Comparative Photometry of the Quiet Quasar PDS 456 and the Radio-Loud Blazar 3C 273

Abstract

A comparative analysis of the photometric variability of the blazar 3C 273 and the quasar PDS 456 using multi-band data from ground- and space-based platforms (2015–2025) reveals contrasting behaviors. For 3C 273, a statistically significant secular dimming was detected in the ATLASc-band light curve $\left((5.6\pm 0.2)\times {10}^{-4}\mathrm{mag}{\mathrm{day}}^{-1}\right)$ and confirmed by Johnson–Cousins V-band photometry. Ten optical flares were identified, two coinciding with Fermi gamma-ray enhancements, suggesting a synchrotron origin linked to jet activity. A significant bluer-when-brighter trend ($\rho =-0.54$) was found relative to the o-band, and several color extrema align with gamma-ray activity, reinforcing the nonthermal interpretation. In contrast, PDS 456 exhibits a statistically significant secular brightening in the o-band $\left((-3.1\pm 0.2)\times {10}^{-5}\mathrm{mag}{\mathrm{day}}^{-1}\right)$ and 75 optical flares, four coinciding with UV flares observed by Swift/UVOT. The $c–o$ color index displays a non-Gaussian distribution with asymmetric reddening and blueing episodes. An extreme reddening event aligns with a strong UV flare, suggesting transient inner-disk heating. These results indicate jet-dominated variability in 3C 273 and disk-driven variability in PDS 456, highlighting distinct physical mechanisms in radio-loud versus radio-quiet active galactic nuclei.

1. Introduction

Quasars and blazars are among the most extreme manifestations of black hole accretion and relativistic jet activity in the universe [1]. As active galactic nuclei (AGNs), they are powered by the accretion of matter onto supermassive black holes at galactic centers [2,3]. Quasars are characterized by broad emission lines and intense multiwavelength radiation, often outshining their host galaxies [4,5]. Blazars are radio-loud AGN—comprising flat-spectrum radio quasars (FSRQs) and BL Lac objects—with relativistic jets oriented close to our line of sight, so their emission is strongly jet-dominated and Doppler-boosted [1,6]. In contrast, radio-quiet quasars do not exhibit prominent relativistic jets, and their emission is primarily powered by the accretion disk and associated winds/outflows [7]. This radio-loud/quiet distinction is fundamental for understanding differences in variability and emission mechanisms.

Blazars, a subclass of radio-loud active galactic nuclei (AGNs) that includes BL Lacertae objects and flat-spectrum radio quasars (FSRQs), are characterized by relativistic jets closely aligned with the line of sight. This geometry leads to strong Doppler boosting, rapid variability, apparent superluminal motion, and high radio and optical polarization [8,9]. While both BL Lacs and FSRQs share this jet orientation, they differ in their spectral and energetic properties: FSRQs display strong, broad optical emission lines and higher luminosities, whereas BL Lacs exhibit weak or absent emission lines (rest-frame EW $\lesssim 5$ Å) and are typically less powerful [1]. These features make blazars valuable probes of jet physics and relativistic beaming and, more broadly, of the impact of radio jets on their environments [10,11]. Within the radio-loud population, blazars represent the aligned end of this orientation sequence [1].

PDS 456 and 3C 273 provide complementary perspectives on AGN behavior. PDS 456, a luminous, relatively nearby radio-quiet quasar [12,13], exhibits ultra-fast outflows (UFOs) with velocities up to ∼0.3c [14,15], consistent with radiatively driven feedback near the Eddington limit [16]. In contrast, 3C 273, a prototypical radio-loud blazar, hosts a powerful relativistic jet oriented close to our line of sight [10,17], producing strong nonthermal emission and pronounced variability [18]. Despite their different jet orientations and structures, both sources are highly luminous and accrete rapidly; disk winds are prominent in PDS 456, while 3C 273 shows ionized absorbers in some epochs [11]. PDS 456 is often considered a radio-quiet analog to 3C 273 (e.g., [19]), making it a useful benchmark for isolating radiative feedback effects without jet contamination. This contrast between radio-loud (often jet-dominated) and radio-quiet (predominantly disk-dominated) quasars is particularly relevant for investigating differences in long-term optical variability, since jets tend to produce larger and sometimes chromatic changes, whereas disk fluctuations are typically lower-amplitude and nearly achromatic.

Although both objects have been studied individually, comparative analyses of their optical variability remain limited. Monitoring their brightness and color evolution can provide insights into the physical origins of variability, separating disk- and wind-driven processes in PDS 456 from jet-dominated variability in 3C 273.

To explore these differences, we analyzed dual-color optical broadband photometric data from the Asteroid Terrestrial-impact Last Alert System (ATLAS) survey (2015–2025) [20,21] for both AGNs, along with BVR observations of 3C 273 acquired in 2024 using a robotic telescope network. Our goal is to characterize the temporal behavior of their magnitudes and color indices (c–o, B–V, and V–R), and to examine how variability correlates with accretion and outflow processes. By comparing a jet-dominated and a wind-dominated AGN through time-domain photometry, we aim to clarify the role of different feedback mechanisms in shaping AGN emission.

2. Observational Data and Photometric Calibration

This section outlines the datasets and reduction procedures used to derive photometric parameters of the active galactic nuclei analyzed in this study. The methodology is organized into two parts, corresponding to observations from the Robotic Observatory and from the ATLAS survey.

Because our analysis targets relative photometric variability (flare detection and long-term trends), we did not apply Galactic-extinction corrections. On the timescales considered, the line-of-sight extinction is effectively constant and therefore introduces only a fixed, band-dependent offset that does not affect variability metrics. We did not attempt to recover absolute fluxes or to compare intrinsic luminosities between the two sources.

Similarly, we did not decompose the host-galaxy light. A static host contributes a constant flux pedestal that, in magnitudes, yields a nearly time-invariant dilution of variability amplitudes rather than a true offset, and is therefore irrelevant to our relative analysis. For PDS 456, the quasar outshines its host by several magnitudes, making the host contribution within the ATLAS photometric aperture negligible for variability metrics. The same reasoning applies to 3C 273, whose nuclear emission dominates the optical flux.

2.1. Robotic Observatory Data

To investigate BVR color variability in the blazar 3C 273, we conducted a monitoring campaign using two robotic telescopes located at the Teide Observatory (Santa Cruz de Tenerife, Spain), namely a PlaneWave CDK20 (0.51 m, f/6.8, T1) and a CDK17 (0.43 m, f/6.8, T2), equipped with FLI PL09000 and PL1603 CCD cameras, respectively.

Observations were scheduled near meridian transit to minimize atmospheric extinction. In nearly all sessions, the zenith distance of the target remained ≤30°, corresponding to an airmass of ≤1.2.

We obtained 15 photometric epochs between 29 December 2023 and 7 February 2024 (UT), with a median sampling interval of ≈1.5 days. We report the median sampling interval as a robust measure of the typical cadence: it is less sensitive to large observational gaps that would bias the mean, suits non-uniform and clustered sampling, and does not assume normality.

All raw RGB images were pre-processed using bias, dark, and dome flat-field corrections applied by the observatory’s internal pipeline. No significant CCD artifacts, such as bad pixels, hot columns, or cosmic ray hits, were detected. These defects can introduce spurious photometric variability, including false flare events, as previously noted for GB 1508 + 5714 [22].

Photometric Calibration and Data Selection

Photometric calibration was carried out using aperture photometry with Python routines from the photutils package of the Astropy project [23]. Aperture radii were matched to the full width at half maximum (FWHM) of stellar profiles to optimize the signal-to-noise ratio.

Instrumental magnitudes were calibrated via differential photometry using 10–15 field stars. The Johnson–Kron–Cousins V-band magnitudes and B–V and V–R color indices of the comparison stars were estimated from 2MASS J–K colors through polynomial transformations based on Landolt standards [24], following the method proposed by Warner [25]. Although standard-star sequences exist for the 3C 273 field [26,27], we adopted a broader set of 10–15 comparison stars within the CCD field of view—including stars from those sequences—to improve photometric precision when the quasar was faint and to avoid saturation of the brighter standards.

To account for differences among telescopes, filters, and CCDs, we derived instrument-specific transformation coefficients ($T_V$, $T_{BV}$, $T_{VR}$) linking instrumental to standard magnitudes and colors via linear least-squares fits, following Benson [28]. The calibration used stars in the M67 and NGC 7790 standard fields.1 These coefficients (together with nightly zero points) were applied consistently to all measurements from each system to ensure cross-instrument homogeneity. Systematic effects due to telescope changes are expected to be minimal, as T1 contributed only two photometric epochs in the dataset.

