An Atlas of Exotic Variability in IGR J17091-3624: A Comparison with GRS 1915+105
James Matthew Christopher Court, Diego Altamirano, Margarita Pereyra, Christopher M. Boon, Kazutaka Yamaoka, Tomaso Belloni, Rudy Wijnands, Mayukh Pahari
MMNRAS , 1–24 (2017) Preprint 29 August 2018 Compiled using MNRAS L A TEX style file v3.0
An Atlas of Exotic Variability in IGR J17091-3624: AComparison with GRS 1915+105
J.M.C. Court (cid:63) , D. Altamirano , M. Pereyra , C.M. Boon , K. Yamaoka ,T. Belloni , R. Wijnands , M. Pahari Department of Physics and Astronomy, University of Southampton, Southampton, SO17 1BJ, UK Department of Physics, Nagoya University, Aichi 464-8602, Japan Osservatorio Astronomico di Brera, Via E. Bianchi 46, 23807 Merate (LC), Italy Anton Pannekoek Institute for Astronomy, University of Amsterdam, Postbus 94249, 1090 GE Amsterdam, The Netherlands Inter-University Center for Astronomy and Astrophysics, Post Bag 4, Ganeshkhind, Pune-411007, India
Accepted 2017 March 27. Received 2017 February 28; in original form 2016 November 25
ABSTRACT
We performed an analysis of all
RXTE observations of the Low Mass X-ray Binary andBlack Hole Candidate IGR J17091-3624 during the 2011-2013 outburst of the source.By creating lightcurves, hardness-intensity diagrams and power density spectra of eachobservation, we have created a set of 9 variability ‘classes’ that phenomenologicallydescribe the range of types of variability seen in this object. We compare our set ofvariability classes to those established by Belloni et al. (2000) to describe the similarbehaviour of the LMXB GRS 1915+105, finding that some types of variability seenin IGR J17091-3624 are not represented in data of GRS 1915+105. We also use allavailable X-ray data of the 2011-2013 outburst of IGR J17091-3624 to analyse its long-term evolution, presenting the first detection of IGR J17091-3624 above 150 keV aswell as noting the presence of ‘re-flares’ during the latter stages of the outburst. Usingour results we place new constraints on the mass and distance of the object, and findthat it accretes at (cid:46) % of its Eddington limit. As such, we conclude that Eddington-limited accretion can no longer be considered a sufficient or necessary criterion for GRS1915+105-like variability to occur in Low Mass X-Ray Binaries. Key words: accretion discs – instabilities – stars: black holes – X-rays: binaries –X-rays: individual: IGR J17091-3624 – X-rays: individual: GRS 1915+105
X-ray binaries are systems in which a black hole or neutronstar accretes matter from a stellar companion, and they pro-vide us with opportunities to test how accretion takes placein the most extreme physical regimes. Some X-ray Binariesare believed to be accreting at very close to the Eddingtonlimit, the limit at which the radiation pressure on accretingmaterial is equal to the force due to gravity. As such, theseobjects can also provide a laboratory with which to exploreaccretion in radiation pressure-dominated systems (White& Zhang 1997).Low Mass X-ray Binaries (hereafter LMXBs) are a sub-class of X-ray binary in which the compact object accretesmatter transferred to it due to a Roche-lobe overflow fromthe companion star (e.g. Paczynski 1979). In general, accre-tion in LMXBs is a variable process, with variability seen (cid:63)
E-mail: [email protected] on timescales from milliseconds to decades. On the shortesttimescales the X-ray lightcurves of these objects can showband-limited noise and low-frequency quasi-periodic oscilla-tions (QPOs) at frequencies from ∼ mHz to ∼
200 Hz (e.g.van der Klis 1989). Black hole binaries also show so-called‘high frequency QPOs’ (e.g. Remillard et al. 1999a,b; Belloniet al. 2002; Belloni & Motta 2016), thought to be caused bymotion of matter in the innermost region of the accretiondisk (e.g. Stefanov 2014).Three sources – GRS 1915+105, IGR J17091-3624 andthe neutron star ‘Rapid Burster’ (MXB 1730-335) – alsoshow a variety of exotic variability on timescales of sec-onds to minutes in addition to the kinds of variability seenin other LMXBs. This exotic variability consists of quasi-periodic flares, dips and other high-amplitude behaviours(e.g Belloni et al. 2000; Altamirano et al. 2011a; Bagnoli& in’t Zand 2015). The second-to-minute-scale lightcurveprofiles of these sources change over timescales of days. InGRS 1915+105 and IGR J17091-3624, this behaviour can be c (cid:13) a r X i v : . [ a s t r o - ph . H E ] M a r J.M.C. Court et al. described as a set of ‘variability classes’. These classes them-selves vary widely in terms of flux, structure, periodicity andspectral properties.GRS 1915+105 (Castro-Tirado et al. 1992), hereafterGRS 1915, is a black hole LMXB which accretes at betweena few tens and 100% of its Eddington Limit (e.g. Vilhu 1999;Done et al. 2004; Fender & Belloni 2004; Reid et al. 2014).Most LMXBs go through periods of low-intensity ‘quies-cence’ and high-intensity ‘outbursts’, the latter consistingof black-body dominated ‘soft’ and power-law dominated‘hard’ spectral states. However, GRS 1915 has been in out-burst since its discovery in 1992 (Castro-Tirado et al. 1992).GRS 1915 is also notable for the incredible variety and com-plexity of variability classes it exhibits (e.g. Yadav et al.2000; Belloni et al. 2000) in addition to the less exotic vari-ability seen in other black hole binary systems. GRS 1915additionally shows high-frequency and low frequency QPOssimilar to those seen in other black hole LMXBs (Morganet al. 1997). In total, 15 distinct variability classes have beenobserved (Belloni et al. 2000; Klein-Wolt et al. 2002; Han-nikainen et al. 2007; Pahari & Pal 2009). This remarkablerange of behaviour is believed to be caused by instability inthe inner accretion disc (e.g. Janiuk et al. 2000; Nayakshinet al. 2000), which is in turn caused by the existence of aradiation pressure dominated regime in the inner disc (e.g.Done et al. 2004). Accounting for this complexity could bekey to our understanding of radiation-dominated accretionregimes.One of the best-studied variability classes of GRS 1915is the highly regular flaring ρ , or ‘heartbeat’, class, so namedfor the similarity of its lightcurve to an electrocardiogram.It has been shown that hard X-ray photons tend to lag softones in this class (e.g. Janiuk & Czerny 2005; Massaro et al.2010). Numerical models derived from Shakura & Sunyaev1973 which reproduce this lag can also reproduce other flar-ing classes seen in GRS 1915 (e.g Nayakshin et al. 2000;Massaro et al. 2014). These numerical models predict thatGRS 1915-like variability should be seen in systems accret-ing with a global Eddington fraction of (cid:38) . (Nayakshinet al. 2000). However, other LMXBs (e.g. GX 17+2, Kuulk-ers et al. 2002, and V404 Cyg, Huppenkothen et al. 2016)have been observed to exceed this Eddington fraction with-out displaying GRS 1915-like variability.Neilsen et al. (2011) proposed a physical scenario, basedon the mathematical model proposed by Nayakshin et al.(2000), to explain the presence of the hard lag in the flaringclasses of GRS 1915. This is outlined schematically in Figure1. First, an overdensity of matter forms via the thermal-viscous Lightman-Eardley Instability (Lightman & Eardley1974) and propagates inwards through the accretion disc.This destabilises the disc, collapsing its inner radius andvastly increasing photon emission. If the local EddingtonLimit in the inner accretion disc is then approached, extremeoutflows are triggered that deplete the inner accretion discand allow the cycle to begin again. As the matter ejectedfrom the disc collides with the non-thermal ‘corona’ abovethe central object, a flash of hard Bremsstrahlung radiationis produced. This causes a hardening of the spectrum andan apparent lag between soft and hard photons. Janiuk &Czerny (2005) instead propose that the lag is caused by thecorona smoothly adjusting to the changing brightness of thedisc after a light travel time. Figure 1.
A schematic diagram illustrating the the process de-scribed by Neilsen et al. (2011) to describe the ρ variability classin GRS 1915+105. 1) The X-ray emission from the system orig-inates from both the accretion disc truncated at an inner radius r in (grey) and a cloud of non-thermal electrons (white ellipse). Atsome time t , an overdensity in the accretion disc (formed by theLightman-Eardley Instability) propagates inwards towards r in . 2)As the inner disc heats up, r in begins to slowly increase due toan increase in photon pressure. This destabilises the disc. 3) Atsome critical density, the disc becomes too unstable and collapsesinwards, greatly decreasing r in and raising the inner disc tem-perature. 4) The sudden increase in emission exceeds the localEddington limit at r in , ejecting matter from the inner accretiondisc in the form of extreme winds. 5) Having been excited bymatter in the winds passing through it, the non-thermal electroncloud emits a hard Brehmsstrahlung ‘pulse’. The black hole candidate LMXB IGR J17091-3624(hereafter J17091) was discovered in outburst by
INTE-GRAL in 2003 (Kuulkers et al. 2003). In 2011, it againwent into outburst (Krimm & Kennea 2011). GRS 1915-likevariability was discovered in its lightcurve, as well as high-frequency QPOs which behave much like the QPOs seen inGRS 1915 (Altamirano et al. 2011a,c; Altamirano & Belloni2012). As IGR J17091 is around a factor of 20 fainter at 2–25 keV than GRS 1915, this object has either a lower blackhole mass M , a lower accretion rate ˙ m or lies at a larger dis-tance D than GRS 1915. Assuming by analogy with GRS1915 that IGR J17091 is accreting at its Eddington rate,the black hole must have a mass of M (cid:46) M (cid:12) or lie at adistance of D (cid:38) kpc (Altamirano et al. 2011a). However,independent estimates based on empirical relationships be-tween black hole mass and high-frequency QPOs have sug-gested values of M between ∼ . and . M (cid:12) (Rebusco et al.2012; Iyer et al. 2015a,b), while multi-wavelength observa-tions of the hard-to-soft state transition have suggested val-ues of D between ∼
11 and ∼
17 kpc (Rodriguez et al. 2011a).This implies that IGR J17091 may have an accretion rate ˙ m that is significantly below its Eddington rate.The suggestion that IGR J17091 accretes at signifi- MNRAS000
17 kpc (Rodriguez et al. 2011a).This implies that IGR J17091 may have an accretion rate ˙ m that is significantly below its Eddington rate.The suggestion that IGR J17091 accretes at signifi- MNRAS000 , 1–24 (2017) xotic Variability in IGR J17091-3624 cantly below the Eddington Limit raises several questions.Despite evidence of disc winds in IGR J17091 faster than0.03c (King et al. 2012), the wind generation proceduredescribed in Neilsen et al. (2011) cannot take place with-out near-Eddington-limited accretion. Additionally, it makesit increasingly unclear what differentiates IGR J17091 andGRS 1915 from the vast majority of LMXBs which do notdisplay such complex behaviour, and what physical systemproperties are required for GRS 1915-like variability to beobserved. This latter point was further complicated by theobservation of GRS 1915-like variability in 2 out of 155 RXTE observations of the Rapid Burster (Bagnoli & in’tZand 2015). As thermonuclear (Type I) X-ray bursts are alsoseen in the Rapid Burster (Hoffman et al. 1978), it is knownto contain a neutron star. As such, the presence of variabilityclasses in this object rules out any black hole-specific effectsas the cause of the complex variability. In addition to this,the persistent luminosity of the Rapid Burster during peri-ods of GRS 1915-like variability is known to be no greaterthan ∼ of its Eddington limit (Bagnoli et al. 2015).Accounting for GRS 1915-like variability is requiredfor a complete understanding of the physics of accretion inLMXBs. As such, Belloni et al. (2000) performed a com-plete model-independent analysis of variability classes inGRS 1915. This work highlighted the breadth and diver-sity of variability in GRS 1915, and allowed these authorsto search for features common to all variability classes. Forexample, Belloni et al. (2000) also found that every variabil-ity class can be expressed as a pattern of transitions betweenthree quasi-stable phenomenological states.Previous works have noted that some of the variabil-ity classes seen in IGR J17091-3624 appear very similar tothose seen in GRS 1915 (e.g. Altamirano et al. 2011a; Zhanget al. 2014). However, although ρ -like classes in the two ob-jects both show lags between hard and soft X-rays photons,these lags appear to possess different signs (Altamirano et al.2011a). Additionally, at least two variability classes havebeen reported in IGR J17091 which have not yet been re-ported in GRS 1915 (Pahari et al. 2012). Previous workshave described some of the behaviour seen in IGR J17091in the context of the variability classes described by Belloniet al. 2000 for GRS 1915 (e.g. Altamirano et al. 2011a; Pa-hari et al. 2014). To further explore the comparison betweenGRS 1915 and IGR J17091, here we perform the first com-prehensive model-independent analysis of variability classesin IGR J17091 using the complete set of RXTE (Bradt et al.1993) data taken of the 2011-2013 outburst of the object. Wealso use data from all other X-ray missions that observed thesource during this time to analyse the long-term evolutionof the outburst.
In this paper, we report data from
RXTE , INTEGRAL , Swift , Chandra , XMM-Newton and
Suzaku covering the2011-2013 outburst of IGR J17091. Unless stated otherwise,all errors are quoted at the 1 σ level.In Figure 2 we present long-term lightcurves from RXTE , INTEGRAL and
Swift with one datapoint per obser-vation, as well as indicating when during the outburst
Chan-dra , XMM-Newton and
Suzaku observations were made.
RXTE
For our variability study, we focus on the data from theProportional Counter Array (
PCA , Jahoda et al. 1996)aboard the Rossi X-Ray Riming Experiment (
RXTE , Bradtet al. 1993). We analysed all 2011
PCA observations of IGRJ17091, corresponding to OBSIDs 96065-03, 96103-01 and96420-01. The observations taken for proposals 96065-03 and96103-01 were contaminated by the nearby X-ray source GX349+2 (Altamirano et al. 2011a; Rodriguez et al. 2011b). Assuch we only use observations performed for proposal 96420-01, corresponding to a total of 243 orbits from 215 separateobservations. These were offset by 25’ such that GX 349+2was not in the ◦ PCA field of view.
