The precessing jets of classical nova YZ Reticuli
Dominic McLoughlin, Katherine M. Blundell, Steven Lee, Chris McCowage
MMNRAS , 1–12 (2020) Preprint 26 February 2021 Compiled using MNRAS L A TEX style file v3.0
The precessing jets of classical nova YZ Reticuli
Dominic McLoughlin, ★ Katherine M. Blundell, Steven Lee, , Chris McCowage , Department of Physics, University of Oxford, Keble Rd, Oxford OX1 3RH, United Kingdom Research School of Astronomy and Astrophysics, Australian National University, Canberra, ACT 2611 Anglo-Australian Telescope, Coonabarabran NSW 2357, Australia
Accepted 2021 February 23. Received 2021 February 23; in original form 2021 January 08
ABSTRACT
The classical nova YZ Reticuli was discovered in July 2020. Shortly after this we com-menced a sustained, highly time-sampled coverage of its subsequent rapid evolution withtime-resolved spectroscopy from the Global Jet Watch observatories. Its H-alpha complexexhibited qualitatively different spectral signatures in the following weeks and months. Wefind that these H-alpha complexes are well described by the same five Gaussian emissioncomponents throughout the six months following eruption. These five components appear toconstitute two pairs of lines, from jet outflows and an accretion disc, together with an addi-tional central component. The correlated, symmetric patterns that these jet/accretion disc pairsexhibit suggest precession, probably in response to the large perturbation caused by the novaeruption. The jet and accretion disc signatures persist from the first ten days after brightening– evidence that the accretion disc survived the disruption. We also compare another classicalnova (V6568 Sgr) that erupted in July 2020 whose H-alpha complex can be described analo-gously, but with faster line-of-sight jet speeds exceeding 4000 km/s. We suggest that classicalnovae with higher mass white dwarfs bridge the gap between recurrent novae and classicalnovae such as YZ Reticuli.
Key words: stars: jets – accretion discs – novae – shock waves – stars: individual:MGAB-V207,Nova Reticuli 2020,YZ Reticuli – stars: individual: V6568 Sgr,PNV J17580848-300537,Nova Sagittarii 2020 No.3
When classical novae erupt, their optical luminosities increase dra-matically and rapid spectral changes are observed which are exem-plified in their evolving Balmer H 𝛼 complexes. Changes occur onrapid and varied timescales ranging from hours to weeks. One suchclassical nova, YZ Reticuli, underwent an outburst in July 2020 andwe implemented a sustained, time-resolved, spectroscopic follow-up programme with the Global Jet Watch observatories, monitoringit extensively throughout the first few months after it was discov-ered. This revealed properties of key dynamical components thatdrive its observed evolution. YZ Reticuli constitutes a particularlyconvenient example of a very energetic classical nova, having broademission lines (FWHM ∼ − ). The high speeds in thisclassical nova make it possible to discern distinctly resolved com-ponents: we reduce the dimensionality of the data by fitting com-posite models of multiple Gaussians to the H 𝛼 profile. This revealsthat much of the dramatic spectral behaviour can be explained bysimple changes to these underlying components giving insights intothe underlying dynamical story.In Section 2, we outline our instrumentation and observations. ★ E-mail: [email protected] (DM)
We present our methodology in Section 3 and the results of ourfits in Section 4. In Section 5 we suggest a physical interpretationfor YZ Reticuli comprising a prominent and precessing accretiondisc, and jets, in the aftermath of the nova event. We confront thismodel for YZ Reticuli with data for another classical nova, PNVJ17580848-300537, which also erupted in July 2020, in Section 6.We summarise our conclusions in Section 7.
It is widely agreed that many classical nova eruptions ejectroughly spherical shells, which move at slow speeds of up to( ∼ − ). The evidence includes remnant shells of past no-vae (Krautter et al. 2002; Santamaría et al. 2020), and absorptionlines in the early spectra around maximum and immediately fol-lowing it. These absorption lines often show slow and fast com-ponents, known as the principal and diffuse systems (Mclaughlin1942; Williams 1992), and are linked to a slow, equatorial outflowand a faster polar outflow respectively. In some novae, these flowsare thought to collide (Williams & Mason 2010), causing shocks.Recent multiwavelength studies have shown correlations betweengamma-ray detections interpreted as shocks between the variousejecta, and changes in the light curve (Aydi et al. 2020). Slavin et al. © a r X i v : . [ a s t r o - ph . H E ] F e b D. McLoughlin et al. (1995) present evidence that in addition to ellipsoidal shells, longtails of emission can appear. A popular idea is that in addition toan initially approximately spherical slowly-ejected shell, there is asignificantly faster outflow along the polar axes which can disruptthe shell; this picture is clearly supported in the case of DQ Her infigure 3 of Toalá et al. (2020).