The resulting median values of V magnitude, B–V, and V–R for 3C 273 are listed in

Table 1. Photometric measurements in the Johnson–Kron–Cousins system obtained using the robotic telescopes described in Section 2.1, filtered subset of the original 15 sessions, with each row representing a distinct photometric JD epoch.

JD	B–V	V–R	V
2,460,316.636	—	$0.21\pm 0.05$	$13.39\pm 0.04$
2,460,318.646	$0.14\pm 0.05$	$0.13\pm 0.02$	$13.61\pm 0.07$
2,460,319.577	$0.1\pm 0.2$	$0.12\pm 0.07$	$13.56\pm 0.09$
2,460,323.817	$0.5\pm 0.1$	$0.18\pm 0.03$	$13.56\pm 0.08$
2,460,329.622	$0.11\pm 0.08$	$0.11\pm 0.04$	$13.6\pm 0.1$
2,460,331.564	$-0.6\pm 0.3$	$0.17\pm 0.03$	$13.51\pm 0.07$
2,460,332.595	—	$-0.10\pm 0.05$	$13.15\pm 0.07$
2,460,342.528	$0.5\pm 0.2$	$0.19\pm 0.04$	$13.3\pm 0.1$
2,460,342.671	$-0.1\pm 0.2$	$0.19\pm 0.07$	$13.42\pm 0.07$
2,460,347.553	$0.4\pm 0.2$	$0.10\pm 0.05$	$13.46\pm 0.04$

, along with their respective median absolute deviations (MADs). These statistics provide robust estimates of central tendency and dispersion and are preferred in this analysis due to their resistance to outliers.

Color and magnitude measurements of 3C 273 were selected based on statistical consistency. Individual values of B–V, V–R, and V were evaluated by their normalized deviation from the sample median using the MAD. Data points were excluded if their uncertainties exceeded three times the MAD of the associated errors, if their color indices deviated by more than $3\times \mathrm{MAD}$ from the median, or if their combined deviation (a quadratically weighted z-score normalized by the individual uncertainties) exceeded 3 in color space. This filtering process is objective, resilient to outliers, and suitable for small or asymmetric samples. After applying these criteria, 10 photometric epochs (out of 15) remained with usable data, and the resulting date range is shorter. This procedure does not suppress the intrinsic variability of 3C 273, as it targets only internally inconsistent or highly uncertain measurements.

2.2. ATLAS Data

The active galactic nuclei 3C 273 and PDS 456 were monitored by the Asteroid Terrestrial-impact Last Alert System (ATLAS) survey. This global network operates four identical 0.5-m telescopes distributed across both hemispheres: Haleakalā and Mauna Loa in Hawaii (USA) in the Northern Hemisphere, and El Sauce (Chile) and Sutherland (South Africa) in the Southern Hemisphere. Although primarily designed to detect potentially hazardous near-Earth objects, ATLAS performs all-sky coverage and routinely reaches limiting magnitudes of approximately 19.7 in the orange (o: 560–820 nm) and cyan (c: 420–650 nm) bands [20]. Observations consist of 30-s exposures, typically taken in sequences of four images spaced over intervals of about one hour [21].

Photometric measurements were obtained using the ATLAS Forced Photometry tool, which performs point spread function (PSF) photometry at fixed celestial coordinates. Only data with uncertainties ≤0.01 mag in either filter were retained. This selection yielded 2046 high-quality measurements for 3C 273 and 3817 for PDS 456, considering both o and c filters. As observed in other ATLAS datasets, approximately 30% of the measurements for each object were obtained in the c band.

The temporal coverage was extensive. 3C 273 was monitored from 28 June 2015 to 12 March 2025, and PDS 456 from 19 July 2015 to 17 March 2025. The median sampling interval between successive observations in either filter was approximately 0.2 h.

The color indices $c–o$ were computed from pairs of consecutive measurements in the c- and o-bands. In each case, the time span between the acquisition of the c- and o-band magnitudes was equal to or less than one day. The corresponding time of occurrence of each color index measurement was defined as the mean Modified Julian Date (MJD) between the two observations.

3. Results

Given the distinct emission mechanisms and datasets, we used different non-optical bands. For 3C 273, a gamma-ray–loud blazar with frequent Fermi-LAT detections [29,30], we examined optical–gamma-ray correlations to probe jet-linked synchrotron and inverse-Compton processes [10,31]. By contrast, PDS 456 is not listed in the Fermi-LAT 4FGL catalog [32] but has been extensively monitored in the ultraviolet with Swift/UVOT [33], enabling UV–optical comparisons that trace thermal disk emission. Although UV data exist for 3C 273, we do not include them here because our aim is to probe jet-linked variability via optical–gamma coupling, whereas the UV band is expected to be disk-dominated.

3.1. 3C 273: Long-Term Optical and Chromatic Variability

3.1.1. Secular Evolution of the c-Band Magnitude

We investigated long-term optical variability in 3C 273 using ATLAS c-band photometry, whose broad passband (∼420–650 nm) samples the blue–green optical continuum and receives contributions from both the thermal accretion disk and the nonthermal synchrotron jet [20]. During flaring states, the jet component can become dominant, consistent with previous studies (e.g., [34,35,36]) and with the bluer-when-brighter trend observed in our data (Section 3.1.4).

A total of 551 individual magnitude measurements were compiled. A weighted least-squares fit of magnitude versus Modified Julian Date (MJD), with weights $w_i=1/{\sigma }_{m,i}^2$, reveals a statistically significant fading trend, with a best-fit slope of $(5.6\pm 0.2)\times {10}^{-4}\mathrm{mag}{\mathrm{day}}^{-1}$ $(\approx$$0.205\pm 0.007\mathrm{mag}{\mathrm{yr}}^{-1})$ (Figure 1). This secular dimming is further supported by a Spearman rank correlation coefficient of $\rho =0.678$ with $p\sim 0$, confirming a strong, highly significant monotonic increase in magnitude over time (i.e., the source is fading). The observed fading is consistent with the long-term optical decline reported by Terrell and Olsen [37], who found a ∼21% luminosity decrease over 80 years based on power-spectral analysis. Taken together, these results support the interpretation that 3C 273’s variability arises from overlapping stochastic flares superimposed on a long-term secular trend. Although a secular optical trend is often attributed to changes in the accretion rate or structural evolution of the disk, alternative geometric scenarios such as jet precession can also modulate long-term optical variability; for example, in OJ 287 a precessing-jet model reproduces its long-term behavior [38]. A comprehensive interpretation therefore requires considering both thermal and geometric contributions.

The observed long-term fading is consistent with optical emission dominated by the thermal accretion disk. Previous multiwavelength studies report weak or absent optical–radio correlations on secular timescales, low optical polarization, and small nonthermal-dominance (NTD) indices, all consistent with a thermal origin of the optical output. The gradual dimming may therefore reflect a secular decline in the accretion rate or structural evolution of the inner disk, as proposed by Fernandes et al. [39] based on radio–optical anticorrelation and spectral diagnostics from 2005–2015.

3.1.2. Long-Term BVR Color Index and Variability Patterns in 3C 273

To complement the secular analysis of the ATLAS c-band light curve, we investigated the long-term variability and color evolution of the blazar 3C 273 using B, V, and R band observations (Table 1).

We derived the median color indices and magnitude from our dataset: $B–V=0.10$, $V–R=0.15$, and $V=13.49$, with corresponding median absolute deviations (MADs) of 0.30, 0.04, and 0.09, respectively. These values indicate a chromatic state consistent with moderately blue optical emission. Although the sample size is limited to ten BVR measurements, the results are consistent with the secular dimming trend observed in the ATLAS c-band data and with previous long-term optical studies of 3C 273 [40,41]. The larger dispersion in $B–V$ relative to $V–R$ suggests only modest chromatic variability during our campaign, while the overall color stability supports a nearly invariant optical spectral shape on seasonal timescales.

In agreement with the ATLAS c-band light curve, our BVR photometry confirms a secular fading trend in the optical brightness of 3C 273. Dai et al. [42], based on multiband observations from 2003 to 2005, reported a mean magnitude $V=12.698$, whereas our 2024 dataset yields a median $V=13.49$. This difference of approximately 0.79 magnitudes over two decades indicates a long-term decline in flux.

