved main emission components (leading and trailingcomponents), which is common to many pulsars. Theyare connected by a pronounced bridge emission. Theouter boundaries of main components are confined at10% of pulse peak of the average profile, while the innerboundary is defined as the minimal intensity betweenthem. The examples of average pulse profile (solid line)and phase boundaries (vertical dotted lines) are shownin Fig. 1. The polarization characteristics of the meanpulse profile provide a framework for understanding theemission processes in pulsars. The averages of linearpolarization (dashed line) is a maximum in the leadingcomponent at 30% compared with 22% in the trailingcomponent. Two components are depolarized relativeto low-frequency observations, but the effect is most no-ticeable in the leading component of the pulse (Cordes et al. 1978; Stinebring et al. 1984). The averages cir-cular waveform (dash-dotted line) is almost negligiblein the leading component and a sense reversal slightlytrails the leading nulls in the linear polarization. Theextrema of the circular polarization occur under thelinear maximum in the trailing component. The posi-tion angle rotation (lower panel) in the saddle regionshows a S-shaped sweep with total swing only about 30degrees. Whereas, two peaks in the leading and trailingregions show up.
Fig. 1
Polarization waveform of PSR B2020+28. The uni-formly weighted values of the Stokes parameters are aver-aged over nearly 8 hours, with total intensity I , circular po-larization V , and the linear polarization L = ( Q + U ) / (where Q and U are the linear Stokes parameters), are dis-played in solid, dot-dashed and dashed lines respectively.The boundaries are plotted with dotted lines to distinguishleading and trailing components. The position angle as afunction of pulse phase, χ = 1 / − ( U/Q ), is given inthe lower panel.
The dynamic spectrum is shown in Fig. 2. It seemslike that the intensity fluctuations with time and fre-quency are very similar to the case of the observa-tions by Wang et al. (2001). The scintles (the redpatches) have a characteristic timescale (the scintilla-tion timescale) of three thousand seconds. The hori-zontal stripes in the dynamic spectrum are because thebad frequency channels are affected by Radio Frequency
Interference (RFI). At the edges of the observing fre-quency, there is no useful data as the bandpass rollsoff. The scintillation bandwidth (the characteristic fre-quency scale of the scintiles) is clearly less than the totalbandwidth of the observation. The effects of such largefluctuations can be addressed by subtracting a runningmean of length one fifth of the scintillation timescale.
Fig. 2
The dynamic spectrum of scintillations for PSRB2020+28. Here the measured signal to noise of the pulsarsignal is plotted as a function of both time and frequency.The horizontal and vertical stripes in the dynamic spectrumare where data were affected by RFI.
The whole 8-hour observations with 912 scintillation-corrected sub-integrations, each is averaged over 30 sec-onds, are shown in Fig. 3. The variation of relativepeak intensity ( R I ) between the leading and trailingcomponents are shown in the right panels. It varies fre-quently in a wide range from 0.5 to 1.6. No strongregularity is found easily by visual inspection. Dis-crete Fourier Transform (FFT) and auto-correlationwere done on the whole 8-hour time sequence of R I val-ues, and no presence of periodic signals was detected ineither methods.The distribution of R I over nearly 8 hours is shownin Fig. 4, which extends over an extremely wide range.The histogram consists of a broad Gaussian componentand a long tail component. As we can see, the best-fit solution gives us a combination of two Gaussian compo-nents corresponding to two emission modes. The pulseswith R I in the range of 0 . ∼ .
13 are classified asnormal mode, the remainder with R I in the range of1 . ∼ .
52 are classified as abnormal mode. The distri-bution of the abnormal mode is considerably narrowerthan that of the normal mode, which may suggest thatit is more stable than the normal mode.