The launch of jets in astrophysical systems undergoing accretionis widely accepted. However, typical cataclysmic variable systemshave been thought to be accreting at too low a rate to launch jetsaccording to established disc theory. Classical novae could thereforepresent an important window on the emergence and survival of thejet phenomenon, because they have recently undergone disruptiveeruption and associated enhanced accretion, which under favourablecircumstances may put these systems over the threshold for the jet-launching mechanism to be triggered (Lasota & Soker 2005).There have been theoretical predictions and suggested obser-vations of jets in classical novae: Retter (2004) discusses the the-oretical viability of jets, using the V1494 Aql eruption in 1999(Iijima & Esenoglu 2003) as an example. Csák et al. (2005) findthat there is evidence in favour of a link between the transitionphase in novae and the establishing of an accretion disc. Kawabataet al. (2006) show jet-like winds in V475 Scuti, a moderately fastbut otherwise unremarkable classical nova that erupted in 2003.They used Okayama Astrophysical Observatory spectropolarimetryto show that the nova had blue and red jets with a constant positionangle of linear polarisation.In Nova V5668 Sgr, Harvey et al. (2018) presents a jet-likebiconical outflow as an explanation for the profiles of [O iii], al-though these model-dependent fits do not capture the asymmetryfully (see their figures 7 & 8). In the following paper in this series(McLoughlin et al, in prep), we present Nova V5668 Sgr in thecontext of the model presented in this paper. We also note Ribeiroet al. (2013) present modelling of a fast bipolar outflow based ontheir fits to emission line structures.There is also evidence of jets in recurrent novae. Sokoloskiet al. (2008) present direct evidence of highly collimated outflowsand lobes in radio images of recurrent nova RS Oph, having milli-arcsecond resolution, captured by the VLBA. These lobes werefound to be moving across the sky with proper motions implyingunderlying jet launch speeds of a few thousand km s − . Figure5 of Taylor et al. (1989) shows a radio map of the object, withclearly aspherical lobes apparent within the first few months of the1985 outburst. Rupen et al. (2008) observe synchrotron jets withthe VLBA, noting that this implies emission is occurring far fromthe central source. This has important consequences, as it impliesthat fast jets may extend beyond the inner system long before thephotosphere fully recedes. Skopal (2008) presents spectra of the H 𝛼 complex of RS Oph, and their figure 3 showing this is qualitativelysimilar to what we see here in YZ Reticuli, although they do notpresent any fitting analysis or decomposition of the complex.Darnley et al. (2017) discuss the jets in the recurrent novaM31N 2008-12a, which is near the Chandrasekhar limit, and thus ac-cretes at a high rate, hence enabling the very rapidly (yearly or faster)recurring eruptions. They found broad shoulders around H 𝛼 (offsetfrom line centre by about 4800 km s − and about 5900 km s − ) andattributed these to fast bipolar outflow - intriguingly faster than thejets we might expect from a lower mass white dwarf such as thosein classical nova systems.V1721 Aquilae, a very fast classical nova, within three days of discovery, was found to host an accretion disc that was either re-established fast, or was indeed never fully disrupted (Hounsell et al.2011). It seems probable then, that there exist generalised jet-likeoutflows in erupting nova systems, with faster and more collimatedjets occurring in recurrent systems containing more massive whitedwarfs, through very fast novae such as V1721 Aquilae with mod-erate jets, to more typical classical novae, such as the one for whichwe present evidence in this paper. YZ Reticuli was discovered by Robert H. McNaught (Coon-abarabran, NSW, Australia) at magnitude 5.3 on 2020 July 15.590UT (CBET 4811). It is also identified as Nova Reticuli 2020, andMGAB-V207, with Right Ascension of 03 58 29.55 and Declination −
54 46 41.2 (J2000). There is a light curve from the All Sky Auto-mated Survey for SuperNovae (hereafter ASAS-SN) for this objectwhich first shows evidence of brightening at 2020-07-08.1708099,or JD 2459038.67081, which is the epoch we take as the zero point.For the rest of this paper, we use the convention that +10.3d shouldbe interpreted as 10.3 days after this zero point. There is a four-daygap in ASAS-SN observations (magnitude 15 at 2020-07-04.18,magnitude 6 at 2020-07-08.17), so it cannot be precluded that itcould have started brightening as early as 2020-07-04.18 (Aydi2020a).Figure 1 shows the AAVSO light curve of YZ Reticuli forthe 125 days following eruption. The decline time of this classicalnova is 𝑡 =
15 days. YZ Reticuli exhibits a plateau-type light curve(Strope et al. 2010) between ∼ +
30 d and ∼ +
60 d, which may re-late to the plateau in the light curve of classical nova V407 Cyg,interpreted by Hachisu & Kato (2012) as a surviving accretion discemerging out of the receding photosphere.This classical nova was detected in gamma-rays between2020-07-10 and 2020-07-15 by Fermi-LAT (Kwan-Lok 2020).(Sokolovsky 2020) detected it in X-rays using NuSTAR, and foundevidence for non-solar abundances of N , O and/or Fe , in 67 ks oftotal exposure commencing 2020-07-17.98. A light curve is avail-able from ASAS-SN for the 2000 days preceding eruption. We didnot detect periodicities in the nova’s quiescent luminosity in thatdataset using the CLEAN algorithm together with deconvolutionfollowing Roberts et al. (1987), but note that it remains at around16 magnitudes.Songpeng (2020a) notes that the X-ray luminosity of YZ Reti-culi is significantly lower than expected, which may hint at theexistence of a spatially extended accretion disc which could be ob-scuring a significant portion of the hot surface of the white dwarf.Coronal line emission is reported by Songpeng (2020b), who alsocomment on the absence of dust and on the appearance of double-peaked profiles with deep troughs at the centre that we explore inSection 5.3. This X-ray phase, deemed by Sokolovsky et al. (2020)to be the super-soft phase, commenced at +
58 d, indicating the epochby which the photosphere has receded enough to see hydrogen burn-ing on the surface of the white dwarf.This classical nova is in the GAIA DR2 catalogue (Prustiet al. 2016; Brown et al. 2018) (source identification number: [ . ± . ] mas, giving a distance of 2 . + . − . kpc (Bailer-Jones et al. aavso.org https://asas-sn.osu.edu/light_curves/ee130388-139d-4f3a-ab8a-7427a7545e21 MNRAS000
58 d, indicating the epochby which the photosphere has receded enough to see hydrogen burn-ing on the surface of the white dwarf.This classical nova is in the GAIA DR2 catalogue (Prustiet al. 2016; Brown et al. 2018) (source identification number: [ . ± . ] mas, giving a distance of 2 . + . − . kpc (Bailer-Jones et al. aavso.org https://asas-sn.osu.edu/light_curves/ee130388-139d-4f3a-ab8a-7427a7545e21 MNRAS000 , 1–12 (2020) ets in classical nova YZ Reticuli Classical nova V6568 Sgr, also known as Nova Sagittarii 2020 −
30 05 37.6 (J2000). It was classified as a classicalnova by Aydi (2020b) using high-resolution optical spectroscopyfrom the Southern African Large Telescope (SALT).
This paper focuses on a sequence of observations of YZ Reticulifrom + .
29 d to the date of submission predominantly from theAquila spectrographs in the Global Jet Watch observatories butsupplemented by some LHires spectra at the Rainbow Observatory(IAU 𝛼 so as to enhance the signal-to-noise in the less intense parts ofthe spectrum, and these spectra are used in the analysis of weakerlines. The Global Jet Watch observatories are a network of five telescopesdistributed in longitude (South Africa [GJW-SA], Chile [GJW-CL], east Australia [GJW-OZ], Western Australia [GJW-WA], India[GJW-IN]), described in Blundell et al (in prep). This configurationallows for sustained observations of an astrophysical target whilethe Earth rotates. Each observatory contains a Ritchey-Cretien car-bon fibre telescope having a 0.5m-diameter primary mirror, and afibre-fed spectrograph known as Aquila (Lee et al, in prep). The predominant data product used in this paper is a set of 5129optical spectra captured by the Aquila spectrographs. Our Aquilaobservations start soon after discovery and last until the present day.The Aquila spectra have a spectral resolution of 𝑅 ∼ − ◦ C by a Peltier cooler.