RXTE was decommis-sioned during a period of Sun constraint centred on MJD55907, and hence the last observation of IGR J17091 wastaken on MJD 55879.We extracted data from the native
FITS format usingour own software . To perform medium- to high-frequency( (cid:38) Hz) timing analysis, we merged files formatted in
PCA ’s ‘Good Xenon’ data mode and extracted their data atthe maximum time resolution ( ∼ . × − s) without ac-counting for the background. We divided these data into128 s segments as this allowed us to reach frequencies be-low ∼ . Hz, partly sampling the high amplitude quasi-periodic flaring behaviour seen in many classes. Using theFast Fourier Transform (FFT), we produced the power spec-trum of each segment separately. We then averaged thesespectra to create a one co-added Power Density Spectrum(PDS) for each observation.For low-frequency ( ≤ Hz) timing and correlated spec-tral/timing analysis, we rebinned the data to 0.5 s and nor-malised count rates by the number of proportional counters(PCUs) active in each observation. Our choice of 1 Hz al-lows us to analyse high amplitude ‘flaring’ behaviour (seenat frequencies (cid:46) . Hz) separately from the lower-amplitudebehaviour seen at (cid:38) Hz.We split the data into three energy bands: A (
PCA channels 0–14, ∼ – keV), B ( PCA channels 15–35, ∼ – keV) and C ( PCA channels 36–255, ∼ – keV). Wechose these energy bands to be consistent with the energybands used by the model-independent classification of vari-ability classes of GRS 1915 in Belloni et al. (2000). Foreach of the energy-filtered lightcurves produced we esti-mated background using pcabackest from the FTOOLS pack-age (Blackburn 1995) with the
PCA faint source backgroundmodel . In all observations, we found that counts in theC band were consistent with background. We then createdLightcurves L A and L B from background-subtracted photonscounted in the A and B bands respectively. We used theselightcurves to define the full-band lightcurve ( L T = L A + L B )and the soft colour ( C = L B / L A ) of each observation. Tocomplement the Fourier spectra, we also constructed Gener-alised Lomb-Scargle Periodograms of L T from each dataset, amodified version of the standard Lomb-Scargle periodogram(Lomb 1976; Scargle 1982) that takes into account errors inthe dataset (Irwin et al. 1989). Using the Lomb-Scargle peri-odogram instead of the Fourier periodogram here allows us https://github.com/jmcourt/pantheon http://heasarc.gsfc.nasa.gov/FTP/xte/calib_data/pca_bkgd/Faint/pca_bkgd_cmfaintl7_eMv20051128.mdl MNRAS , 1–24 (2017)
J.M.C. Court et al. F l u x ( m C r a b ) a) RXTE/PCA
ChandraXMM-NewtonSuzaku -2 -1 F l u x ( m C r a b ) b) Swift/XRT
Windowed TimingPhoton Counting F l u x ( m C r a b ) c) Swift/BAT55600 55700 55800 55900 56000 56100 56200 56300 56400 56500 56600Time (MJD)020406080100120140160 F l u x ( m C r a b ) d) INTEGRAL/ISGRI F e b M a r A p r M a y J un J u l A u g S e p O c t N o v D e c J a n F e b M a r A p r M a y J un J u l A u g S e p O c t N o v D e c J a n F e b M a r A p r M a y J un J u l A u g S e p O c t N o v -4 -3 -2 -1 F l u x ( m C r a b ) Figure 2.
RXTE (Panel a),
Swift/XRT (Panel b),
Swift/BAT (Panel b) and
INTEGRAL /IBIS (Panel d) lightcurves of IGR J17091-3624during its 2011-2013 outburst. Arrows mark times at which
XMM-Newton (blue),
Chandra (red) or
Suzaku (magenta) observed IGRJ17091-3624. The cyan line represents MJD 55963, the approximate time IGR J17091-3624 transitions from the soft to the hard state(Drave et al. 2012).
RXTE/PCA (Jahoda et al. 1996) data are for the 2–16 keV energy band and taken from (Altamirano et al. 2011a),
Swift/BAT (Barthelmy 2000) data are for 15–50 keV,
Swift/XRT (Burrows et al. 2003) data are for 0.3–10 keV and
INTEGRAL/ISGRI (Ubertini et al. 2003) data are for 20-40 keV. Note that the data from
Swift/XRT (Panel B) are shown with a logarithmic y -axis tobetter show the late time progression of the outburst. Data points are coloured according to the observing mode used. The Swift/XRT data from times later than MJD 56422 are shown to a different scale to better represent the post-outburst evolution of the source. Alldata are presented in 1 day bins, except for data from
Swift/BAT which is presented in 4 day bins. See also Figure 3, in which datafrom
RXTE/PCA is presented on a smaller scale. The Crab count rates used to normalise these data were 2300 cts s − PCU − , 747.5cts s − , 0.214 cts s − and 183.5 cts s − for RXTE , Swift/XRT , Swift/BAT and
INTEGRAL/ISGRI respectively.
RXTE data have notbeen corrected for the 25’ offset to avoid contamination from GX 349+2, and for all instruments we implicitly assume that IGR J17091presents a Crab-like spectrum. MNRAS000
RXTE data have notbeen corrected for the 25’ offset to avoid contamination from GX 349+2, and for all instruments we implicitly assume that IGR J17091presents a Crab-like spectrum. MNRAS000 , 1–24 (2017) xotic Variability in IGR J17091-3624 to sample the low-frequency behaviour of lightcurves withdata gaps. This is important, for example, in lightcurveswhich show two populations of flares, as it allows each pop-ulation to be studied independently by cropping the otherfrom the lightcurve.We also used data from Altamirano et al. 2011a to sam-ple the long-term colour evolution of IGR J17091. We use 2hardness ratios defined by Altamirano et al.: H A and H A ,corresponding to the ratios of the 2–3.5 keV band againstthe 3.5–6 keV band and the 6–9.7 keV band against the 9.7–16 keV band respectively.When possible, if low-frequency peaks were present inthe Lomb-Scargle spectrum of an observation, we used theposition of the highest peak to define a value for a period.This period was then used to rebin data by phase (or ‘fold’the data) to search for reccurent hysteretic patterns in thehardness-Intensity diagram (hereafter HID , a plot of L T against C ). We found that quasi-periodic oscillations in ourobservations tended to show significant frequency shifts ontimescales shorter than the length of the observations. Assuch, we employed the variable-period folding algorithm out-lined in Appendix A where appropriate. For cases in whichthis algorithm was not appropriate, we considered small sec-tions of each lightcurve, with a length equivalent to smallnumber of periods, before performing folding.Additionally, in observations which showed a pattern ofhigh-amplitude X-ray flaring in L T , we used our own algo-rithm to find individual flares (this algorithm is described inAppendix A) and collect statistics on the amplitude, dura-tion and profile of these events.A list of all observations used in this study can be foundin Appendix B. Swift
In this paper, we consider data from the Burst Alert Tele-scope (
BAT , Barthelmy 2000), and the X-ray Telescope(
XRT , Burrows et al. 2003) aboard Swift Gamma-ray BurstMission (Gehrels 2004). IGR J17091-3624 was observed with
XRT for a total of 172 pointed XRT observations betweenMJDs 55575 and 56600, corresponding to Target IDs 31921,34543, 30967, 30973, 31920, 35096, 67137, 81917, 522245,677582 and 677981. These observations were interruptedduring sun constraints centred on MJDs 55907 and 56272.We created a long-term 0.3–10 keV
Swift/XRT light curve,with one bin per pointed observation, using the online light-curve generator provided by the UK Swift Science Data Cen-tre (UKSSDC; Evans et al. 2007). We have also created along-term 15–50 keV lightcurve using the publicly available
Swift/BAT daily-averaged light curve . These are shown inFigure 2 Panels (b) and (c) respectively. INTEGRAL
The INTErnational Gamma Ray Astrophisical Laboratory(
INTEGRAL , Winkler et al. 2003) is a medium-sized ESAmission launched in 2002. Unique hard X-ray (15–1000 keVwith the ISGRI detector plane) sensitivity and wide field of http://swift.gsfc.nasa.gov/results/transients/weak/IGRJ17091-3624/ view make INTEGRAL ideally suited to surveying the hardX-ray sky (see Bird et al. (2016) and Krivonos et al. (2015)for recent surveys).
INTEGRAL observations are divided into short ( ∼ INTE-GRAL /IBIS (Ubertini et al. 2003) between MJD 55575–55625 where the source is less than 12 degrees from thecentre of the field of view and where there is more the 1 ksof good ISGRI time per ScW. This corresponds to the spec-trally hardest period of the 2011-2013 outburst. The filter-ing of observations results in a total of 188 Science Win-dows which were processed using the Offline Science Anal-ysis (OSA) software version 10.2 following standard datareduction procedures in four energy bands (20–40, 40–100,100–150, 150–300 keV). These bands were selected as theyare standard energy bands used in the surveys of Bird et al.(2016) and Bazzano et al. (2006) and allow comparison tothese previous works. Images were created at the ScW levelas well as a single mosaic of all Science Windows in eachenergy band. XMM-Newton
In this paper we only consider data from the European Pho-ton Imaging Camera (
EPIC
Bignami et al. 1990) aboard
XMM-Newton (Jansen et al. 2001).
EPIC consists of onetelescope with a pn-type CCD (
EPIC-pn , Str¨uder et al.2001) and two telescopes with MOS CCDs (
EPIC-MOS1 and -MOS2 , Turner et al. 2001).
XMM/Newton observed IGR J17091 thrice during theperiod from 2011–2013 (represented by the blue arrows inFigure 2). One of these (OBSID 0721200101) was made on12 September 2013; we do not consider this observation fur-ther as IGR J17091 had returned to quiescence by this time(Altamirano et al. 2013). The remaining two observations,corresponding to OBSIDs 0677980201 and 0700381301 re-spectively, were taken on March 27 2011 (MJD 55647) andSeptember 29 2012 (MJD 56199).During observation 0677980201,
EPIC-pn was operat-ing in burst mode and
EPIC-MOS was operating in tim-ing mode. Given the low efficiency of burst mode, we onlyconsider data from
EPIC-MOS for this observation. Duringobservation 0700381301,
EPIC-pn was operating in timingmode, and thus we use data from
EPIC-pn for this observa-tion.We used the
XMM-Newton
Science Analysis Softwareversion 15.0.0 (
SAS , see Ibarra et al. 2009) to extract cali-brated event lists from
EPIC in both observations. We usedthese to construct lightcurves to study the X-ray variability,following standard analysis threads . Chandra
In this paper we consider data from the Advanced CCDImaging Camera (
ACIS , Nousek et al. 1987) and the HighResolution Camera (
HRC , Murray et al. 1987) aboard
Chan-dra (Weisskopf 1999).
Chandra made 7 observations of IGR MNRAS , 1–24 (2017)
J.M.C. Court et al.
OBSID Instrument Grating Exposure (ks) Mode MJD
HRC-I
NONE 1.13 I ACIS-S
HETG 31.21 C ACIS-S
HETG 27.29 T Table 1.
Chandra observations log covering the three observa-tions considered in this paper. I refers to Imaging mode, C refersto CC33 Graded mode and T refers to Timed Exposure Faintmode. HETG refers to the High Energy Transmission Grating. J17091 during the period 2011–2013. Four of these observa-tions were taken after IGR J17091 returned to quiescence,and we do not consider these further in this paper. TheChandra observations log is reported in Table 1.We analysed these data using
CIAO version 4.8 (Frus-cione et al. 2006), following the standard analysis threads.In order to apply the most recent calibration files (CALDB4.7.0, Graessle et al. 2006), we reprocessed the data fromthe three observations using the chandra_repro script , andused this to produce data products following standard pro-cedures.The first Chandra observation(OBSID 12505) of thissource was made shortly after it went into outburst in Febru-ary 2011. It was a 1 ks observation performed to refine theposition of the X-Ray source, using the High-ResolutionCamera in Imaging mode (HRC-I). We created the 0.06–10 kev light curve accounting for the Dead Time Factor(DTF) to correct the exposure time and count rate, dueto the deviation of the detector from the standard detectionefficiency, using the dmextract tool in the CIAO software.Two additional observations (OBSIDs 12405 and 12406)were performed within 214 days of this first observation,using the High Energy Transmition Grating Spectrometer(HETGS) on board
Chandra . The incident X-Ray flux wasdispersed onto
ACIS using a narrow strip array configu-ration (ACIS-S). Continuous Clocking and Time Exposuremodes were use in each observation respectively (see Kinget al. 2012 for further details). We exclude any events below0.4 keV, since the grating efficiency is essentially zero belowthis energy. In the case of the OBSID 12405 observations wealso excluded the Flight Grade 66 events in the event file, asthey were not appropriately graded. We extracted the 0.5-10 kev HEGTS light curves, excluding the zeroth-order flux,adopting standard procedures.
Suzaku
In this paper, we only consider data from the X-ray Imag-ing Spectrometer (
XIS , Koyama et al. 2007) aboard
Suzaku (Mitsuda et al. 2007).
Suzaku observed IGR J17091 twiceduring the period 2011–2013; a 42.1 ks observation on Oc-tober 2–3, 2012 (MJD 56202–56203, ObsID: 407037010)and an 81.9 ks observation on February 19–21, 2013 (MJD56342–56344, ObsID: 407037020).
XIS consists of four X-rayCCDs (
XIS
0, 1, 2 and 3), and all them except for XIS 2 See e.g. http://cxc.harvard.edu/ciao/ahelp/chandra_repro.html were operating in the 1/4 window mode which has a mini-mum time resolution of 2 seconds.We analysed the
Suzaku data using
HEASOFT aepipeline and the latest calibration database(version 20160607). We extracted
XIS light curves in the0.7–10 keV range, and subtracted background individuallyfor XIS 0, 1 and 3 and then summed these to obtain thetotal background. We created Power density spectra (PDS)using powspec in the
XRONOS package.
The 2011-2013 outburst of IGR J17091-3624 was first de-tected with
Swift/XRT on MJD 55595 (3 Feb 2011) (Krimm& Kennea 2011), and was observed by
RXTE , Swift/XRT , Swift/BAT and
INTEGRAL/ISGRI (see Figure 2, Panelsa, b, c, d respectively). There were also pointed observationsby
XMM-Newton , Chandra and
Suzaku during this time (de-noted by coloured arrows in Figure 2).
RXTE data were taken within the first week of the out-burst, but they were heavily contiminated by the nearbysource GX 349+2 (Rodriguez et al. 2011b). Thus these dataare not considered here.The onset of the outburst can be seen in the
Swift/BAT lightcurve (Figure 2 Panel c). In a 22 day period betweenMJDs 55584 and 55608, the 15–50 keV intensity from IGRJ17091 rose from ∼ mCrab to a peak of ∼ mCrab.This onset rise in intensity can also be seen in 0.3–10 keV Swift/XRT data and 20–40 keV
INTEGRAL/ISGRI data.After peak intensity, the 15–50 keV flux (
Swift/BAT )began to steadily decrease, until returning to a level of ∼
20 mCrab by MJD 55633. A similar decrease in flux can beseen in the data obtained by
INTEGRAL at this time (Fig-ure 2 Panel (d). However, there was no corresponding fallin the flux at lower energies; both the long-term 2–16 keV
RXTE data and
Swift/XRT data (Panels a and b respec-tively) show relatively constant fluxes of 45 mCrab betweenMJDs 55608 and 55633.The significant decrease in high-energy flux during thistime corresponds to IGR J17091 transitioning from a hardstate to a soft intermediate state (Pahari et al. 2014). Thistransition coincides with a radio flare reported by Rodriguezet al. (2011a) which was observed by the Australian Tele-scope Compact Array (
ATCA ).Altamirano et al. 2011b first reported a 10 mHz QPOin
RXTE data on MJD 55634 , evolving into ‘Heartbeat-like’ flaring by MJD 55639 (Altamirano et al. 2011c). Be-tween MJDs 55634 and 55879, the global
RXTE lightcurveshows large fluctuations in intensity on timescales of daysto weeks, ranging from a minimum of ∼ mCrab on MJD55731 to a maximum of ∼ mCrab on MJD 55756. The Swift/XRT lightcurve shows fluctuations that mirror thoseseen by
RXTE during this period, but the amplitude of thefluctuations is significantly reduced.
Swift/XRT was unable to observe again until MJD55952. Between this date and MJD 55989,
Swift/XRT ob-served a gradual decrease in intensity corresponding to areturn to the low/hard state (Drave et al. 2012).