Fig. 4 R I distribution versus integration time of 30 secondsby using the data of nearly 8 hours observation. The solidline represents the fitting based on the combination of twoGaussian components (dashed lines). The lower panel showsthe fit residuals. Representative example of a sequence of 15 sub-integrated pulse profiles, arbitrarily normalized, isshown in the left panel of Fig. 5. The mode switchingphenomenon is presented apparently as R I varies withtime, which is shown in the right panel. The integratedpulse profiles from 445 to 447 switch from normal modeto abnormal mode. The averaged pulse profiles of bothmodes are shown in the insets. Through a careful in-spection on the pulse profiles, it is noted that an ad-ditional component appears on the leading edge of thefirst main component for the abnormal pulse profiles.The integrated polarization properties of the normal(left panel) and abnormal (right panel) modes are givenin Fig. 6. The flux density in the abnormal mode is 1.3times stronger than that of the normal mode. Note Fig. 3
The 912 scintillation-corrected sub-integrations observed at 1556 MHz, each was integrated over 30 seconds. Two-hour blocks are plotted horizontally. The left panels plot the sub-integrations with a colorful density proportional to thereceived mean flux. Time runs left to right across the pulse window (pulse phase) and bottom to top with sub-integrationnumber. The distribution of the peak intensity ratio between the leading and the trailing components versus the sub-integration number are shown in the right panels. It varies from 0.5 to 1.6. The threshold of normal and abnormal modesare indicated with dashed lines. that the polarization profiles are slightly different aswell. For the abnormal mode, the linear polarizationintensity in the leading component increases by a fac-tor of almost 45%, the circular polarization intensitydecreases by a factor of almost 11%, compared withthose of the normal mode. There are no significantvariations in PA between both modes.To further investigate the relative properties of nor-mal and abnormal modes, the durations in each modeare obtained. The histograms for the timescales of thenormal (left) and abnormal (right) modes are shownin Fig. 7. A curve of best fit was calculated assum-ing a single power law (dashed line). The exponent, α , derived from the power-law best fits for the normaland abnormal modes are -1.0(1), -2.7(1) respectively.There is a significant difference in the power-law indexof the duration distribution between normal and abnor-mal modes.In order to further to analysis the pulse-to-pulse fluc-tuation properties. The sub-integrated pulse profileswhich are in the same range of R I are superposed ina group. The details of the division are shown in Ta-ble 1. The time proportions of seven groups are all morethan 6 per cent of the total observation time, which al-low high S/N to analysis the polarization properties.The intervals of R I between two adjacent groups arenearly 0.1. The averaged polarization waveforms ofseven groups are shown in Fig. 8. It is found thatthe seven groups all have the same pulse widths. Theintermediate ”saddle” regions in total and linear po-larization profiles are relatively broad compared withthe widths of the lobes, especially the linear polariza-tion profiles. With the increase of R I , the fractionalcircular polarization is decreased by a factor of almost50%, and a factor of 26% in the linear polarization isincreased. There is no much difference between theseseven groups in PA variation.Following to the calculation of R I , the linear po-larization intensity ratio ( R L ) is derived between twopeaks in the linear polarization profile. The errors of R I and R L are derived from the standard deviation foreach group. By means of the error transfer formula,the errors of R L /R I are calculated. The intensity ra-tio ( R L ) is also changing with that of averaged pulseprofiles. And R I and R L are positive correlating afterfitting with a straight line (see Fig. 9), which indicatesthat the fractional linear polarization is constant. Ifthis correlation is confirmed in other pulsars, R L maybe also used as an indicator to identify the mode switch-ing phenomenon and to figure out emission mechanismmore than R I . Fig. 9