The Shelyak LHires instrument is a slit spectrograph, which uses anNe lamp for wavelength calibration arcs. It uses a ZWO ASI camera(model ASI294MM) cooled to − ◦ C. While the slit width is se-lectable between several preset widths, it was set to 23 µ m for theseobservations. This instrument was commissioned to support theAquila spectrographs, complementing them with high-resolution( 𝑅 ∼ , high level of confidence in our fits that deconstruct the highly time-sampled Aquila spectra presented in Sections 3 and 4. It has been afruitful strategy to obtain a high temporal density sustained datasetusing Aquila, with occasional supplementation from LHires. We present the epochs at which we have data in Figure 2, markedagainst the AAVSO light curve on the back wall. The colour of thelight curve at the back is mapped to Julian Day using the same colourmap as that of the Aquila observations themselves, shown as spectrain the foreground; this light curve is shown in Figure 1. The LHiresspectra are also shown, interspersed in gold. This gives an generalimpression of our dataset, as a continuous follow-up programmewith highly regular observations and indicates how such an extendedyet thorough campaign is made possible by the network of GlobalJet Watch observatories, as there is typically at least one locationwith both local night-time and favourable observing conditions.
The spectra were reduced with a bespoke data reduction pipelinecalled endeavour , a full end-to-end image-processing solutionwhich takes data captured at the telescopes, performs the reduc-tion, and delivers 1D spectra as a product to the fitting applications.It is a bespoke pipeline suited to our specific instrumentation, butbuilt on widely-used open-source scientific libraries available for thePython programming language. The packages used include NumPy(Harris et al. 2020), SciPy (Virtanen et al. 2020), Pandas (Mckinney2010), scikit-learn (Pedregosa et al. 2011) and Astropy (Robitailleet al. 2013; Price-Whelan et al. 2018). The pipeline ensures ev-ery spectrum is dark and bias corrected, and calibrates wavelengthsagainst Thorium-Krypton bulb arcs. The spectra are heliocentricallycorrected to the barycentre of the solar system.
Post-processing analysis is performed by poirot , which takes asan input the reduced 1D spectra from endeavour . In this paper, wefocus on the H 𝛼 complex, and so data are typically normalised tothe peak of H 𝛼 . In the fits presented in this paper, the backgroundhas been fit with a simple constant offset, valid as the backgrounddoes not vary strongly over the H 𝛼 region. This is performed at thepoint of fitting the physical model, and is allowed to freely vary. Wefind that the background is consistently responsible for < poirot , it removes any spec-tra with saturation on the feature of interest, or adverse weathereffects (determined by manual inspection). In the later stages of theeruption when changes are happening on longer timescales, we sumthe spectra taken at a given observatory on a given night. Whennecessary, as the nova became fainter, to further improve signal-to-noise, we reject any spectra which are not the longest exposuretaken on a particular night. Deducing the underlying physical model from spectral complexesis notoriously difficult in astronomy, in part because of the intrin-sic richness of the phenomena, and in part because of the trade-offbetween spectral resolution and signal-to-noise. Our H 𝛼 spectra MNRAS , 1–12 (2020)
D. McLoughlin et al.
Figure 1.
The AAVSO light curve of YZ Reticuli ranging from magnitude 4.5 to 10.1 in the visible band over the first 125 days since discovery.
Figure 2.
This plot indicates the epochs at which we have spectra. TheLHires spectra are depicted in gold, while the light curve and the Aquilaepochs are coloured according to their Julian Date. The rest wavelength ofH 𝛼 is projected to the floor as a fiducial marker. We reproduce for contextthe AAVSO light curve of YZ Reticuli against the back wall, coloured as inFigure 1 of classical novae are no exception, with several dynamical compo-nents evolving rapidly. There are different strategies for deconstruct-ing the underlying components that comprise emission lines profilese.g. Schmidt et al. (2019). We preferred to experiment with a sim-ple set of Gaussians. The merits of this approach are demonstratedelsewhere e.g. Blundell et al. (2008); Grant et al. (2020). We used multiple one-dimensional Gaussian models (and a subse-quently subtracted constant offset background, discussed in Section2.6) in a composite fit to our data, using the Levenberg-Marquardtalgorithm with a least squares statistic (Price-Whelan et al. 2018) tominimise the difference between observation and model. Residualswere analysed, and inspected to ensure that there only remainedvariation on velocity scales smaller than or comparable to the limitfrom spectral resolution. See Section 3.3 for a discussion of therobustness of the fits. Figure 3 demonstrates typical results of ourfitting template, with residuals plotted in the panel below. Figure 4
Figure 3.
Fitted model showing two pairs of emission lines, and a centralcomponent. Spectrum (shown as pink points) taken at the GJW-OZ obser-vatory, at + .
56 d since discovery, with a 1000 s exposure time. illustrates how the qualitatively changing profile of YZ Reticuli isnonetheless well-modelled by these five Gaussian components.
The extended coverage spanning 180 days together with the consis-tency of the succession of fits were key to our confidence in thesefits. While there are many possible fits which might reasonably ap-proximate any one individual spectrum, we were able to reject most a priori possible fits by comparing across time ranges within ourdataset, shown in Figure 2 alongside the light curve.In Section 4.2 we discuss various changes which occur in theevolution of the fits to YZ Reticuli H 𝛼 profiles. We are confident ofvariation timescales that are significantly sub-night, which wouldhave been missed with sparser sampling. We now consider different signals that might give a minor influenceto our determinations of the underlying H 𝛼 emission components. MNRAS000
The extended coverage spanning 180 days together with the consis-tency of the succession of fits were key to our confidence in thesefits. While there are many possible fits which might reasonably ap-proximate any one individual spectrum, we were able to reject most a priori possible fits by comparing across time ranges within ourdataset, shown in Figure 2 alongside the light curve.In Section 4.2 we discuss various changes which occur in theevolution of the fits to YZ Reticuli H 𝛼 profiles. We are confident ofvariation timescales that are significantly sub-night, which wouldhave been missed with sparser sampling. We now consider different signals that might give a minor influenceto our determinations of the underlying H 𝛼 emission components. MNRAS000 , 1–12 (2020) ets in classical nova YZ Reticuli Figure 4.