MNRAS000
MNRAS000 , 1–24 (2017) xotic Variability in IGR J17091-3624 Between MJD 55989 and the end of the outburst onMJD 56445, we see secondary peaks in the
Swift/XRT , Swift/BAT and
INTEGRAL/ISGRI lightcurves that evolveover timescales of (cid:46) days. Similar humps have beenseen before in lightcurves from other objects, for examplethe black hole candidate XTE J1650-500 (Tomsick et al.2003) and the neutron stars SAX J1808.4-3658 (Wijnandset al. 2001) and SAX J1750.8-2900 (Allen et al. 2015). Thesehumps are referred to as ‘re-flares’ (also as ‘rebrightenings’,‘echo-outbursts’, ‘mini-outbursts’ or a ‘flaring tail’, e.g. Pa-truno et al. 2016). We see a total of 3 apparent re-flares inthe
Swift/BAT data, centred approximately at MJDs 56100,56220 and 56375.The observation with
XMM-Newton/EPIC-pn on MJD56547 (12 September 2013) recorded a rate of 0.019 cts s − .An observation with EPIC-pn in 2007, while IGR J17091was in quiescence (Wijnands et al. 2012), detected a similarcount rate of 0.020 cts s − . Therefore we define MJD 56547as the upper limit on the endpoint of the 2011-2013 outburst.As such the outburst, as defined here, lasted for (cid:46)
952 days.After the end of the 2011-2013 outburst, IGR J17091remained in quiescence until the start of a new outburstaround MJD 57444 (26 February 2016, Miller et al. 2016).
RXTE
Using the data products described in Section 2, we assigneda model-independent variability class to each of the 243
RXTE /PCA orbits. To avoid bias, this was done withoutreference to the classes defined by Belloni et al. (2000) todescribe the behaviour of GRS 1915.Classes were initially assigned based on analysisof lightcurve profiles, count rate, mean fractional RMS(Vaughan et al. 2003), Fourier and Lomb-scargle power spec-tra and hardness-intensity diagrams. For observations withsignificant quasi-periodic variability at a frequency lowerthan ∼ Hz, we also attempted to fold lightcurves to anal-yse count rate and colour as a function of phase. When flareswere present in the lightcurve, we used our algorithm (de-scribed in Appendix A) to sample the distribution of pa-rameters such as peak flare count rate, flare rise time andflare fall time. All parameters were normalised per activePCU, and fractional RMS values were taken from 2–60 keVlightcurves binned to 0.5 s. We identify nine distinct classes,labelled I to IX; we describe these in the following sections.Although the criteria for assigning each class to an ob-servation was different, a number of criteria were given themost weight. In particular, the detection, q -value and peakfrequency of a QPO in the range 2 Hz–10 Hz were used ascriteria for all classes, as well as the presence or absence ofhigh-amplitude quasi-periodic flaring with a frequency be-tween 0.01–1 Hz. The folded profile of these flares, as wellas the presence of associated harmonics, were also used asclassification diagnostics in observations. Additionally, thepresence or absence of low count-rate ’dips’ in a lightcurvewas used as a criterion for Classes VI, VIII and IX. Detailedcriteria for each individual class are given below in Sections3.2.1 to 3.2.9.For hardness-intensity diagrams, we describe loopingbehaviour with the terms ‘clockwise’ and ‘anticlockwise’; inall cases, these terms refer to the direction of a loop plotted in a hardness-intensity diagram with colour on the x -axisand intensity on the y -axis.In Appendix B, we present a list of all orbits used inthe study along with the variability classes we assigned tothem.In Figure 3, we show global 2–16 keV lightcurves of IGRJ17091 during the 2011-2013 outburst. In each panel, all ob-servations of a given class are highlighted in red. A charac-teristic lightcurve is also presented for each class. In Figure4 panel (a), we show a plot of average hardness H A against H A for each observation, showing the long-term hysteresisof the object in colour-colour space. Again, observations be-longing to each variability class are highlighted. In Figure4 panels (b) and (c), we show global hardness-intensity dia-grams for H A and H A respectively.In Figure 4 Panel (a), we see that IGR J17091-3624traces a two branched pattern in colour-colour space corre-sponding to a branch which is soft ( ∼ . ) in H A and vari-able in H A and a branch which is soft ( ∼ . ) in H A andvariable in H A . The ‘soft’ HID shown in Figure 4 Panel (b)is dominated by a branch with a wide spread in H A andintensities between ∼ – mCrab. A second branch existsat lower intensities, and shows an anticorrelation betweenintensity and H A . Finally, the ‘hard’ HID shown in Figure4 Panel (c) shows an obvious anticorrelation between H A and intensity, but there is also a secondary branch between H A ≈ . – . at a constant intensity of ∼ mCrab.For characteristic count rates and colours in each class,we quote the upper and lower quartile values (Kenney 1939)instead of the mean. This is due to the presence of high-amplitude but short-lived flares in many of the classes wedescribe. Using the upper and lower quartiles as our measureof average and distribution means that our values will beless susceptible to outlier values of count rate and colourpresent in these flares. All count rates have been backgroundcorrected (see Section 2.1).We have obtained mean values for these count rate quar-tiles, as well as values for colour C and fractional RMS,by calculating these values individually for each orbit. His-tograms were then constructed from these datasets for eachclass, such that the mean and standard deviation of thesevalues could be measured for each class. These values arepresented in Table 2.We describe QPOs in terms of their q -value; a measureof coherence defined by the ratio of peak frequency and full-width half-maximum of each QPO. We collected these valuesby fitting our power spectra with Lorentzians.For each class, we present three standard data prod-ucts; a 500 s lightcurve, a variable-length lightcurve wherethe length has been selected to best display the variabilityassociated with the class and a Fourier PDS. Unless other-wise stated in the figure caption, the 500 s lightcurve and theFourier PDS are presented at the same scale for all classes.In Table 3 we present a tally of the number of times weassigned each Variability Class to an RXTE orbit.
In the 2 s binned lightcurve of a Class I observation, there isno structured second-to-minute scale variability. The FourierPDS of all observations in this class shows broad band noise
MNRAS , 1–24 (2017)
J.M.C. Court et al. I n t e n s i t y ( m C r a b ) Class V204060 Class VI204060 Class VII204060 Class VIII0 50 100 150 200 250 300Date (MJD-55620)0204060 Class IX
Figure 3.
Global 2–3.5 keV Lightcurves of IGR J17091-3524 during the 2011-2013 outburst, with each point corresponding to the meanCrab-normalised count rate of a single
RXTE observation of the object (in turn corresponding to between 0.4 and 3.6 ks of data). Ineach lightcurve, every observation identified as belonging to a particular class (indicated on the plot) is highlighted. These are presentedalong with a characteristic lightcurve (inset) from an observation belonging to the relevant class. Each lightcurve is 250 s in length, andhas a y -scale from 0 to 250 cts s − PCU − . Data taken from Altamirano et al. 2011a. MNRAS000
RXTE observation of the object (in turn corresponding to between 0.4 and 3.6 ks of data). Ineach lightcurve, every observation identified as belonging to a particular class (indicated on the plot) is highlighted. These are presentedalong with a characteristic lightcurve (inset) from an observation belonging to the relevant class. Each lightcurve is 250 s in length, andhas a y -scale from 0 to 250 cts s − PCU − . Data taken from Altamirano et al. 2011a. MNRAS000 , 1–24 (2017) xotic Variability in IGR J17091-3624 A1 (3.5-6keV / 2-3.5keV)0.40.50.60.70.80.91.0 H A ( . - k e V / - . k e V ) IIIIIIIVVVIVIIVIIIIX (a)
Colour-Colour Diagram A1 (3.5-6keV / 2-3.5keV)010203040506070 F l u x ( m C r a b ) (b) ”Soft” ( H A ) Hardness-Intensity Diagram A2 (9.7-16keV / 6-9.7keV)010203040506070 F l u x ( m C r a b ) (c) ”Hard” ( H A ) Hardness-Intensity Diagram Figure 4.
A global colour-colour diagram (a), ”soft” hardness-intensity diagram (b) and ”hard” hardness-intensity diagram (c)of the 2011-2013 outburst of IGR J17091, using the colours H A and H A defined previously. Observations belonging to differentclasses have been highlighted in different colours. Data taken fromAltamirano et al. 2011a. between ∼ – Hz, as well as a weak QPO (with a q -valueof ∼ ) which peaks at around 5 Hz. Class II observations are a factor of ∼ fainter in the L T band than Class I observations. They also occupy a differentbranch in a plot of hardness H A against flux (see Figure 4c).The PDS shows no significant broad band noise above ∼ Hz Class LQ Rate UQ Rate Frac. RMS Median C (cts s − ) (cts s − ) I 84–108 106–132 0.13–0.19 0.4–0.68II 43–57 59–71 0.15–0.23 0.4–0.68III 64–84 80–110 0.17-0.23 0.35–0.45IV 63–81 92–122 0.27–0.37 0.32–0.4V 49–67 88–134 0.44–0.54 0.28–0.46VI 64–98 111–155 0.29–0.47 0.33–0.61VII 65–79 128–140 0.45–0.57 0.32–0.42VIII 62–88 142–178 0.42–0.52 0.36–0.49IX 87–111 114–144 0.16–0.24 0.42-0.6
Table 2.
Lower and upper quartile count rates, fractional RMSand median colour averaged across all observations belonging toeach class. Count rates and fractional RMS are taken from thefull energy range of
RXTE/PCA , and fractional RMS values are2–60 keV taken from lightcurves binned to 0.5 s. Count rates arenormalised for the number of PCUs active during each observa-tion. All values are quoted as σ ranges. Class Orbits Total Time (s) Fraction
I 31 69569 14.8%II 26 50875 10.8%III 14 26228 5.6%IV 31 69926 14.9%V 35 72044 15.3%VI 29 54171 11.5%VII 11 19241 4.1%VIII 16 26553 5.7%IX 50 81037 17.3%
Table 3.
A tally of the number of times we assigned each ofour nine Variability Classes to an
RXTE orbit. We have alsocalculated the amount of observation time corresponding to eachclass, and thus inferred the fraction of the time that IGR J17091spent in each class. Note: the values in the Total Time columnassume that each orbit only corresponds to a single variabilityClass.
Figure 5.
Plots of the Class I observation 96420-01-01-00, orbit 0.
Top-left : 1000 s lightcurve binned on 2 seconds to show lightcurveevolution.
Top-right : Fourier Power Density Spectrum.
Bottom :500 s lightcurve binned on 2 seconds.MNRAS , 1–24 (2017) J.M.C. Court et al.
Figure 6.
Plots of the Class II observation 96420-01-11-00, or-bit 0.
Top-left : 1000 s lightcurve binned on 2 seconds to showlightcurve evolution.
Top-right : Fourier Power Density Spectrum.
Bottom : Lightcurve binned on 2 seconds.
Figure 7.
Plots of the Class III observation 96420-01-04-01, or-bit 0.
Top-left : 1000 s lightcurve binned on 2 seconds to showlightcurve evolution.
Top-right : Fourier Power Density Spectrum.
Bottom : Lightcurve binned on 2 seconds. Note that, to emphasisethe behaviour of the lightcurve in this class, we have magnifiedthe 500 s lightcurve y-scale by a factor of 2 compared with thelightcurves presented for other classes. unlike that which is seen in Class I. The ∼ Unlike Classes I & II, Class III lightcurves show structuredflaring, with a peak-to-peak recurrence time of – s. Mostflares consist of a steady ∼ s rise in count rate and then anadditional and sudden rise to a peak count rate at (cid:38) ctss − PCU − which lasts for (cid:46) G e n e r a li s e d L o m b - S c a r g l e P o w e r Figure 8.
The Lomb-Scargle periodogram of observation 96420-19-01, orbit 0, with significance levels of 1, 2 and 3 σ plotted. Thepeak at 0.31 Hz was used to define a QPO frequency when foldingthe data from this observation. from every flare feature. No 5Hz QPO is present in the PDSand there is no significant variability in the range between ∼ – Hz.As this class has a well-defined periodicity, we foldeddata in each observation to improve statistics using thebest-fit period obtained from generalised Lomb-Scargle Pe-riodogram Analysis; we show a representative Lomb-Scargleperiodogram in Figure 8. We find an anticlockwise hystereticloop in the folded HID of all 15 Class III orbits. In Figure9 we show an example of one of these loops. The lightcurves in this class show regular variability with apeak-to-peak recurrence time of – s. We performed peakanalysis (see Appendix A) on observations belonging to thisclass, finding that each peak has a rise time with lower andupper quartile values of . and . s, a fall time with lowerand upper quartile values of . and . s and a peak countrate of – cts s − PCU − . There are no prominentsignificant QPOs in the Fourier PDS above ∼ Hz.We folded individual Class IV lightcurves and found an-ticlockwise hysteretic loops in the HID of 14 out of 30 ClassIV observations. In the top panel of Figure 11 we show anexample of one of these loops. However, we also find clock-wise hysteretic loops in 6 Class IV observations, and in 10orbits the data did not allow us to ascertain the presence ofa loop. We provide an example of both of these in the lowerpanels of Figure 11. We note that the structure of clockwiseloops are more complex than anticlockwise loops in Class IV,consisting of several lobes rather than a single loop (Figure11, bottom-left).Compared with Class III, the oscillations in Class IVoccur with a significantly lower period, with a mean peak- In HIDs with multiple lobes, the loop direction we assign to theobservation corresponds to the direction of the largest lobe.MNRAS000
The Lomb-Scargle periodogram of observation 96420-19-01, orbit 0, with significance levels of 1, 2 and 3 σ plotted. Thepeak at 0.31 Hz was used to define a QPO frequency when foldingthe data from this observation. from every flare feature. No 5Hz QPO is present in the PDSand there is no significant variability in the range between ∼ – Hz.As this class has a well-defined periodicity, we foldeddata in each observation to improve statistics using thebest-fit period obtained from generalised Lomb-Scargle Pe-riodogram Analysis; we show a representative Lomb-Scargleperiodogram in Figure 8. We find an anticlockwise hystereticloop in the folded HID of all 15 Class III orbits. In Figure9 we show an example of one of these loops. The lightcurves in this class show regular variability with apeak-to-peak recurrence time of – s. We performed peakanalysis (see Appendix A) on observations belonging to thisclass, finding that each peak has a rise time with lower andupper quartile values of . and . s, a fall time with lowerand upper quartile values of . and . s and a peak countrate of – cts s − PCU − . There are no prominentsignificant QPOs in the Fourier PDS above ∼ Hz.We folded individual Class IV lightcurves and found an-ticlockwise hysteretic loops in the HID of 14 out of 30 ClassIV observations. In the top panel of Figure 11 we show anexample of one of these loops. However, we also find clock-wise hysteretic loops in 6 Class IV observations, and in 10orbits the data did not allow us to ascertain the presence ofa loop. We provide an example of both of these in the lowerpanels of Figure 11. We note that the structure of clockwiseloops are more complex than anticlockwise loops in Class IV,consisting of several lobes rather than a single loop (Figure11, bottom-left).Compared with Class III, the oscillations in Class IVoccur with a significantly lower period, with a mean peak- In HIDs with multiple lobes, the loop direction we assign to theobservation corresponds to the direction of the largest lobe.MNRAS000 , 1–24 (2017) xotic Variability in IGR J17091-3624 Figure 9.
The hardness-intensity diagram (HID ) of the Class IIIobservation 96420-01-04-01, orbit 0. The data have been foldedover a period of 79.61 s, corresponding to the peak frequency inthe Lomb-Scargle spectrum of this observation. Inset is the foldedlightcurve of the same data. Figure 10.
Plots of the Class IV observation 96420-01-05-00,orbit 0.
Top-left : 1000 s lightcurve binned on 2 seconds to showlightcurve evolution.
Top-right : Fourier Power Density Spectrum.
Bottom : Lightcurve binned on 0.5 seconds. to-peak recurrence time of ∼ s compared to ∼ s in ClassIII. In Figure 4 we show that Classes III and IV can alsobe distinguished by average hardness, as Class III tends tohave a greater value of H A than Class IV. Figure 11.
Top : The hardness-intensity diagram (HID ) of theClass IV observation 96420-01-05-00, orbit 0 showing an anti-clockwise loop. The data have been folded over a variable periodfound with the algorithm described in Appendix A. Inset is thefolded lightcurve of the same data. Bottom Left : The hardness-intensity diagram of Class IV observations 96420-01-24-02 orbit0, an example of a clockwise loop.
