betaFeedback

WorksPhysics and Astronomy

X-ray Flare Evolution and Changing-Look Nature of the TeV Blazar OQ 334

Abstract

The blazar OQ 334 (B2 1420+326, z = 0.682) is one of the few jetted AGN with repeated changing-look (CL) transitions between FSRQ and BL Lac states. We present a multi-wavelength study covering 2017 December to 2024 September, focusing on three X-ray flares observed with Swift-XRT (X1: 2019 June 26–July 24, X2: 2020 January 4–February 21, X3: 2024 June 24–August 2) and on the spectral properties of the CL states. The spectrum hardens from flare to flare, with ΓΓ decreasing from 1.97±0.091.97 \pm 0.09 to 1.71±0.041.71 \pm 0.04 and 1.34±0.061.34 \pm 0.06, while the hysteresis evolves from an initial clockwise loop, consistent with cooling-dominated emission, to multi-loop trajectories in the later flares. The X-ray spectra separate the CL states, with FSRQ states hardest (ΓX≈1.3Γ_X \approx 1.3–1.51.5), BL Lac states softest (ΓX≈1.8Γ_X \approx 1.8–2.42.4), and transition states intermediate. Within the CL states, flux and photon index are positively correlated (r=0.59r = 0.59, p=2.5×10−2p = 2.5 \times 10^{-2}), producing a softer-when-brighter trend. The first two flares coincide with FSRQ →\rightarrow BL Lac transitions, whereas the third, the brightest, remains FSRQ-like. One-zone leptonic modelling of the broadband SEDs reproduces the flares with a weak, nearly constant magnetic field and a particle-dominated region, and the X-ray emission evolves from a substantial synchrotron contribution in the first flare to complete inverse-Compton dominance in the third. The X-ray band traces the changing balance between the synchrotron and inverse-Compton components in the flares and provides a sensitive probe of the CL behaviour of the source.

Introduction

Blazars are a subclass of radio-loud active galactic nuclei (AGN) powered by a central supermassive black hole, characterised by highly relativistic jets directed at small angles to the observer’s line of sight ([102, 20]). Their emission is non-thermal, spanning the entire electromagnetic spectrum from radio to very-high-energy (VHE, E≳100E \gtrsim100 GeV) γ\gamma-rays, and is significantly enhanced by relativistic beaming ([17]). Blazars are among the most luminous persistent sources in the universe, dominating the extragalactic γ\gamma-ray sky. They exhibit extreme observational properties, including strong flux variability across all wavelengths on timescales ranging from minutes to years ([60, 15]), high optical polarisation, and strong γ\gamma-ray emission ([74, 57, 1]). The broadband spectral energy distribution (SED) of blazars displays a characteristic double-humped structure. The low-energy hump, spanning from radio to optical/X-ray frequencies, is attributed to synchrotron radiation from relativistic electrons in the jet ([101]). The origin of the high-energy hump, peaking in the MeV–TeV range, remains a subject of active investigation, with two broad classes of models proposed. In the leptonic framework, this emission is produced via inverse-Compton (IC) scattering of low-energy photons by the relativistic electron population in the jet. The seed photons for this process may be internal to the jet, as in the synchrotron self-Compton (SSC) model ([69, 70]), or may originate from external photon fields such as the accretion disk, the broad-line region (BLR), or the dusty torus, as in the external Compton (EC) model ([34, 96, 18]). son 1995; Ghisellini & Madau [49]). Hadronic models, on the other hand, attribute the high-energy emission to proton synchrotron radiation or to secondary particles produced in photo-pion and photo-pair interactions (Aharonian [4]; Böttcher et al. [21]).

Blazars are generally classified in two ways: by the position of their synchrotron peak and by their optical/UV emissions. Based on the rest-frame equivalent width (EW) of their broad emission lines (BELs), they are divided into Flat Spectrum Radio Quasars (FSRQs), which exhibit strong lines with EW >> 5 Å, and BL Lacertae objects (BL Lacs), which show featureless spectra or weak lines with EW << 5 Å ([97, 102]). This observational distinction often corresponds to different accretion regimes, where FSRQs are associated with radiatively efficient accretion and BL Lacs with radiatively inefficient modes ([48, 47]). Blazars are also categorised by the frequency of their low-energy synchrotron peak (νp\nu_p). They are classified as Low Synchrotron Peaked (LSPs) when νp≤1014\nu_p \le10^{14} Hz, Intermediate Synchrotron Peaked (ISPs) when 101410^{14} Hz <νp<< \nu_p < 101510^{15} Hz, and High Synchrotron Peaked (HSPs) when νp≥1015\nu_p \ge10^{15} Hz (Abdo et al. [1]).

The discovery of changing-look blazars (CLBs) has challenged these traditional categories, as these transitional sources switch between FSRQ and BL Lac states over timescales of months to years ([103, 52, 51, 90, 76, 80, 88, 78]). These transitions, such as those observed in OQ 334, involve dramatic variations in the EW of BELs (Mishra et al. [76]). The physical mechanisms triggering these changes are still under investigation, with primary theories suggesting either sudden shifts in the accretion rate ([43, 106, 38]) or the dilution of emission lines by a highly variable, Doppler-boosted jet continuum during active flares ([19, 90]). Other recent studies propose that CLBs represent a specific, short-lived phase in the evolutionary process of blazars ([67, 64, 86]).

OQ 334, also referred to as B2 1420+32 or QSO B1420+326 (RA: 215.6265, Dec: 32.3862), is a blazar located at a redshift of z=0.682z = 0.682 (Hewett & Wild [61]). Based on its radio spectrum, the source has been historically classified as an FSRQ (Healey et al. [59]) and is listed as such in the Fermi-LAT Fourth Source Catalogue (4FGL; Abdollahi et al. [2]). The central black hole has an estimated mass of MBH≃4×108 M⊙M_{\rm BH} \simeq4 \times10^{8}\,M_{\odot}, inferred from its luminosity and broad-line width (Brocktherton et al. 2015). After nearly a decade of relative quiescence, the source underwent a dramatic optical flare of more than 2 magnitudes in December 2017 (Mishra et al. [76]), marking the beginning of a prolonged period of elevated activity across the electromagnetic spectrum, including optical, infrared, and γ\gamma-ray bands (Ciprini [31]; Mishra et al. [76]; Prince et al. [84]). During this active period, Mishra et al. [76] found evidence of repeated transitions between the FSRQ and BL Lac states, beginning in January 2018, establishing OQ 334 as a changing-look blazar. The continued high state in late 2019 triggered follow-up observations by the MAGIC telescopes, leading to the discovery of VHE γ\gamma-ray emission in January 2020 (Mirzoyan [75]; MAGIC Collaboration et al. [68]).

The paper is organised as follows. In § 2, we describe the multi-wavelength dataset and associated data reduction procedures. § 3 presents the analysis and results, including the timing and optical analysis, X-ray spectral analysis, changing-look nature, and broadband SED modelling. In § 4, we discuss the physical implications of our findings, and we summarise our conclusions in § 5.

Data Reduction

MAGIC

MAGIC is a stereoscopic system consisting of two 17 m diameter ground-based Imaging Atmospheric Cherenkov Telescopes (IACTs) located on the island of La Palma, Spain (Aleksić et al. [6]). The telescopes record Cherenkov light from extensive air showers initiated by gamma rays, covering an energy range from ∼\sim50 GeV to ∼\sim100 TeV. With an integral sensitivity of (0.66±0.03)%(0.66 \pm0.03)\% of the Crab Nebula flux above 220 GeV for point-like sources with a Crab Nebula-like spectrum, achieved in 50 h of observations (Aleksić et al. [6]), MAGIC is among the most sensitive instruments for VHE blazar studies.

The observations analysed span MJD 58869.22 – 58885.25, corresponding to periods C (VHE flare) and D (post-flare) reported by MAGIC Collaboration et al. [68]. Calibration, image cleaning, and event reconstruction were performed with the MARS package (Zanin [108]), after which the analysis was carried out from the DL3 level (Abe et al. [3]) using Gammapy (Donath et al. 2023). The signal was extracted from a point-like region centred on the source, with the extraction radius taken from the energy-dependent RAD_MAX values stored in the IRFs, and the background was estimated using three reflected regions placed at the same camera offset. The individual observations were stacked into a single ON-OFF dataset, and the fit was restricted to the 50–500 GeV range. The intrinsic spectrum was modelled with a power law,

dNdE=N0(EE0)−Γ,(1)\frac{\mathrm{d}N}{\mathrm{d}E}=N_0\left(\frac{E}{E_0}\right)^{-\Gamma}, \tag*{(1)}

multiplied by an extragalactic background light (EBL) absorption term e−τ(E,z)e^{-\tau(E,z)} following the model of

Domínguez et al. [35]. The normalisation and photon index were left free, while the reference energy (E0=0.1E_0 = 0.1 TeV) and redshift were fixed. The fit gives N0=(1.37±0.18)×10−9N_0 = (1.37 \pm0.18) \times10^{-9} TeV−1^{-1} cm−2^{-2} s−1^{-1} and Γ=3.48±0.32\Gamma= 3.48 \pm0.32. The intrinsic points were obtained by referencing to the unabsorbed power law. The intrinsic and observed spectra of the source are shown in Figure 2.

Figure 2. MAGIC very high-energy γ\gamma-ray spectra of OQ 334 during the flare period MJD 58869.22–58885.25. The observed and intrinsic spectra are shown together with their respective fits. The intrinsic spectrum has been corrected for attenuation by the extragalactic background light, and the shaded bands indicate the statistical uncertainties of the corresponding fits.