Although the fading appears consistent across independent datasets, we caution that the long-term trend may not be strictly linear. The difference between our median magnitude and that of Dai et al. corresponds to a mean fading rate of approximately $1.1\times {10}^{-4}\mathrm{mag}{\mathrm{day}}^{-1}$, which is lower than the ATLAS-derived slope. This discrepancy suggests that the fading could be episodic or subject to variation over time, rather than continuous. Nonetheless, the overall decline in optical brightness is robust and well supported by both contemporary and archival observations.

Our results support the long-term photometric monitoring of 3C 273 by Fan et al. [41], who obtained 1118 VRI measurements between 2000 and 2008. They reported a magnitude range in the V band from $12.204\pm 0.062$ to $13.567\pm 0.062$, which encompasses both the average brightness reported by Dai et al. [42] and the dimmer state observed in our 2024 dataset ($V=13.49$). This continuity in magnitude evolution over two decades reinforces the evidence for a secular fading trend. The average color index $\langle V–R\rangle =0.18\pm 0.10$ reported by Fan et al. is also consistent with our median value of 0.15, but our dataset shows significantly lower dispersion (MAD = 0.04), suggesting chromatic stability during the observational window. Moreover, while Fan et al. detected intra-day variability on timescales from 13 to 245 min, the absence of significant short-term variability in our data supports the interpretation that the observed fading is a long-term phenomenon rather than the cumulative effect of rapid flares. Together with the results of Zeng et al. [43], who reported minimal variability and no systematic reddening across BVRI bands, these findings suggest that the secular evolution of 3C 273 involves a gradual decrease in total flux without significant alteration of the optical spectral slope.

This long-term chromatic evolution, inferred from multi-decade observations, supports the interpretation that the secular fading reflects not only a decline in total flux from the accretion disk, but also episodic shifts in the dominant emission component toward a higher-frequency synchrotron component, meaning a relative shift of the synchrotron spectral energy distribution (SED) peak toward shorter optical wavelengths, consistent with a bluer-when-brighter trend. Such behavior may be driven by internal jet restructuring or changes in particle-acceleration efficiency; in the presence of spectral curvature or breaks, geometric effects such as modest viewing-angle variations can produce comparable signatures [44].

In conclusion, the combined evidence from ATLAS, our BVR photometry, and archival studies supports a model in which 3C 273’s long-term optical variability arises from the interplay between a gradual fading of the synchrotron emission from the jet and a stable thermal continuum from the accretion disk, resulting in an optical spectral energy distribution whose relative component contributions evolve over time, even though no statistically significant intrinsic changes are detected in the thermal part.

The apparent constancy of the $B–V$ and $V–R$ color indices in our BVR dataset does not contradict the statistically significant bluer-when-brighter trend reported in Section 3.1.4 from the ATLAS data. The BVR observations comprise only 10 epochs over a 40-day interval and show low dispersion, consistent with a short-term quiescent phase. By contrast, the ATLAS dataset spans nearly a decade at high cadence and reveals long-term chromatic evolution. The two datasets are therefore complementary in temporal baseline and cadence.

3.1.3. Detection of Optical Flares in the c-Band

In addition to the secular fading observed in the ATLAS c-band light curve of 3C 273, we searched for transient brightenings indicative of flaring activity. A LOWESS (Locally Weighted Scatterplot Smoothing) algorithm was applied to the 550-point time series with photometric uncertainties ≤0.01 mag. Residuals were computed relative to the smoothed curve, and outliers were identified as epochs with $|r_i|>3\sigma$ ($\sigma \approx 0.210$ mag). We then classified flares as those points with $r_i<-3\sigma$, i.e., brightenings that lie ≳0.63 mag below the local trend in magnitude space.

We flag outliers (flares and fadings) as points whose residuals relative to a non-parametric LOWESS trend satisfy $\left|r\right|\ge 3\sigma$, with $\sigma$ estimated from the residual scatter. This combination of detrending and $3\sigma$ clipping is standard in astronomical time-series—explicitly used to identify or excise events in large light-curve samples (e.g., [45,46]) and in optical blazar analyses (e.g., [47])—while LOWESS provides a flexible baseline without imposing a parametric model [48]. Under near-Gaussian residuals, a $3\sigma$ threshold yields a low per-point false-positive rate, ensuring that events selected from our 550-point dataset are statistically significant.

This criterion revealed ten distinct flares, listed in

Table 2. Detected optical flares in the ATLAS c-band light curve of 3C 273. Residuals represent the difference between observed magnitudes and the LOWESS-smoothed trend.

MJD	c-Band Magnitude	Residual (mag)
57,896.372	$13.995\pm 0.004$	$-0.667$
58,846.555	$14.191\pm 0.004$	$-0.998$
58,846.557	$14.368\pm 0.004$	$-0.821$
58,846.562	$14.372\pm 0.004$	$-0.817$
58,846.573	$14.477\pm 0.004$	$-0.713$
59,224.594	$14.523\pm 0.004$	$-0.659$
60,145.704	$14.429\pm 0.004$	$-0.795$
60,145.707	$14.570\pm 0.004$	$-0.654$
60,145.713	$14.565\pm 0.004$	$-0.659$
60,444.948	$14.421\pm 0.004$	$-0.764$

and highlighted in red in Figure 2. These events appear as sharp deviations from the global trend and are consistent with episodes of enhanced synchrotron emission in the relativistic jet. Their stochastic distribution over time suggests that they arise from localized and transient processes, such as internal shocks resulting from collisions between plasma shells of different velocities [49] or magnetic reconnection, where twisted magnetic fields rearrange and explosively release energy [50]. Both mechanisms naturally account for the rarity, intensity, and brevity of the detected flares. Less probable causes include geometric variations or turbulent multizone activity, which would produce smoother or more frequent modulations not observed in this dataset.

To investigate whether these optical flares had high-energy counterparts, we examined the public gamma-ray light curves of 3C 273 from the Fermi Science Support Center (FSSC).2 The dataset spans MJD∼54,000 to ∼62,000 and reports integrated gamma-ray flux in the 100 MeV–300 GeV energy range, fully encompassing all ten flaring epochs. One optical flare cluster—centered around MJD 58,846.56 (late 2019)—coincides with an interval of nonzero gamma-ray activity. A second flare, at MJD 60,444.95 (May 2024), is isolated but temporally aligned with a modest gamma-ray enhancement. Although these gamma-ray features are not as intense as the early-mission outbursts, they exceed the quiescent baseline, particularly during the May 2024 event. This is consistent with a common origin for both emissions in relativistic jet activity. The simultaneous detection of optical and gamma-ray variability favors leptonic emission models [51], where synchrotron radiation in the optical regime and inverse Compton scattering in the gamma-ray regime arise from the same population of relativistic electrons. The relatively weak gamma-ray counterparts suggest that the flaring regions may be synchrotron-dominated, or that they are located in zones of the jet where the density of external seed photons from the accretion disk and dusty torus is locally reduced, thereby limiting the efficiency of Compton upscattering despite the strong optical–UV and infrared bumps seen in the broadband SED. These findings disfavor purely thermal origins for the flares and reinforce the interpretation of the flares as signatures of efficient, localized particle acceleration within the jet of 3C 273.

3.1.4. Chromatic Evolution of the Blazar 3C 273 Based on ATLAS Data

We analyzed the temporal evolution of the ATLAS color index $c–o$ using a dataset of 29 epochs with associated photometric uncertainties (

Table 3. Color index and magnitude measurements for 3C 273.