The positive correlation between R I and R L afterfitting with a straight line. Corresponding error bars denote3 σ errors. We report new results on the emission properties ofpulsar B2020+28 based on detailed analysis of 8-hourobservations at a center frequency 1556 MHz using theNanshan 25-m radio telescope. A total of 76 modeswitching events are detected. It spends 89% in thenormal mode and 11% in the abnormal mode.The major difference between the normal and ab-normal modes is the intensity ratio between the lead-ing and trailing components and the length of modingtimescales. The variation of R I has no strong regular-ity, and the mode switching phenomenon seems a ran-dom process. The distribution of R I for the abnormalmode is narrower than that of the normal mode by afactor of 65%, which may indicate that the abnormalmode is more stable than the normal mode. The intrin-sic timescale distributions, constrained for this pulsarfor the first time, provide valuable information to un-derstand the physics of mode switching phenomenon.The durations of abnormal mode are extremely short,which are less than 250 pulse periods. They may indi-cate that the timescale for the abnormal mode to getstable is shorter than that for the normal mode. Theshort durations of both modes are very common and long durations are much less, which may be representedby an elementary Markov process (Cordes 2013).The strong linear polarization is an outstandingcharacteristics of pulsars, which is usually associatedwith cone emission around magnetic field lines (Gedalin& Dzigan 2005). The stronger linear polarization inthe leading and trailing components than that in theintermediate ”saddle” region reveals the double-lobedstructure in the total intensity is from the conal emis-sions. The circular polarization often accompanies coreemission and is generally the strongest in the central or’core’ regions of a profile. The sense reversal often oc-curs near the middle of the profile (Rankin 1983). Butfor PSR B2020+28, the circular polarization changessenses at the leading component and reaches peak valueat the trailing component.The mechanisms of mode switching phenomenon areproposed by many authors. Pulsar switches betweendifferent magnetospheric states are likely to be causedby changes of particle current flow in the pulsar mag-netosphere (Lyne et al. 1971; Bartel et al. 1982). Insome cases, the pulse profile changes are also correlatedwith large changes in spin-down rates. (Lyne et al.2010). Changes between two distinct emission states inPSR J0742 − R I with two Gaus-sian components has smaller residuals than that withone. Furthermore, the whole data set are examinedcarefully, approximately third of pulse profiles show anadditional leading component in the intervals of abnor-mal mode. And this additional component is also de-tected in several normal pulse profiles. As shown in Fig.7, the duration of abnormal mode is relatively short, thedetected abnormal pulse profiles may be contaminatedby some normal pulses, so the additional weak leadingemission becomes blurred. Therefore, the association between the additional component with intervals of ab-normal mode is expected to exist in PSR B2020+28.The relative pulse-to-pulse total intensity fluctuationis identified in variable linear and circular polarization.The evolution of fractional polarization with the in-crease of R I was also identified. The stable PA vari-ations of different groups imply that the geometry ofthe pulsar emission beam remains constant, while theemission strength varies in different emission regions,as a result of the intensity variations of total and linearpolarization. It is noted that most points deviate fromthe straight line shown in Fig. 9. Such deviation maybedue to the error in the systematic polarization calibra-tion, or perhaps caused by the intrinsic variation.Mode switching is expected to occur over a broad-range of wavelengths (Sobey et al. 2015), even up toX-rays. Therefore, we would require multi-frequencysimultaneous single pulse observations to better under-stand these emission characteristics, which may lead toa better understanding of the magnetospheric emissionmechanism. Acknowledgements
We are grateful to the refereefor valuable suggestions. This work was supportedby National Basic Research Program of China grants973 Programs 2015CB857100 and 2012CB82180, thePilot-B project grant XDB09010203, and the WestLight Foundation of Chinese Academy of Sciences(WLFC) No.XBBS201422. WMY is supported byNSFC (11203063, 11273051) andWLFC (XBBS201123). JPY is supported by NSFC2012CB821801 and 11173041. We thank members ofthe Pulsar Group at Xinjiang Astronomical Observa-tory for helpful discussions.
Fig. 5
Representative example of a sequence of sub-integrated pulse profiles which contained both the normal and abnormalmodes, each averaged over 30 seconds. The 445th to the 447th integrated pulse profiles show switches to the abnormal mode.The dotted lines are the boundaries that divide the profiles into two components, leading and trailing. The right panelshows the distribution of the intensity ratio between the leading and the trailing components versus the sub-integrationnumber. The dashed line shows the threshold of both modes. The averaged pulse profiles of both modes are shown in theinsets.