Fermi

The Fermi γ\gamma-ray Space Telescope, launched by NASA on June 11, 2008, is designed to study the high-energy gamma-ray sky. The observatory carries two instruments onboard: the Large Area Telescope (LAT) and the Gamma-ray Burst Monitor (GBM), together enabling gamma-ray observations spanning from a 8 keV to beyond 300 GeV [12, 73]. The LAT is the primary instrument, a pair-conversion telescope designed to detect gamma rays in the energy range of ∼\sim20 MeV to more than 300 GeV, featuring a large effective area of >8000>8000 cm2^2 at ∼\sim1 GeV and a wide field of view of 2.4 sr [12]. Operating in survey mode, the LAT covers approximately 20% of the sky at any given time and scans the entire sky every ∼\sim3 hours, making it exceptionally well-suited for monitoring the long-term gamma-ray variability and spectral properties of blazars over a broad range of timescales.

The LAT data were downloaded from the Fermi Science Support Center1 for a 20∘^\circ region of interest (ROI) centred on the source position. The analysis utilised Pass 8 (P8R3) data [11] to search for gamma-ray emission in the energy range of 0.1-300 GeV, employing FermiTools 2.2.0 and Fermipy 1.2.2 [105]. The SOURCE class events (evclass = 128 and evtype = 3) were selected within a 10∘^\circ ROI centred on the source position, adopting the P8R3_SOURCE_V3 instrument response functions. Standard data quality cuts were applied, including a zenith angle cut of zmax⁡<90∘z_{\max}<90^\circ to reduce contamination from the Earth’s limb, and the selection of good time intervals based on (DATA_QUAL>0)&&(LAT_CONFIG==1). The significance of the γ\gamma-ray signal was evaluated using a maximum-likelihood test statistic (TS) defined as TS = 2(log⁡L1−log⁡L0)2(\log L_1-\log L_0), where L1L_1 and L0L_0 are the likelihoods with and without a point source at the position of the target [72]. All point sources from the 4FGL catalog [2] within 15∘^\circ of the ROI centre were included in the initial model file, along with the Galactic (gll_iem_v07) and extragalactic isotropic diffuse emission (iso_P8R3_SOURCE_V3_v1) components.

A binned likelihood analysis was performed using eight energy bins per decade and a spatial binning of 0.1∘^\circ per pixel to derive the spectral parameters. After initial optimisation, sources with TS <10<10 were fixed, and the normalisation of sources with TS >10>10 was allowed to vary in the model file. The normalisation parameters for all sources within 3∘^\circ of the ROI centre were additionally freed, and all parameters associated with the isotropic and Galactic diffuse components were allowed to vary freely. To investigate the temporal variability of the source, a weekly binned light curve was generated. In each time bin, the normalisation parameters of all sources within 3∘^\circ of the ROI centre were left free to vary, along with the diffuse components, while the spectral parameters of all remaining sources were fixed to the values obtained from the global fit. Only time bins with TS >9>9 were considered statistically significant detections and retained in the light curve.

Swift

Swift is a multi-wavelength observatory launched in 2004 by NASA for Gamma-Ray Burst (GRB) astronomy, which is a medium-sized explorer (MIDEX) mission including three telescopes: Burst Alert Telescope (BAT), X-Ray Telescope (XRT), and Ultraviolet/Optical Telescope (UVOT) [46]. While primarily designed for studying GRBs, Swift has proven valuable for multi-wavelength studies of various compact objects and facilitates the detection and analysis of diverse sources in high-energy astrophysics. The observatory’s capability for rapid slewing and simultaneous multi-wavelength observations makes it invaluable for studying the highly variable nature of blazars across ultraviolet, optical, and X-ray bands.

The XRT is a focusing X-ray imaging spectrometer with >120>120 cm2^2 effective area at 1.5 keV, 23′' field of view, and 18′′'' angular resolution (half-power diameter), operating across the 0.2-10 keV energy range [25]. The telescope provides sensitive X-ray spectroscopy and photometry capabilities through an auto-exposure mode that dynamically adjusts readout mode to optimise data quality as source flux varies. The X-ray spectrum and light curve of the source were generated using the Swift-XRT Online Data Products Generator tool2. The data reduction process followed by this online tool can be found in Evans et al. [40] (2009). Due to the low number of counts, the 0.3-10 keV X-ray spectra were grouped to a minimum of one count per bin using the GRPPHA task of FTOOLS. The spectra were then fitted in XSPEC version 12.14.0 [10] using Cash statistics [27].

The UVOT has a modified Ritchey-Chrétien telescope design with a 30 cm diameter aperture, featuring 11 filters that cover a wavelength range of 170-600 nm across a 17×\times17 arcmin2^2 field of view [89]. The source is observed with six standard photometric filters (Optical: B, V, U; UV: UVM2, UVW1, UVW2). The observations were downloaded from the SWIFTMAS-TER catalog and data analysis was performed using the HEASoft package version 6.33 and the CALDB version 20220331. The data reduction was performed by following the standard procedures in the Swift UVOT Software Guide3. The source counts were extracted using a 6″ radius centred on the source, while background counts were measured from a 20″ radius circle placed in a nearby source-free region. Images across all extensions were combined for each observation using uvotimsum, with processing performed independently for each filter. Aperture photometry was subsequently conducted using uvotsource for individual filters within each observation. Galactic extinction corrections were applied to the magnitudes using E(B-V) value (1.07±0.08)×10−2(1.07 \pm0.08) \times10^{-2} ([94, 62]) and the extinction laws of Cardelli et al. [26], and the corrected magnitudes were transformed into flux densities using zero points from Breeveld et al. [22] and conversion factors from Poole et al. [83].

Optical Archival Data

The All-Sky Automated Survey for Supernovae (ASAS-SN) ([95, 56]) is a global network of 24 telescopes that has been routinely monitoring the entire visible sky since 2012. The survey operated in the V-band until late 2018, after which it transitioned to g-band observations. The archival optical light curve data for the source were retrieved from the ASAS-SN Sky Patrol4 V2.0 Photometry Database.

The Zwicky Transient Facility (ZTF) ([71, 14]) is a high-cadence optical time-domain survey operating on the Palomar 48-inch Schmidt telescope since 2018. ZTF observes the northern sky in three optical bands (g, r, and i), reaching typical depths of ∼20.5\sim20.5 mag with a nominal cadence of 3 days for fields in the main survey area. Archival optical light curve data for the source were obtained from the ZTF Data Release5 using point spread function (PSF) fitting photometry.

While ASAS-SN and ZTF were primarily designed to discover bright optical transients and identify variable phenomena, their continuous all-sky coverage, systematic observing cadence, and photometric precision make them well-suited for optical variability studies of blazars. The downloaded light curve data were filtered using instrument-specific quality cuts, corrected the magnitudes for Galactic extinction adopting the E(B-V) value (1.07±0.08)×10−2(1.07 \pm0.08) \times10^{-2} ([94, 62]) and extinction laws from Cardelli et al. [26], and converted the corrected magnitudes into flux densities.

Analysis & Results

The multi-wavelength light curves of OQ 334 spanning MJD 54769.0 to 60824.0 is presented in Figure 1. For this work, we use the period from MJD 58095.0 to 60580.0, during which multi-wavelength data are available across all bands. Prior to MJD 58095.0, the source was in a quiescent state, as shown by Fermi-LAT and ASAS-SN V-band observations.

Figure 1. Multi-wavelength light curves of the FSRQ OQ 334 (B2 1420+326) from MJD 54769.0 to 60824.0. From top to bottom, the panels show: ASAS-SN optical, ZTF optical, Swift-UVOT optical, Swift-UVOT Ultraviolet, Swift-XRT X-ray, and Fermi-LAT γ\gamma-ray observations. The shaded region indicates the time period analysed in this work (MJD 58095.0–60580.0). In the Swift-XRT panel, the orange lines mark the change points identified using Bayesian-block analysis, while the red horizontal line indicates the mean X-ray flux over the analysed period.

Timing & Optical Analysis

To quantify the inter-band correlations and the associated time lags, we performed a cross-correlation analysis of the optical and gamma-ray light curves. The Swift XRT and UVOT light curves were not included, as their sampling is too sparse for a meaningful correlation analysis. The z-transformed Discrete Correlation Function (ZDCF; Alexander [8, 9]) was used, which is a modification of the standard Discrete Correlation Function (DCF; Edelson & Krolik [39]; see also Bhatta [15]). The ZDCF applies Fisher’s z-transformation to the correlation coefficient rr to stabilize the skewed sampling distribution (Alexander [8]). The correlation is computed from the unbinned discrete correlation function (UDCF) for each flux measurement pair (xi,yj)(x_i,y_j):

UDCF⁡ij=(xi−xˉ)(yj−yˉ)(Sx2−σerr,x2)(Sy2−σerr,y2)(2)\operatorname{UDCF}_{ij} = \frac{(x_i-\bar{x})(y_j-\bar{y})}{\sqrt{(S_x^2-\sigma_{\mathrm{err},x}^2)(S_y^2-\sigma_{\mathrm{err},y}^2)}} \tag*{(2)}

where xˉ\bar{x} and yˉ\bar{y} are the mean fluxes, Sx2S_x^2 and Sy2S_y^2 are their respective variances, and σerr,x\sigma_{\mathrm{err},x} and σerr,y\sigma_{\mathrm{err},y} are the flux uncertainties. These UDCF pairs are subsequently binned in equal-width time bins to obtain the DCF:

DCF⁡(τ)=1M∑UDCF⁡ij,(3)\operatorname{DCF}(\tau) = \frac{1}{M}\sum\operatorname{UDCF}_{ij}, \tag*{(3)}

where MM is the number of pairs within the bin.

The ZDCF analysis was implemented using the pyZDCF Python package [63], and results are presented in Figure 3. The cross-correlation between the Fermi-LAT light curve and three optical bands (ZTF ii, ZTF rr, and ASAS-SN gg) reveals a consistent peak near zero time lag. The peak occurs at τ≈−7\tau\approx-7 days for the LAT-ZTF ii pair, with DCF =0.74±0.04= 0.74 \pm0.04, and at comparable lags for ZTF rr and ASAS-SN gg, though those pairs do not exceed the DCF >0.5> 0.5 threshold.