Mean MJD	$\mathit{c}–\mathit{o}$	c	o
57,506.96778	$0.40\pm 0.01$	$13.839\pm 0.004$	$13.436\pm 0.004$
57,808.61823	$0.46\pm 0.01$	$14.015\pm 0.004$	$13.554\pm 0.004$
57,808.64202	$0.40\pm 0.01$	$13.954\pm 0.005$	$13.553\pm 0.005$
57,812.60939	$0.29\pm 0.01$	$13.786\pm 0.003$	$13.499\pm 0.005$
57,826.57846	$-0.25\pm 0.01$	$13.858\pm 0.004$	$14.108\pm 0.006$
57,841.02690	$0.40\pm 0.01$	$13.940\pm 0.004$	$13.542\pm 0.005$
57,857.00976	$-0.03\pm 0.01$	$13.931\pm 0.004$	$13.963\pm 0.005$
57,882.92872	$-0.07\pm 0.01$	$13.995\pm 0.004$	$14.067\pm 0.006$
57,898.86425	$1.48\pm 0.01$	$14.667\pm 0.005$	$13.191\pm 0.002$
58,252.45468	$0.33\pm 0.01$	$14.762\pm 0.005$	$14.437\pm 0.005$
58,971.96334	$0.01\pm 0.02$	$14.751\pm 0.006$	$14.741\pm 0.009$
59,608.87673	$0.91\pm 0.01$	$15.157\pm 0.007$	$14.242\pm 0.005$
59,642.90558	$1.53\pm 0.02$	$16.037\pm 0.016$	$14.506\pm 0.005$
59,712.60368	$0.45\pm 0.02$	$15.531\pm 0.009$	$15.085\pm 0.007$
59,964.87374	$0.55\pm 0.02$	$15.812\pm 0.009$	$15.266\pm 0.009$
59,968.79979	$0.00\pm 0.02$	$15.755\pm 0.010$	$15.756\pm 0.011$
60,020.69771	$0.17\pm 0.03$	$15.597\pm 0.009$	$15.430\pm 0.019$
60,350.73672	$-0.34\pm 0.02$	$15.591\pm 0.009$	$15.931\pm 0.013$
60,432.62362	$-1.13\pm 0.06$	$14.751\pm 0.050$	$15.879\pm 0.012$
60,436.61584	$-0.63\pm 0.02$	$15.248\pm 0.008$	$15.874\pm 0.012$
60,458.70910	$0.87\pm 0.04$	$15.966\pm 0.017$	$15.098\pm 0.027$
60,462.69647	$0.53\pm 0.03$	$15.932\pm 0.026$	$15.407\pm 0.008$
60,488.58022	$-0.30\pm 0.02$	$14.579\pm 0.011$	$14.875\pm 0.007$
60,496.50709	$0.22\pm 0.02$	$15.279\pm 0.008$	$15.062\pm 0.008$
60,684.83287	$-0.10\pm 0.02$	$15.594\pm 0.010$	$15.695\pm 0.014$
60,702.77204	$0.24\pm 0.03$	$15.594\pm 0.018$	$15.352\pm 0.010$
60,706.79951	$0.17\pm 0.02$	$15.639\pm 0.009$	$15.468\pm 0.010$
60,712.52941	$-0.51\pm 0.03$	$15.554\pm 0.011$	$16.062\pm 0.014$
60,732.72881	$-0.21\pm 0.02$	$15.826\pm 0.011$	$16.032\pm 0.013$

). This analysis was motivated by the hypothesis that geometric effects such as jet precession could induce periodic modulations in the observed color index, reflecting changes in the spectral energy distribution of the relativistic jet. Our goal was therefore to test for periodic signals and secular trends in $c–o$.

The Lomb–Scargle periodogram applied to the unevenly sampled dataset suggests a best-fit period of $P\approx 12.7$ days. However, given the small number of observations (29 epochs) and the long temporal baseline of 3225 days, this signal lacks statistical robustness and is likely spurious. We report it here for completeness, but caution that no physical interpretation should be inferred without further confirmation from densely sampled monitoring. No previously reported periodicities for 3C 273 match this timescale: prior studies have identified much longer quasi-periodic signals, including a period of ∼8.8 years in the radio band [52,53], and periodicities ranging from ∼0.64 to 13 years in the optical domain [54,55]. These comparisons support the conclusion that the 12.7-day signal does not correspond to any known or persistent variability pattern in this blazar. To explore nonlinear trends, we applied LOWESS smoothing, which uncovered long-term chromatic modulations not captured by the linear model. The smoothed curve exhibits alternating phases of reddening and blueing, potentially linked to changes in jet dominance and thermal contributions, and—when spectral curvature or breaks are present—to geometric effects such as variations of the viewing angle [56,57,58] (Figure 3). A Shapiro–Wilk test indicates that the distribution of $c–o$ values is compatible with a normal distribution ($W=0.966$, $p=0.454$). The fitted Gaussian has mean $\mu =0.20$ and standard deviation $\sigma =0.56$ (Figure 4). We define reddening as epochs with $c–o>\mu +0.5\sigma \approx 0.48$, and blueing as $c–o<\mu -0.5\sigma \approx -0.08$. To ensure consistency with the classification adopted for PDS 456, we also define a neutral phase for intermediate values within the interval $-0.08\le c–o\le 0.48$. These criteria were used to identify chromatic episodes throughout the monitoring period. However, given the sparse temporal sampling (29 measurements over 8.8 years), the phase structure inferred from smoothing should be interpreted with caution.

To quantify the relationship between chromatic behavior and brightness, we computed Spearman rank correlations. A statistically significant anti-correlation was found between $c–o$ and the o-band magnitude ($\rho =-0.54$, $p=2.3\times {10}^{-3}$) (Figure 5), supporting the presence of a bluer-when-brighter (BWB) trend, as previously reported for 3C 273 by Dai et al. [42] and commonly observed in many blazars [59]. No significant correlation was found between $c–o$ and the c-band magnitude ($\rho =0.15$, $p=0.44$), mitigating concerns about artificial correlations introduced by the definition of the color index. Despite the limited cadence, the BWB correlation remains statistically robust, reinforcing the interpretation that synchrotron processes dominate the observed variability in this blazar.

Among the ten identified optical flares, the events at MJD 58,846.5 and MJD 60,145.7 are particularly notable for exhibiting strong blueing ($c–o<\mu -0.5\sigma$) and large negative residuals in the c-band light curve (i.e., brightenings relative to the LOWESS trend). Although both events share similar photometric characteristics, they are separated by nearly 1300 days (≃3.56 years), far exceeding the duration of any plausible synchrotron flare; we therefore interpret them as independent outbursts, consistent with localized particle-acceleration episodes in the relativistic jet. Similar chromatic and temporal features have been reported for synchrotron flares in radio-loud AGNs (e.g., [35,60]).

While the BWB trend is evident on short timescales, its persistence over longer periods remains uncertain. This behavior is consistent with the findings of [36], who reported that the BWB trend in 3C 273 is prominent on intraday and intermediate timescales, but absent when considering the full 12-year monitoring dataset. To investigate potential long-term chromatic evolution, we examined the secular behavior of the color index as a function of time.

Additionally, sinusoidal fitting yielded poor agreement with the data, reinforcing the absence of statistically significant periodic behavior. A linear regression revealed a modest secular trend (slope $\approx -1.32\times {10}^{-4}{\mathrm{day}}^{-1}$, $p\approx 0.136$), weakly pointing toward a gradual shift to bluer colors over time. However, this trend is not statistically significant and must be interpreted with caution given the small sample size and low temporal resolution.

No periodicities were found to be statistically robust. Although a weak secular trend toward bluer colors was suggested by linear fitting, it lacked statistical significance and should be interpreted with caution. Nevertheless, the observed chromatic variability on shorter timescales may reflect intrinsic changes in jet physics, potentially modulated by Doppler boosting, geometric effects, or multi-zone synchrotron emission scenarios.

The stronger correlation of the c–o color index with the o-band magnitude—and its lack of correlation with the c-band magnitude—may initially seem unexpected, since blazars typically show tightly coupled variability across optical bands. However, this behavior can be explained by a combination of two observational effects.

First, the ATLAS o-band (560–820 nm) spans a broader and redder wavelength range than the c-band (420–650 nm), and thus serves as a more stable tracer of the overall optical flux. In contrast, the c-band is more sensitive to high-frequency fluctuations or short-lived brightenings, which may introduce excess noise into the color index when paired asynchronously.

Second, the c- and o-band measurements used to compute the c–o index are not strictly simultaneous: they are acquired within 24 h of each other. Any short-term variations—such as intra-day flares—can therefore decorrelate the c-band from the color index, even while the o-band, being smoother, retains a meaningful chromatic relationship.

This interpretation is consistent with prior work showing wavelength- and cadence-dependent color–magnitude behavior in 3C 273.

These findings may at first appear contradictory: while the ATLAS c–o color index reveals a statistically significant bluer-when-brighter (BWB) trend, our BVR photometry of 3C 273 shows no significant chromatic variability. However, this apparent tension is resolved by recognizing that the datasets probe distinct temporal regimes and physical mechanisms. The BVR observations span a brief, quiescent interval in early 2024 and are consistent with thermal emission from the accretion disk, whereas the ATLAS monitoring covers a decade and captures multiple optical flares dominated by synchrotron radiation from the relativistic jet. The BWB behavior detected in the ATLAS data is episodic and linked to these flaring events, rather than a global trend. Additionally, the weak secular bluing observed over the full ATLAS baseline is statistically insignificant and occurs on much longer timescales, distinct from the short-term BWB chromaticity. Together, these results suggest that the long-term fading of 3C 273 reflects structural evolution in the jet or accretion rate, while short-term chromatic excursions arise from localized, transient events within the jet.