Bottom Right : The hardness-intensity diagram of Class IV observation 96420-01-06-00 orbit 0,in which we were unable to ascertain the presence of a loop.
The lightcurves in this class, like in Classes III and IV, showflaring behaviour, with flares separated by a few tens of sec-onds. At higher frequencies, the PDS shows a prominentQPO centred at ∼ Hz with as q -value of ∼ . There is alsosignificant broad band noise between ∼ . – HzIn Figure 13 we show that the flaring in this class ismore complex than that seen in Classes III and IV. ClassV lightcurves consist of short strongly peaked symmetricalflares (hereafter Type V ) and a longer more complex type offlare (hereafter Type V ). The Type V flare consists of a fastrise to a local maximum in count rate, followed by a ∼ speriod in which this count rate gradually reduces by ∼ and then a much faster peak with a maximum count ratebetween 1 and 2 times that of the initial peak. In both typesof flare, we find that the increase in count rate correspondswith an increase in soft colour. The two-population nature MNRAS , 1–24 (2017) J.M.C. Court et al.
Figure 12.
Plots of the Class V observation 96420-01-06-03,orbit 0.
Top-left : 750 s lightcurve binned on 2 seconds to showlightcurve evolution.
Top-right : Fourier Power Density Spectrum.
Bottom : Lightcurve binned on 0.5 seconds.
Figure 13.
A portion of the lightcurve of observation 96420-01-06-03, orbit 0, showing Type V flares (highlighted in cyan) andType V flares (highlighted in red). of flares in Class V can also clearly be seen in Figure 14,where we show a two-dimensional histogram of flare peakcount rate against flare duration.We folded all individual Class V lightcurves, in eachcase cropping out regions of V flaring. We find clockwisehysteretic loops in the HID of 30 out of 33 Class V ob-servations, suggesting a lag in the aforementioned relationbetween count rate and soft colour. In the upper panel Fig-ure 15 we present an example of one of these loops. In oneobservation however, we found an anticlockwise loop in theHID (shown in Figure 15 lower-left panel). We were unableto ascertain the presence of loops in the remaining 2 orbits;for the sake of completeness, we show one of these in thelower-right panel of Figure 15. Figure 14.
Every flare in all observations identified as Class V,plotted in a two-dimensional histogram of flare peak count rateagainst flare duration to show the two-population nature of theseevents.
The lightcurves of observations of this class show large dipsin count rate; this can be seen in Figure 16 at, for example, t ≈ – s . These dips vary widely in duration, from ∼ to ∼ seconds, and the count rate in both L A and L B fallto a level consistent with background. The dips’ rise and falltimes are fast, both lasting no longer than a second. Theydo not appear to occur with any regular periodicity.Aside from the dips, Class VI observations show otherstructures in their lightcurves. Large fluctuations in countrate, by factors of (cid:46) , occur on timescales of ∼ – s; no pe-riodicity in these oscillations could be found. This behaviouris reflected in the PDS, which shows high-amplitude broadband noise below ∼ . Hz with RMS-normalized power (Bel-loni & Hasinger 1990) of up to ∼ . Hz − . As can be seenin Figure 16, this feature takes the form of a broad shoulderof noise which shows a either weak peak or no clear peak atall. The ∼ Hz QPO seen in the PDS of other classes is notpresent in Class VI observations.We attempted to fold all individual Class VI lightcurves,ignoring the sections of data corresponding to the largecount rate dips described above. In general, foldinglightcurves belonging to this class is difficult; many orbitsshowed low-amplitude oscillations which were difficult tofold using our flare-finding algorithm (see Appendix A),while many others only showed oscillatory behaviour for asmall number of periods between each pair of dips. As such,we only succesfully folded 23 of the 40 Class VI orbits. Ofthese, 19 showed clockwise loops in the HID (top panel,Figure 17), 3 showed anticlockwise loops (bottom-left panel,Figure 17). In the remaining 1 observation, the data did notallow us to ascertain the presence of loops (bottom-rightpanel, Figure 17).Like in Class VI, we note that the clockwise loops inClass VI appear more complex than clockwise loops. Again,the clockwise loop shown in Figure 17 appears to have a 2-lobe structure; this is repeated in all clockwise loops foundin this class. MNRAS000
The lightcurves of observations of this class show large dipsin count rate; this can be seen in Figure 16 at, for example, t ≈ – s . These dips vary widely in duration, from ∼ to ∼ seconds, and the count rate in both L A and L B fallto a level consistent with background. The dips’ rise and falltimes are fast, both lasting no longer than a second. Theydo not appear to occur with any regular periodicity.Aside from the dips, Class VI observations show otherstructures in their lightcurves. Large fluctuations in countrate, by factors of (cid:46) , occur on timescales of ∼ – s; no pe-riodicity in these oscillations could be found. This behaviouris reflected in the PDS, which shows high-amplitude broadband noise below ∼ . Hz with RMS-normalized power (Bel-loni & Hasinger 1990) of up to ∼ . Hz − . As can be seenin Figure 16, this feature takes the form of a broad shoulderof noise which shows a either weak peak or no clear peak atall. The ∼ Hz QPO seen in the PDS of other classes is notpresent in Class VI observations.We attempted to fold all individual Class VI lightcurves,ignoring the sections of data corresponding to the largecount rate dips described above. In general, foldinglightcurves belonging to this class is difficult; many orbitsshowed low-amplitude oscillations which were difficult tofold using our flare-finding algorithm (see Appendix A),while many others only showed oscillatory behaviour for asmall number of periods between each pair of dips. As such,we only succesfully folded 23 of the 40 Class VI orbits. Ofthese, 19 showed clockwise loops in the HID (top panel,Figure 17), 3 showed anticlockwise loops (bottom-left panel,Figure 17). In the remaining 1 observation, the data did notallow us to ascertain the presence of loops (bottom-rightpanel, Figure 17).Like in Class VI, we note that the clockwise loops inClass VI appear more complex than clockwise loops. Again,the clockwise loop shown in Figure 17 appears to have a 2-lobe structure; this is repeated in all clockwise loops foundin this class. MNRAS000 , 1–24 (2017) xotic Variability in IGR J17091-3624 Figure 15.
Top : The hardness-intensity diagram (HID ) of atype V flaring region in Class V observation 96420-01-07-00, orbit0 showing a clockwise loop. The data have been folded over avariable period found with the algorithm described in AppendixA. Inset is the folded lightcurve of the same data. Bottom Left :The hardness-intensity diagram of Class V observation 96420-01-25-05 orbit 0, an example of an anticlockwise loop.
Bottom Right :The hardness-intensity diagram of Class V observation 96420-01-25-06 orbit 0, in which we were unable to ascertain the presenceof a loop.
Class VII shows high-amplitude flaring behaviour with apeak-to-peak recurrence time of – s. In Figure 19 we showa dynamical Lomb-Scargle spectrogram of a Class VII ob-servation, showing that the fast flaring behaviour has a fre-quency which moves substantially over time. This in turnaccounts for the large spread in the value of the flare peak-to-peak recurrence time.In Figure 19 we show that the peak frequency of theQPO also varies in a structured way. We also suggest thatthe variabilitity of the frequency is itself a QPO with a pe-riod of ∼ .At higher frequencies, the PDS shows a weak QPOscentred at ∼ Hz, with a q -values of ∼ .We used our flare-finding algorithm (see Appendix A)to perform variable-frequency folding of Class VII orbits. We Figure 16.
Plots of the Class VI observation 96420-01-09-00,orbit 0.
Top-left : 750 s lightcurve binned on 2 seconds to showlightcurve evolution.
Top-right : Fourier Power Density Spectrum.
Bottom : Lightcurve binned on 1 second. find clockwise loops in 9 out of 11 Class VII orbits. In theremaining two observations, the oscillations were extremelyfast. As a result, the errors in the HID of these too observa-tions were too large to succesfully select peaks, and we areunable to confirm or reject the presence of loops. The lightcurve of this variability class shows the dipping be-haviour seen in Class VI, as can be seen in Figure 20 at t ≈ – s. The dips are less frequent than in Class VI.The behaviour outside of the dips is dominated by highlystructured high-amplitude oscillations consisting of flareswith a peak to peak separation of . ± . s. The PDS showsthis behaviour as a very significant ( q -value >
20) QPO; twoharmonics of this QPO are also visible. The PDS also showsa strong ( q -value = 4.7) QPO at ∼ Hz.We attempted to fold Class VIII lightcurves, ignoringthe portions of data corresponding to dips, using our flare-finding algorithm. The high frequency of the dominant os-cillation in Class VIII resulted in large errors in the peaktimes of individual flares, which translated to large errors inall HID s; however, we were able to ascertain the presence inloops in 8 out of 16 orbits. All 8 of these loops are clockwise. The 1 s lightcurve of a Class IX observation is superficiallysimilar to the lightcurve of a Class I observation, with littleobvious structured variability at timescales larger than 2 s;however, large count rate dips like those seen in Classes VIand VIII (e.g. the feature at t ≈ s in the lightcurve ofFigure 21) are very occasionally observed. These dips mayin turn be coupled to short second-scale flares in which countrate briefly increases by a factor of 2–3.Outside of these dips and flares, the lightcurve of a ClassIX observation is indistinguishable from the lightcurve of aClass I or Class II observation. However, in Figure 4, we showthat Class IX occupies a very different part of the global MNRAS , 1–24 (2017) J.M.C. Court et al.
Figure 17.
Top : The hardness-intensity diagram (HID ) of theClass VI observation 96420-01-30-03, orbit 0 showing a clockwiseloop. The data have been folded over a variable period foundwith the algorithm described in Appendix A. Inset is the foldedlightcurve of the same data. Bottom Left : The hardness-intensitydiagram of Class VI observation 96420-01-30-04 orbit 0, an ex-ample of an anticlockwise loop.
Bottom Right : The hardness-intensity diagram of Class VI observation 96420-01-09-03 orbit0, in which we were unable to ascertain the presence of a loop. H A / H A colour-colour diagram. Class IX observations showa significantly larger H A than Class I and II observations,but a significantly lower H A .The PDS reveals significant broad band noise peaked at ∼ ∼ Hz QPO seen in other classes is absent.Altamirano & Belloni (2012) discovered high frequency ( ∼ Hz) QPOs in observations corresponding to this variabilityclass.
Observations with
Swift took place throughout the 2011-2013 outburst of IGR J17091-3624. Between MJDs 55622and 55880, 17
Swift/XRT were at least partly simultaneouswith an
RXTE observation, corresponding to at least oneobservation of all 9 classes. In each case, the
Swift and
RXTE lightcurves were similar. The remainder of the
Swift/XRT
Figure 18.
Plots of the Class VII observation 96420-01-18-05,orbit 0.
Top-left : 750 s lightcurve binned on 2 seconds to showlightcurve evolution.
Top-right : Fourier Power Density Spectrum.
Bottom : Lightcurve binned on 0.5 seconds.
Figure 19.
A sliding window Lomb-Scargle spectrogram of ClassVII observation 96420-01-18-05, showing power density spectrafrom an overlapping 32 s window moved 1 s at a time. The peakfrequency of this low frequency QPO itself appears to oscillatewith a frequency of ∼ mHz. observations during this time were also consistent with be-longing to one of our nine classes. Given that the RXTE datahave higher count rate and time resolution, we do not fur-ther discuss the
Swift observations taken before MJD 55880.A more detailed comparison of
RXTE and
Swift data is be-yond the scope of this paper.Between MJD 55952 and 56445,
Swift observationsshowed IGR J17091-3624 decreasing in flux. For all obser-vations longer than 500 s, we rebinned the lightcurves to10 s and calculated the RMS. We find the lower and upperquartiles of the fractional RMS in these measurements tobe 18.3% and 21.7% respectively.
INTEGRAL observationstaken as part of a scan programme of the Galactic Plane(Fiocchi et al. 2012) and reported by Drave et al. (2012)suggest that IGR J17091-3624 returned to the hard state
MNRAS000
MNRAS000 , 1–24 (2017) xotic Variability in IGR J17091-3624 Figure 20.
Plots of the Class VIII observation 96420-01-19-03,orbit 0.
Top-left : 300 s lightcurve binned on 2 seconds to showlightcurve evolution.
Top-right : Fourier Power Density Spectrum.
Bottom : Lightcurve binned on 0.5 seconds. Inset is a zoom of the25 s portion of the lightcurve highlighted in cyan, to show thesecond-scale structure in the lightcurve.
Figure 21.
Plots of the Class IX observation 96420-01-35-02,orbit 1.
Top-left : 1200 s lightcurve binned on 2 seconds to showlightcurve evolution.
Top-right : Fourier Power Density Spectrum.
Bottom : Lightcurve binned on 2 seconds. between MJDs 55952 and 55989. Therefore these observa-tions sample IGR J17091-3624 the hard state.
The results of the
INTEGRAL /IBIS analysis are presentedin Table 4. We see clear detections of IGR J17091-3624 in allenergy bands during the hardest period (MJD 55575–55625)of the 2011–2013 outburst. Conversion from detected countsto flux was achieved using an
INTEGRAL /IBIS observationof the Crab taken between MJD 57305.334 and 57305.894.Conversion from Crab units to standard flux units was ob-tained by conversion factors listed in Bird et al. (2016) andBazzano et al. (2006).
Figure 22.
INTEGRAL /ISGRI 150–300 keV significance map ofa ◦ region centred on the position of IGR J17091-3624, showingthe first significant detection of this source above 150 keV. Thedetection significance is 7.6 σ . Comparing these results with those of Bazzano et al.(2006), we see that IGR J17091 is detected for the firsttime above 150 keV with a detection significance of 7.6 σ ,corresponding to a flux of . ± . × − ergs s − cm − (Figure 22). In Figure 23, we present lightcurves from the three
Chandra observations considered in this paper (see also Table 1 fordetails of these observations).Observation 12505 was performed within 24 hours of
RXTE observation 96420-01-02-01, which showed Class Ivariability. No structured variability is seen in the lightcurveof OBSID 12505 (Figure 23, upper panel), which is consis-tent with Class I. Note that we consider the energy range0.06-10 keV for this observation but 0.5-10 keV for observa-tions 12405 and 12406.Observation 12405 was performed within 24 hours of
RXTE observation 96420-01-23-03, which showed Class Vvariability. The two observations were not simultaneous;OBSID 12405 began ∼ . ks after OBSID 96420-01-2303finished. The lightcurve of Chandra
OBSID 12405 (shownin Figure 23, middle panel) shows a mean count rate of41 cts s − . The lightcurve shows fast flaring behaviour (witha recurrence time on the order of 10s of seconds) in whichthe frequency changes widely on timescales of ∼ s. Thisobservation strongly resembles a Class VII lightcurve, butwith its characteristic timescales increased by a factor of ∼ .This leads to the possibility that the low number of ClassVII RXTE observations we identify is due to a selection ef-fect; we would not have been able to see this observation’slong-term Class VII-like behaviour if the observation hadbeen shorter than ∼ ks.Observation 12406 was performed within 24 hours of MNRAS , 1–24 (2017) J.M.C. Court et al.
Energy Intensity Significance Exposure Flux Flux(keV) (cts/s) σ (ks) (mCrab) (10 − ergs s − cm − )20–40 12.39 ± ± ± ± ± ± ± ± ± ± ± ± Table 4.
Results from the IBIS/ISGRI analysis of the 2011–2013 Outburst of IGR J17091. The 20–40 keV flux is given in units ofmCrab and (10 − ergs s − cm − ). Conversion between counts and mCrab was obtained using an observation of the Crab taken duringRevolution 1597 between MJD 57305.334 and 57305.894 and the conversion factors of Bird et al. (2016) and Bazzano et al. (2006). C o un t s / s OBSID 124050 200 400 600 800 1000Time (s)020406080100 OBSID 12406
Figure 23.