Figure 3. zz-transformed discrete correlation functions (ZDCF) for OQ 334 between pairs of light curves across multiple wavebands. Panels (a)-(c) show the cross-correlation of the Fermi-LAT light curve with the ZTF ii-band, ZTF rr-band, and ASAS-SN gg-band optical light curves, respectively.

The spectral behaviour of the optical emission is studied using colour-magnitude and colour-colour diagrams in ZTF gg, rr, and ii bands, shown in Figure 4. Each colour is plotted against the magnitude of the band not entering it, so that the uncertainties on the two axes remain independent. The g−rg-r colour is correlated with the ii band magnitude, with a Pearson coefficient of −0.67-0.67 and p=4.4×10−53p=4.4 \times10^{-53}. The r−ir-i colour shows the same trend against the gg band magnitude, though more weakly, with a coefficient of −0.25-0.25 and p=2.9×10−7p=2.9 \times10^{-7}. In both cases, the colour index decreases as the magnitude increases, and the optical spectrum becomes redder as the source brightens [29]. The colour-colour diagram shows a positive correlation between g−rg-r and r−ir-i, with a correlation coefficient of 0.300.30, and p=5.5×10−10p=5.5 \times10^{-10}.

Figure 4. Colour-magnitude and colour-colour diagrams for OQ 334 constructed from ZTF gg, rr, and ii band photometry. Left and middle panels show the g−rg-r colour against the ii band magnitude and the r−ir-i colour against the gg band magnitude, respectively. The right panel shows g−rg-r against r−ir-i. Solid lines are linear least-squares fits, Pearson correlation coefficient rr, and the corresponding pp value given in each panel.

X-ray Analysis

Spectral Analysis

A total of 64 Swift-XRT observations were available during the studied time period. We performed spectral fitting in the 0.3-10 keV energy range using three different absorbed (tbabs) spectral models: power-law, log-parabola, and broken power-law. When more than one model is used to fit a source, a statistical comparison among those models is required to select the best result. To identify the best fit for each observation, model comparison was carried out using the corrected Akaike Information Criterion (AICc_{\mathrm{c}}; [5, 98, 66]), defined as

AICc=AIC+2k(k+1)N−k−1(4)\mathrm{AIC}_{\mathrm{c}} = \mathrm{AIC} + \frac{2k(k+1)}{N-k-1} \tag*{(4)}

where AIC=2k+C\mathrm{AIC}=2k+C, C=−2ln⁡(L)C=-2\ln(L) is the Cash statistic value, LL is the maximum likelihood for the model, kk is the number of free parameters, and NN is the number of spectral bins. Given the relatively small number of spectral bins in some observations, the finite-sample correction is necessary [66].

The AICc_{\mathrm{c}} difference, defined as ΔAICc=AICc−AICc,min\Delta\mathrm{AIC}_{\mathrm{c}}=\mathrm{AIC}_{\mathrm{c}}-\mathrm{AIC}_{\mathrm{c,min}}, allows for a comparison of the models with respect to the model that has the lowest AICc_{\mathrm{c}} value. Following the criteria presented in [24], models with ΔAICc>10\Delta\mathrm{AIC}_{\mathrm{c}}>10 relative to the best-fit model can be ruled out [85]. None of the observations showed ΔAICc>10\Delta\mathrm{AIC}_{\mathrm{c}}>10 for any of the three models, indicating that all models provide statistically acceptable fits. Using the minimum AICc_{\mathrm{c}} criterion, the power-law model provides the best description for 47 of the 64 observations, the log-parabola model for 13, and the broken power-law model for 4. The Swift-XRT observation log and the best-fitting spectral parameters obtained are presented in Table 1, and Figure 5 shows the correlation between flux and photon index from the power-law fit. The photon index varies between 1.15 ±\pm 0.22 and 2.5 ±\pm 0.28 without a systematic dependence on the flux (r=−0.14r=-0.14, p=0.27p=0.27).

Figure 5. Correlation between X-ray flux and photon index from the absorbed power-law model for all OQ 334 Swift-XRT observations. The black line indicates the linear fit.

ObsIDDateExposure (s)powlaw Γ\Gammapowlaw FX−rayF_{\mathrm{X-ray}}logpar α\alphalogpar β\betabknpow Γ1\Gamma_1bknpow EbE_b (keV)bknpow Γ2\Gamma_2
000105200012018-01-201029.62.54±0.292.54 \pm0.292.17±0.672.17 \pm0.672.48±0.232.48 \pm0.23−1.03±0.53-1.03 \pm0.532.62±0.302.62 \pm0.303.96±1.493.96 \pm1.49−2.50-2.50†
000105200022018-02-222881.31.49±0.141.49 \pm0.142.15±0.532.15 \pm0.531.29±0.231.29 \pm0.230.52±0.410.52 \pm0.411.50±0.151.50 \pm0.157.127.12†−2.46-2.46†
⋅ ⋅ ⋅\cdot\ \cdot\ \cdot
000105200722024-07-311602.21.21±0.081.21 \pm0.0814.1±1.814.1 \pm1.81.21±0.141.21 \pm0.140.004±0.220.004 \pm0.221.26±0.101.26 \pm0.105.00±2.145.00 \pm2.140.140.14†

Table 1. Swift-XRT observation log and best-fitting spectral parameters for OQ 334 in the 0.3–10 keV band. Note—All models include Galactic absorption fixed at NH=1.08×1020 cm−2N_{\mathrm{H}} = 1.08 \times10^{20}\ \mathrm{cm}^{-2}, and use cstat. FX−rayF_{\mathrm{X-ray}} is the 0.3–10 keV flux in units of 10−12 erg cm−2s−110^{-12}\ \mathrm{erg\ cm}^{-2}\mathrm{s}^{-1}. Parameters marked † had errors that were not constrained by the fit. The full table is available in the supplementary material.

Spectral Hysteresis

Spectral hysteresis describes the loop-like trajectories traced in the hardness–intensity or flux–photon-index plane during a flare. This phenomenon is driven by the interplay of particle injection, energy-dependent acceleration, and radiative cooling timescales within the jet ([65, 42, 100]). The direction of this loop (clockwise vs. counter-clockwise) and its geometric morphology serve as a diagnostics for the physical conditions of the emitting region. A clockwise (CW) loop is conventionally associated with a “soft lag”, where the high-energy emission leads the low-energy emission, indicating a regime dominated by synchrotron cooling processes ([99, 100]). Conversely, a counter-clockwise (CCW) loop is linked to a “hard lag”, which is predicted when the particle acceleration timescale is comparable to or dominates the cooling timescale, causing particles to be gradually accelerated into the radiating window and resulting in low-energy variations leading the high-energy ([42, 100]).

To investigate the presence of spectral hysteresis in our source, we examined the relationship between the X-ray flux and the photon index Γ\Gamma for three flares (X1, X2, and X3) identified in the X-ray light curve. These flare periods were selected using the Bayesian Blocks algorithm (Scargle et al. [93]), a statistically robust method for adaptive binning that identifies change points in time series data by segmenting observations into blocks of constant average flux based on a likelihood function. We used the Astropy implementation with a false-alarm probability of p0=0.05p_0=0.05 to detect different flux states, with flaring periods defined by the threshold FBB≥2×FˉF_{\mathrm{BB}}\geq2\times\bar{F}, where Fˉ\bar{F} is the long-term mean count rate. The identified flares occurred during MJD 58660–58688 (X1), MJD 58852–58900 (X2), and MJD 60485–60524 (X3). The flare light curve and hysteresis plots are shown in Figure 6.

Figure 6. Spectral hysteresis and X-ray light curves of OQ 334 during flares X1, X2, and X3, shown from left to right, respectively. In each column, the upper panel presents the light curve of the corresponding flare, while the lower panel shows its evolution in the flux-photon index plane. Arrows in the lower panels trace the hysteresis loops and denote the direction of temporal evolution. Both panels are colour coded by the time elapsed since the first observation of the flare.

The X1 flare exhibits a clockwise (CW) loop. The average count rate is 0.125±\pm0.036 cts s−1^{-1}, with the photon index spanning a wide range from 1.151 to 2.352. The X2 flare exhibits a complex, multi-loop trajectory, reflecting multi-peaked light curve morphology. The average count rate during this period is 0.145±\pm0.014 cts s−1^{-1}, with the photon index ranging between 1.364 and 2.203. The trajectory initiates with a counter-clockwise (CCW) loop, transitions into a clockwise (CW) loop near the main peak, and reverts back to a CCW loop during the subsequent peak. The first peak in the light curve is spectrally softer, while the second peak is notably harder. The X3 flare is the brightest of the three, with an average count rate of 0.237±0.0220.237 \pm0.022 cts s−1^{-1} and highest brightness of 0.358±0.0240.358 \pm0.024 cts s−1^{-1} at MJD 60489.12. The photon index remained persistently hard throughout the flare, ranging from 1.192 to 1.739. Its flux-photon index trajectory is incomplete, as the observations began near the flux peak and did not sample the full rise or decay of the flare. The observed trajectory traces a CW pattern from the high-flux peak, followed by a CCW loop associated with the subsequent peak.

Hardness Ratio

To investigate the variability of the X-ray emission, we extracted light curves in three energy bands: soft band (0.3-2.0 keV), hard band (2.0-10.0 keV), and the full band (0.3-10.0 keV). These light curves are displayed in the first panel of Figure 7, and the second panel shows the hardness ratio (HR, defined as the ratio of hard to soft counts). The HR provides a model-independent probe of spectral variations across different flux states.