3.1.5. Temporal Correlation Between Optical Color Index and Gamma-Ray Flares

The variability of the color index c–o, documented in Table 3, shows associations with gamma-ray activity. The most extreme values of the color index c–o occur at MJD 57,898.864 ($1.48\pm 0.01$), MJD 59,642.905 ($1.53\pm 0.02$), MJD 60,432.623 ($-1.13\pm 0.06$), and MJD 60,436.615 ($-0.63\pm 0.02$). The first of these aligns with the same gamma-ray flare near MJD 57,900, reinforcing the notion of contemporaneous optical and gamma-ray activity. The last two events coincide with a broader interval of elevated flux between MJD 60,400 and 60,500, suggesting a possible link between chromatic transitions and energetic episodes.

Although not all optical flares or color extrema correspond to gamma-ray peaks, and vice versa, the identified overlaps support a scenario in which a fraction of the observed optical activity is connected to changes in the high-energy emission regime. The diversity of correlations also implies that multiple mechanisms may be at play, including geometric effects (e.g., Doppler boosting due to jet orientation) and varying dominance of synchrotron versus thermal components.

Finally, the Johnson–Kron–Cousins BVR measurements listed in Table 1, fall within the range MJD 60,316 to 60,447. Most of these measurements occur during periods of relatively low gamma-ray flux. However, the observation at MJD 60,432.623 corresponds to the strongest chromatic outlier in the dataset (c–o = −1.13), coinciding with the gamma-ray activity in that interval. This suggests that the broad-band photometry captures real spectral changes that may be associated with variations in the emission zones or mechanisms.

In summary, cross-comparison of the optical and gamma-ray datasets reveals multiple instances of temporal alignment between chromatic anomalies and gamma-ray flux enhancements. These findings provide empirical support for multiwavelength variability driven by jet dynamics in 3C 273.

Among all intervals analyzed, the period around MJD 60,430–60,500 represents the strongest observed episode of multiwavelength correlation. The most extreme value of the $c–o$ color index ($-1.13\pm 0.06$ at MJD 60,432.62) occurs during a phase of elevated gamma-ray flux, coinciding with a cluster of optical flares detected in the ATLAS c-band (Table 2). This interval spans nearly the full dynamic range of chromatic variability and includes several additional outliers. The temporal coincidence between intense blueing and increased gamma-ray activity is consistent with a common leptonic origin in the relativistic jet—synchrotron in the optical and inverse-Compton in gamma rays, produced by the same electron population. We identify this phase as the most compelling observational evidence for color–gamma coupling in 3C 273.

However, this temporal overlap does not, by itself, constitute definitive evidence of a physical connection. Given the stochastic nature of AGN variability and the sparser optical sampling relative to gamma-ray monitoring, the coincidences should be viewed as suggestive rather than conclusive. A dedicated cross-correlation analysis with appropriate significance testing would be required to establish a causal link, which is beyond the scope of this work.

4. PDS 456: Radio-Quiet Quasar Analysis

4.1. Secular Evolution of the o-Band Magnitude

The ATLAS o-band is especially well suited for investigating long-term optical variability in PDS 456 due to its combination of photometric depth, sampling cadence, and spectral coverage. Centered at ∼680 nm with a broad red-optical response, the o-band is less sensitive to UV-driven, short-term variability often seen in active galactic nuclei, and is therefore sensitive to gradual changes in the red optical continuum. This continuum is generally associated with thermal emission from the cooler, outer regions of the accretion disk, although additional contributions from broad emission lines, scattering, or reprocessing by disk winds may also be present. In addition, ATLAS provides high-cadence, homogeneous photometric monitoring over multiple years, with typical uncertainties ≤0.01 mag in this band for PDS 456. These characteristics mitigate short-term noise and enhance the detection of subtle brightness variations on secular timescales.

Based on this, we analyzed the long-term behavior of PDS 456 using a dataset of 2942 individual photometric measurements in the o-band, each consisting of the Modified Julian Date (MJD), the apparent magnitude m, and its associated uncertainty $\Delta m$. To identify potential secular trends, we performed a weighted least-squares linear regression.

The analysis revealed a statistically significant secular brightening trend in the optical flux of PDS 456 over the observed time span, characterized by a slope of $(-3.1\pm 0.2)\times {10}^{-5}\mathrm{mag}{\mathrm{day}}^{-1}$. This corresponds to a long-term brightening rate of $(-0.0113\pm 0.0007)\mathrm{mag}{\mathrm{yr}}^{-1}$ (Figure 6). The fitted negative slope indicates that the apparent magnitude of PDS 456 has been decreasing over time, which corresponds to a gradual increase in its optical brightness.

The coefficient of determination was $R^2=0.121$, suggesting that about 12.1% of the variance in the o-band magnitude is explained by the linear temporal evolution. Despite this modest proportion, the associated p-value of the slope (p∼0) confirms the trend’s statistical robustness.

To validate the presence of a monotonic trend independently of any linearity assumption, we applied the non-parametric Spearman rank correlation test between MJD and m. We find a Spearman rank correlation of $\rho =-0.38$ (p∼0), indicating a statistically significant secular brightening of PDS 456 over the monitoring period.

This result places PDS 456 among the rare cases of quasars exhibiting statistically significant secular photometric evolution. Although long-term optical variability in quasars has been reported in earlier studies, the identification of linear brightening trends in radio-quiet quasars has remained elusive due to limited temporal coverage or large photometric uncertainties. In particular, Smith [61] reported secular changes of 0.2 to 0.5 magnitudes over 15 years in a subset of radio-quiet quasars, while Angione and Smith [62] identified several quasars exhibiting non-random, long-term optical trends.

This observed brightening may reflect long-term changes in the accretion rate on the central engine [63], modulations in reprocessed disk emission [64], or evolving geometric configurations such as disk inclination or line-of-sight extinction [65]. Continued multi-band monitoring will help disentangle these scenarios and constrain the physical processes driving the secular variability.

To our knowledge, no published, dedicated optical monitoring of PDS 456 with cadence and photometric precision comparable to our dataset has been reported. Prior optical/UV observations exist but are sparse or campaign-focused rather than long-term, high-cadence (e.g., early V-band light-curve segments and seasonal rms; UV/optical photometry within X-ray campaigns) [15,19,33,66]. Accordingly, the secular brightening and short-term variability amplitudes presented here represent, to our knowledge, the first long-baseline optical variability characterization of this source at this level of fidelity.

4.1.1. Detection of Optical Flares in the o-Band

Superimposed on the secular brightening trend described above, the o-band light curve of PDS 456 also exhibits short-term variability on timescales of days to weeks. To isolate these rapid events, we applied a LOWESS (Locally Weighted Scatterplot Smoothing) algorithm to the time series, using a dataset of 2942 measurements with photometric uncertainties below 0.01 mag.

We then computed residuals relative to the smoothed curve and identified optical flares as epochs with brightness excursions more than three standard deviations (>3$\sigma$) from the trend, corresponding to sudden brightenings or dimmings. The standard deviation of the residuals was found to be $\sigma \approx 0.045\mathrm{mag}$.

This criterion yielded 75 flaring events, listed in

Table 4. Detected optical flares in PDS 456 (Part I). Magnitudes with positive residuals indicate dimming events.