Chandra observations 12505, 12405 and 12406, showing Class I, Class VIIand Class IX variability respectively. The lightcurve presentedfor observation 12505 is for the energy range 0.06-10 keV, whilethe other two lightcurves are for the energy range 0.5-10 keV. Allthree lightcurves are binned to 0.5 s.
RXTE observation 96420-01-32-06, which showed Class IXvariability. The lightcurve presented for
Chandra
OBSID12406 shows a mean count rate (36 cts s − ), which is con-sistent with IGR J17091 being harder in this observationthan in Observation 12505. This, combined with the lack ofvariability seen in its lightcurve, suggests that Observation12505 is consistent with Class IX. In Figure 24 we show lightcurves from two
XMM-Newton observations. The lightcurve of
XMM-Newton observation0677980201, shown in the upper panel of Figure 24, showsthe regular flares characteristic of Class IV variability. Asimultaneous
RXTE observation (OBSID 96420-01-05-000)also showed Class IV variability.
XMM-Newton observation 070038130, shown in thelower panel of Figure 24, was made after the end of
RXTE observations IGR J17091-3624. As such it cannot be com-pared with contemporaneous
RXTE data. The 5 s binned R a t e ( c t s / s ) OBSID 06779802010 200 400 600 800 1000Time (s)050100150200 R a t e ( c t s / s ) OBSID 0700381301
Figure 24.
Lightcurves of
XMM-Newton observations0677980201 and 0700381301, showing Class IV variabilityand the hard state respectively. Both lightcurves binned to 2 s.Data for observation 0677980201 is taken from
EPIC-MOS2 anddata for observation 0700381301 is taken from
EPIC-pn . lightcurve shows no apparent variability, but a Fourier PDSof the observation (shown in Figure 25) reveals a QPO cen-tred at around ∼ . Hz and a broad band noise componentat lower frequencies. Drave et al. (2012) reported that IGRJ17091 transited to the hard state in February 2012, sevenmonths before this observation was taken. As such, we findthat observation 0677980201 samples the hard state in IGRJ17091 and is thus beyond the scope of our set of variabilityclasses.
Suzaku
The two
Suzaku observations of IGR J17091-3624 consid-ered, OBSIDs 407037010 and 407037020, were performedduring the 2nd and 3rd re-flares of the hard state phase ofthe 2011–2013 outburst. OBSID 407037010 was taken simul-taneously with
XMM-Newton observation 0700381301. TheXIS 0 count rates are 7.8 cts s − and 2.5 cts s − respectively.Neither lightcurve shows ‘heartbeats’ or any other typeof GRS 1915-like variability. However, we find evidence ofa low frequency QPO feature at ∼ XMM-Newton observa-tion 0700381301 (Figure 25). The presence of a QPO below1 Hz and flat-topped power density spectrum confirm thatIGR J17091 was in the hard state at this time.
MNRAS000
MNRAS000 , 1–24 (2017) xotic Variability in IGR J17091-3624 -3 -2 P o w e r ( ν P ( ν ) ) XMM-NewtonSuzaku
Figure 25. ν P ( ν ) -normalised co-added power density spectraof XMM-Newton observation 0700381301 and
Suzaku observa-tion 407037010. Both observations were taken simultaneouslyon September 29 2012 (MJD 56199). We sample observation0700381301 up to a frequency of 10 Hz, while the 2 s time res-olution of observation 407037010 results in a Nyquist frequencyof 0.25 Hz.
Figure 26.
A lightcurve of observation 96420-01-06-02, orbit 0,showing a transition in behaviour between Class IV (in cyan, seeSection 3.2.4) and Class V (in red, see Section 3.2.5).
Using observations from
XMM-Newton , RXTE and
Chan-dra , we describe the complex variability seen in IGR J17091as a set of nine variability ‘classes’, labelled I to IX. Theseclasses are distinguished from each other by values of upperand lower quartile (i.e. 25 th and 75 th percentile) count rates,mean RMS, the presence of QPOs in Fourier PDS, the shapeof flare and dip features in the lightcurve and the presence ofloops in the 6–16/2–6 keV hardness-intensity diagram HID .See Section 3 for a full description of these classes.The classification of some observations is clearer thanothers. Some orbits were too short to definitively quantifythe behaviour of the source, whereas some other orbits con-tain a transition between two classes. An example lightcurveshowing a transition from Class III to Class IV is presentedin Figure 26. Table 5.
The nine variability classes of IGR J17091-3624, show-ing the name of the closest corresponding variability class in GRS1915+105. The names of GRS 1915+105 classes are taken fromBelloni et al. (2000), where more detailed descriptions can befound. Eight additional classes of GRS 1915+105 have been de-scribed; we do not find analogies to these classes in IGR J17091-3624. IGR J17091-3624 Class GRS 1915+105 ClassI χ II φ III ν IV ρ V µ VI λ VII
None
VIII
None IX γ Our set of classes is analogous to, but not based upon,the set of variability classes defined by Belloni et al. 2000 todescribe the behaviour of the similarly complex LMXB GRS1915. This ensures that our set of classes is not biased by an apriori assumption that the two objects are similar. Howeverif we do assume that wide range of variability seen in thesetwo objects are driven by the same physical processes, adirect comparison between the variability classes in the twosystems can further our understanding of the physics thatdrive these exotic objects.We also use all 2011-2013 IGR J17091-3624 data from
RXTE , XMM-Newton , Chandra , Swift , INTEGRAL and
Suzaku to analyse the long-term evolution of the 2011–2013outburst. This in turn corresponds to all available X-raydata taken during this outburst.
As observations of IGR J17091 and GRS 1915 suffer fromdifferent values of interstellar absorption N H , we cannot di-rectly compare the absolute colours of these two objects.However, we can compare the evolution of colour both overtime and as a function of count rate. We therefore use theseparameters, along with power spectra and lightcurve mor-phology, when comparing GRS 1915 with IGR J17091.For seven of our classes, we were able to assign the clos-est matching class described by Belloni et al. 2000 for GRS1915 (see Table 5). We are unable to find analogues to ourclasses VII and VIII in observations of GRS 1915, and wesuggest that these classes are unique to IGR J17091.Below, we evaluate our mapping between GRS 1915 andIGR J17091 classes, and interpret the differences betweeneach matched pair. Classes I and II both show low count rates and little struc-ture in their lightcurves. The two classes in GRS 1915 thatalso show this lightcurve behaviour are Class χ and Class Note that, in GRS 1915+105, Class χ is further subdivided intofour classes based on hard colour (Belloni et al. 2000; Pahari et al.MNRAS , 1–24 (2017) J.M.C. Court et al. φ . Belloni et al. 2000 differentiate between Classes φ and χ based on the hard colour (corresponding to C ), as Class χ has a significantly higher value for this colour than Class φ .Data from RXTE indicates that the transition fromthe hard state to the soft intermediate state between MJDs55612 and 55615 (Drave et al. 2012). This was confirmedby a radio spectrum taken on MJD 55623 which was consis-tent with an observation of discrete ejecta (Rodriguez et al.2011a). This observation of discrete ejecta at the transitionbetween the hard state and the intermediate state has beenreported in other LMXBS (e.g. XTE J1550-564, Rodriguezet al. 2003), and has also been associated with transitionsto the χ Class in GRS 1915 (Rodriguez et al. 2008, see alsoreview by Fender 2006).Using Fourier PDS, we conclude that Class I is analo-gous to Class χ in GRS 1915, while Class II is analogous toClass φ . In Class χ observations of GRS 1915, broad bandnoise between ∼ − Hz and a QPO at around 5Hz are seenin the PDS. We find that both of these are present in Class Iobservations of IGR J17091. On the other hand, we find thatClass φ observations of GRS 1915 do not show this broadband noise, and show either a weak ( q -value (cid:46) ) QPO at ∼ Hz or no QPO at all. We find that the weak QPO andlack of broad band noise are also seen in the PDS of ClassII observations.
Classes III and IV both show highly regular flaring activityin their lightcurves, but they differ in terms of timescale andpulse profile. As can be seen in lightcurves in Figure 10, flaresin Class IV occur every ∼ s and are nearly identical to eachother in shape. On the other hand, as can be seen in Figure7, flares in Class III occur every ∼ s and may or maynot end in a much faster sharp peak which is never seen inClass IV. In Figure 27 we show a two-dimensional histogramof flare peak count rate against flare duration, showing allflares in all observations classified as Class III or Class IV.In this figure, we can see that flares tend to group in one oftwo regions in count rate-duration space; a region between ∼ – cts s − PCU − and ∼ – s, corresponding to flaresseen in Class III, and a region between ∼ – cts s − PCU − and ∼ – s, corresponding to flares seen in ClassIV. From this plot, we conclude that the flares seen in ClassIII exist in a different population to the flares seen in ClassIV. The GRS 1915 classes that show behaviour most similarto these are ρ and ν ; both produce similar structures in theirlightcurve, but Class ν is differentiated from Class ρ by thepresence of a secondary count rate peak which occurs ∼ safter the primary (Belloni et al. 2000).The secondary peak is present in most Class III observa-tions and some Class IV observations (Figure 28), suggest-ing that both classes consist of a mix of ρ -like and ν -likeobservations. However, the poor statistics sometimes makethe presence of this secondary peak difficult to detect. Assuch, we do not use the presence or absence of this peak as χ as a single variability class here. Figure 27.
Every flare in all observations identified as Class IIIor Class IV, plotted in a two-dimensional histogram of flare peakcount rate against flare duration to show the two-population na-ture of these events. Flares belonging to Class IV occupy the dis-tribution at higher peak rate and lower duration, whereas flaresbelonging to Class III occupy the distribution at lower peak rateand higher duration.
Figure 28.
Lightcurve from Class III observation 96420-01-10-01of IGR J17091-3624, with pairs of primary and secondary countrate spikes highlighted in cyan and red respectively. The yellowregion highlights a primary count rate spike that did not producea secondary. a criterion when assigning classes. Instead we choose to sep-arate Classes III and IV based on the larger-scale structurein their lightcurves (see Section 3.2.4). Due to the afore-mentioned difference in burst populations between the twoclasses, we suggest that classes III and IV do represent twodistinct classes rather than a single class with a period thatdrifts over time. We suggest that Classes ρ and ν in GRS1915 could also be re-partitioned in this way.However, HID loops are found to generally execute inan anticlockwise direction in Classes III and IV (previouslynoted by e.g. Altamirano et al. 2011a); the opposite direc-tion to the clockwise loops in Classes ρ and ν reported by MNRAS000
Lightcurve from Class III observation 96420-01-10-01of IGR J17091-3624, with pairs of primary and secondary countrate spikes highlighted in cyan and red respectively. The yellowregion highlights a primary count rate spike that did not producea secondary. a criterion when assigning classes. Instead we choose to sep-arate Classes III and IV based on the larger-scale structurein their lightcurves (see Section 3.2.4). Due to the afore-mentioned difference in burst populations between the twoclasses, we suggest that classes III and IV do represent twodistinct classes rather than a single class with a period thatdrifts over time. We suggest that Classes ρ and ν in GRS1915 could also be re-partitioned in this way.However, HID loops are found to generally execute inan anticlockwise direction in Classes III and IV (previouslynoted by e.g. Altamirano et al. 2011a); the opposite direc-tion to the clockwise loops in Classes ρ and ν reported by MNRAS000 , 1–24 (2017) xotic Variability in IGR J17091-3624 e.g. Belloni et al. 2000 and repeated by us using the samemethod we apply to data from IGR J17091-3624 (see Section2). This suggests that Classes III and IV could be generatedby a different physical mechanism to Classes ρ and ν . Alter-natively, Classes III and IV could be generated by the samemechanism as ρ and ν if some other unknown process wasable to alter the spectral evolution of flares in these classes. The lightcurve of a Class V observation appears similar tothat of a Class µ observation of GRS 1915, as both are char-acterised by rapid ρ -like flares which occur less regularlythan in Class ρ . In addition to this, flares in Class µ fallinto two clear populations, as do the flares in Class V. How-ever, significant differences exist between Class V and Class µ . Class µ observations are characterised by long ( ∼ s)excursions to plateaus of high count rate, a behaviour whichis not seen in any Class V observation thus far.We note that the HID in Class V observations displaysa loop in the clockwise direction; the opposite direction tothe looping seen in Classes III and IV but the same directionseen in Class µ .Regarding the two-population nature of flares seen inthis class (see Section 3.2.5), we suggest that V flares maysimply be two V flares that occur close together in time,such that the second flare starts during the decay of thefirst flare. This would result in an apparent two-peaked flarestructure, as we see in type V flares. This interpretationalso accounts for the bimodal distribution of flare duara-tions shown in the 2D histogram of Figure 14, as this couldbe caused by the misinterpretation of two-flare V eventsas a single event. This also accounts for the Gaussian dis-tribution of peak flare intensities seen in Figure 14), as theconstituents of each V event would be from the same pop-ulation as V flares. Class VI is dominated by long flaring periods which separateperiods of low count rate, as can be seen in the lightcurvepresented in Figure 16. Similar behaviour is seen in thelightcurves of observations of GRS 1915 belonging to Classes λ and ω (Klein-Wolt et al. 2002). However, the long countrate ‘dips’ are far less regular in Class VI than in Classes λ and ω , and we also note long periods of medium countrate during which neither flares nor dips occur. This vari-ability class is noted by Pahari et al. (2012) who suggestthat this class is unique to IGR J17091 . However, Pahariet al. (2013a) show that, in a plot of burst decay time againstburst rise time, Classes VI and λ fall in a straight line, sug-gesting a similar physical origin for both.While it is cetainly true that Class VI is not a perfectanalogue of either Class λ or Class ω , Class VI only differsnoticeably from Class λ during the extended low-variabilityportions of its lightcurves. As such, we associate Class VIwith Class λ . Pahari et al. (2012) refers to Class VI as Class C2.
We are unable to find an analogue of Class VII in obser-vations of GRS 1915. This class, and its apparent unique-ness, have previously been noted by Pahari et al. 2012 .Pahari et al. found that the C hard colour in this class in-creases during count rate dips and decreases during countrate peaks. Here we reproduced the results of Pahari et al.and found that the anti-correlation between hard-colour andintensity is not physical, but due to the definition of C : thecount rate in band L C is approximately constant and con-sistent with background, and therefore C = L C / L A ∝ L − A ,which will naturally anticorrelate with intensity.Although a correlation between QPO frequency andcount rate has been noted in the ∼ Hz QPO seen in GRS1915 (e.g. Markwardt et al. 1999; Vignarca et al. 2003), thisQPO is also seen in Class VII observations at the same timeas the ∼ . Hz QPO. As such, the flux-frequency relation-ship in the very low frequency ( ∼ . Hz) QPO in Class VII isapparently unique amongst the classes of both IGR J17091and GRS 1915.