Figure 7. X-ray temporal and spectral variability of OQ 334. From top to bottom, the panels show the light curves in the full (0.3-10.0 keV), hard (2.0-10.0 keV), and soft (0.3-2.0 keV) energy bands; the hardness ratio, defined as the ratio of the hard to soft band count rates; and the photon index obtained from spectral fits. The three flares (X1, X2, and X3) are indicated by shaded regions.

The average count rates 0.085 ±\pm 0.006 cts s−1^{-1} in the soft band, 0.038 ±\pm 0.003 cts s−1^{-1} in the hard band, and 0.123 ±\pm 0.009 cts s−1^{-1} in the full band. The soft band dominates the total X-ray flux throughout the observation period. The fractional variability measurements of 57.32 ±\pm 1.46%, 60.29 ±\pm 2.16%, and 55.89 ±\pm 1.21% for the soft, hard, and full bands respectively, indicate strong variability across all energy ranges, with the hard band showing slightly higher variability than the soft band.

Figure 8 presents the correlations between energy bands and HR in two panels. A strong positive correlation is found between the hard and soft-band count rates (Pearson’s r=0.77r = 0.77, p=4.02×10−14p = 4.02 \times10^{-14}), while no significant correlation is found between the full-band count rate and HR (Pearson’s r=−0.03r = -0.03, p=7.97×10−1p = 7.97 \times10^{-1}), consistent with the absence of any trend between flux and photon index. Table 2 quantifies the spectral evolution across the three flares. The first two flares (X1 and X2) exhibit relatively similar spectral characteristics, with soft-band-dominated emission (HR ∼\sim 0.32-0.37) and photon indices (Γ=1.97±0.09\Gamma= 1.97 \pm0.09 and 1.71 ±\pm 0.04). In contrast, the third flare (X3) displays different behaviour. While it reaches the highest flux level, it is significantly harder (HR = 0.554 ±\pm 0.027, Γ=1.34±0.06\Gamma= 1.34 \pm0.06).

EpochTime PeriodF‾Soft\overline{F}_{\mathrm{Soft}} (cts s−1^{-1})F‾Hard\overline{F}_{\mathrm{Hard}} (cts s−1^{-1})HRΓ\Gamma
X158660.0 – 58688.00.094 ± 0.0330.030 ± 0.0030.323 ± 0.0301.97 ± 0.09
X258852.0 – 58900.00.105 ± 0.0110.039 ± 0.0030.374 ± 0.0191.71 ± 0.04
X360485.0 – 60524.00.153 ± 0.0160.085 ± 0.0090.554 ± 0.0271.34 ± 0.06

Table 2. X-ray properties of OQ 334 during the three flaring periods.

Figure 8. Correlation plots from the Swift-XRT hardness ratio (HR) analysis of OQ 334. The left panel shows the hard-band (2.0–10.0 keV) versus soft-band (0.3–2.0 keV) count rates, while the right panel shows the full-band (0.3–10.0 keV) count rate versus HR. The black lines indicate the corresponding linear fits.

Changing-look Nature

To investigate the changing-look (CL) behaviour of the source, we adopted the state classification from Table 1 of Ren et al. [86] and combined the Swift-XRT observations corresponding to each CL states. The spectral analysis shown in Figure 9 reveals a clear and systematic separation of the three spectral states, BL Lac, FSRQ, and Transition, in the X-ray flux-photon index space. This evolution of the X-ray photon index (ΓX\Gamma_{\mathrm{X}}) closely mirrors the behaviour of the γ\gamma-ray photon index (Γγ\Gamma_{\gamma}) during CL states quantified by Ren et al. [86]. In their work, CL states are defined via the combined evolution of broad-line EW and Γγ\Gamma_{\gamma}, with characteristic ranges: Γγ≳2.2\Gamma_{\gamma} \gtrsim2.2 for FSRQ states, 2.0<Γγ<2.22.0 < \Gamma_{\gamma} < 2.2 for transition states, and Γγ≲2.0\Gamma_{\gamma} \lesssim2.0 for BL Lac states. The *Note—*X1, X2, and X3 denote the three X-ray flares, and the time periods are given in MJD. F‾Soft\overline{F}_{\mathrm{Soft}} and F‾Hard\overline{F}_{\mathrm{Hard}} are the mean count rates in the soft (0.3-2.0 keV) and hard (2.0-10.0 keV) bands, respectively. HR is the hardness ratio, defined as the ratio of the hard to soft band count rates, and Γ\Gamma is the photon index from the absorbed power-law fits.

Figure 9. Swift-XRT analysis of the changing-look (CL) states of OQ 334 identified by Mishra et al. [76] and Ren et al. [86]. The left panel shows the photon index (top) and the X-ray flux (bottom) for each CL state, while the right panel shows the correlation between the flux and photon index during these states. The black line indicates the corresponding linear fit.

X-ray spectral evolution follows an opposite trend to that observed in γ\gamma rays: FSRQ states exhibit the hardest X-ray spectra, with ΓX\Gamma_{\mathrm{X}} clustered at ΓX≈1.32−1.52\Gamma_{\mathrm{X}} \approx1.32 - 1.52, while BL Lac states show softer X-ray spectra, with ΓX≈1.77−2.36\Gamma_{\mathrm{X}} \approx1.77 - 2.36. Transition states occupy an intermediate space with ΓX≈1.49−1.94\Gamma_{\mathrm{X}} \approx1.49 - 1.94. This clustering indicates that the X-ray emission is coupled to the CL state of the blazar. Within CL states, we find a positive correlation between FX−rayF_{\mathrm{X-ray}} and ΓX\Gamma_{\mathrm{X}} (r=0.59r = 0.59, p=2.53×10−2p = 2.53 \times10^{-2}), indicating a “softer-when-brighter” trend in the X-rays.

The X-ray flares X1 and X2 coincide with the CL states (Figure 10). Flare X1 overlaps the BL Lac state B2 and the transition states T3 and T4, while flare X2 spans T5, B3, T6, and F4. Flare X3 occurred after the time period covered by the previous CL studies. In both flares X1 and X2, ΓX\Gamma_{\mathrm{X}} increases from the transition state into the BL Lac state and then decreases again in the subsequent transition state, tracing a rise-and-fall pattern across the CL sequence. Using long-term ZTF photometry, Chand & Gopal-Krishna [29] showed that FSRQ →\to BL Lac transitions were accompanied by an optical flare of ≈1.5−2.5\approx1.5 - 2.5 mag. For Flare X1, the optical counterpart has amplitudes of Δmg=1.734\Delta m_g = 1.734 mag, Δmr=1.119\Delta m_r = 1.119 mag, and Δmi=0.939\Delta m_i = 0.939 mag. Flare X2 shows exceptional optical variability by Δmg=3.651\Delta m_g = 3.651 mag, Δmr=3.465\Delta m_r = 3.465 mag, and Δmi=2.983\Delta m_i = 2.983 mag. Flare X3 flare with Δmg=2.097\Delta m_g = 2.097 mag, no rr- or ii-band data are available.

Figure 10. Swift-XRT and ZTF light curves of the OQ 334 changing-look (CL) states of OQ 334 identified by Mishra et al. [76] and Ren et al. [86]. Shaded regions indicate CL states: BL Lac (blue), transition (green), and FSRQ (red).

Broadband SED

The broadband SEDs of X-ray flares were modelled with the standard one-zone leptonic framework using the open-source Python package agnpy (v0.5.1, Nigro et al. [77]). The model assumes a single homogeneous spherical emission region of radius RbR_{\mathrm{b}}, filled isotropically with a uniform magnetic field BB and a population of relativistic electrons. The blob moves along the jet with bulk Lorentz factor Γ\Gamma and velocity βc\beta c, with the observed emission Doppler boosted by δD=[Γ(1−βcos⁡θ)]−1\delta_{\mathrm{D}} = [\Gamma(1 - \beta\cos\theta)]^{-1}, where θ\theta is the viewing angle. The size of the emission region is further constrained by the shortest observed variability timescale tvart_{\mathrm{var}} through the causality condition Rb≲cδDtvar/(1+z)R_{\mathrm{b}} \lesssim c \delta_{\mathrm{D}} t_{\mathrm{var}}/(1 + z). For OQ 334, the shortest γ\gamma-ray variability timescale reported is ∼9\sim9 hours (Prince et al. [84]). The relativistic electron population is described by a broken power-law energy distribution,

n(γ)=k[(γγb)−p1H(γ;γmin⁡,γb)+(γγb)−p2H(γ;γb,γmax⁡)],(5)n(\gamma) = k \left[ \left(\frac{\gamma}{\gamma_{\mathrm{b}}}\right)^{-p_1} H(\gamma;\gamma_{\min},\gamma_{\mathrm{b}}) + \left(\frac{\gamma}{\gamma_{\mathrm{b}}}\right)^{-p_2} H(\gamma;\gamma_{\mathrm{b}},\gamma_{\max}) \right], \tag*{(5)}

where γ\gamma is the electron Lorentz factor in the comoving frame, kk is the normalisation, and p1p_1 and p2p_2 are the indices below and above the break (γb\gamma_{\mathrm{b}}). The Heaviside functions (HH) restrict each power-law component to its respective energy range.

The broadband emission is reproduced by three non-thermal radiative components: synchrotron radiation, synchrotron self-Compton (SSC) emission, and external Compton (EC) scattering from the broad-line region (BLR) and dusty torus (DT), along with thermal components from the accretion disk and dusty torus. The low-energy hump of the SED is modelled as synchrotron emission from relativistic electrons in the emitting blob, accounted for synchrotron self-absorption. The same electron population upscatters both the synchrotron photons via IC scattering, producing the SSC component, and photons from the external radiation fields, producing the EC components responsible for the high-energy hump of the SED. The SED of each component is calculated using the equations given in Appendix A of Nigro et al. [77].