MJD	Magnitude	Residual
59,781.13982	$14.251\pm 0.005$	$0.28224$
58,919.60627	$14.219\pm 0.005$	$0.17888$
59,785.16742	$14.138\pm 0.004$	$0.17108$
59,655.37404	$14.190\pm 0.008$	$0.16712$
59,421.42573	$14.224\pm 0.008$	$0.16446$
58,919.58674	$14.196\pm 0.007$	$0.15587$
59,742.35348	$14.138\pm 0.006$	$0.14984$
59,785.14536	$14.112\pm 0.004$	$0.14507$
59,785.14904	$14.107\pm 0.004$	$0.14007$
58,919.58997	$14.172\pm 0.006$	$0.13187$
60,739.34464	$13.780\pm 0.005$	$-0.13307$
59,819.83821	$13.830\pm 0.005$	$-0.13338$
60,739.36349	$13.778\pm 0.005$	$-0.13507$
60,458.26818	$13.746\pm 0.004$	$-0.13553$
60,735.39055	$13.777\pm 0.005$	$-0.13670$
60,442.28403	$13.740\pm 0.004$	$-0.13759$
60,442.24279	$13.737\pm 0.004$	$-0.14058$
60,721.36584	$13.775\pm 0.006$	$-0.14085$
60,360.65913	$13.762\pm 0.004$	$-0.14096$
59,709.27420	$13.841\pm 0.004$	$-0.14152$
60,442.24095	$13.736\pm 0.004$	$-0.14157$
60,442.27024	$13.736\pm 0.004$	$-0.14158$
59,358.38563	$13.909\pm 0.006$	$-0.14215$
60,558.07891	$13.766\pm 0.004$	$-0.14425$
60,721.38508	$13.764\pm 0.006$	$-0.15185$
60,739.35427	$13.758\pm 0.004$	$-0.15507$
59,358.42088	$13.895\pm 0.006$	$-0.15614$
60,721.38882	$13.756\pm 0.005$	$-0.15985$
59,515.20129	$13.941\pm 0.006$	$-0.16019$
60,721.35842	$13.754\pm 0.005$	$-0.16185$
59,819.85561	$13.801\pm 0.005$	$-0.16239$
59,358.38837	$13.888\pm 0.006$	$-0.16315$
59,819.84137	$13.800\pm 0.005$	$-0.16338$
60,187.13460	$13.769\pm 0.006$	$-0.17352$
60,398.31558	$13.707\pm 0.005$	$-0.17465$
60,398.28632	$13.707\pm 0.005$	$-0.17466$
60,558.08450	$13.735\pm 0.004$	$-0.17525$
60,187.11838	$13.763\pm 0.006$	$-0.17952$

and

Table 5. Detected optical flares in PDS 456 (Part II).

MJD	Magnitude	Residual
60,558.07566	$13.723\pm 0.004$	$-0.18724$
59,710.36622	$13.793\pm 0.005$	$-0.18926$
60,187.11242	$13.753\pm 0.006$	$-0.18952$
60,558.09510	$13.714\pm 0.004$	$-0.19625$
59,741.16267	$13.791\pm 0.004$	$-0.19658$
60,742.37770	$13.714\pm 0.004$	$-0.19858$
60,747.31812	$13.713\pm 0.006$	$-0.19878$
60,742.38047	$13.712\pm 0.004$	$-0.20058$
59,741.16965	$13.784\pm 0.004$	$-0.20359$
59,515.19806	$13.897\pm 0.007$	$-0.20418$
60,750.34293	$13.707\pm 0.005$	$-0.20429$
59,710.35596	$13.776\pm 0.005$	$-0.20625$
60,742.38964	$13.705\pm 0.004$	$-0.20758$
60,742.37493	$13.705\pm 0.004$	$-0.20759$
60,739.34831	$13.703\pm 0.005$	$-0.21007$
59,742.17138	$13.774\pm 0.005$	$-0.21419$
59,741.16618	$13.772\pm 0.004$	$-0.21559$
60,398.29174	$13.666\pm 0.005$	$-0.21566$
60,750.32962	$13.694\pm 0.005$	$-0.21729$
60,750.34843	$13.689\pm 0.005$	$-0.22228$
60,750.33289	$13.689\pm 0.005$	$-0.22229$
60,398.28812	$13.657\pm 0.006$	$-0.22466$
59,741.17910	$13.759\pm 0.005$	$-0.22860$
60,562.04650	$13.681\pm 0.005$	$-0.23089$
59,710.34664	$13.750\pm 0.007$	$-0.23225$
59,515.20820	$13.860\pm 0.006$	$-0.24119$
59,742.18804	$13.738\pm 0.005$	$-0.25019$
60,747.31447	$13.652\pm 0.006$	$-0.25978$
60,377.38048	$13.617\pm 0.006$	$-0.27625$
60,747.32545	$13.632\pm 0.006$	$-0.27978$
60,751.38583	$13.624\pm 0.005$	$-0.28711$
60,751.40451	$13.608\pm 0.005$	$-0.30311$
60,751.40165	$13.599\pm 0.005$	$-0.31211$
60,747.33965	$13.599\pm 0.005$	$-0.31278$
60,562.04329	$13.583\pm 0.007$	$-0.32889$
59,680.11169	$13.608\pm 0.005$	$-0.38342$
59,819.82223	$13.514\pm 0.005$	$-0.44938$

and marked in red in Figure 7. These events span a broad temporal range and may be associated with episodic enhancements in accretion activity or transient structures in the inner accretion disk. Given the absence of strong jet signatures in PDS 456—such as radio or gamma-ray counterparts—it is plausible that these flares originate from internal thermal processes or clumpy outflows rather than relativistic jet activity.

To assess the reliability and physical nature of the most significant dimming events, we examined the ten largest positive deviations from the LOWESS-smoothed light curve. These events show magnitude increases of up to 0.28 mag (≃6.3$\sigma$), with photometric uncertainties below 0.008 mag, indicating high measurement quality. The events are temporally dispersed, with no clustering that might suggest instrumental issues. While sudden brightenings are more typical of AGN variability, abrupt dimmings can also occur due to transient obscuration by clumpy, ionized outflows [67], thermal instabilities, or geometric changes in the inner accretion disk. Given the lack of jet-related activity in PDS 456, a thermal or disk-origin scenario is plausible.

Four optical flaring events in the quasar PDS 456 exhibit striking temporal coincidence with observations from the Swift satellite’s UVOT instrument, with time differences of less than 60 min between optical and ultraviolet detections. The Swift data were obtained from the High Energy Astrophysics Science Archive Research Center (HEASARC).3

These flares occurred around MJD 59,819.8 (corresponding to 28 August 2022) and are characterized by anomalously bright magnitudes of $13.514\pm 0.005$, $13.800\pm 0.005$, $13.830\pm 0.005$, and $13.801\pm 0.005$. The associated photometric residuals ($-0.449$, $-0.163$, $-0.133$, and $-0.162$) indicate significant deviations from the modeled baseline, identifying these as strong optical brightening episodes.

Swift/UVOT obtained two exposures in the V (5468 Å) and two in the UVW1 (2600 Å) filters [68,69,70]. V-band:total counts = 25,440,902 and 19,082,684 with $t_{\exp }=841.3$ and $634.6$ s, yielding count rates of $30,240.0$ and $30,070.4\mathrm{ct}{\mathrm{s}}^{-1}$. UVW1:total counts = 2,651,393 and 1,548,750 with $t_{\exp }=806.9$ and $462.7$ s, yielding $3285.9$ and $3347.2\mathrm{ct}{\mathrm{s}}^{-1}$. Relative to the historical medians (V: $33,775\mathrm{ct}{\mathrm{s}}^{-1}$; UVW1: $3721\mathrm{ct}{\mathrm{s}}^{-1}$; July 2021–October 2024), both the V- and UVW1-band rates are below the median; these exposures alone therefore do not indicate stronger flaring at longer wavelengths.

Converting the optical magnitudes into relative fluxes using $F\propto {10}^{-0.4m}$ [71], we find a clear correspondence with contemporaneous, normalized Swift/UVOT count rates. This multiwavelength agreement suggests that the optical and ultraviolet variations likely arise from the same physical process—plausibly transient temperature fluctuations in the accretion disk or localized magnetic reconnection. However, the UVOT count rates in these exposures lie below their historical medians, so these data alone do not constitute a UV “flare”. No concurrent X-ray enhancements were found in the available Swift/XRT data for the same epochs [72], implying that the variability energy may have been preferentially radiated in the optical–UV domain or that any X-ray response was below the detection threshold at the time.

Interestingly, the optical–UV variability observed in 2022 contrasts with a previously reported X-ray flare detected in September 2018 by Swift and NuSTAR, during which no significant optical or ultraviolet variability was found [73]. That event was attributed to energetic processes confined to the X-ray corona. In 2022, by contrast, no concurrent X-ray activity was detected despite clear optical brightenings and only modest UV changes, suggesting that different physical mechanisms or emission regions were responsible. This duality may indicate that PDS 456 hosts multiple, temporally variable zones of activity—such as the accretion disk and the compact corona—whose relative contributions evolve independently over time.

Due to the limited temporal sampling around individual events, it is not possible to determine whether the corresponding light curves are symmetric or asymmetric. Most events are represented by only one or two data points beyond the detection threshold, which precludes any reliable morphological characterization.

Although PDS 456 is a radio-quiet quasar and would typically be expected to exhibit fewer flares than a jet-dominated blazar such as 3C 273, a larger number of events (75) were detected in its o-band light curve, compared to only 10 flares in the c-band light curve of 3C 273. This apparent discrepancy is most likely due to the substantially greater number of o-band observations for PDS 456 (2942 versus ∼550; i.e., ∼5.3× more data points), which increases the probability of capturing short-lived deviations in a threshold-based search; consequently, raw event counts are not directly comparable across sources [74]. A fair comparison would normalize by the number of observations or observing time (and, ideally, match bandpass and cadence).