Fig. 6
Integrated polarization profiles display for PSR B2020+28 in the normal (left) and abnormal modes (right).
Fig. 7
Histograms of the timescales of the normal mode (left) and the abnormal mode (right). The curves stand for theconstrained optimal power-law distributions. Table 1
The total intensity ratio ( R I ) and the linear polarization intensity ratio ( R L ) between the leading and the trailingcomponents of seven different groups. Also present their ratios and errors Group No. R I Range percentage R I σ R L σ R L /R I σ (a) < . . − . . − . . − . . − . . − . > . Fig. 8
Average Stokes parameter profiles of seven groups of sub-integrations according to the peak intensity ratio ( R I )between the leading and the trailing components of the total intensity. The intensity ratios of total and linear polarizationhave been given in Table 1. References
Backer D. C., 1970, Nature, 228, 1297Backer D. C., Rankin J. M., 1980, Astrophys. J. Suppl. Ser.,42, 143Bartel N., Morris D., Sieber W., Hankins T. H., 1982, As-trophys. J., 258, 776Bilous A. V., Hessels J. W. T., Kondratiev V. I., vanLeeuwen J., Stappers B. W., Weltevrede P., Falcke H.,Hassall T. E., Pilia M., Keane E., Kramer M., GrießmeierJ.-M., Serylak M., 2014, Astron. Astrophys., 572, A52Bonsignori-Facondi S. R., Salter C. J., Sutton J. M., 1973,Astron. Astrophys., 27, 67Chen J. L., Wang H. G., Wang N., Lyne A., Liu Z. Y.,Jessner A., Yuan J. P., Kramer M., 2011, Astrophys. J.,741, 48Cordes J. M., 2013, Astrophys. J., 775, 47Cordes J. M., Rankin J., Backer D. C., 1978, Astrophys. J.,223, 961Gangadhara R. T., 1997, Astron. Astrophys., 327, 155Gedalin M., Dzigan Y., 2005, Astron. Astrophys., 439, 23Gould D. M., Lyne A. G., 1998, Mon. Not. R. Astron. Soc.,301, 235Han J. L., Demorest P. B., van Straten W., Lyne A. G.,2009, Astrophys. J. Suppl. Ser., 181, 557Hotan A. W., van Straten W., Manchester R. N., 2004,Proc. Astron. Soc. Aust., 21, 302Keith M. J., Shannon R. M., Johnston S., 2013, Mon. Not.R. Astron. Soc., 432, 3080Komesaroff M. M., 1970, Nature, 225, 612Lorimer D. R., Kramer M., 2005, The Observatory, 125, 338Lyne A., Hobbs G., Kramer M., Stairs I., Stappers B., 2010,Science, 329, 408Lyne A. G., Smith F. G., 1968, Nature, 218, 124Lyne A. G., Smith F. G., Graham D. A., 1971, Mon. Not.R. Astron. Soc., 153, 337Manchester R. N., Taylor J. H., Huguenin G. R., 1975, As-trophys. J., 196, 83McKinnon M. M., Stinebring D. R., 1998, Astrophys. J.,502, 883Petrova S. A., 2001, Astron. Astrophys., 378, 883Radhakrishnan V., Cooke D. J., 1969, Astrophys. Lett., 3,225Rankin J. M., 1983, Astrophys. J., 274, 333Rankin J. M., 1986, Astrophys. J., 301, 901Sobey C., Young N. J., Hessels J. W. T., Weltevrede P.,Noutsos A., Stappers B. W., Kramer M., Bassa C., LyneA. G., Kondratiev V. I., Hassall T. E., Keane E. F., BilousA. V., Breton R. P., Grießmeier J.-M., Karastergiou A.,Pilia M., Serylak M., Veen S. t., van Leeuwen J., AlexovA., Anderson J., Asgekar A., Avruch I. M., Bell M. E.,Bentum M. J., Bernardi G., Best P., Bˆırzan L., BonafedeA., Breitling F., Broderick J., Br¨uggen M., Corstanje A.,Carbone D., de Geus E., de Vos M., van Duin A., DuschaS., Eisl¨offel J., Falcke H., Fallows R. A., Fender R., FerrariC., Frieswijk W., Garrett M. A., Gunst A. W., HamakerJ. P., Heald G., Hoeft M., H¨orandel J., J¨utte E., KuperG., Maat P., Mann G., Markoff S., McFadden R., McKay-Bukowski D., McKean J. P., Mulcahy D. D., Munk H.,Nelles A., Norden M. J., Orr`u E., Paas H., Pandey-Pommier M., Pandey V. N., Pietka G., Pizzo R., Polatidis A. G., Rafferty D., Renting A., R¨ottgering H., RowlinsonA., Scaife A. M. M., Schwarz D., Sluman J., SmirnovO., Steinmetz M., Stewart A., Swinbank J., Tagger M.,Tang Y., Tasse C., Thoudam S., Toribio C., VermeulenR., Vocks C., van Weeren R. J., Wijers R. A. M. J., WiseM. W., Wucknitz O., Yatawatta S., Zarka P., 2015, Mon.Not. R. Astron. Soc., 451, 2493Stinebring D. R., Cordes J. M., Rankin J. M., WeisbergJ. M., Boriakoff V., 1984, Astrophys. J. Suppl. Ser., 55,247Timokhin A. N., 2010, Mon. Not. R. Astron. Soc., 408, L41Wang N., Wu X.-J., Manchester R. N., Zhang J., Yusup A.,Zhang H.-B., 2001, Chin. J. Astron. Astrophys., 1Yan W. M., Manchester R. N., van Straten W., ReynoldsJ. E., Hobbs G., Wang N., Bailes M., Bhat N. D. R.,Burke-Spolaor S., Champion D. J., Coles W. A., HotanA. W., Khoo J., Oslowski S., Sarkissian J. M., VerbiestJ. P. W., Yardley D. R. B., 2011, Mon. Not. R. Astron.Soc., 414, 2087Zhang B., Qiao G. J., Lin W. P., Han J. L., 1997, Astro-phys. J., 478, 313