We are unable to find an analogue of Class VIII in observa-tions of GRS 1915. When it is flaring, the lightcurve wave-form is similar to that seen in Class ρ , with rapid regularspikes in count rate. The lightcurve also shows irregular dipsin count rate similar to those seen in Class VI and in Class λ in GRS 1915.However, the amplitude of the flares in Class VIII ismuch larger, and the frequency much higher, than in ClassesVI or λ . The amplitude of the flares in Class VIII can ap-proach ∼ cts s − PCU − , while the flare separation timeof 4–5 s makes Class VIII the fastest flaring activity seen inany class of IGR J17091 or GRS 1915. As such, we considerthis variability class distinct from both Class VI and Class λ . Class IX is defined by long periods of high amplitude butunstructured variability (with a broad peaked noise compo-nent in the Fourier spectrum peaked at ∼ ∼ – . A similarity be-tween this Class and Class γ in GRS 1915 has been previ-ously noted by Altamirano & Belloni (2012). However, theirregular spikes seen in some Class IX lightcurves are notreproduced in Class γ lightcurves of GRS 1915. Overall, variability in IGR J17091 tends to be faster thanstructurally similar variability in GRS 1915, as can be notedin Classes III and IV compared to Classes ρ and ν (see alsoAltamirano et al. 2011a). Additionally, IGR J17091 also dis-plays highly structured variability unlike anything yet seen Pahari et al. (2012) refers to Class VII as Class C1.MNRAS , 1–24 (2017) J.M.C. Court et al. in GRS 1915, with classes VII and VIII in particular showingvery fine detail in their lightcurves.In total we find 2 variability classes which are seen inIGR J17091 but not in GRS 1915, compared with 8 that areseen in GRS 1915 but not in IGR J17091. As relatively lit-tle data exists on GRS 1915-like variability in IGR J17091,the presence of classes in GRS 1915 that are not seen inIGR J17091 could simply be an observational effect. It isunknown how long each variability class lasts for and, assuch, additional variability classes could have occurred en-tirely while IGR J17091 was not being observed (however,see Huppenkothen et al. 2017 for a study on GRS1915 basedon more than 16 years of data). However, GRS 1915 hasdisplayed variability classes consistently since its discoveryin 1992, implying that the two classes seen only in IGRJ17091 are either completely absent in GRS 1915 or thatthey occur with a much lower probability. In either case,this implies physical differences between methods of gener-ating GRS 1915-like variability in the two objects.As noted in sections 4.1.1 to 4.1.7, variability classesseen in both IGR J17091 and GRS 1915 show differencesin the different objects. In particular, we note the pres-ence of irregular flares in Class IX which are not seen inthe analogous Class γ . If these classes are indeed generatedby the same processes in both objects, the differences be-tween them must represent physical differences between theobjects themselves.It has previously been noted that, while the hardnessratios in IGR J17091 and GRS 1915 during ρ -like classes aredifferent, the fractional hardening between the dip and peakof each flare is consistent with being the same in both objects(Capitanio et al. 2012). This suggests that the same physicalprocess is behind the ‘heartbeats’ seen in both objects.We note the presence of hysteretic HID loops in someclasses of both objects. Although these loops are alwaysclockwise in GRS 1915, they can be executed in either direc-tion in IGR J17091. Classes in IGR J17091 that show loopsall have a preferred loop direction: anticlockwise in ClassesIII and IV and clockwise in classes V, VI, VII and VIII.In cases where the loop direction was opposite to that ex-pected for a given class, loop detections were generally onlymarginally significant. In particular, we note that ClassesIV and V tend to show loops in opposite directions, despitethe similarities between their lightcurves and the ρ , ν and µ classes in GRS 1915. The fact that IGR J17091 can showHID loops in both directions suggests that an increase insoft emission can either precede or lag a correlated increasein hard emission from IGR J17091. Whether soft emissionprecedes or lags hard emission is in turn is dependent on thevariability class.There are also non-trivial similarities between variabil-ity in the two objects. We note the presence of a ∼ HzQPO in many of the classes seen in IGR J17091, and thissame 5Hz QPO is seen in lightcurves of GRS 1915. Simi-larly Altamirano & Belloni (2012) reported the discovery ofa 66Hz QPO in IGR J17091; a very similar frequency to the67Hz QPO observed in GRS 1915 (Morgan et al. 1997). It isnot clear why these QPOs would exist at roughly the samefrequencies in both objects when other variability in IGRJ17091 tend to be faster.
Table 6.
The six OBSIDs explicitly classified in Altamirano et al.(2011a). We also present the GRS 1915 class with which we im-plicitly label each OBSID in this paper.OBSID Altamirano et al.
Class Court et al.
Class(implied)96420-01-04-03 α ρ / ν ν ρ / ν ρ ρ / ν ρ µ β / λ λ µ λ In 2015, Bagnoli & in’t Zand (2015) discovered the existenceof two GRS 1915-like variability classes in the neutron starbinary MXB 1730-335, also known as the ‘Rapid Burster’.Specifically, Bagnoli & in’t Zand (2015) note the presence ofvariability similar to Classes ρ and θ in GRS 1915.Class θ -like variability, seen in RXTE observation92026-01-20-02 of the Rapid Burster, is not closely matchedby any of the classes we identify for IGR J17091. However,the lightcurves of a Class θ observation feature large dips incount rate similar to those seen in Classes VI and VIII inIGR J17091.Conversely, Class ρ -like variability is seen in all threeobjects. Bagnoli & in’t Zand (2015) note that the variabilityof the ρ -like flaring is slower in the Rapid Burster than ineither GRS 1915 or IGR J17091. It has previously been sug-gested that the maximum rate of flaring in LMXBs should beinversely proportional to the mass of the central object (e.g.Belloni et al. 1997; Frank et al. 2002). In this case, the factthat variability is faster in IGR J17091 than in GRS 1915could simply be due to a lower black hole mass in the for-mer object (Altamirano et al. 2011a). However if variabilityin the Rapid Burster is assumed to be physically analogousto variability in these two black hole objects, then we notethat a correlation between central object mass and variabil-ity timescale no longer holds. et al. Altamirano et al. (2011a) identify 5 GRS 1915 variabilityclasses in a subset of observations from the 2011-2013 out-burst of IGR J17091: six of these observations are presentedin Table 6 along with the best-fit GRS 1915 class that weassign it in this paper (see also Table 5).We acknowledge differences between the classificationsassigned by this paper and by Altamirano et al. (2011a). Weascribe these differences to the different approaches we haveused to construct our classes. In particular while we haveconstructed an independent set of variability classes for IGRJ17091 which we have then compared to the Belloni et al.classes for GRS 1915, Altamirano et al. applied the Belloniet al. classes for GRS 1915 directly to IGR J17091.In general, the variability classes we find to be present inIGR J17091 are broadly the same as those noted by Altami-rano et al. (2011a). We do not associate any class with Class α in GRS 1915, but we find examples of all of the other vari- MNRAS000
Class(implied)96420-01-04-03 α ρ / ν ν ρ / ν ρ ρ / ν ρ µ β / λ λ µ λ In 2015, Bagnoli & in’t Zand (2015) discovered the existenceof two GRS 1915-like variability classes in the neutron starbinary MXB 1730-335, also known as the ‘Rapid Burster’.Specifically, Bagnoli & in’t Zand (2015) note the presence ofvariability similar to Classes ρ and θ in GRS 1915.Class θ -like variability, seen in RXTE observation92026-01-20-02 of the Rapid Burster, is not closely matchedby any of the classes we identify for IGR J17091. However,the lightcurves of a Class θ observation feature large dips incount rate similar to those seen in Classes VI and VIII inIGR J17091.Conversely, Class ρ -like variability is seen in all threeobjects. Bagnoli & in’t Zand (2015) note that the variabilityof the ρ -like flaring is slower in the Rapid Burster than ineither GRS 1915 or IGR J17091. It has previously been sug-gested that the maximum rate of flaring in LMXBs should beinversely proportional to the mass of the central object (e.g.Belloni et al. 1997; Frank et al. 2002). In this case, the factthat variability is faster in IGR J17091 than in GRS 1915could simply be due to a lower black hole mass in the for-mer object (Altamirano et al. 2011a). However if variabilityin the Rapid Burster is assumed to be physically analogousto variability in these two black hole objects, then we notethat a correlation between central object mass and variabil-ity timescale no longer holds. et al. Altamirano et al. (2011a) identify 5 GRS 1915 variabilityclasses in a subset of observations from the 2011-2013 out-burst of IGR J17091: six of these observations are presentedin Table 6 along with the best-fit GRS 1915 class that weassign it in this paper (see also Table 5).We acknowledge differences between the classificationsassigned by this paper and by Altamirano et al. (2011a). Weascribe these differences to the different approaches we haveused to construct our classes. In particular while we haveconstructed an independent set of variability classes for IGRJ17091 which we have then compared to the Belloni et al.classes for GRS 1915, Altamirano et al. applied the Belloniet al. classes for GRS 1915 directly to IGR J17091.In general, the variability classes we find to be present inIGR J17091 are broadly the same as those noted by Altami-rano et al. (2011a). We do not associate any class with Class α in GRS 1915, but we find examples of all of the other vari- MNRAS000 , 1–24 (2017) xotic Variability in IGR J17091-3624 ability classes posited by Altamirano et al. to exist in IGRJ17091.Altamirano et al. 2011a noted the presence of an an-ticlockwise loop in the HID of ‘heartbeat’-like observationsof IGR J17091, opposed to the clockwise loop seen in HIDof ρ -class observations of GRS 1915. This is consistent withour finding that hysteretic loops in classes III and IV alsotend to execute in an anticlockwise direction. However, weadditionally find that hysteretic loops in classes V, VI, VIIand VIII tend to execute in a clockwise direction. This isalso different from GRS 1915, in which the loop is executedin the same direction in all classes. We also additionally re-port that clockwise loops tend to be more complex thananticlockwise loops seen in IGR J17091, with many showinga multi-lobed structure not seen in GRS 1915. This appar-ent inconsistency between the objects strengthens the sug-gestion in Altamirano et al. 2011a that the heartbeat-likeclasses in GRS 1915 and IGR J17091 may be generated byphysically different mechanisms. The constraints that Altamirano et al. 2011a placed on themass and distance of IGR J17091 assumed that the ob-ject emitted at its Eddington luminosity at the peak ofthe 2011–2013 outburst. They report a peak 2–50 keV fluxof × − ergs s − cm − during flares in ‘heartbeat’-likelightcurves during this time. The correction factor C Bol , Peak to convert 2–50 keV flux to bolometric flux is not well con-strained, but Altamirano et al. 2011a suggest an order-of-magnitude estimate of (cid:46) , corresponding to a peak bolo-metric flux of (cid:46) . × − ergs s − cm − .Maccarone 2003 performed a study of the soft to hardtransitions in 10 LMXBs. They found that all but one per-form this transition at a luminosity consistent with between1% and 4% of the Eddington limit. We use Swift observa-tion 00031921058 taken on MJD 55965 to create a spectrumof IGR J17091 during the approximate time of its transi-tion from a soft to a hard state (Drave et al. 2012). We fitthis spectrum above 2 keV with a power-law, and extrapo-late to find a 2–50 keV flux of . × − ergs s − cm − .Assuming that the transition bolometric correction factor C Bol , Tran is also (cid:46) , this corresponds to a bolometric flux of (cid:46) . × − ergs s − cm − .By comparing this with the results of Maccarone 2003and Altamirano et al. 2011a, we find that IGR J17091-3624was likely emitting at no more than ∼ –20% of its Edding-ton Limit at its peak. This number becomes ∼ – % if weinstead use C Bol , Tran = . , or ∼ – % if C Bol , Tran = . . Withthis new range of values, we are able to re-derive the com-pact object mass as the function of the distance (Figure 29).We find that for a black hole mass of ∼ M (cid:12) , as suggestedby Iyer et al. 2015a, IGR J17091 is within the galaxy at adistance of 6–17 kpc. This is consistent with the estimateddistance of ∼ – kpc estimated by Rodriguez et al. 2011afor a compact object mass of 10M (cid:12) . Figure 29.
Mass of the compact object in IGR J17091-3624 plot-ted against its distance, for values of peak Eddington fractions of F Edd = We have found that hysteretic HID loops can execute inboth directions in IGR J17091 (e.g. Section 4.4), as wellas found a revised estimate that IGR J17091 accretes at (cid:46) % Eddington (Section 4.5). Both of these findings haveimplications for physical models of GRS 1915-like variabilityin this source.Firstly, we find that Eddington-limited accretion is nei-ther necessary nor sufficient for GRS 1915-like variabil-ity. The discovery of GRS 1915-like variability in the sub-Eddington Rapid Burster (Bagnoli & in’t Zand 2015; Bag-noli et al. 2015) provided the first evidence that Eddington-limited accretion may not be a driving factor in this typeof variability. We strengthen this case by finding that IGRJ17091-3624 is also likely sub-Eddington. As such, we fur-ther rule out any scenario in which Eddington-limited accre-tion is required for GRS 1915-like variability in black holeLMXBs specifically.Secondly, by using the direction of hysteretic HID loops,we find that hard photon lag in ‘heartbeat’-like classes ofIGR J17091 can be either positive or negative. This couldmean that we must rule out the causal connection betweensoft and hard emission being common to all classes.In either case, we find that scenarios that require highglobal accretion rates or predict a consistent hard photonlag (e.g. Neilsen et al. 2011; Janiuk & Czerny 2005), arenot able to explain GRS 1915-like variability in IGR J17091unless they also feature geometric obscuration in a subset ofvariability classes. We note that simulations by Nayakshinet al. 2000 require an Eddington fraction of (cid:38) . beforeGRS 1915-like variability, a value which falls in the range ∼ . – . that we find for the peak Eddington fraction ofIGR J17091.In addition to being near its Eddington limit GRS 1915also has the largest orbit of any known LMXB (e.g. McClin-tock & Remillard 2006). S¸adowski 2016 have also shown thatthin, radiation dominated regions of disks in LMXBs requirea large-scale threaded magnetic field to be stable, and the MNRAS , 1–24 (2017) J.M.C. Court et al. field strength required to stabilise such a disk in GRS 1915is higher than for any other LMXB they studied. We sug-gest that one of these parameters is more likely to be thecriterion for GRS 1915-like variability. If better constraintscan be placed on the disk size and minimum stabilising fieldstrength in IGR J17091, it will become clear whether ei-ther of these parameters can be the unifying factor behindLMXBs that display GRS 1915-like variability.
We have constructed the first model-independent set of vari-ability classes for the entire portion of the 2011–2013 out-burst of IGR J17091 that was observed with
RXTE . We findthat the data are well-described by a set of 9 classes; 7 ofthese appear to have direct counterparts in GRS 1915, whiletwo are, so far, unique to IGR J17091. We find that variabil-ity in IGR J17091 is generally faster than in the correspond-ing classes of GRS 1915, and that patterns of quasi-periodicflares and dips form the basis of most variability in bothobjects. Despite this, we find evidence that ‘heartbeat’-likevariability in both objects may be generated by differentphysical processes. In particular, while hard photons alwayslag soft in GRS 1915, we find evidence that hard photonscan lag or precede soft photons in IGR J17091 depending onthe variability class.We also report on the long-term evolution of the 2011–2013 outburst of IGR J17091, in particular noting the pres-ence of 3 re-flares during the later part of the outburst. Usingan empirical relation between hard-soft transition luminos-ity and Eddington luminosity (Maccarone 2003), we esti-mate that IGR J17091 was likely accreting at no greaterthan ∼ % of its Eddington limit at peak luminosity.We use these result to conclude that any model of GRS1915-like variability which requires a near-Eddington globalaccretion rate is insufficient to explain the variability we seein IGR J17091. As such we suggest that an extreme valueof some different parameter, such as disk size or minimumstabilising large-scale magnetic field, may be the unifyingfactor behind all objects which display GRS 1915-like vari-ability. This would explain why sub-Eddington sources suchas IGR J17091 and the Rapid Burster do display GRS 1915-like variability, while other Eddington-limited sources suchas GX 17+2 and V404 Cyg do not. ACKNOWLEDGEMENTS
J.C. and C.B. thank the Science & Technology FacilitiesCouncil and the Royal Astronomical Society for their finan-cial support. D.A. thanks the Royal Society, and the In-ternational Space Science Institute (
ISSI ) for its supportduring the ‘The extreme physics of Eddington and superEddington accretion onto Black Hole’ meetings ( ). T.B. alsothanks the
ISSI for their support. R.W. is supported by aNWO Top grant, Module 1. M. Pereyra gratefully acknowl-edges the Committee on Space Research (COSPAR) Capac-ity Building and Schlumberger Foundation Faculty for theFuture Fellowship Programmes for their financial support. M.Pahari acknowledges the support of the UGC/UKIERIthematic partnership grant UGC 2014-15/02.The authors acknowledge the use of AstroPy(AstropyCollaboration et al. 2013), APLpy (Robitaille & Bressert2012), NumPy (Jones et al. 2001) and MatPlotLib (Hunter2007) libraries for Python. We also acknowledge the use ofpublic data from the
RXTE (Bradt et al. 1993) archive, aswell as
FTOOLS (Blackburn 1995) for data manipulation.This work made use of data supplied by the UK SwiftScience Data Centre at the University of Leicester.
REFERENCES
Allen J. L., Linares M., Homan J., Chakrabarty D., 2015, ApJ,801, 10Altamirano D., Belloni T., 2012, ApJ, 747, L4Altamirano D., et al., 2011a, ApJ, 742, L17Altamirano D., et al., 2011b, The Astronomer’s Telegram, 3225Altamirano D., et al., 2011c, The Astronomer’s Telegram, 3230Altamirano D., Wijnands R., Belloni T., 2013, The Astronomer’sTelegram, 5112Astropy Collaboration et al., 2013, A&A, 558, A33Bagnoli T., in’t Zand J. J. M., 2015, MNRAS, 450, L52Bagnoli T., in’t Zand J. J. M., D’Angelo C. R., Galloway D. K.,2015, MNRAS, 449, 268Barthelmy S. D., 2000, in Flanagan K. A., Siegmund O. H., eds,Proc. SPIEVol. 4140, X-Ray and Gamma-Ray Instrumenta-tion for Astronomy XI. pp 50–63Bazzano A., et al., 2006, ApJ, 649, L9Belloni T., Hasinger G., 1990, A&A, 230, 103Belloni T. M., Motta S. E., 2016, preprint, ( arXiv:1603.07872 )Belloni T., M´endez M., King A. R., van der Klis M., van ParadijsJ., 1997, ApJ, 479, L145Belloni T., Klein-Wolt M., M´endez M., van der Klis M., vanParadijs J., 2000, A&A, 355, 271Belloni T., Psaltis D., van der Klis M., 2002, ApJ, 572, 392Bignami G. F., et al., 1990, in Siegmund O. H. W., Hudson H. S.,eds, Proc. SPIEVol. 1344, EUV, X-ray, and Gamma-ray in-strumentation for astronomy. pp 144–153Bird A. J., et al., 2016, ApJS, 223, 15Blackburn J. K., 1995, in Shaw R. A., Payne H. E., Hayes J. J. E.,eds, Astronomical Society of the Pacific Conference Series Vol.77, Astronomical Data Analysis Software and Systems IV.p. 367Bradt H. V., Rothschild R. E., Swank J. H., 1993, A&AS, 97, 355Burrows D. N., et al., 2003, in Truemper J. E., TananbaumH. D., eds, Proc. SPIEVol. 4851, X-Ray and Gamma-RayTelescopes and Instruments for Astronomy.. pp 1320–1325,doi:10.1117/12.461279Capitanio F., Del Santo M., Bozzo E., Ferrigno C., De Cesare G.,Paizis A., 2012, MNRAS, 422, 3130Castro-Tirado A. J., Brandt S., Lund N., 1992, IAU Circ., 5590Done C., Wardzi´nski G., Gierli´nski M., 2004, MNRAS, 349, 393Drave S. P., Fiocchi M., Sguera V., Bazzano A., Bird A. J., SidoliL., Kuulker E., 2012, The Astronomer’s Telegram, 3916Evans P. A., et al., 2007, A&A, 469, 379Fender R., 2006, Jets from X-ray binaries. pp 381–419Fender R., Belloni T., 2004, ARA&A, 42, 317Fiocchi M. T., Natalucci L., GPS Team 2012, in Proceedings of”An INTEGRAL view of the high-energy sky (the first 10years)” - 9th INTEGRAL Workshop and celebration of the10th anniversary of the launch (INTEGRAL 2012). 15-19 Oc-tober 2012. Bibliotheque Nationale de France, Paris, France.Published online at http://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=176 , id.82. p. 82 MNRAS000
Allen J. L., Linares M., Homan J., Chakrabarty D., 2015, ApJ,801, 10Altamirano D., Belloni T., 2012, ApJ, 747, L4Altamirano D., et al., 2011a, ApJ, 742, L17Altamirano D., et al., 2011b, The Astronomer’s Telegram, 3225Altamirano D., et al., 2011c, The Astronomer’s Telegram, 3230Altamirano D., Wijnands R., Belloni T., 2013, The Astronomer’sTelegram, 5112Astropy Collaboration et al., 2013, A&A, 558, A33Bagnoli T., in’t Zand J. J. M., 2015, MNRAS, 450, L52Bagnoli T., in’t Zand J. J. M., D’Angelo C. R., Galloway D. K.,2015, MNRAS, 449, 268Barthelmy S. D., 2000, in Flanagan K. A., Siegmund O. H., eds,Proc. SPIEVol. 4140, X-Ray and Gamma-Ray Instrumenta-tion for Astronomy XI. pp 50–63Bazzano A., et al., 2006, ApJ, 649, L9Belloni T., Hasinger G., 1990, A&A, 230, 103Belloni T. M., Motta S. E., 2016, preprint, ( arXiv:1603.07872 )Belloni T., M´endez M., King A. R., van der Klis M., van ParadijsJ., 1997, ApJ, 479, L145Belloni T., Klein-Wolt M., M´endez M., van der Klis M., vanParadijs J., 2000, A&A, 355, 271Belloni T., Psaltis D., van der Klis M., 2002, ApJ, 572, 392Bignami G. F., et al., 1990, in Siegmund O. H. W., Hudson H. S.,eds, Proc. SPIEVol. 1344, EUV, X-ray, and Gamma-ray in-strumentation for astronomy. pp 144–153Bird A. J., et al., 2016, ApJS, 223, 15Blackburn J. K., 1995, in Shaw R. A., Payne H. E., Hayes J. J. E.,eds, Astronomical Society of the Pacific Conference Series Vol.77, Astronomical Data Analysis Software and Systems IV.p. 367Bradt H. V., Rothschild R. E., Swank J. H., 1993, A&AS, 97, 355Burrows D. N., et al., 2003, in Truemper J. E., TananbaumH. D., eds, Proc. SPIEVol. 4851, X-Ray and Gamma-RayTelescopes and Instruments for Astronomy.. pp 1320–1325,doi:10.1117/12.461279Capitanio F., Del Santo M., Bozzo E., Ferrigno C., De Cesare G.,Paizis A., 2012, MNRAS, 422, 3130Castro-Tirado A. J., Brandt S., Lund N., 1992, IAU Circ., 5590Done C., Wardzi´nski G., Gierli´nski M., 2004, MNRAS, 349, 393Drave S. P., Fiocchi M., Sguera V., Bazzano A., Bird A. J., SidoliL., Kuulker E., 2012, The Astronomer’s Telegram, 3916Evans P. A., et al., 2007, A&A, 469, 379Fender R., 2006, Jets from X-ray binaries. pp 381–419Fender R., Belloni T., 2004, ARA&A, 42, 317Fiocchi M. T., Natalucci L., GPS Team 2012, in Proceedings of”An INTEGRAL view of the high-energy sky (the first 10years)” - 9th INTEGRAL Workshop and celebration of the10th anniversary of the launch (INTEGRAL 2012). 15-19 Oc-tober 2012. Bibliotheque Nationale de France, Paris, France.Published online at http://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=176 , id.82. p. 82 MNRAS000 , 1–24 (2017) xotic Variability in IGR J17091-3624 Frank J., King A., Raine D. J., 2002, Accretion Power in Astro-physics: Third EditionFruscione A., et al., 2006, in Society of Photo-Optical Instru-mentation Engineers (SPIE) Conference Series. p. 62701V,doi:10.1117/12.671760Gehrels N., 2004, in Schoenfelder V., Lichti G., Winkler C., eds,ESA Special Publication Vol. 552, 5th INTEGRAL Workshopon the INTEGRAL Universe. p. 777Graessle D. E., Evans I. N., Glotfelty K., He X. H., Evans J. D.,Rots A. H., Fabbiano G., Brissenden R. J., 2006, in Society ofPhoto-Optical Instrumentation Engineers (SPIE) ConferenceSeries. p. 62701X, doi:10.1117/12.672876Hannikainen D. C., Hjalmarsdotter L., Rodriguez J., Vilhu O.,Zdziarski A. A., Belloni T., 2007, in ESA Special Publication.p. 353Heidke P., 1926, Geogr. Ann., 8, 301349Hoffman J. A., Marshall H. L., Lewin W. H. G., 1978, Nature,271, 630Hunter J. D., 2007, Computing In Science & Engineering, 9, 90Huppenkothen D., et al., 2016, preprint, ( arXiv:1610.08653 )Huppenkothen D., Heil L. M., Hogg D. W., Mueller A., 2017,MNRAS, 466, 2364Ibarra A., Calle I., Gabriel C., Salgado J., Osuna P., 2009, inBohlender D. A., Durand D., Dowler P., eds, AstronomicalSociety of the Pacific Conference Series Vol. 411, AstronomicalData Analysis Software and Systems XVIII. p. 322Irwin A. W., Campbell B., Morbey C. L., Walker G. A. H., YangS., 1989, PASP, 101, 147Iyer N., Nandi A., Mandal S., 2015a, in Astronomical Society ofIndia Conference Series.Iyer N., Nandi A., Mandal S., 2015b, ApJ, 807, 108Jahoda K., Swank J. H., Giles A. B., Stark M. J., Strohmayer T.,Zhang W., Morgan E. H., 1996, in Siegmund O. H., GumminM. A., eds, Proc. SPIEVol. 2808, EUV, X-Ray, and Gamma-Ray Instrumentation for Astronomy VII. pp 59–70Janiuk A., Czerny B., 2005, MNRAS, 356, 205Janiuk A., Czerny B., Siemiginowska A., 2000, ApJ, 542, L33Jansen F., et al., 2001, A&A, 365, L1Jones E., Oliphant T., Peterson P., et al., 2001, SciPy: Opensource scientific tools for Python,
Kenney J., 1939, Mathematics of Statistics. No. v. 1 in Mathe-matics of Statistics, D. Van Nostrand Company, Incorporated, https://books.google.co.uk/books?id=Vs1AAAAAIAAJ
King A. L., et al., 2012, ApJ, 746, L20Klein-Wolt M., Fender R. P., Pooley G. G., Belloni T., MigliariS., Morgan E. H., van der Klis M., 2002, MNRAS, 331, 745van der Klis M., 1989, ARA&A, 27, 517Kok C., 2000, On the Behaviour of a Few Popular VerificationScores in Yes No Forecasting. Scientific report, KoninklijkNederlands Meteorologisch Insutuut, https://books.google.co.uk/books?id=DIBSAAAACAAJ
Koyama K., et al., 2007, PASJ, 59, 23Krimm H. A., Kennea J. A., 2011, The Astronomer’s Telegram,3148, 1Krivonos R., Tsygankov S., Lutovinov A., Revnivtsev M., Chu-razov E., Sunyaev R., 2015, MNRAS, 448, 3766Kuulkers E., Homan J., van der Klis M., Lewin W. H. G., M´endezM., 2002, A&A, 382, 947Kuulkers E., Lutovinov A., Parmar A., Capitanio F., Mowlavi N.,Hermsen W., 2003, The Astronomer’s Telegram, 149Lightman A. P., Eardley D. M., 1974, ApJ, 187, L1Lomb N. R., 1976, Ap&SS, 39, 447Maccarone T. J., 2003, A&A, 409, 697Markwardt C. B., Swank J. H., Taam R. E., 1999, ApJ, 513, L37Massaro E., Ventura G., Massa F., Feroci M., Mineo T.,Cusumano G., Casella P., Belloni T., 2010, A&A, 513, A21Massaro E., Ardito A., Ricciardi P., Massa F., Mineo T., D’A`ıA., 2014, Ap&SS, 352, 699 McClintock J. E., Remillard R. A., 2006, Black hole binaries. pp157–213Miller J. M., Reynolds M., Kennea J., King A. L., Tomsick J.,2016, The Astronomer’s Telegram, 8742Mitsuda K., et al., 2007, PASJ, 59, 1Morgan E. H., Remillard R. A., Greiner J., 1997, ApJ, 482, 993Murray S. S., Chappell J. H., Elvis M. S., Forman W. R., GrindlayJ. E., 1987, Astrophysical Letters and Communications, 26,113Nayakshin S., Rappaport S., Melia F., 2000, ApJ, 535, 798Neilsen J., Remillard R. A., Lee J. C., 2011, ApJ, 737, 69Nousek J. A., Garmire G. P., Ricker G. R., Collins S. A., ReiglerG. R., 1987, Astrophysical Letters and Communications, 26,35Paczynski B., 1979, in Baity W. A., Peterson L. E., eds, X-rayAstronomy. pp 251–255Pahari M., Pal S., 2009, preprint, ( arXiv:0906.4611 )Pahari M., Bhattacharyya S., Yadav J. S., Pandey S. K., 2012,MNRAS, 422, L87Pahari M., Yadav J. S., Rodriguez J., Misra R., BhattacharyyaS., Pandey S. K., 2013a, ApJ, 778, 46Pahari M., Neilsen J., Yadav J. S., Misra R., Uttley P., 2013b,ApJ, 778, 136Pahari M., Yadav J. S., Bhattacharyya S., 2014, ApJ, 783, 141Patruno A., Maitra D., Curran P. A., D’Angelo C., FridrikssonJ. K., Russell D. M., Middleton M., Wijnands R., 2016, ApJ,817, 100Rebusco P., Moskalik P., Klu´zniak W., Abramowicz M. A., 2012,A&A, 540, L4Reid M. J., McClintock J. E., Steiner J. F., Steeghs D., RemillardR. A., Dhawan V., Narayan R., 2014, ApJ, 796, 2Remillard R. A., McClintock J. E., Sobczak G. J., Bailyn C. D.,Orosz J. A., Morgan E. H., Levine A. M., 1999a, ApJ, 517,L127Remillard R. A., Morgan E. H., McClintock J. E., Bailyn C. D.,Orosz J. A., 1999b, ApJ, 522, 397Robitaille T., Bressert E., 2012, APLpy: Astronomical Plot-ting Library in Python, Astrophysics Source Code Library(ascl:1208.017)Rodriguez J., Corbel S., Tomsick J. A., 2003, ApJ, 595, 1032Rodriguez J., et al., 2008, ApJ, 675, 1436Rodriguez J., Corbel S., Caballero I., Tomsick J. A., Tzioumis T.,Paizis A., Cadolle Bel M., Kuulkers E., 2011a, A&A, 533, L4Rodriguez J., Corbel S., Tomsick J. A., Paizis A., Kuulkers E.,2011b, The Astronomer’s Telegram, 3168S¸adowski A., 2016, MNRAS, 462, 960Scargle J. D., 1982, ApJ, 263, 835Shakura N. I., Sunyaev R. A., 1973, A&A, 24, 337Stefanov I. Z., 2014, MNRAS, 444, 2178Str¨uder L., et al., 2001, A&A, 365, L18Tomsick J. A., Kalemci E., Corbel S., Kaaret P., 2003, ApJ, 592,1100Turner M. J. L., et al., 2001, A&A, 365, L27Ubertini P., et al., 2003, A&A, 411, L131Vaughan S., Edelson R., Warwick R. S., Uttley P., 2003, MNRAS,345, 1271Vignarca F., Migliari S., Belloni T., Psaltis D., van der Klis M.,2003, A&A, 397, 729Vilhu O., 1999, in Poutanen J., Svensson R., eds, AstronomicalSociety of the Pacific Conference Series Vol. 161, High En-ergy Processes in Accreting Black Holes. p. 82 ( arXiv:astro-ph/9811441 )Weisskopf M. C., 1999, ArXiv Astrophysics e-prints,White N. E., Zhang W., 1997, ApJ, 490, L87Wijnands R., M´endez M., Markwardt C., van der Klis M.,Chakrabarty D., Morgan E., 2001, ApJ, 560, 892Wijnands R., Yang Y. J., Altamirano D., 2012, MNRAS, 422, 91Winkler C., et al., 2003, A&A, 411, L1MNRAS , 1–24 (2017) J.M.C. Court et al.
Yadav J. S., Agrawal P. C., Paul B., Rao A. R., Seetha S., Kas-turirangan K., 2000, Advances in Space Research, 25, 441Zhang Z., Qu J. L., Gao H. Q., Zhang S., Bu Q. C., Ge M. Y.,Chen L., Li Z. B., 2014, A&A, 569, A33
APPENDIX A: FLARE-FINDING ALGORITHM
The algorithm used to find flares is performed as such (seealso Figure A1):(i) Choose some threshold values T L and T H . Set the valueof all datapoints below T L to zero.(ii) Retrieve the x-co-ordinate of the highest value re-maining in the dataset. Call this value x m and store it ina list.(iii) Set the value of point at x m to zero.(iv) Scan forwards from x m . If the selected point has anonzero value, set it to zero and move to the next point. Ifthe selected point has a zero value, move to step (v).(v) Scan backwards from x m . If the selected point has anonzero value, set it to zero and move to the previous point.If the selected point has a zero value, move to step (vi).(vi) Retrieve the y-co-ordinate of the highest value re-maining in the dataset. Call this y m .(vii) If y m > T H , repeat steps (ii)–(vi). If y m < T H , proceedto step (viii).(viii) Restore the original dataset.(ix) Retrieve the list of x m values found in step (ii). Sortthem in order of size.(x) For each pair of adjacent x m values, find the x-coordinate of the datapoint between them with the lowesty-value. Call these values x c .(xi) This list of x c can now be used to demarcate theborder between peaks.The values T L and T H can also be procedurally generatedfor a given piece of data:(i) Select a small section of the dataset or a similardataset (containing ∼ peaks by eye) and note the location x e of all peaks found by eye.(ii) Let P L and P H be two arbitrary values in the range [ , ] .(iii) Let T L ( T H ) be the P L th ( P H th) percentile of the y-values of the subsection of dataset.(iv) Run the flare-finding algorithm up to step (ix). Savethe list of x m .(v) Split the dataset into bins on the x-axis such as thebin width b (cid:28) p , where p is the rough x-axis separationbetween peaks.(vi) For each bin, note if you found any value in x m fallsin the bin and note if any value of x e falls in the bin.(vii) Using each bin as a trial, compute the Heidke SkillScore (Heidke 1926) of the algorithm with the method offinding peaks by eye: HSS = ( AD − BC )( A + B )( B + D ) + ( A + C )( C + D ) (A1)Where A is the number of bins that contain both x e and x m , B ( C ) is the number of bins that contain only x m ( x e ) and D is the number of bins which contain neither (Kok 2000). C o un t s C o un t s T L C o un t s T L T H C o un t s T L T H C o un t s C o un t s Figure A1.
From top-left: (i) An untouched data-set. (ii) Thedataset with all y < T L removed. (iii) The dataset with all con-tiguous nonzero regions with max ( y ) < T H removed. (iv) The peakx-values x m . (v) The restored dataset with the peak x-values x m highlighted. (vi) The boundaries between adjacent peaks. (viii) Repeat steps (iii)–(vii) for all values of P H > P L for P L and P H in [ , ] . Use a sensible value for the resolutionof P L and P H . Save the HSS for each pair of values(ix) Locate the maximum value of HSS, and note the P L and P H values used to generate it. Use these values to gen-erate your final T L and T H values.We show an example of Heidke skill score grid for thisalgorithm, applied to a Class IV observation, in Figure A2. APPENDIX B: MODEL-INDEPENDENTCLASSIFICATION OF EACH OBSERVATIONOF IGR J17091-3624
Observation IDs, and orbit IDs, for every observation andobservation segment that was used in our analysis are pre-sented in Table B1. Note that not all of every observationwas used; in many cases, large spikes caused by
PCA
PCUsswitching off or on rendered ∼ s unusable. As these oftenoccurred very close to the beginning or end of an observationsegment, small sections of data before or after these spikeswas also sometimes discarded. Every observation segment ispresented along with the variability class assigned to it bythis study. This paper has been typeset from a TEX/L A TEX file prepared bythe author. MNRAS000
PCUsswitching off or on rendered ∼ s unusable. As these oftenoccurred very close to the beginning or end of an observationsegment, small sections of data before or after these spikeswas also sometimes discarded. Every observation segment ispresented along with the variability class assigned to it bythis study. This paper has been typeset from a TEX/L A TEX file prepared bythe author. MNRAS000 , 1–24 (2017) xotic Variability in IGR J17091-3624 Table B1.
Here is listed the Observation IDs for every
RXT E observation that was used in this analysis, along with the variability classwhich has been assigned to it.
Orb. is the orbit ID (starting at 0) of each observation segment,
Exp. is the exposure time in seconds and X is the prefix 96420-01. This table is continued overleaf in Table B2.MJD OBSID Orb.
Class
Exp.
MJD OBSID
Orb.
Class
Exp.
MJD OBSID
Orb.
Class
Exp. X -01-00 0 I 1840 55676 X -09-06 0 VI 3540 55741 X -18-05 0 VII 78255622 X -01-000 0 I 3480 55677 X -09-01 0 V 1676 55743 X -19-00 0 VII 141255622 X -01-000 1 I 1656 55678 X -09-04 0 V 2090 55744 X -19-01 0 VIII 193855622 X -01-000 2 I 3384 55679 X -09-02 0 V 2306 55745 X -19-02 0 VII 217255622 X -01-000 3 I 3400 55680 X -10-02 0 V 952 55747 X -19-03 0 VIII 169155622 X -01-000 4 I 3384 55681 X -10-00 0 V 3725 55748 X -19-04 0 VI 128355623 X -01-01 0 I 1240 55682 X -10-03 0 V 1157 55749 X -19-05 0 VIII 141755623 X -01-01 1 I 752 55684 X -10-01 0 III 1504 55751 X -20-05 0 VI 172655623 X -01-01 2 I 992 55686 X -10-04 0 III 1127 55752 X -20-01 0 VIII 107955623 X -01-01 3 I 1184 55686 X -10-05 0 II 2179 55753 X -20-02 0 VIII 143355623 X -01-01 4 I 1056 55687 X -11-00 0 II 3537 55754 X -20-03 0 VII 112255623 X -01-010 0 I 2080 55688 X -11-01 0 II 1153 55756 X -20-04 0 VIII 148655623 X -01-010 1 I 1832 55690 X -11-02 0 II 1408 55757 X -21-00 0 VIII 337255623 X -01-010 2 I 1648 55691 X -11-03 0 II 886 55758 X -21-01 0 VIII 338355623 X -01-010 4 I 1424 55692 X -11-04 0 II 3566 55759 X -21-02 0 VI 193855623 X -01-010 5 I 400 55693 X -11-05 0 II 1817 55761 X -21-04 0 VII 149755623 X -01-02 0 I 3056 55694 X -12-00 0 II 2761 55762 X -21-05 0 VII 154855623 X -01-02 1 I 2792 55695 X -12-01 0 II 1374 55763 X -21-06 0 VII 220255623 X -01-02 2 I 2432 55695 X -12-02 0 II 2041 55764 X -22-00 0 VII 168255623 X -01-020 0 I 3456 55696 X -12-03 0 II 1456 55765 X -22-01 0 VII 122155623 X -01-020 1 I 3464 55698 X -12-04 0 II 1916 55766 X -22-02 0 V 72055623 X -01-020 2 I 3512 55698 X -12-05 0 II 3139 55767 X -22-03 0 V 180155623 X -01-020 3 I 3520 55700 X -12-06 0 II 1189 55768 X -22-04 0 VIII 198355623 X -01-020 4 I 3512 55701 X -13-00 0 II 1214 55769 X -22-05 0 VIII 99955623 X -01-020 5 I 464 55702 X -13-01 0 II 980 55770 X -22-06 0 VIII 66755624 X -02-00 0 I 1758 55704 X -13-02 0 II 732 55771 X -23-00 0 VIII 207555626 X -02-01 0 I 1380 55705 X -13-03 0 III 1217 55772 X -23-01 0 VII 338555628 X -02-02 0 I 3305 55706 X -13-04 0 III 1161 55773 X -23-02 0 VII 221855630 X -02-03 0 I 1876 55707 X -13-05 0 IV 2763 55774 X -23-03 0 V 181155632 X -03-00 0 I 1712 55708 X -14-00 0 IV 1188 55775 X -23-04 0 V 335655634 X -03-01 0 III 3590 55709 X -14-01 0 IV 3342 55776 X -23-05 0 V 260355639 X -04-00 0 IV 3099 55710 X -14-02 0 IV 1094 55777 X -23-06 0 IV 91255642 X -04-02 0 IV 2972 55712 X -14-03 0 IV 1404 55777 X -23-06 1 IV 154455643 X -04-01 0 III 1190 55713 X -14-04 0 V 871 55778 X -24-00 0 IV 130955644 X -04-03 0 III 2903 55714 X -14-05 0 V 1311 55779 X -24-01 0 IV 359955645 X -05-02 0 I 3578 55715 X -15-00 0 IV 1241 55779 X -24-02 0 IV 201355647 X -05-00 0 IV 2872 55716 X -15-01 0 IV 1262 55782 X -24-03 0 V 176155647 X -05-000 0 IV 3472 55717 X -15-02 0 III 1557 55782 X -24-04 0 V 172555647 X -05-000 1 IV 3520 55718 X -15-03 0 III 1334 55784 X -24-05 0 V 314455647 X -05-000 2 IV 3512 55720 X -15-04 0 IV 1486 55784 X -24-06 0 V 259155647 X -05-000 3 IV 3520 55721 X -15-05 0 IV 1500 55785 X -25-00 0 V 236655647 X -05-000 4 IV 3512 55722 X -16-00 0 IV 900 55786 X -25-01 0 V 180455647 X -05-000 5 IV 648 55723 X -16-01 0 III 1004 55787 X -25-02 0 V 195155649 X -05-03 0 IV 2409 55724 X -16-02 0 II 1923 55788 X -25-03 0 V 161955650 X -05-01 0 IV 1473 55725 X -16-03 0 II 1919 55789 X -25-04 0 V 260155651 X -05-04 0 IV 2954 55726 X -16-04 0 III 1935 55790 X -25-05 0 V 147355653 X -06-00 0 IV 2723 55727 X -16-05 0 II 730 55791 X -25-06 0 V 92255654 X -06-01 0 IV 3388 55728 X -16-06 0 II 1953 55792 X -26-00 0 V 233655656 X -06-02 0 IV 2908 55729 X -17-00 0 II 2735 55794 X -26-01 0 V 138555657 X -06-03 0 V 1842 55730 X -17-01 0 II 3556 55795 X -26-02 0 VIII 145855661 X -07-00 0 V 1754 55731 X -17-02 0 II 3605 55796 X -26-03 0 VI 132555662 X -07-01 0 V 3365 55732 X -17-03 0 II 1647 55798 X -26-04 0 VI 207555663 X -07-02 0 V 3373 55733 X -17-04 0 II 1459 55799 X -27-00 0 VI 139655666 X -08-00 0 V 3338 55734 X -17-05 0 III 1736 55800 X -27-01 0 VI 268455669 X -08-01 0 V 3368 55735 X -17-06 0 III 3653 55801 X -27-02 0 VI 101655670 X -08-03 0 VI 2489 55736 X -18-00 0 III 2317 55802 X -27-03 0 VI 117955671 X -08-02 0 VI 2609 55737 X -18-01 0 IV 1387 55803 X -27-04 0 VI 130455673 X -09-03 0 VI 1011 55738 X -18-02 0 V 1291 55805 X -27-05 0 VI 166355674 X -09-00 0 VI 1386 55739 X -18-03 0 V 2178 55806 X -28-00 0 VI 145655675 X -09-05 0 IX 1148 55740 X -18-04 0 V 1478 55808 X -28-01 0 VIII 577MNRAS , 1–24 (2017) J.M.C. Court et al.
Table B2.
A continuation of Table B1.
Orb. is the orbit ID (starting at 0) of each observation segment,
Exp. is the exposure time inseconds and X is the prefix 96420-01.MJD OBSID Orb.
Class
Exp.
MJD OBSID
Orb.
Class
Exp.
MJD OBSID
Orb.
Class
Exp. X -28-02 0 VI 1251 55836 X -32-02 0 IX 1591 55857 X -35-01 0 IX 191255811 X -28-03 0 VI 2000 55837 X -32-03 0 IX 2155 55859 X -35-02 0 IX 20055813 X -29-00 0 VIII 1309 55838 X -32-04 0 IX 2641 55859 X -35-02 1 IX 129655819 X -29-04 0 VIII 1686 55838 X -32-05 0 IX 2077 55860 X -35-03 0 IX 137255820 X -30-00 0 VI 1488 55840 X -32-06 0 IX 3392 55861 X -35-04 0 IX 83655821 X -30-01 0 VI 1503 55840 X -32-06 1 IX 3512 55862 X -36-00 0 IX 114555822 X -30-02 0 VI 1417 55840 X -32-06 2 IX 3934 55863 X -36-01 0 IX 132255823 X -30-03 0 VI 1290 55840 X -32-06 3 IX 3880 55865 X -36-03 0 IX 148555824 X -30-04 0 VI 1489 55840 X -32-06 4 IX 1896 55866 X -36-04 0 IX 179555825 X -30-05 0 VI 2581 55841 X -33-00 0 IX 1188 55867 X -36-05 0 IX 173255826 X -30-06 0 VI 2747 55842 X -33-01 0 IX 855 55868 X -36-06 0 IX 165755827 X -31-00 0 VI 1559 55843 X -33-02 0 IX 1156 55871 X -37-00 0 IX 81555828 X -31-01 0 VI 2954 55845 X -33-04 0 IX 1713 55871 X -37-02 0 IX 146055829 X -31-02 0 IX 3005 55846 X -33-05 0 IX 934 55872 X -37-03 0 IX 168355830 X -31-03 0 IX 1472 55847 X -33-06 0 IX 717 55873 X -37-04 0 IX 140255830 X -31-03 1 IX 288 55848 X -34-00 0 IX 1159 55874 X -37-05G 0 IX 153655831 X -31-04 0 IX 1586 55849 X -34-01 0 IX 973 55875 X -37-06 0 IX 153655832 X -31-05 0 VI 3812 55851 X -34-02 0 IX 2261 55876 X -38-00 0 IX 149755833 X -31-06 0 IX 3675 55852 X -34-03 0 IX 1092 55877 X -38-01 0 IX 113455834 X -32-00 0 IX 1217 55853 X -34-04 0 IX 741 55878 X -38-02 0 IX 128955835 X -32-01 0 IX 1445 55856 X -35-00 0 IX 797 55879 X -38-03 0 IX 1433 L o w T h r e s h o l d L =4 P H =94 Heidke Skill Score for Peak Finding Algorithm in Class IV Data,Scanning for Peaks in a 5s Sliding Window H SS Figure A2.
The Heidke Skill score of a Class IV observation ofIGR J17091-3624 for a selection of different values P L and P H . MNRAS000