4.1.2. Chromatic Analysis of PDS 456 Based on ATLAS Data

Building on the results of the o-band light curve analysis, where PDS 456 exhibited a statistically significant secular brightening and a sequence of 75 optical flares, we extended our investigation to its chromatic behavior. This complementary analysis seeks to determine whether the observed flux variability is accompanied by color changes, which could offer additional insight into the physical origin of the variability.

We analyzed 46 epochs of simultaneous photometric measurements in the ATLAS c and o bands. While the o-band primarily traces the redder, thermal component of the optical continuum, the c-band is more sensitive to emission from the hotter, inner regions of the accretion disk. The color index $c–o$ thus serves as a proxy for the spectral slope and can reveal episodes of disk heating, obscuration, or structural changes.

Statistical Characterization of the Color Index

The color index $c–o$ was computed for all selected epochs and tested for normality using the Shapiro–Wilk test. The result ($W=0.761,p\approx 2.98\times {10}^{-7}$) indicates a strong departure from a Gaussian distribution. Therefore, we adopted robust estimators: the median as the central location and the median absolute deviation (MAD) as the scale parameter. The median color index is $\tilde{x}=0.483$, with $\mathrm{MAD}=0.0255$.

Based on this distribution, we defined three chromatic phases: reddening ($c–o>\tilde{x}+\mathrm{MAD}\approx 0.509$), blueing ($c–o<\tilde{x}-\mathrm{MAD}\approx 0.458$), and neutral. This classification yielded 8 reddening epochs, 15 blueing epochs, and 23 neutral epochs (Figure 8;

Table 6. Color index and magnitude measurements for PDS 456 (Part I).

Mean MJD	$\mathit{c}–\mathit{o}$	c	o
57,895.08	$0.438\pm 0.009$	$14.346\pm 0.004$	$13.908\pm 0.005$
57,899.08	$0.399\pm 0.011$	$14.336\pm 0.004$	$13.937\pm 0.007$
57,903.06	$0.410\pm 0.011$	$14.364\pm 0.004$	$13.954\pm 0.007$
57,980.84	$0.458\pm 0.008$	$14.402\pm 0.004$	$13.944\pm 0.004$
58,253.04	$0.469\pm 0.009$	$14.462\pm 0.005$	$13.993\pm 0.004$
58,332.90	$0.502\pm 0.011$	$14.485\pm 0.005$	$13.983\pm 0.006$
58,396.76	$0.455\pm 0.013$	$14.468\pm 0.005$	$14.013\pm 0.008$
58,573.10	$0.482\pm 0.010$	$14.538\pm 0.006$	$14.056\pm 0.004$
58,965.03	$0.487\pm 0.015$	$14.501\pm 0.005$	$14.014\pm 0.010$
58,969.02	$0.488\pm 0.009$	$14.512\pm 0.005$	$14.024\pm 0.004$
58,985.05	$0.470\pm 0.009$	$14.490\pm 0.004$	$14.020\pm 0.005$
58,988.96	$0.513\pm 0.009$	$14.533\pm 0.005$	$14.020\pm 0.004$
58,997.03	$0.408\pm 0.010$	$14.496\pm 0.004$	$14.088\pm 0.006$
59,015.96	$0.488\pm 0.009$	$14.528\pm 0.005$	$14.040\pm 0.004$
59,040.89	$0.482\pm 0.009$	$14.533\pm 0.005$	$14.051\pm 0.004$
59,048.87	$0.457\pm 0.009$	$14.538\pm 0.005$	$14.081\pm 0.004$
59,069.84	$0.496\pm 0.009$	$14.524\pm 0.005$	$14.028\pm 0.004$
59,080.81	$0.438\pm 0.011$	$14.503\pm 0.007$	$14.065\pm 0.004$
59,371.04	$0.376\pm 0.010$	$14.445\pm 0.004$	$14.069\pm 0.006$
59,375.03	$0.490\pm 0.009$	$14.529\pm 0.005$	$14.039\pm 0.004$

and

Table 7. Color index and magnitude measurements for PDS 456 (Part II).

Mean MJD	$\mathit{c}–\mathit{o}$	c	o
59,378.90	$0.512\pm 0.008$	$14.532\pm 0.004$	$14.020\pm 0.004$
59,408.89	$0.503\pm 0.010$	$14.560\pm 0.006$	$14.057\pm 0.004$
59,453.83	$0.456\pm 0.010$	$14.549\pm 0.005$	$14.093\pm 0.005$
59,680.33	$0.827\pm 0.015$	$14.435\pm 0.010$	$13.608\pm 0.005$
59,787.51	$0.459\pm 0.009$	$14.414\pm 0.005$	$13.955\pm 0.004$
59,812.08	$0.499\pm 0.009$	$14.443\pm 0.005$	$13.944\pm 0.004$
59,812.86	$0.497\pm 0.009$	$14.437\pm 0.005$	$13.940\pm 0.004$
59,814.96	$0.473\pm 0.008$	$14.438\pm 0.004$	$13.965\pm 0.004$
59,818.91	$0.499\pm 0.009$	$14.467\pm 0.005$	$13.968\pm 0.004$
59,820.06	$0.619\pm 0.010$	$14.420\pm 0.005$	$13.801\pm 0.005$
59,822.65	$0.485\pm 0.009$	$14.489\pm 0.005$	$14.004\pm 0.004$
Mean MJD	$\mathit{c}–\mathit{o}$	c	o
59,836.91	$0.499\pm 0.010$	$14.456\pm 0.006$	$13.957\pm 0.004$
59,850.10	$0.536\pm 0.011$	$14.468\pm 0.005$	$13.932\pm 0.006$
60,034.75	$0.489\pm 0.009$	$14.427\pm 0.005$	$13.938\pm 0.004$
60,142.52	$0.433\pm 0.008$	$14.376\pm 0.004$	$13.943\pm 0.004$
60,193.44	$0.488\pm 0.015$	$14.425\pm 0.005$	$13.937\pm 0.010$
60,436.72	$0.452\pm 0.009$	$14.311\pm 0.005$	$13.859\pm 0.004$
60,442.64	$0.611\pm 0.009$	$14.351\pm 0.005$	$13.740\pm 0.004$
60,448.02	$0.448\pm 0.014$	$14.324\pm 0.005$	$13.876\pm 0.009$
60,466.48	$0.441\pm 0.008$	$14.346\pm 0.005$	$13.905\pm 0.003$
60,466.51	$0.440\pm 0.008$	$14.349\pm 0.005$	$13.909\pm 0.003$
60,490.58	$0.556\pm 0.009$	$14.336\pm 0.005$	$13.780\pm 0.004$
60,554.48	$0.441\pm 0.009$	$14.375\pm 0.005$	$13.934\pm 0.004$
60,558.51	$0.484\pm 0.009$	$14.411\pm 0.005$	$13.927\pm 0.004$
60,732.89	$0.471\pm 0.009$	$14.366\pm 0.005$	$13.895\pm 0.004$
60,739.73	$0.619\pm 0.010$	$14.397\pm 0.005$	$13.778\pm 0.005$

).

Distribution and Variability Patterns

The histogram of $c–o$ values (Figure 9) confirms a non-Gaussian distribution with mild left-skew (an excess of blueing excursions). Despite these episodic departures, there is no evidence for a persistent long-term chromatic drift over the monitoring baseline.

The color–magnitude diagram in Figure 10 illustrates the relationship between the $c–o$ index and the o-band magnitude. Although the LOWESS trend suggests mild nonlinearity, no statistically significant correlation is found.

UV–Optical Chromatic Trends Post-Flaring

We identified a temporal coincidence between an ATLAS color index measurement and a Swift/UVOT UVW2 observation of PDS 456. The ATLAS data at MJD 60,442.64 yielded $c–o=0.611\pm 0.009$, among the reddest epochs in our sample. Less than a day earlier, Swift/UVOT observed the source in the UVW2 filter (central wavelength $1928\AA$) with an exposure time of $t_{\exp }=1411.6\mathrm{s}$ and a total photon count of 6,651,476. This coincidence provides a snapshot of the spectral slope between the near-ultraviolet and optical regimes during a reddened phase. Within the optical alone, the elevated $c–o$ denotes a redder color (a steeper optical slope). When combined with the contemporaneous UV measurement, the broad UV–optical SED is better described as curved or UV-hardened rather than uniformly flattened or steepened; i.e., the UV appears relatively stronger than implied by the optical slope. This pattern is consistent with a temporary heating of the inner disk and/or short-timescale reprocessing that favors shorter wavelengths. This behavior is consistent with enhanced emission from the accretion disk, where UV light originates from hotter inner regions and optical light from cooler, outer zones.