The luminosities and characteristic radii of the two external fields scale with the accretion disk luminosity (LdiskL_{\mathrm{disk}}) following Ghisellini & Tavecchio [50]. The BLR radius is given by RBLR=1017Ld,451/2 cmR_{\mathrm{BLR}} = 10^{17}L_{\mathrm{d,45}}^{1/2}\ \mathrm{cm}, where Ld,45=Ldisk/(1045 erg s−1)L_{\mathrm{d,45}} = L_{\mathrm{disk}}/(10^{45}\ \mathrm{erg\ s}^{-1}), while the DT radius is RDT=2.5×1018Ld,451/2 cmR_{\mathrm{DT}} = 2.5 \times10^{18}L_{\mathrm{d,45}}^{1/2}\ \mathrm{cm}. The BLR is modelled as an infinitesimally thin spherical shell of radius RBLRR_{\mathrm{BLR}} reprocessing a fraction ξline\xi_{\mathrm{line}} of the disk radiation and re-emitting it as a monochromatic Lyman-α\alpha line (Finke [44]). The DT is modelled as an infinitesimally thin ring of radius RDTR_{\mathrm{DT}} lying in the plane perpendicular to the jet axis, which reprocesses a fraction ξDT\xi_{\mathrm{DT}} of the disk radiation and re-emits it as thermal infrared radiation characterised by a single blackbody temperature TDTT_{\mathrm{DT}} (Finke [44]). The disk luminosity is fixed to Ldisk=2×1046 erg s−1L_{\mathrm{disk}} = 2 \times10^{46}\ \mathrm{erg\ s}^{-1}, estimated from the broad Mg II emission line flux measured during the 2020 multi-wavelength campaign (MAGIC Collaboration et al. [68]).

during the 2020 multi-wavelength campaign (MAGIC Collaboration et al. [68]). The corresponding radii of the two external fields are RBLR=4.4×1017 cmR_{\mathrm{BLR}} = 4.4 \times10^{17}\ \mathrm{cm} and RDT=1.1×1019 cmR_{\mathrm{DT}} = 1.1 \times10^{19}\ \mathrm{cm}. The reprocessing fractions are set to ξline=0.1\xi_{\mathrm{line}} = 0.1 for the BLR and ξDT=0.6\xi_{\mathrm{DT}} = 0.6 for the DT, and the torus dust is modelled as a blackbody at TDT=1000 KT_{\mathrm{DT}} = 1000\ \mathrm{K}.

The SED data are supplemented with nonsimultaneous Planck radio measurements retrieved from the Markarian Multi-wavelength Data Center (Sahakyan et al. [92]; Planck Collaboration et al. [82]). Because the high-energy emission region in OQ 334 is physically offset from the radio-emitting core by ≈11\approx11 pc, the observed radio emission is dominated by the slowly varying larger-scale jet (Prince et al. [84]), and these measurements are therefore used only to define the low-frequency part of the synchrotron component. The SED of the flares are shown in Figure 11. The radiative processes are combined into a single spectral model through the ExternalComptonModel class of agnpy, which interfaces with the modeling sub-package of gammapy (Donath et al. 2023). We first obtain a best-fit solution by minimising the χ2\chi^2 statistic with the iminuit optimizer (Dembinski et al. [33]), and then use the resulting parameter values to initialise a Markov Chain Monte Carlo (MCMC) sampling performed with emcee (Foreman-Mackey et al. [45]). Uniform priors are assigned to all free model parameters, and the resulting posterior probability distributions are used to derive the parameter values and their uncertainties.

Figure 11. Broad-band SEDs of OQ 334 during the three X-ray flaring periods X1, X2, and X3, shown in blue, green, and red, respectively. The Planck data points, plotted in black, are common to all three flares.

The electron Lorentz factors are held fixed throughout, at γmin⁡=10\gamma_{\min} = 10, following MAGIC Collaboration et al. [68], which showed that such a value is necessary to correctly reproduce the valley observed between the synchrotron and IC humps, and at γmax⁡=1×107\gamma_{\max} = 1 \times10^{7}. The location of the emission region along the jet axis was left free in the initial MCMC sampling. The sampling consistently converged to r≈1.5×1019r \approx1.5 \times10^{19} cm, far outside the BLR boundary and beyond the characteristic radius of the DT, where the BLR contribution is negligible, and the EC scattering on DT becomes the dominant high-energy process [68, 81]. The value of rr is therefore fixed in the final sampling, leaving six free parameters: kek_e, p1p_1, p2p_2, γb\gamma_b, δD\delta_D, BB.

The broadband SED and the best-fit model are shown in Figure 12. Table 3 reports the fitted values of the free parameters, and Table 4 reports the derived SED quantities. The break Lorentz factor rises from X1 to X2 and stays at a comparable level at X3, whereas the electron normalisation shows no clear flare-to-flare trend. The low-energy index is almost the same between X1 and X2 but softens significantly at X3, while the high-energy index increases from X1 to X2 and then remains at this higher value at X3. As a result, the spectral break p2−p1p_2-p_1 increases from X1 to the later epochs. The Doppler factor is consistent between X1 and X2 and drops at X3. The magnetic field remains low throughout, being identical within errors at X1 and X3 and decreasing slightly at X2. The synchrotron peak frequency varies by a factor **Note—**Quoted values are posterior medians with 16th and 84th percentile uncertainties. kek_{\mathrm{e}} is the electron distribution normalization (cm−3^{-3}); p1p_{1} and p2p_{2} are the low- and high-energy electron spectral indices; γb\gamma_{\mathrm{b}} is the break Lorentz factor; δD\delta_{\mathrm{D}} is the Doppler factor; and BB is the magnetic field strength (G).

Epochkek_{\mathrm{e}}p1p_{1}p2p_{2}γb\gamma_{\mathrm{b}}δD\delta_{\mathrm{D}}BB
X17.09−2.46+3.39×10−47.09^{+3.39}_{-2.46} \times10^{-4}1.95−0.08+0.081.95^{+0.08}_{-0.08}4.05−0.06+0.084.05^{+0.08}_{-0.06}5.24−0.68+0.84×1035.24^{+0.84}_{-0.68} \times10^{3}34.20−1.37+1.4334.20^{+1.43}_{-1.37}0.136−0.007+0.0080.136^{+0.008}_{-0.007}
X23.33−0.88+1.13×10−43.33^{+1.13}_{-0.88} \times10^{-4}1.94−0.04+0.041.94^{+0.04}_{-0.04}4.75−0.10+0.124.75^{+0.12}_{-0.10}9.23−0.91+1.11×1039.23^{+1.11}_{-0.91} \times10^{3}34.69−1.01+0.9634.69^{+0.96}_{-1.01}0.113−0.004+0.0040.113^{+0.004}_{-0.004}
X35.99−2.06+3.21×10−45.99^{+3.21}_{-2.06} \times10^{-4}2.20−0.06+0.052.20^{+0.05}_{-0.06}4.87−0.18+0.104.87^{+0.10}_{-0.18}8.15−0.97+0.96×1038.15^{+0.96}_{-0.97} \times10^{3}28.01−1.21+1.1828.01^{+1.18}_{-1.21}0.136−0.008+0.0090.136^{+0.009}_{-0.008}

Table 3. Fitted SED model parameters of OQ 334 during the X-ray flares

Epochνsynpeak\nu_{\mathrm{syn}}^{\mathrm{peak}} (Hz)νICpeak\nu_{\mathrm{IC}}^{\mathrm{peak}} (Hz)CDUeU_{\mathrm{e}} (erg cm−3^{-3})UBU_{\mathrm{B}} (erg cm−3^{-3})Ue/UBU_{\mathrm{e}}/U_{\mathrm{B}}PjetP_{\mathrm{jet}} (erg s−1^{-1})
X12.92×10142.92 \times10^{14}3.08×10233.08 \times10^{23}6.079.24×10−29.24 \times10^{-2}7.31×10−47.31 \times10^{-4}1261.05×10471.05 \times10^{47}
X26.06×10146.06 \times10^{14}8.57×10238.57 \times10^{23}9.281.38×10−11.38 \times10^{-1}5.04×10−45.04 \times10^{-4}2741.48×10471.48 \times10^{47}
X33.91×10143.91 \times10^{14}6.16×10226.16 \times10^{22}5.574.63×10−14.63 \times10^{-1}7.35×10−47.35 \times10^{-4}6315.87×10475.87 \times10^{47}

Table 4. Derived SED properties of OQ 334 during the X-ray flares

Figure 12. Broadband SED modelling of OQ 334 during the three X-ray flares X1, X2, and X3, shown from top to bottom, respectively. In each panel, the individual radiative components and the observed data points are identified in the legend.

Note—νsynpeak\nu_{\mathrm{syn}}^{\mathrm{peak}} and νICpeak\nu_{\mathrm{IC}}^{\mathrm{peak}} are the synchrotron and inverse-Compton peak frequencies of the best-fit model; CD is the Compton dominance, defined as the ratio of the inverse-Compton to synchrotron peak νFν\nu F_{\nu}. UeU_{\mathrm{e}} and UBU_{\mathrm{B}} are the comoving electron and magnetic energy densities, and Ue/UBU_{\mathrm{e}}/U_{\mathrm{B}} is the departure from equipartition. PjetP_{\mathrm{jet}} is the total jet power, summing the electron, cold-proton and magnetic contributions, computed assuming one cold proton per emitting electron.

tor of ∼2\sim2 across the three epochs, while the IC peak shifts by more than an order of magnitude, reaching its highest value at X2 and dropping at X3. Compton dominance follows a similar pattern, peaking at X2 and declining at X3. The energy densities and jet powers indicate a particle-dominated emission region that departs strongly from equipartition in all three flaring epochs. The magnetic energy density remains relatively stable, whereas the electron energy density increases progressively from X1 to X3. Consequently, the departure from equipartition strengthens across the flares, with Ue/UBU_{\mathrm{e}}/U_{\mathrm{B}} rising from X1 to X3. This increase in particle density is reflected in the total jet power, computed as Pjet=2πR22Γ22βcUiP_{\mathrm{jet}} = 2\pi R_{2}^{2}\Gamma_{2}^{2}\beta c U_{i}, where UiU_{i} represents the energy density of the components: electrons, cold protons, and magnetic field. The resulting jet power grows from X1 to X3, assuming one cold proton per relativistic electron [28].