Sequence of H-alpha profiles. An example LHires spectrum (fourthfrom the bottom) covers a shorter wavelength range than Aquila. Theseprofiles can be deconstructed into a similar set of five emission Gaussiancomponents. Section 4 presents these components in more detail.
Figure 5 shows Earth’s own atmospheric lines (bottom panel), whichconsist of many narrow absorption lines. These lines do not havea significant impact on the fits, as the second panel shows clearly.Many of these lines are in doublets, although these are rarely fullyresolved, even by the LHires instrument, as they absorb against therelatively dim nova at later epochs.Another feature usually observed in the early spectra of clas-sical novaeis blue-shifted absorption. Denoted the low- and high-velocity components respectively, the hundreds of km s − and thethousands of km s − absorption systems are thought to be due tothe complex ejecta (Arai et al. 2016). These are often significantand usually need to be fitted for early spectra. In the particularcase of our observations however, by the time of our our first spec-trum and thereafter, only emission components dominate; any suchblue-shifted absorption components had already disappeared.We also carefully consider the possible presence of forbidden[N ii] lines at 6548 Å and 6584 Å. These lines could be very prob-lematic for fitting the broad H 𝛼 complexes in classical novae suchas YZ Reticuli or V6568 Sgr as a full width at half maximum of ∼ − at late times is easily enough to encompass thesewavelengths. However, the nitrogen lines would correspond to H 𝛼 speeds of −
685 km s − and 959 km s − , and should exhibit similardynamics to H 𝛼 . Indeed, Figure 5 shows that the profiles of two for-bidden [O iii] lines (4959 Å, 5007 Å) in the spectrum are strikinglysimilar to H 𝛼 (the [O iii] emission at 5007 Å is a factor of three lessstrong than H 𝛼 at late times for our spectra). If the nitrogen 6548 Åand 6584 Å lines were strong, we would expect to see evidence ofa similarly spaced pattern — but there is no such evidence, and noris there any emission component at the rest wavelengths either. When experimenting with fits to the YZ Reticuli H 𝛼 spectra withmultiple Gaussians, there was one solution set which persistedthroughout the entire time series. While individual parameterschanged over time as the nova progressed, two pairs of lines be- haved in a correlated fashion, leading us to believe that these fitsdo indeed reflect an underlying physical reality rather than a math-ematical degeneracy. The qualitative shape of the overall spectraseems to change dramatically over the course of our observations(see Figure 4 for examples), but we show that only small changes tothe Gaussian components reasonably explain these variations in anintuitive way. While of course it is possible to model classical novaemission line profiles with spatial models of the ejecta, we believethat the results emerging here show that a much simpler picturecan explain a lot of the spectral behaviour without presuming anygeometry in advance. The template which best fits our H 𝛼 spectra consists of five Gaussiancomponents in emission. In Figure 6, the left plot is a convenientrepresentation of the evolving characteristics of each fit becauseit represents the Doppler-shifting movements in the centroids ofeach of the five emission components, and reveals any time-varyingpatterns where these are present. The colour code for the five com-ponents is persistent throughout this paper: red and blue denotingthe red- and blue-shifted double peaks of an inner pair of emissionlines, and pink and turquoise correspondingly for an outer (higher-speed) pair. Orange is consistently the colour of the componentroughly stationary with respect to the system, which we call thecentral component. There seems to be a natural pairing of the twooutermost emission components, since the movement of their cen-troids is associated throughout time. The same is true for the innerpair (coloured red and blue). This leaves the single central com-ponent, close to the H 𝛼 rest wavelength, which seems to vary onlyslightly in wavelength, though independently of the other four emis-sion components. This is a persistent trait of the spectra throughoutour spectroscopic monitoring.What is truly remarkable about this five emission componentmodel is that it fits the data well over a large range of dates, andfor a variety of seemingly qualitatively different spectral complexes.The overall shape of the H 𝛼 complex at early times displays broademission, then from about +
15 d to +
95 d we observe a generallyflat peak, with a surplus of red emission at about 1000 km s − .After +
95 d (long after the SSS switch-on) we see a sudden tran-sition to a doubly-peaked profile with emission at − − and 1000 km s − . Our five peaks fit all of these rich data acrossthe whole time series, with only minimal changes to the modelparameters, which preserve the relationships between the pairs. In the context of Figure 6 (left panel), we now consider not just howeach emission component is behaving, but the relationships betweenthem. We calculate relative metrics from the fitted parameters foreach spectrum, and we plot these values over time in Figure 7. Thistechnique is essential for reducing the dimensionality of this highlycomplex dataset, and gives insights as to the underlying physics.The pertinent derived parameters include the differences betweenvelocity centroids of the inner (blue, red) and outer (turquoise, pink)pairs of components, and the integrated area under each Gaussiancomponent. The right panel of Figure 6 shows how the area ofeach Gaussian component varies with time. The areas of the inner(blue, red) pair of components are remarkably steady relative toone another at every epoch. The central component, depicted inorange, shows a step-change decrease in area just before +
100 d. The
MNRAS , 1–12 (2020)
D. McLoughlin et al.
Figure 5.
On the first three panels, we draw vertical lines marking peaks at radial velocity − − , −
680 km s − and 760 km s − for the relevant transition.Top panel: H 𝛼 complex taken on an Aquila instrument over the course of several nights ( +
120 d to +
160 d) plotted in purple, with the mean of these overlaidin gold. Second panel: [O iii] (4959 Å and 5007 Å) complex in the same spectra, with corresponding sets of markers at the same radial velocities for each ofthese transitions. Third panel: an example H 𝛼 spectrum from LHires at roughly the same time but significantly higher resolution shown in purple, with thesum of five fitted Gaussian emission components overlaid in teal. The vertical markers match the centroids of three of these peaks. Fourth panel: a close-upof the top of the previous panel’s H 𝛼 complex, with known telluric lines denoted with vertical lines given by Curcio et al. (1964). Several of these have onlyslightly weaker doublet partners whose wavelengths are not known as accurately as the ones depicted, hence some of these lines appearing wider than others. The close matching of the fit despite the telluric features confirms the appropriateness of our template model to explore the dynamics of this classical nova.The wavelength alignment required no additional correction beyond the calibration as part of the endeavour pipeline. Each spectrum is normalised to thepeak of the complex being plotted, but the oxygen 5007 Å peak is a factor of three weaker than that of H 𝛼 . redshifted outer component, depicted in pink, initially fluctuatesto larger and then steadily smaller areas by +
50 d, after which itgradually increases in area.Figure 7 depicts the time-variation of the difference in velocitybetween the red- and blue-shifted component centroids within eachpair. We smooth this quantity (henceforth the separation) for eachpair over time using a mean filter, and observe pronounced quasi-periodic oscillations in both the inner and outer pairs. Note thatthis is a separation in velocities rather than pure distances. Thesepairs behave in a similar fashion to each other, with large velocity separations at early times ( < +
50 d) which reach a minimum sepa-ration within the first few weeks. They then both recover slightly,but continue to oscillate. The oscillations happen at different timesfor each pair, and the timescale of recurrence decreases over time,which we discuss further in Section 4.3.At around +
100 d there is an abrupt change in behaviour,whereby the oscillation of the inner pair decreases in amplitude(moving to lower radial speeds) but the outer pair maintains itssteady progression to higher-magnitude radial speeds. After thisdate, the predominant general trend is of much more steady in-
MNRAS000
MNRAS000 , 1–12 (2020) ets in classical nova YZ Reticuli Figure 6.