Compared with the historical median UVW2 source count rate of $2059{\mathrm{s}}^{-1}$, the count rate inferred from the quoted exposure and counts, 6,651,476$/1411.6\simeq 4712{\mathrm{s}}^{-1}$, constitutes a marked increase in near-ultraviolet emission (≈2.29× the median). For clarity, we report the UVOT source count rate and whether coincidence-loss corrections were applied; values derived from “total counts” outside the source aperture can overestimate the source rate.

Notably, this episode occurred shortly after a cluster of four optical flares detected in the o-band light curve at MJDs 60,442.24095, 60,442.24279, 60,442.27024, and 60,442.28403. These events exhibited significant negative residuals in magnitude ($\Delta m<0$), indicating transient brightening, relative to the LOWESS trend, and preceded the color-index measurement by $\Delta t\simeq 8.5–9.6\mathrm{h}$. The close temporal proximity of these flares with the UV and color-index measurements suggests that the observed reddening coincides with a rapidly evolving post-flare spectral state, with enhanced ultraviolet emission from the inner accretion flow. The temporal alignment across independent instruments strengthens the interpretation of coordinated central-engine activity.

5. Comparative Analysis of 3C 273 and PDS 456

The blazar 3C 273 and the radio-quiet quasar PDS 456 exhibit distinct but complementary patterns of optical variability, as revealed by long-term multi-band photometric monitoring.

In 3C 273, a statistically significant secular dimming was observed in the ATLAS c-band light curve, with a fading rate of $(5.6\pm 0.2)\times {10}^{-4}\mathrm{mag}{\mathrm{day}}^{-1}$. This trend is consistent with the historical decline observed in Johnson–Cousins $BVR$ data, confirming a long-term reduction in optical flux. We interpret this secular dimming primarily as a gradual decline in the accretion rate and/or slow structural evolution of the thermal disk; the relativistic jet chiefly modulates short-term, flare-like variability rather than the long-term trend. Conversely, PDS 456 shows a statistically significant secular brightening in the ATLAS o-band, with a rate of $(-3.1\pm 0.2)\times {10}^{-5}\mathrm{mag}{\mathrm{day}}^{-1}$. This long-term increase in optical flux may indicate an evolutionary phase characterized by enhanced accretion activity or thermal instabilities in the inner disk, resulting in sustained luminosity growth.

Short-term variability differs markedly between the two sources. In 3C 273, ten optical flares were identified in the c-band light curve, with one cluster temporally associated with gamma-ray enhancements (MJD 58,846.56) and one isolated event aligned with a modest enhancement (MJD 60,444.95) observed by Fermi. These flares are interpreted as synchrotron events linked to internal shocks or magnetic reconnection in the relativistic jet. In contrast, PDS 456 exhibited 75 optical outliers (brightenings and dimmings) in the o-band, including one episode with a clear near-UV enhancement (UVW2) measured by Swift/UVOT; other cases showed only modest UV response. The absence of simultaneous X-ray flares suggests a disk origin for these events, possibly due to transient heating episodes or clumpy outflows.

Color variability also displays contrasting behaviors. In 3C 273, the $c–o$ index revealed alternating reddening and blueing phases and a statistically significant bluer-when-brighter correlation with the o-band magnitude ($\rho =-0.54$, $p=2.3\times {10}^{-3}$), consistent with synchrotron-dominated variability from a relativistic jet. In PDS 456, the $c–o$ distribution is non-Gaussian and mildly asymmetric (with a modest excess of blueing excursions), and shows no evidence of a persistent chromatic trend or a significant correlation with brightness. The phase classification based on the median and MAD yields 8 reddening, 15 blueing, and 23 neutral epochs.

Temporal alignments between color extrema and high-energy events further illustrate their distinct physical mechanisms. In 3C 273, extreme color values such as $c–o=-1.13$ coincided with enhanced gamma-ray flux, supporting a jet-based origin. In PDS 456, an extreme reddening event ($c–o=0.611$) occurred within a day of a UVW2 flare measured by Swift/UVOT, indicating UV hardening (spectral curvature) consistent with inner-disk heating. These findings suggest that while structural differences (i.e., jet-dominated vs. disk-dominated emission) account for much of the short-term variability, the observed secular trends in brightness are best explained by evolutionary changes in the accretion dynamics of each AGN. 3C 273 may be transitioning toward a lower-luminosity state, whereas PDS 456 is undergoing a phase of increasing radiative output. These differences reinforce the interpretation that optical variability in 3C 273 is primarily jet-driven, while in PDS 456 it is dominated by processes on the accretion disk.

While the contrasting secular trends observed in 3C 273 and PDS 456 are statistically robust and supported by long-term photometric monitoring, we emphasize that AGN variability is intrinsically stochastic and shows considerable dispersion across sources of similar luminosity and type. Our objective here is not to propose a unified physical mechanism underlying the observed differences, but rather to document their empirical behavior over decade-long timescales using internally homogeneous datasets. A more detailed interpretation—for example, via structure-function analysis, power-spectral density (PSD) fitting, or placement in variability–luminosity scaling relations—would require broader samples and extended modeling, which lie beyond the scope of the present work. Nonetheless, the detection of a statistically significant brightening in PDS 456 is notable given its classification as a luminous, radio-quiet quasar, which are typically less variable on such baselines. We hope this observational baseline contributes to future efforts to map variability mechanisms across the AGN population.

Finally, we note that the contrasting photometric behaviors observed in PDS 456 and 3C 273 align with their expected positions along the Eigenvector 1 (E1) main sequence of quasars [75]. PDS 456 is often interpreted as a high-luminosity analog of extreme Population A sources, showing radiation-driven ultra-fast outflows, a soft X-ray excess, and accretion rates near the Eddington limit [15,19,76]. In contrast, 3C 273— a prototypical radio-loud quasar with broad emission lines and a powerful jet—lies naturally within the Population B domain, consistent with the prevalence of radio-loud sources in Pop B along the quasar main sequence [77]. While our analysis does not attempt a full spectral characterization, the secular brightening of PDS 456 and the long-term dimming of 3C 273 are qualitatively compatible with the E1 framework, reinforcing the view that distinct accretion regimes and emission mechanisms are at play.

6. Conclusions

The main results obtained from the analysis of multi-spectral data from ground- and space-based platforms for the blazar 3C 273 and the quasar PDS 456 are summarized as follows:

• Secular trends: ATLAS c-band photometry of 3C 273 (551 points) shows a robust secular dimming with a slope of $(5.6\pm 0.2)\times {10}^{-4}\mathrm{mag}{\mathrm{day}}^{-1}$, consistent with archival $BVR$ data. In contrast, PDS 456 shows a statistically significant secular brightening in the o-band $(-3.1\pm 0.2)\times {10}^{-5}\mathrm{mag}{\mathrm{day}}^{-1}$, placing it among the few radio-quiet quasars with measurable long-term optical evolution.
• Short-term flares: We detect ten optical flares in 3C 273 and 75 outliers in PDS 456 using $3\sigma$ clipping on LOWESS residuals. Two of the 3C 273 flares coincide temporally with Fermi-LAT gamma-ray enhancements, favoring a leptonic scenario where optical synchrotron and gamma-ray inverse-Compton emissions originate from the same electron population.
• Chromatic variability: In 3C 273, a significant bluer-when-brighter trend is detected with the o-band magnitude ($\rho =-0.54$, $p=2.3\times {10}^{-3}$), while long-term color drift remains weak. PDS 456, by contrast, shows a non-Gaussian $c–o$ color distribution with low-amplitude, intermittent chromatic events and no persistent correlation between color and brightness.
• UV–optical coupling: In PDS 456, one brightening episode shows a UVW2 flux enhancement measured by Swift/UVOT within $\Delta t<1\mathrm{h}$ of ATLAS o-band flares, suggesting disk-related variability. However, most optical events exhibit only modest UV response, implying multiple variability drivers.
• Physical interpretation: Overall, 3C 273’s optical variability is primarily jet-driven on short timescales, superposed on a disk-dominated secular dimming. PDS 456, instead, shows disk-dominated variability and is currently in a phase of increasing radiative output, consistent with its expected location along the E1 main sequence.