Discussion

Timing & Optical Analysis

The cross-correlation analysis provides evidence for a close temporal connection between the inter-band emission. The near-zero lag suggests that the optical and γ\gamma-ray emitting components respond to the same underlying variations in the relativistic jet. Given the weekly binning of the LAT light curve, this lag indicates that the optical and γ\gamma-ray emission originate in the same region. A radio–γ\gamma-ray connection is also present in this source, with radio variations leading the γ\gamma-ray emission by ∼70\sim70 days [84]. The redder-when-brighter behaviour observed is commonly reported in FSRQs and is generally attributed to the increasing dominance of the variable synchrotron emission from the jet, which has a redder spectrum, over the comparatively stable thermal emission from the accretion disc as the source brightens [29]. The correlation between the two colour indices shows that the colour changes are consistent across the optical bands, in line with the colour–magnitude diagrams.

X-ray Analysis

The AICc_{\mathrm{c}}-based model comparison indicates that, although all three spectral models provide statistically acceptable descriptions of the X-ray spectra, the majority of the observations are best represented by a simple power-law shape. This suggests that a single-component spectrum suffices to describe the X-ray emission of the source in most observations, with no strong requirement for additional curvature or spectral breaks. The absence of a clear correlation between the X-ray flux and the photon index implies that variations in the source brightness are not accompanied by systematic spectral hardening or softening over long timescales. In BL Lacs, the photon index generally shows a dependence on flux, with a harder-when-brighter trend when the X-ray band samples the synchrotron tail and a softer-when-brighter trend when the synchrotron and IC components overlap in the band [54, 104]. FSRQs, in which the X-ray emission is generally dominated by the IC component, show a harder-when-brighter trend [55, 58, 91] or no correlation over long timescales [32, 107]. The absence of a long-term correlation in the source is closer to the behaviour reported for FSRQs than to that of BL Lacs.

The distinct morphologies and directions of the hysteresis loops observed across the three flaring periods reveal a dynamic emitting environment with shifting physical regimes. During Flare X1, the CW loop in the flux-Γ\Gamma plane is consistent with energy-dependent radiative cooling, in which the highest-energy electrons cool more rapidly and spectral softening follows the flux decline [65, 42]. Flares X2 and X3 display complex, multi-loop trajectories that transition between CCW and CW directions, signifying a departure from the simple cooling-dominated state. For Flare X3, the hard photon index (Γ\Gamma ranging from 1.192 to 1.739) suggests that the high-energy electron population was continuously replenished by rapid particle acceleration or sustained injection, preventing the cooling-induced softening seen in X1.

The hardness ratio analysis provides a robust, direct probe of spectral variability that complements the spectral fitting. The strong positive correlation between the hard-band and soft-band count rates shows that flux variations across the X-ray spectrum are tightly coupled. Three flaring periods reveal a progressive, monotonic hardening of the X-ray band. The photon index decreases systematically across the three flares, Γ=1.97±0.09\Gamma= 1.97 \pm0.09 (X1) →1.71±0.04\rightarrow1.71 \pm0.04 (X2) →1.34±0.06\rightarrow1.34 \pm0.06 (X3), while the mean HR increases in lock-step, HR =0.323±0.030→0.374±0.019→0.554±0.027= 0.323 \pm0.030 \rightarrow0.374 \pm0.019 \rightarrow0.554 \pm0.027. This monotonic evolution implies that the spectral changes are not purely stochastic but trace a physical transition in the X-ray flares. The progression from a soft flaring state to a hard flaring state indicates that Figure 13, as the source becomes increasingly populated by high energy particles, the X-ray emitting region undergoes a systematic change in its underlying physical conditions.

Figure 13. Swift-XRT spectra of OQ 334 during the three X-ray flares X1, X2, and X3, fitted with an absorbed power-law model. Vertical offsets have been applied to the spectra for clarity of presentation, and each spectrum is accompanied by a colour-matched vertical axis.

Changing-look Nature

The flux-ΓX\Gamma_{\mathrm{X}} analysis of the CL states shows that the X-ray spectral state of the source is directly coupled to its CL classification. The measured photon index reflects the relative contribution of the synchrotron and IC components to the X-ray band. The hard spectra of the FSRQ states (ΓX≲1.5\Gamma_{\mathrm{X}} \lesssim1.5) are consistent with IC-dominated emission, whereas the soft spectra of the BL Lac states (ΓX≳1.8\Gamma_{\mathrm{X}} \gtrsim1.8) indicate an increased contribution of the synchrotron tail to the soft X-ray band, with the transition states occupying an intermediate regime. The “softer-when-brighter” trend within the CL states confirms this interpretation, implying that the flux enhancements during these states are driven by the synchrotron component, whose steep spectrum progressively softens the composite emission as it brightens.

Flares X1 and X2 includes the BL Lac states B2 and B3 from the CL classification (Figure 9). The BL Lac states correspond to phases of elevated jet activity rather than a quiescence. The optical flares reported by Chand & Gopal-Krishna [29] coincide with the flares X1 and X2, thereby extending the flare-CL transition association to the X-ray band. This association can be examined only for B2 and B3, as X-ray observations are not available for the other BL Lac states. This enhanced jet continuum can dilute the broad MgII line below the EW threshold defining the BL Lac state. Although simultaneous optical spectroscopy is needed

to confirm this scenario directly, the optical and X-ray behaviour during B2/X1 and B3/X2 is consistent with jet-continuum dilution of the BELs rather than an intrinsic change in the accretion disk ([76, 29]). This is consistent with spectroscopic studies of other CLB sources, which show strong variability in the optical continuum and high-energy emission while the BELs remain comparatively stable ([30, 13, 79]). Flare X3 challenges the interpretation that the CL transitions trace a systematic evolution of the central engine ([87 a]). If the association between X-ray flaring and spectral softening observed during the CL states extended to the full observation period, the brightest outburst would be expected to display the most pronounced softening. Instead, Flare X3 has the hardest photon index of the three flares (ΓX=1.34±0.06\Gamma_{\mathrm{X}} = 1.34 \pm0.06), characteristic of the FSRQ states (ΓX≲1.5\Gamma_{\mathrm{X}} \lesssim1.5). The persistence of a hard spectrum at peak brightness indicates that the flux enhancement during X3 is driven by the IC component.

Broadband SED

The one-zone leptonic model successfully reproduces the broadband emission across all three flares, with external seed photons supplied by the thermal radiation of the dusty torus. The X-ray emission is produced by the SSC component, which is strong and extends towards the IC peak, whereas the γ\gamma-ray emission is dominated by EC-dust scattering. SED modelling results are plotted in Figure 12. The magnetic field remains weak and approximately constant (B≈0.11B \approx0.11–0.140.14), indicating that the flaring activity is driven by changes in the particle population rather than magnetic-field variations. The rise in PjetP_{\mathrm{jet}} and in Ue/UBU_{\mathrm{e}}/U_{\mathrm{B}} at X3 is due to the increase in UeU_{\mathrm{e}} from the injection of kinetic energy into the emitting electron population, consistent with the particle-dominated flaring scenario (Prince et al. [84]). The emission region remains far from equipartition in all three epochs, with Ue/UB≫1U_{\mathrm{e}}/U_{\mathrm{B}} \gg1, so a magnetically dominated region is not required to reproduce the broadband SEDs. For the VHE flare, which falls within X2, MAGIC Collaboration et al. [68] located the emission region inside the DT ring, which yields a higher magnetic field and an emission region close to equipartition. The jet power is high in all three flares and exceeds LeddL_{\mathrm{edd}}. The power carried by FSRQ jets is often larger than the disc luminosity and can surpass the Eddington luminosity ([53]).

The synchrotron peak frequencies derived for X1, X2, and X3 all lie within the intermediate-synchrotron-peaked (ISP) band under the standard definition νsynpeak=1014\nu_{\mathrm{syn}}^{\mathrm{peak}} = 10^{14}–101510^{15} Hz, indicating that the synchrotron component does not migrate along the blazar sequence across these three flares. Although the source is classified as LSP in the 4LAC catalogue on the basis of a time-averaged synchrotron peak at 3.16×10133.16 \times10^{13} Hz, our measured peak frequencies are an order of magnitude higher and are consistent with the ISP classification of Prince et al. [84], whereas Ren et al. [87] place the peak below 101410^{14} Hz. The high-energy peak shows a different behaviour, with the IC peak displaced well above the long-term average during X1 and X2 but returning close to it during X3.

For an ISP blazar like OQ 334, the Swift-XRT band (≈1017\approx10^{17}–101810^{18} Hz) lies in the spectral valley between the declining high-energy tail of the synchrotron hump and the rising branch of the IC component. In this regime, the observed X-ray spectrum is a superposition of synchrotron and IC emission, and its shape is sensitive to their relative contributions. The three X-ray flares reveal a clear spectral evolution that reflects a changing balance between these two components. Flare X1 shows the softest X-ray spectrum, with SED model indicating a substantial synchrotron contribution in addition to the IC component. During X2, the spectrum hardens relative to X1, with a reduced synchrotron fraction and a correspondingly larger IC contribution. By X3, the spectrum is the hardest of the three, and the SED modelling shows that the X-ray emission is dominated entirely by the IC component.

Conclusion

We have presented a multi-wavelength study of the CL TeV blazar OQ 334 (B2 1420+326) spanning MJD 54769.0 to 60824.0, focusing on the X-ray regime. The analysis centres on the CL nature of the source and the X-ray flaring activity identified over this interval. The main results are summarised as follows.