The left plot shows the centroid velocity of each of the five emission components relative to the centroid of the inner pair over time, ensuring that nosubtle wavelength calibration trends are at play. The widths of the lines are proportional to the full width at half maximum of that component. The right plotshows how the area of each component develops with time, and the identified pairs of lines are plotted back-to-back to more easily see correlations. The widthof each bar in this figure is proportional to the FWHM of the line, with an overall scaling that avoids horizontal overlap.
Figure 7.
The two left panels show how the separations in velocity between each of the inner components and each of the outer components respectively varyover the full time range. These are smoothed using a mean filter with a moving window size of three observations. Each observation is the median of thehighest-quality spectra we have for that observatory on each night. The right two plots represent the same underlying data as the left two plots, but over limiteddate ranges, and with the linear trend removed from each to show the oscillations more clearly. The date ranges used for the right-hand plots are represented byfiducial horizontal lines on their respective left-hand analogues. crease in velocity by ∼ . − per day, but superimposed onthis is the distinct but much more subtle oscillation of ∼
10 km s − with a characteristic timescale of ∼
10 days.On removing the overall trends in the two leftmost panels ofFigure 7 by subtracting a straight-line fit from each, we recoverthe underlying oscillations more clearly and these are plotted inthe two rightmost panels for a restricted time range (marked inthe left plots with horizontal lines). This representation shows theoscillations in the clearest manner. Upon folding by the strongestpeaks in the Fourier transform, we could see that the period ofboth these pairs was in fact changing by several days, so this isin fact a quasi-periodic oscillation in the velocity separation. Therewere no statistically significant periodicities, most likely because the duration of each cycle was quite different from that of the precedingcycle. However, the pattern of irregular oscillations with similarvelocity amplitude is clear, and we take this to mean that thesepairings of lines reflect pairings in the underlying physical reality.In the two rightmost panels of Figure 7, there also appear to be hintsof smaller-scale nutations superimposed on the large oscillations,but we would need even finer time resolution to quantify theseaccurately and fully separate the various perturbations.The inner pair velocity separation decreases by 120 km s − be-tween +
100 d and +
150 d, while the outer pair separation steadilyincreases from 2200 km s − to 2600 km s − between +
60 d and +
150 d, which we discuss in more detail in Section 5.1.
MNRAS , 1–12 (2020)
D. McLoughlin et al.
What is fascinating about the oscillatory behaviour of the innerand outer pair seen in both Figure 6 and Figure 7 is that it is notfixed in either amplitude or period, but instead these quantities bothdecline over the course of the first six months. The implication hereis that this reflects some property of the underlying system whichis returning to an equilibrium. Furthermore, these timescales aresubstantially longer than both the canonical classical nova binaryorbit timescale ( ∼ ∼
76 s), meaning this effect takes tens to hundredsof orbital periods to perform a cycle.
The data clearly fit well with a template of 5 Gaussian emissioncomponents: a central component, an inner pair of componentswith radial velocity relative to the centre of ∼
600 km s − , and anouter pair of ∼ − .There are several possible models which may initially appearto explain this clear signal. The key components in the system arethought to be the accretion disc, the hot white dwarf, the secondarystar, and various forms of ejecta. We disregard the H 𝛼 flux from tothe secondary itself, since the magnitude of the whole system was > =
15 during quiescence (ASAS-SN data in Section 1.3).
The natural interpretation of the outermost widely-separated andbroad emission lines is as oppositely-directed collimated outflows,which may reasonably be described as jets (Iijima & Esenoglu2003). Given the prevalence of accretion discs within jet-ejectingsystems, we posit that the inner pair of lines could arise from theaccreting material either because of its rotation or winds off theaccretion disc.We propose that the five Gaussian emission components withinthe H 𝛼 complex are best explained by a model involving jets, anaccretion disc and an additional component henceforth the ‘jets’model. This model therefore posits that all five observed Gaussianshave some connection with an accretion disc. The outer pair of lines(turquoise, pink) is caused by bipolar anti-parallel jets launched as aconsequence of an accretion process. We therefore suggest that theinner pair of lines (blue, red in Figure 6) may be a manifestation ofthe standard double-peaked rotating disc signature or the wind offthis rotating matter. This is a simple suggestion, and is exactly whatis seen arising from the accretion disc in the microquasar SS433(Blundell et al. 2008). As noted by Long & Knigge (2002), theevidence for winds off the discs in cataclysmic variable systems iscompelling. The fifth, central component has a significantly lowerline-of-sight Doppler shift and may be a shell arising from localmatter that is not dynamically important.This model is supported by the fact that there seem to bedistinctive epochs where characteristics change in a related way ineach of the components, sometimes after a short delay; these changesoccur most notably around +
100 d. This is explained naturally underthe jets model, because the lines all have a common origin in theaccretion disc. Furthermore, the evolution of the pairs of lines andtheir oscillatory behaviour makes most sense when consideringthem as being derived from a precessing disc.After about +
100 d, we observe that the inner pair moves tosteadily lower speeds, while the outer pair increases its velocity separation. Classical nova events differ from normal cataclysmicvariables by their enhanced mass-transfer rates. The increased ac-cretion rate may lead to their exceeding the threshold necessary forlaunching jets (Lasota & Soker 2005), a situation that can persistfor at least a few months following eruption. Since not all the ejectareceive enough energy to be fully ejected, some gas can be trappedand forms an envelope surrounding the inner system. As this gassteadily rejoins the accretion disc, we expect that there exists a re-gion of temporary overdensity, which would then feed the jet. Whatwe propose is that the trends in pair speeds are caused by suchan effect. The inner pair decreases in speed, as the accretion discis depleted at its inner radius (where it is fastest). The outer pairincreases in speed, as the accretion disc processes this extra mate-rial and launches increasingly powerful jets. This assumes the outerpair of lines are produced near the base of the jet, which is expectedsince at high rates of adiabatic expansion into relative vacuum, theplasma becomes optically thin.Simulations conducted by Figueira et al. (2018) under partic-ular conditions result in the accretion disc being fully disrupted bythe ejecta within the first forty minutes after outburst. This occurswhen the mass of the ejecta is assumed to be high compared withthe mass of the accretion disc. A typical timescale for reforming adisrupted accretion disc has been estimated at around hundreds ofyears. However, Figueira et al mention, accretion discs have beenobserved in novae within months of eruption, implying that theyeither are re-established more rapidly than theoretical predictionsor indeed are never fully disrupted. Moreover, it is possible thataccretion can be efficacious even when the disc has not yet attaineda state of perfect equilibrium. Hernanz & Sala (2002) show the re-sumption of accretion after only a brief period of disruption. Also,the best-fitting model to the progenitor spectral energy distributionin ASASSN-18fv includes an accretion disc, shown comprehen-sively in Wee et al. (2020) and used to explain changes in the colourevolution of that classical nova around 50 days after outburst. Thespectral signature of the accretion disc, or potentially the wind offits surface, is visible beyond the photosphere. The very early time-series data on YZ Reticuli shows a rapidly-evolving regime whichmay be caused by an accelerated resumption of accretion processes.A succession of over-pressured jet ejecta could plausibly ex-pand into each other, causing shocks and hence particle accelerationgiving rise to the X-ray shocks observed by NuSTAR reported inSokolovsky (2020). This is exactly analogous to the behaviour ex-hibited in the microquasar SS433 as its H 𝛼 -emitting jet bolidescollide into one another, causing non-thermal, polarised emissionafter shock acceleration (Blundell et al. 2018). The leftmost teal panel in Figure 7 which plots the velocity separa-tion of the inner pair with time shows oscillations on two differentcharacteristic timescales, one being about 40 days prior to about +
100 d, and thereafter about +
10 d (zoom-in in the third panel).These oscillations show a general tendency to decrease in both am-plitude and period for the entire duration. We conjecture that the firstphase of oscillations might be a response of the accretion disc to therapid influx of material in the immediate aftermath of the eruption.The existence of the inner lines only a few days after the initial clas-sical nova event indicates a pre-existing (albeit perturbed) accretiondisc in the YZ Reticuli system. After about +
100 d, the amplitudeof the oscillations is significantly smaller, suggesting equilibrationhas happened to some extent.The parameter space occupied by cataclysmic variables, and
MNRAS000
MNRAS000 , 1–12 (2020) ets in classical nova YZ Reticuli a discussion of the precessing accretion discs in these systems, isoutlined in Patterson (2001). Charles et al. (2008) give a summaryof the observational properties of accretion discs in X-ray binaries,many of whose time-varying/periodic characteristics have been in-terpreted as precession on a wide variety of different timescales.It is plausible that the oscillations seen in the leftmost tealpanel in Figure 7 may arise from reactive precession in a (likelyasymmetrically) perturbed accretion disc. It is therefore interestingto consider that the oscillations seen in the gold panels (2 and 4)of Figure 7 arise from a change in axis of the jet ejecta given theongoing re-orientation of its launch platform i.e. accretion disc. Inthe microquasar SS433, its accretion disc has been shown to ex-hibit a persistent periodic pattern of precession and nutation tightlycoupled to that exhibited by its famous precessing jets (figure 2 ofBlundell et al. (2008)). In contrast to the quasi-persistent periodicprecession of SS433, YZ Reticuli’s accretion disc appears to precesson shortening timescales as a response to the disruption followingthe classical nova outburst, so a tight link between this and conse-quent jet precession might not be expected. This can be especiallyhard to discern when the underlying period is itself decreasing (per-haps due to mass-loss through the jets themselves). Regardless ofthe duration of the precession period, the accretion disc and jetsare expected to exhibit a quarter-phase lag because of their perpen-dicularity from one another. It is therefore interesting to note inthe leftmost two panels of Figure 7 that there are distinct hints ofsuch a phase difference between jet oscillations and accretion discoscillations that are consistent with such a picture. One phenomenon we observe in the succession of YZ Reticulispectra is that several emission species appear to transition from aprofile with a central peak to one characterised by wider double-peaks. The first such transition event occurs suddenly at + .
21 din H 𝛼 . It then recurs multiple times, most dramatically at + .
93 d(lasting until + .
92 d). At +
96 d, the transition repeats, steadilydeepening across the next few hours and drastically deepening againat + . 𝛼 complex remains in a split state.This unsteady approach to eventual stability could be due to thetemperature of the emitting body of gas dropping below a relativelysharp cutoff, with some fluctuations in temperature causing it totemporarily return above the threshold before settling below.Figure 8 shows the velocity profiles of the most prominent linesin our spectra. Some of the lines (shown in the left panel) displaythis split feature, and others do not (shown in the right panel). Thecentral panel shows the split velocity (difference between the twopeaks, zero if only one peak) against the upper energy level of theatomic transition responsible, giving a strong positive correlation.The transitions that do not appear to split show no qualitative signof significant changes before and after +
95 d.By close examination of the fits to the H 𝛼 complex in a timeseries around this epoch, we note that in fact the apparent splitting iscaused by the central dynamical component growing weaker relativeto the inner and outer pair. While the other lines are fainter and assuch have lower signal to noise than H 𝛼 they correspond veryclosely in velocity space, so we assume that they are also appearingto split because a central component is growing dimmer.Since the splitting seems to only occur in the higher upperenergy level transitions, we believe that this is evidence that theorigin of the central component has cooled to below 10 eV by +
100 d.
The existence of a jet whose radiant gas gives rise to H 𝛼 emissiondoes not necessarily imply the existence of a radio-emitting syn-chrotron jet. In the case of the jets from the microquasar SS433,the former leads to the latter only because of the expansion of thejet bolides into one another causing shocks and consequent particleacceleration (Blundell et al 2018) evinced by particular signaturesin the synchrotron polarisation coinciding with where the expansionand collision happens. In the case of a binary whose H 𝛼 emitting jetejecta either (i) did not expand sufficiently as to collide into one an-other causing a shock-accelerated synchrotron emitting populationor (ii) were not propagating through a magnetised region of space,no synchrotron radio jet would be observed.We note that Toalá et al. (2020) require a non-thermal compo-nent to fit the spectral shape of elongated X-ray emission of DQ Hershown in their figure 3; we conjecture that this X-ray emission arisesfrom inverse Compton emission of spent synchrotron electrons. Given the resolved images of approximately spherical shells aroundmany classical novae it might be tempting to associate the innerpair of emission lines with a spherically ejected shell discussed inSection 1.1. However, a ballistically-ejected outflow would not con-tinue to modulate in an apparently symmetric way nor change speedseveral months after the eruption (Figure 7, panel 3). Moreover, thiscould not explain the sometimes correlated changes in the charac-teristics of the inner pair and the outer pair of emission components.As described in Section 4.3, the magnitude of oscillations in the rel-ative velocity of the inner pair components consistently decreasesover time. While this is expected behaviour for a precessing accre-tion disc re-establishing equilibrium, there is no clear explanationfor why this would happen to opposite sides of a spatially-extendedejected shell.In addition, a spherical shell would be too hot within the firstfew weeks of the eruption ( ∼ K) to emit H 𝛼 , which requires amore modest temperature regime (a few 10 K). Indeed, the reasonwhy we are able to see the jets in H 𝛼 spectra is because they arefar cooler, which is consistent with their having been launched bythe accretion disc rather than being the products of thermonuclearrunaway. We considered several models before settling on the jets model.The primary such alternative model is that the inner pair is a slow-moving outflow, considered in Section 5.5.We also considered whether one of the pairs could be formedin a circumbinary disc. While we have seen preliminary evi-dence pointing towards circumbinary discs in other novae such asASASSN-18fv (McLoughlin et al. 2020), the speeds involved hereare significantly faster, implying either rotation within the binarysystem or high-speed outflow.Another alternative model is that one pair of lines is formed inan accretion disc while the other pair are the fast bipolar ballisticejecta. The difference between this and our primary model is that inthis case, the ejecta are not launched by the accretion, but are directejecta from the surface of the white dwarf. This model acknowledgesthat there has been a recent explosion and as such matter should beflying outwards, but it fails to explain why the inner pair and theouter pair should continue (i) at all and (ii) to oscillate, sometimes in
MNRAS , 1–12 (2020) D. McLoughlin et al.
Figure 8.
Average spectra of different emission species taken at +
94 d of YZ Reticuli. The left panel shows certain species which display the splittingphenomenon, the right panel shows the lines which do not exhibit this. The central panel shows the excitation energy of the upper level of the identifiedtransitions responsible for each line, plotted against the velocity difference of the two peaks in the case of the split lines. Note that this is not to be confusedwith the quantum effect of line splitting, it is purely dynamical. related ways, to late times. Under this scenario, it would be difficultto explain why the bipolar outflow would be visible in H 𝛼 , as itwould be far too hot. Given the theoretical basis for jets in classical novae set out in Kato& Hachisu (2003), it is tempting to consider whether this modelapplies in a prevalent manner, or if YZ Reticuli was an unusualevent. In the same month, July 2020, there was another classicalnova, V6568 Sgr, of the very fast variety. We confronted this withthe same model, and discovered that while largely similar, it was,notably, missing the central component. The speeds of the outermostpair are − − and 4500 km s − , while the speeds of theinner pair are − − and 1900 km s − . Although V6568Sgr is substantially faster than YZ Reticuli, it nonetheless displaysstrikingly similar pairings. Figure 9 shows that our model fits tothis other classical nova well, despite the vast difference in speedclass. Note that the spectrum appears to be noisier than those ofYZ Reticuli — this is expected, because it is so much faster andbecause this was taken with the LHires instrument which has a muchsmaller angstrom-to-pixel ratio. As such, there are fewer photons perwavelength bin, and the telluric absorption lines play a bigger role.Notwithstanding this, the same model fits, with the exception of agreatly diminished central component, which may be interpreted asany optically thick shell having disappeared. The two pairs of linesare very symmetric, and we could not obtain such a good fit withoutthe outer pair; despite their relatively lower amplitude in comparisonto the inner pair, they are nonetheless crucial. We conjecture thatV6568 Sgr bridges the gap in parameter space, specifically that ofWD mass, between YZ Reticuli and the recurrent novae. Figure 9.
V6568 Sgr spectrum taken at +1.03d after discovery. This fitsthe same model as YZ Reticuli, apart from the notable absence of a centralcomponent.
We have investigated the rapidly evolving spectral characteristicsof the classical nova YZ Reticuli, especially its H 𝛼 complex, sinceshortly after its discovery in July 2020 with time-resolved spec-troscopy from the Global Jet Watch. We find that, remarkably, amodel for the H 𝛼 complex comprising the same family of Gaus-sians throughout the first six months fits the data well across a varietyof qualitatively different spectral shapes. These five components ap-pear to be naturally categorised as two pairs of lines, representingan accretion disc and jet outflows, together with a small centralcomponent. MNRAS000
We have investigated the rapidly evolving spectral characteristicsof the classical nova YZ Reticuli, especially its H 𝛼 complex, sinceshortly after its discovery in July 2020 with time-resolved spec-troscopy from the Global Jet Watch. We find that, remarkably, amodel for the H 𝛼 complex comprising the same family of Gaus-sians throughout the first six months fits the data well across a varietyof qualitatively different spectral shapes. These five components ap-pear to be naturally categorised as two pairs of lines, representingan accretion disc and jet outflows, together with a small centralcomponent. MNRAS000 , 1–12 (2020) ets in classical nova YZ Reticuli Correlated and symmetric changes in these two pairs of linessuggest that the accretion disc and jets appear to be precessing,probably in response to the recent large perturbation caused by thenova eruption.The jet and accretion disc signatures are followed from thefirst spectra at around +
10 d until the present day, which we takeas evidence that the accretion disc survived the blast, albeit in anon-equilibrium state for the first ∼
100 days.Confirmation of this interpretation for YZ Reticuli could bepossible from future VLBI observations giving milli-arcsecond-scale resolution radio imaging akin to the images of the radio jetsin recurrent nova RS Oph observed by Sokoloski et al. (2008) butonly if the H 𝛼 jets give rise to particle acceleration and if there is asufficiently strong magnetic field present. However, it is undoubtedlyfruitful to consider time-resolved spectroscopy from other classicalnovae to establish the generality of H 𝛼 jets, and we will present theresults of such an analysis in a forthcoming paper. ACKNOWLEDGEMENTS
Gaia ( ), processed by the Gaia
Data Processing and Analy-sis Consortium (DPAC, ). Funding for the DPAC has been pro-vided by national institutions, in particular the institutions partic-ipating in the
Gaia
Multilateral Agreement. This work has madeuse of data from the All-Sky Automated Survey for Supernovae(ASAS-SN)(Shappee et al. 2014; Kochanek et al. 2017).
DATA AVAILABILITY
GAIA data publicly available at https://gea.esac.esa.int/archive/.ASAS-SN data publicly available at https://asas-sn.osu.edu/light_curves/ee130388-139d-4f3a-ab8a-7427a7545e21. AAVSO data publicly available at https://aavso.org.The data underlying this article were provided by the Global JetWatch by permission. Data will be shared on request to thecorresponding author with permission of the Global Jet Watch. REFERENCES
Arai A., Kawakita H., Shinnaka Y., Tajitsu A., 2016, The AstrophysicalJournal, 830Aydi E., 2020a, ATel Aydi E., et al., 2020, The Astrophysical Journal, 905, 62Bailer-Jones C. A. L., Rybizki J., Fouesneau M., Mantelet G., Andrae R.,2018, The Astronomical Journal, 156, 58Blundell K. M., Bowler M. G., Schmidtobreick L., 2008, The AstrophysicalJournal, 678, 47Blundell K. M., Laing R., Lee S., Richards A., 2018, The AstrophysicalJournal Letters, 867, 25Brown A. G. A., et al., 2018, Astronomy & Astrophysics, 47Charles P., Clarkson W., Cornelisse R., Shih I C., 2008, New AstronomyReviews, 51, 768Csák B., Kiss L. L., Retter A., Jacob A., Kaspi S., 2005, Astronomy andAstrophysics, 429, 599Curcio J. A., Drummeter L. F., Knestrick G. L., 1964, Applied Optics, 3,1401Darnley M. J., et al., 2017, The Astrophysical Journal, 847, 35Figueira J., José J., García-Berro E., Campbell S. W., García-Senz D., Mo-hamed S., 2018, Astrophysics A&A, 613, 8Grant D., Blundell K., Matthews J., 2020, Monthly Notices of the RoyalAstronomical Society, 494, 17Hachisu I., Kato M., 2012, Baltic Astronomy, 21, 68Harris C. R., et al., 2020, Array programming with NumPy,doi:10.1038/s41586-020-2649-2Harvey E. J., et al., 2018, Astronomy & Astrophysics, 611Hernanz M., Sala G., 2002, Science, 298, 393Hounsell R., Darnley M. J., Bode M. F., Harman D. J., Helton L. A., SchwarzG. J., 2011, Astronomy & Astrophysics, 530, A81Iijima T., Esenoglu H. H., 2003, Astronomy and Astrophysics, 404, 997Kato M., Hachisu I., 2003, The Astrophysical Journal, 587, 39Kawabata K. S., et al., 2006, The Astronomical Journal, 132, 433Kochanek C. S., et al., 2017, Publications of the Astronomical Society ofthe Pacific, 129, 104502Krautter J., et al., 2002, The Astronomical Journal, 124, 2888Kwan-Lok L., 2020, ATel
Lasota J.-P., Soker N., 2005, in The Astrophysics of Cataclysmic Variablesand Related Objects ASP Conference.Long K. S., Knigge C., 2002, The Astrophysical JournalMcLoughlin D., Blundell K. M., Lee S., 2020, Monthly Notices of the RoyalAstronomical SocietyMckinney W., 2010, PROC. OF THE 9th PYTHON IN SCIENCE CONF,pp 56–61Mclaughlin D. B., 1942, Technical report, SPECTRAL STAGES OF NO-VAE. University of MichiganPatterson J., 2001, Publications of the Astronomical Society of the Pacific,113, 736Pedregosa F., et al., 2011, Journal of Machine Learning Research, 12, 2825Price-Whelan M., et al., 2018, The Astronomical Journal, 156, 123Prusti T., et al., 2016, Astronomy & Astrophysics, 595Retter A., 2004, The Astrophysical Journal, 615, 125Ribeiro V. A. R. M., Munari U., Valisa P., 2013, The Astrophysical Journal,768, 49Roberts D. H., Lehar J., Dreher J. W., 1987, The Astronomical Journal, 93,968Robitaille T. P., et al., 2013, A&A, 558, 33Rupen M. P., Mioduszewski A. J., Sokoloski J. L., 2008, The AstrophysicalJournal, 688, 559Santamaría E., Guerrero M. A., Ramos-Larios G., Toalá J. A., Sabin L.,Rubio G., Quino-Mendoza J. A., 2020, The Astrophysical Journal, 892,60Schmidt K. B., et al., 2019, Astronomy & Astrophysics, 628, 1Shappee B. J., et al., 2014, The Astrophysical Journal, 788, 48Skopal A., 2008, in ASP Conference Series. Elsevier, pp 128–138,doi:10.1016/j.newast.2013.12.005MNRAS , 1–12 (2020) D. McLoughlin et al.
Slavin A. J., O’Brien T. J., Dunlop J. S., 1995, 353S Mon. Not. R. Astron.Soc, 276, 353Sokoloski J. L., Rupen M. P., Mioduszewski A. J., 2008, The AstrophysicalJournal, 685, 137Sokolovsky K. V., 2020, ATel
Sokolovsky K., Aydi E., Chomiuk L., Kawash A., 2020, ATel
Songpeng P., 2020a, ATel
Songpeng P., 2020b, ATel
Strope R. J., Schaefer B. E., Henden A. A., 2010, Astronomical Journal,140, 34Taylor A. R., Davis R. J., Porcas R. W., Bode M. F., 1989, Monthly Noticesof the Royal Astronomical Society, 237, 89Toalá J. A., Guerrero M. A., Santamaría E., Ramos-Larios G., Sabin L.,2020, Monthly Notices of the Royal Astronomical Society, 495, 4372Virtanen P., et al., 2020, Nature Methods, 17, 261Wee J., et al., 2020, The Astrophysical Journal, 899, 162Williams R. E., 1992, The Astronomical Journal, 104, 725Williams R., Mason E., 2010, Astrophysics and Space Science, 327, 207This paper has been typeset from a TEX/L A TEX file prepared by the author. MNRAS000