  1. The source exhibits intense, high-amplitude non-thermal variability across all observed wavebands, with tight optical, X-ray, and γ\gamma-ray coupling, while the optical emission shows significant colour-colour and colour-magnitude correlations.

  1. The AICc comparison shows that all three spectral models provide statistically acceptable fits, with a simple power law favoured for the majority of Swift-XRT observations and log-parabolic or broken power-law models preferred in fewer cases.

  1. Spectral hysteresis analysis reveals distinct evolution across the three flares: X1 shows a clockwise loop consistent with cooling-dominated emission, while X2 and X3 exhibit complex multi-loop trajectories with changing directions, indicating shifts in the dominant physical processes.

  1. The hardness ratio analysis reveals a systematic hardening across the three X-ray flares, with the photon index decreasing from Γ=1.97±0.09\Gamma= 1.97 \pm0.09 (X1) to 1.71±0.041.71 \pm0.04 (X2) and 1.34±0.061.34 \pm0.06 (X3), indicating a progressive change in the physical conditions of the X-ray flaring.

  1. The X-ray spectra systematically distinguish the CL states, with FSRQ states being harder, BL Lac states softer, and transition states intermediate. The CL behaviour is therefore consistent with jet-driven dilution of the BELs by a variable synchrotron continuum.

  1. The broadband SEDs are well described by a one-zone leptonic model, with the X-ray emission evolving from a substantial synchrotron contribution in X1 to increasing IC dominance in X2 and complete IC dominance in X3. The flares are particle dominated, with their energetics driven primarily by changes in the electron population rather than the magnetic field.

These results reveal a systematic evolution of the X-ray flaring in OQ 334, from soft, synchrotron-influenced flares associated with the apparent FSRQ →\rightarrow BL Lac transitions to a hard, IC-dominated flare. The X-ray band provides a sensitive probe of the changing balance between the emission components and the CL behaviour.

Acknowledgments: This paper is supported by the European Union’s Horizon Europe research and innovation programme under grant agreement No 101131928, project ACME. This work was partially supported by a program of the Polish Ministry of Science under the title “Regional Excellence Initiative”, project No. RID/SP/0050/2024/1.

Facilities: MAGIC, Fermi (LAT), Swift (XRT and UVOT), ASAS-SN, ZTF, Plank/MMDC.

Software: Astropy, Gammapy, Fermipy, HEASoft, pyZDCF, agnpy, emcee, iminuit.

References

  1. [1]Abdo, A. A., Ackermann, M., Agudo, I., et al. 2010, ApJ, 716, 30, doi: 10.1088/0004-637X/716/1/30DOI
  2. [2]Abdollahi, S., Acero, F., Ackermann, M., et al. 2020, ApJS, 247, 33, doi: 10.3847/1538-4365/ab6bcbDOI
  3. [3]Abe, S., Abhir, J., Abhishek, A., et al. 2024, Journal of High Energy Astrophysics, 44, 266, doi: 10.1016/j.jheap.2024.09.011DOI
  4. [4]Aharonian, F. A. 2002, MNRAS, 332, 215, doi: 10.1046/j.1365-8711.2002.05292.xDOI
  5. [5]Akaike, H. 1974, IEEE Transactions on Automatic Control, 19, 716, doi: 10.1109/TAC.1974.1100705DOI
  6. [6]Aleksić, J., Ansoldi, S., Antonelli, L. A., et al. 2016a, Astroparticle Physics, 72, 61, doi: 10.1016/j.astropartphys.2015.04.004DOI
  7. [8]Alexander, T. 1997, in Astrophysics and Space Science Library, Vol. 218, Astronomical Time Series, ed. D. Maoz, A. Sternberg, & E. M. Leibowitz, 163, doi: 10.1007/978-94-015-8941-3_14
  8. [9]Alexander, T. 2013, arXiv e-prints, arXiv:1302.1508, doi: 10.48550/arXiv.1302.1508
  9. [10]Arnaud, K. A. 1996, 101, 17
  10. [11]Atwood, W., Albert, A., Baldini, L., et al. 2013, arXiv e-prints, arXiv:1303.3514, doi: 10.48550/arXiv.1303.3514DOI
  11. [12]Atwood, W. B., Abdo, A. A., Ackermann, M., et al. 2009, ApJ, 697, 1071, doi: 10.1088/0004-637X/697/2/1071DOI
  12. [13]Azzollini, A., Buson, S., & Coleiro, A. 2025, A&A, 703, L13, doi: 10.1051/0004-6361/202557347DOI
  13. [14]Bellm, E. C., Kulkarni, S. R., Graham, M. J., et al. 2019, PASP, 131, 018002, doi: 10.1088/1538-3873/aaecbeDOI
  14. [15]Bhatta, G. 2021a, ApJ, 923, 7, doi: 10.3847/1538-4357/ac2819DOI
  15. [17]Blandford, R. D., & Königl, A. 1979, ApJ, 232, 34, doi: 10.1086/157262DOI
  16. [18]Blandford, R. D., & Levinson, A. 1995, ApJ, 441, 79, doi: 10.1086/175338DOI
  17. [19]Blandford, R. D., & Rees, M. J. 1978, in BL Lac Objects, ed. A. M. Wolfe, 328–341
  18. [20]Blandford, R. D., & Znajek, R. L. 1977, MNRAS, 179, 433, doi: 10.1093/mnras/179.3.433DOI
  19. [21]Böttcher, M., Reimer, A., Sweeney, K., & Prakash, A. 2013, ApJ, 768, 54, doi: 10.1088/0004-637X/768/1/54DOI
  20. [22]Breeveld, A. A., Landsman, W., Holland, S. T., et al. 2011, in American Institute of Physics Conference Series, Vol. 1358, Gamma Ray Bursts 2010, ed. J. E. McEnery, J. L. Racusin, & N. Gehrels (AIP), 373–376, doi: 10.1063/1.3621807
  21. [24]Burnham, K. P., & Anderson, D. R., eds. 2002, Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach, 2nd edn. (New York, NY: Springer Science+Business Media), doi: 10.1007/b97636
  22. [25]Burrows, D. N., Hill, J. E., Nousek, J. A., et al. 2005, SSRv, 120, 165, doi: 10.1007/s11214-005-5097-2DOI
  23. [26]Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245, doi: 10.1086/167900DOI
  24. [27]Cash, W. 1979, ApJ, 228, 939, doi: 10.1086/156922DOI
  25. [28]Celotti, A., & Ghisellini, G. 2008, MNRAS, 385, 283, doi: 10.1111/j.1365-2966.2007.12758.xDOI
  26. [29]Chand, K., & Gopal-Krishna. 2025, MNRAS, 541, L52, doi: 10.1093/mnrasl/slaf053DOI
  27. [30]Chen, G., Yang, W., Liu, Y., et al. 2025, MNRAS, 544, 1926, doi: 10.1093/mnras/staf1818DOI
  28. [31]Ciprini, S. 2018, The Astronomer’s Telegram, 12277, 1
  29. [32]D’Ammando, F., Raiteri, C. M., Villata, M., et al. 2019, MNRAS, 490, 5300, doi: 10.1093/mnras/stz2792DOI
  30. [33]Dembinski, H., & et al., P. O. 2020, doi: 10.5281/zenodo.3949207DOI
  31. [34]Dermer, C. D., Schlickeiser, R., & Mastichiadis, A. 1992, A&A, 256, L27
  32. [35]Domínguez, A., Primack, J. R., Rosario, D. J., et al. 2011, MNRAS, 410, 2556, doi: 10.1111/j.1365-2966.2010.17631.xDOI
  33. [38]Duffy, L., Eracleous, M., Ruan, J. J., Yang, Q., & Runnoe, J. C. 2025, ApJ, 989, 102, doi: 10.3847/1538-4357/adeb73
  34. [39]Edelson, R. A., & Krolik, J. H. 1988, ApJ, 333, 646, doi: 10.1086/166773DOI
  35. [40]Evans, P. A., Beardmore, A. P., Page, K. L., et al. 2007, A&A, 469, 379, doi: 10.1051/0004-6361:20077530DOI
  36. [42]Falcone, A. D., Cui, W., & Finley, J. P. 2004, ApJ, 601, 165, doi: 10.1086/380447DOI
  37. [43]Feng, H.-C., Hu, C., Li, S.-S., et al. 2021, ApJ, 909, 18, doi: 10.3847/1538-4357/abd851DOI
  38. [44]Finke, J. D. 2016, ApJ, 830, 94, doi: 10.3847/0004-637X/830/2/94DOI
  39. [45]Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman, J. 2013, PASP, 125, 306, doi: 10.1086/670067
  40. [46]Gehrels, N., Chincarini, G., Giommi, P., et al. 2004, ApJ, 611, 1005, doi: 10.1086/422091DOI
  41. [47]Ghisellini, G. 2011, in American Institute of Physics Conference Series, Vol. 1381, 25th Texas Symposium on Relativistic AstroPhysics (Texas 2010), ed. F. A. Aharonian, W. Hofmann, & F. M. Rieger (AIP), 180–198, doi: 10.1063/1.3635832DOI
  42. [48]Ghisellini, G., & Celotti, A. 2001, A&A, 379, L1, doi: 10.1051/0004-6361:20011388
  43. [49]Ghisellini, G., & Madau, P. 1996, MNRAS, 280, 67, doi: 10.1093/mnras/280.1.67DOI
  44. [50]Ghisellini, G., & Tavecchio, F. 2009, MNRAS, 397, 985, doi: 10.1111/j.1365-2966.2009.15007.xDOI
  45. [51]Ghisellini, G., Tavecchio, F., Foschini, L., Bonnoli, G., & Tagliaferri, G. 2013, MNRAS, 432, L66, doi: 10.1093/mnrasl/slt041DOI
  46. [52]Ghisellini, G., Tavecchio, F., Foschini, L., & Ghirlanda, G. 2011, MNRAS, 414, 2674, doi: 10.1111/j.1365-2966.2011.18578.xDOI
  47. [53]Ghisellini, G., Tavecchio, F., & Ghirlanda, G. 2009, MNRAS, 399, 2041, doi: 10.1111/j.1365-2966.2009.15397.xDOI
  48. [54]Giommi, P., Perri, M., Capalbi, M., et al. 2021, MNRAS, 507, 5690, doi: 10.1093/mnras/stab2425DOI
  49. [55]Gu, M., & Ai, Y. L. 2011, Journal of Astrophysics and Astronomy, 32, 87, doi: 10.1007/s12036-011-9051-2
  50. [56]Hart, K., Shappee, B. J., Hey, D., et al. 2023, arXiv e-prints, arXiv:2304.03791, doi: 10.48550/arXiv.2304.03791
  51. [57]Hartman, R. C., Bertsch, D. L., Bloom, S. D., et al. 1999, ApJS, 123, 79, doi: 10.1086/313231DOI
  52. [58]Hayashida, M., Madejski, G. M., Nalewajko, K., et al. 2012, ApJ, 754, 114, doi: 10.1088/0004-637X/754/2/114DOI
  53. [59]Healey, S. E., Romani, R. W., Taylor, G. B., et al. 2007, ApJS, 171, 61, doi: 10.1086/513742DOI
  54. [60]Heidt, J., & Wagner, S. J. 1996, A&A, 305, 42, doi: 10.48550/arXiv.astro-ph/9506032DOI
  55. [61]Hewett, P. C., & Wild, V. 2010, MNRAS, 405, 2302, doi: 10.1111/j.1365-2966.2010.16648.xDOI
  56. [62]IRSA. 2022, Galactic Dust Reddening and Extinction, NASA IPAC DataSet, IRSAS537, IPAC, doi: 10.26131/IRSA537DOI
  57. [63]Jankov, I., Kovačević, A. B., Ilić, D., Sánchez-Sáez, P., & Nikutta, R. 2022, pyZDCF: Initial Release, v1.0.0, Zenodo, doi: 10.5281/zenodo.7253034DOI
  58. [64]Kang, S.-J., Zheng, Y.-G., & Wu, Q. 2023, MNRAS, 525, 3201, doi: 10.1093/mnras/stad2456DOI
  59. [65]Kirk, J. G., Rieger, F. M., & Mastichiadis, A. 1998, A&A, 333, 452, doi: 10.48550/arXiv:astro-ph/9801265arxiv.org/abs/astro-ph/9801265
  60. [66]Liddle, A. R. 2007, MNRAS, 377, L74, doi: 10.1111/j.1745-3933.2007.00306.xDOI
  61. [67]Lyu, B., Wu, Q., Yan, Z., Yu, W., & Liu, H. 2022, ApJ, 927, 227, doi: 10.3847/1538-4357/ac5256DOI
  62. [68]MAGIC Collaboration, Acciari, V. A., Ansoldi, S., et al. 2021, A&A, 647, A163, doi: 10.1051/0004-6361/202039687
  63. [69]Maraschi, L., Ghisellini, G., & Celotti, A. 1992, ApJL, 397, L5, doi: 10.1086/186531DOI
  64. [70]Marscher, A. P., & Gear, W. K. 1985, ApJ, 298, 114, doi: 10.1086/163592DOI
  65. [71]Masci, F. J., Laher, R. R., Rusholme, B., et al. 2019, PASP, 131, 018003, doi: 10.1088/1538-3873/aae8acDOI
  66. [72]Mattox, J. R., Bertsch, D. L., Chiang, J., et al. 1996, ApJ, 461, 396, doi: 10.1086/177068DOI
  67. [73]Meegan, C., Lichti, G., Bhat, P. N., et al. 2009, ApJ, 702, 791, doi: 10.1088/0004-637X/702/1/791DOI
  68. [74]Miller, J. S., French, H. B., & Hawley, S. A. 1978, ApJL, 219, L85, doi: 10.1086/182612DOI
  69. [75]Mirzoyan, R. 2020, The Astronomer’s Telegram, 13412, 1
  70. [76]Mishra, H. D., Dai, X., Chen, P., et al. 2021, ApJ, 913, 146, doi: 10.3847/1538-4357/abf63dDOI
  71. [77]Nigro, C., Sitarek, J., Gliwny, P., et al. 2022, A&A, 660, A18, doi: 10.1051/0004-6361/202142000DOI
  72. [78]Paiano, S., Falomo, R., Treves, A., Scarpa, R., & Sbarufatti, B. 2024, ApJ, 968, 81, doi: 10.3847/1538-4357/ad4a56DOI
  73. [79]Pandey, A., Hu, C., Wang, J.-M., et al. 2025, ApJ, 978, 120, doi: 10.3847/1538-4357/ad9b7cDOI
  74. [80]Peña-Herazo, H. A., Massaro, F., Gu, M., et al. 2021, AJ, 161, 196, doi: 10.3847/1538-3881/abe41d
  75. [81]Peer, A., Shah, Z., Akbar, S., Maqbool, B., & Misra, R. 2026, arXiv e-prints, arXiv:2607.20375, doi: 10.48550/arXiv.2607.20375
  76. [82]Planck Collaboration, Akrami, Y., Argüeso, F., et al. 2018, A&A, 619, A94, doi: 10.1051/0004-6361/201832888
  77. [83]Poole, T. S., Breeveld, A. A., Page, M. J., et al. 2008, MNRAS, 383, 627, doi: 10.1111/j.1365-2966.2007.12563.xDOI
  78. [84]Prince, R., Khatoon, R., & Stalin, C. S. 2021, MNRAS, 502, 5245, doi: 10.1093/mnras/stab369DOI
  79. [85]Psaradaki, I., Costantini, E., Mehdipour, M., et al. 2020, A&A, 642, A208, doi: 10.1051/0004-6361/202038749DOI
  80. [86]Ren, S. S., Zhou, R. X., Zheng, Y. G., & Kang, S. J. 2024a, ApJ, 976, 124, doi: 10.3847/1538-4357/ad83ceDOI
  81. [87]Ren, S. S., Zhou, R. X., Zheng, Y. G., Kang, S. J., & Wu, Q. 2024b, A&A, 685, A140, doi: 10.1051/0004-6361/202347312DOI
  82. [88]Ricci, C., & Trakhtenbrot, B. 2023, Nature Astronomy, 7, 1282, doi: 10.1038/s41550-023-02108-4DOI
  83. [89]Roming, P. W. A., Kennedy, T. E., Mason, K. O., et al. 2005, SSRv, 120, 95, doi: 10.1007/s11214-005-5095-4DOI
  84. [90]Ruan, J. J., Anderson, S. F., Plotkin, R. M., et al. 2014, ApJ, 797, 19, doi: 10.1088/0004-637X/797/1/19DOI
  85. [91]Sahakyan, N. 2021, MNRAS, 504, 5074, doi: 10.1093/mnras/stab1135DOI
  86. [92]Sahakyan, N., Vardanyan, V., Giommi, P., et al. 2024, AJ, 168, 289, doi: 10.3847/1538-3881/ad8231DOI
  87. [93]Scargle, J. D., Norris, J. P., Jackson, B., & Chiang, J. 2013, ApJ, 764, 167, doi: 10.1088/0004-637X/764/2/167DOI
  88. [94]Schlafly, E. F., & Finkbeiner, D. P. 2011, ApJ, 737, 103, doi: 10.1088/0004-637X/737/2/103DOI
  89. [95]Shappee, B. J., Prieto, J. L., Grupe, D., et al. 2014, ApJ, 788, 48, doi: 10.1088/0004-637X/788/1/48DOI
  90. [96]Sikora, M., Begelman, M. C., & Rees, M. J. 1994, ApJ, 421, 153, doi: 10.1086/173633DOI
  91. [97]Stocke, J. T., Morris, S. L., Gioia, I. M., et al. 1991, ApJS, 76, 813, doi: 10.1086/191582DOI
  92. [98]Sugiura, N. 1978, Communications in Statistics - Theory and Methods, 7, 13, doi: 10.1080/03610927808827599
  93. [99]Takahashi, T., Tashiro, M., Madejski, G., et al. 1996, ApJL, 470, L89, doi: 10.1086/310302DOI
  94. [100]Tramacere, A., Giommi, P., Perri, M., Verrecchia, F., & Tosti, G. 2009, A&A, 501, 879, doi: 10.1051/0004-6361/200810865DOI
  95. [101]Urry, C. M. 1998, Advances in Space Research, 21, 89, doi: 10.1016/S0273-1177(97)00619-4DOI
  96. [102]Urry, C. M., & Padovani, P. 1995, PASP, 107, 803, doi: 10.1086/133630DOI
  97. [103]Vermeulen, R. C., Ogle, P. M., Tran, H. D., et al. 1995, ApJL, 452, L5, doi: 10.1086/309716DOI
  98. [104]Williamson, K. E., Jorstad, S. G., Marscher, A. P., et al. 2014, ApJ, 789, 135, doi: 10.1088/0004-637X/789/2/135DOI
  99. [105]Wood, M., Caputo, R., Charles, E., et al. 2017, 301, 824, doi: 10.22323/1.301.0824DOI
  100. [106]Xiao, H., Fan, J., Ouyang, Z., et al. 2022, ApJ, 936, 146, doi: 10.3847/1538-4357/ac887fDOI
  101. [107]Zacharias, M., & Wierzcholska, A. 2026, Journal of High Energy Astrophysics, 54, 100689, doi: 10.1016/j.jheap.2026.100689
  102. [108]Zanin, R. 2013, in 33rd International Cosmic Ray Conference, 0773

Paper details

vlnc:W_53xnkn88
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents