Multiwavelength variability of gamma-ray emitting narrow-line Seyfert 1 galaxies

As one of the drivers of feedback in active galactic nuclei (AGNs), the jets launched from supermassive black holes (SMBHs) are important for understanding the co-evolution of SMBHs and their host galaxies. However, the formation of AGN jets is far from clear. The discovery of gamma-ray narrow-line Seyfert 1 (NLS1) galaxies during the past two decades has provided us with a new means of studying the link between jets and accretion processes and the formation of jets. Here, we explore the coupling of jet and accretion discs in seven bright gamma-ray NLS1 galaxies by studying simultaneous optical/ultraviolet and X-ray observations of these systems taken by Swift. The results show that, except for 1H 0323+342 in which the X-rays are significantly contributed from the accretion disc, the observed X-ray emission of the other sources is dominated by the jet, and accretion process makes little contribution if not absent. Although the origin of the X-ray emission is different, the broad-band spectral shape characterized by alpha_ox and the X-ray flux is found to follow the same evolutionary trend in 1H 0323+342, PMN J0948+0022, and PKS 1502+036. For the remaining sources, the trend is not observed or the sampling is not dense enough.


INTRODUCTION
Powered by supermassive black holes (SMBHs, in the mass range 10 6 -10 10 M ⊙ ) at their centers, active galactic nuclei (AGNs) are among the most luminous long-lived sources in the universe. Collimated outflows of matter in form of jets, launched from the vicinity of the central accreting SMBH, are found in many AGNs. Jets and outflows are suggested to be one of the fundamental mechanisms by which the SMBH interacts with the host galaxy, regulating its evolution (Di Matteo et al. 2005;Wagner et al. 2012). However, the physical mechanism of jet formation and the role of black hole mass, magnetic fields and accretion disc properties are still not well understood (Blandford et al. 2019). The energetic particles in AGNs emit radiation at radio frequencies via the synchrotron process and make their jets detectable as luminous radio sources. But only less than 20 percent of all AGNs are found to be radio-loud (Urry & Padovani 1995;Ivezić et al. 2002), while the rest are radio-quiet. The majority of the radio-quiet sources do have faint jets, and only few AGNs are not detected at all in the radio regime. However, the question is still open as to which parameters drive the most powerful, bright, relativistic jets in the radio-loud population in particular.
The conventional picture established during the first few decades after the discovery of radio quasar was that the relativistic AGN jets are preferentially associated with the heavier SMBHs residing in giant elliptical/bulge-dominated galaxies (Bahcall et al. 1997;Dunlop et al. 2003;Floyd et al. 2004). This knowledge has been updated by the discoveries in the last two decades of samples of radio-loud ★ E-mail: syao@mpifr-bonn.mpg.de † E-mail: astrokomossa@gmx.de narrow-line Seyfert 1 (NLS1) galaxies and the more massive radioloud narrow-line type 1 quasars (Komossa et al. 2006) that showed SMBH masses below 10 8 M ⊙ , below the classical radio-loud regime. The vast majority of the latter still lacks host images, but among the very small number of them that do have host imaging, a few disc-like galaxies have been identified (e.g. Zhou et al. 2007;Järvelä et al. 2018;Olguín-Iglesias et al. 2020).
In particular, some radio-loud NLS1 galaxies show persistent or flaring -ray emission independently re-confirming their launch of powerful relativistic jets (e.g. Abdo et al. 2009a,b). The presence of such jets was already earlier found from radio observations (e.g., Komossa et al. 2006;Yuan et al. 2008). NLS1 galaxies generally host less massive but rapidly growing SMBHs and NLS1s galaxies are suggested to be located at the end of the AGN parameter space opposite to the classical blazars (Boroson 2002). Thus, NLS1 galaxies provide us with new laboratories for studying open questions related to the formation and evolution of relativistic jets in AGNs.
For this study, we selected bright radio-loud, gamma-ray detected NLS1 galaxies with multiyear coverage obtained by the Neil Gehrels Swift observatory (Swift hereafter; Gehrels et al. 2004). The targets of our study (Table 1) have been studied in selected time intervals and/or wavebands before (e.g., Liu et al. 2010;Calderone et al. 2011;Jiang et al. 2012;Foschini et al. 2012;Eggen et al. 2013;Itoh et al. 2013;Yao et al. 2015a,b;Foschini et al. 2015;Gabanyi et al. 2018;Angelakis et al. 2018;Yang et al. 2018;Yao et al. 2019;D'Ammando 2020a,b;Berton et al. 2021a;Mao & Yi 2021;Ojha et al. 2022). Here, we combine archival Swift data and our own Swift data until 2021 December. Focus is on recent data since 2019, but we also add pre-2019 data to the analysis, for comparison purposes and in order to identify systematic trends. In particular, we address the broad-band spectral evolution using the simultaneous optical/UV and X-ray observations obtained by the X-ray telescope (XRT; Burrows et al. 2005) and ultra-violet-optical telescope (UVOT; Roming et al. 2005) onboard Swift. We note that several of the sources in our study are formally quasars, however, for simplicity we continue to refer to all of them as "Seyfert" galaxies hereafter, as is a common habit when discussing NLS1 galaxies. This paper is structured as follows: In Section 2 we describe the data reduction for the XRT and UVOT, respectively. The results and the analysis based on the optical/UV and X-ray data are provided in Section 3. In Section 4 we discuss the results of the Swift observations and finally a brief summary is given in Section 5. Throughout this work, we use a cosmology with 0 =70 km s −1 Mpc −1 , Ω M =0.3 and Ω Λ =0.7.
2 Swift OBSERVATIONS AND DATA REDUCTION

Data reduction
Data taken with the Swift/XRT are reprocessed following standard procedures using the task xrtpipeline. The photon counting (PC) mode (Hill et al. 2004) data are selected for the analysis. The images are extracted first and visually inspected to exclude those observations in which the source is at the edge or outside of the field of view. To obtain the spectra, the source counts are extracted from a circle of 50 ′′ radius centered on the source position and the background is estimated from an annulus with inner radius of 60 ′′ and three times the area of the source region, except for 1H 0323+342 which sometimes has brightened to a count rate higher than 0.5 counts s −1 . These data are affected by pile-up. In order to determine at which point the data and the XRT point spread function (PSF) model diverge, the XRT image from the observation in which 1H 0323+342 has highest count rate is modeled by a King function PSF( ) = 1 + ( / c ) 2 , where c = 5.8, = 1.55 and is the distance to the PSF center. The King function is fitted to the outer wings of the PSF at distance larger than 15 ′′ . The best-fit model is then visually compared with the data, and the diverge appears below ∼8 ′′ . Thus, for all the XRT observations on 1H 0323+342 with count rate > 0.5 counts s −1 the source events are extracted from an annulus with inner/outer radius of 8 ′′ /60 ′′ , and the backgrounds are estimated from an annulus with inner/outer radius of 70 ′′ /90 ′′ .
The observations in which the sources are not detected (< 5 ) are excluded. Full grade selection (0-12) is adopted to extract the source and background spectra. The ancillary response files are built using task xrtmkarf with the corresponding exposure maps to correct the PSF if any. The relevant response matrix files given in the output of xrtmkarf are used. The spectra with more than 125 counts are grouped such that they have at least 25 counts in each bin, and the 2 minimization is used to fit model to the spectra. For the spectra with 125 or less counts, the spectra are grouped such that each spectrum has 5 to 10 bins, and the Cash statistic is adopted to fit the spectra. The software package XSPEC (Arnaud 1996) was used for spectral analysis. A summary of the X-ray observations used in our analysis is given in Table 2.

Spectral fitting and flux estimation
All the spectra with 125 or less counts, and most of the spectra with more than 125 counts can be well fitted by a single power law with Galactic neutral hydrogen absorption, except for some spectra of 1H 0323+342 when the 2 -minimization gives a very poor, unacceptable fit, indicating an extra component other than a single power law is required. The spectra revealing a 2 higher than the 90% quantile of the 2 distribution given their degrees of freedom (dof) are then selected and fitted with a blackbody component plus a single power-law model. The absorption is fixed at the Galactic value and the blackbody temperature is forced to be in the range of 0.03 keV < Tk < 0.2 keV during the fitting. The absorptioncorrected flux is estimated by cflux. The fluxes in the 0.3-2 keV band ( soft ) and 2-10 keV ( hard ) band of the sources are shown in Figure 1-7. A summary of the results obtained from XRT observations is given in Table 3.

UVOT
The NLS1 galaxies of this study have also been observed with the Swift/UVOT, in up to all three optical and up to all three UV bands [with filters: (5468Å), (4392Å), (3465Å), 1 (2600Å), 2 (2246Å), and 2 (1928Å), where values in brackets represent the filter central wavelengths (Poole et al. 2008)]. Some sources have only been observed in selected bands. For each observation, data in each filter are co-added using the task uvotimsum after aspect correction. The images are visually inspected to exclude those in which the source is on the edges or out of the field of view. The source counts in each available filter were extracted in a circle with radius of 5 ′′ , while the background was selected in a nearby source-free region of radius 15 ′′ . The background-corrected counts are then converted into fluxes based on the latest calibration as described by Poole et al. (2008) and Breeveld et al. (2010) using the task uvotsource. A summary of the UVOT observations and results are given in Table 2 and  Table 3, respectively. The total duration of each UVOT observation is the same as the corresponding XRT duration. Each UVOT band : : : 1: 2: 2 is observed with a ratio of 1:1:1:2:3:4 of the total exposure time, respectively. Correction of the UVOT fluxes for Galactic reddening towards the individual NLS1 galaxies was carried out using the values of (B−V) from Schlafly & Finkbeiner (2011) and the reddening curves of Fitzpatrick (1999). We use the extinction-corrected optical/UV flux in the following analysis unless mentioned otherwise.

Broadband spectral slope ox
To quantify the broadband spectral slope, the spectral index ox between optical/UV and X-ray is calculated as ox = −0.384 log ( 2500 / 2 keV ), where 2500 and 2 keV are the flux densities at 2500 Å and 2 keV, respectively. Here the flux density obtained from 1 band centered at ∼ 2600 Å is adopted. The flux density at 2 keV is obtained from the best-fit model to each XRT spectrum.

UV-X-ray SEDs
For visually inspecting the correlation between the evolution of optical/UV and X-ray spectra, we build the spectral energy distributions for these sources with, if any, the UVOT data and the best-fit model of the simultaneous X-ray spectra in the same colors ( Figure 9). Data are represented as simple powerlaw connections between the respective UVOT fluxes: For 1H 0323+342, PMN J0948+0022 and SDSS J2118−0732, the extinction-corrected and 2 band data are used. For SDSS J1222+0413, the band is used instead of as there is only two -band exposures. For PKS 1502+036 and PKS 2004-447, the and band data are used, respectively, as the data points in other optical band are too few. For SDSS J0946+1017, the 1 and 2 band data are used as in the optical band this source is only detected in the band in one of the observations by Swift.
As can be seen in Figure 9, the broad-band spectral shape varies along with the flux variability. The optical/UV and X-ray flux of 1H 0323+342 reveal a simultaneous increase or decrease behavior. Its optical/UV emission is becoming bluer when brighter indicating the brightening of the source in the optical/UV being mainly due to the emergence and increase of the emission component in the ultraviolet. In comparison, the optical/UV emission of PMN J0948+0022 shows a redder when brighter trend indicating that the increase of the flux is more dominated by the emission in the infrared band. For PKS 1502+036, the relation of the optical/UV and X-ray spectral variability is not obvious by visual inspection. The X-ray spectra vary from steep to flat, but without strong constraints given the low pho- 2 0 1 9 -1 0 -0 1 2 0 1 9 -1 0 -1 5 2 0 1 9 -1 1 -0 1 2 0 1 9 -1 1 -1 5 5 5 6 9 6 5 5 7 6 0 5 5 7 7 5 5 5 7 9 0 5 5 8 0 5 5 5 8 2 0 5 5 8 8 0 5 6 1 0 5 5 6 1 2 0 5 6 1 8 5 5 6 2 0 0 5 6 4 8 3 5 6 5 6 0 5 6 5 7 5 5 6 5 9 0 5 6 6 0 5 5 6 7 2 8 5 7 5 0 0 5 7 5 1 5 5 8 7 6 0 5 8 7 7 5 5 8 7   (5): factor of flux variation in the optical, UV, soft X-rays (0.3-2 keV) and hard X-rays (2-10 keV), respectively. A lower limit on the variability in the optical or UV is obtained when a flux upper limit given by uvotsource is lower than all the detections. Column (6): X-ray photon index range from spectral fitting. Column (7) and (8): isotropic X-ray peak luminosity in the soft and hard X-ray band, respectively, in units of 10 44 erg s −1 . Column (9): spectral index between optical and UV, uv , where ∝ − uv , calculated using the band and the 2 band. When there is no or too few or 2 data, a nearby band is used (see Section 3.2). Column (10): ox calculated using the extinction corrected flux density at the 1 band and at 2 keV. ton statistics. For PKS 2004-447, the optical/UV and X-ray fluxes show a coordinated increasing/decreasing behavior, and the main contributing emission to the flux variability is likely in the infrared and in the X-ray/ -ray bands. It is hard to see any trends in the other three sources as they have very few simultaneous SEDs. Both SDSS J0946+1017 and SDSS J2118−0732 have very red optical/UV spectra, indicating a peak at infrared wavelengths. SDSS 1222+0413 shows strong optical/UV spectral variability, between a bluer spectrum in the bright state and a redder one in the fainter state, while the X-ray spectra were hard.

Stacked X-ray spectra
The spectra are stacked as the statistics of each individual XRT spectrum is low. But simply stacking all the spectra may produce incorrect or misleading results in those cases where the sources go through multiple different spectral states. As can be seen, the sources have shown obvious large amplitude flux variability (Figure 1-7) and significant spectral variability (Figure 9). The spectral indices are different within the uncertainties between individual observations. For example, although the largest and smallest Γ of PMJ J0948+0022 have large errors, the tightest constraints are Γ = 1.2 +0.2 −0.3 and Γ = 2.2 +0.3 −0.2 , implying significant spectral variability within the measurement errors.
For 1H 0323+342 and PMN J0948+0022, there is an obvious correlation between the optical/UV and X-ray emission, and the evolution of broad-band spectral shape ox and the X-ray flux. For PKS 1502+036, the correlation is not visually obvious from the light curve. There is still an trend between the ox and the 2-10 keV flux, but this trend disappears for the soft X-rays. While for PKS 2004-447 the correlation is not visually obvious either for the optical/UV and X-ray flux, or for the ox and X-ray flux. So we firstly perform the stacking of the X-ray spectra for 1H 0323+342 and PMN J0948+0022. For 1H 0323+342, as shown in literature, the X-ray flux below 10 keV is usually dominated by emission from the accretion disc and the jet only starts to dominate at ≳10 keV (Yao et al. 2015a). For PMN J0948+0022, the accretion disc emission seems to only detectable in the soft X-ray band and that the jet likely dominates the hard X-ray band according to the decomposition of its broadband X-ray spectrum (Berton et al. 2019). Thus, the spectra for stacking are selected according to the soft X-ray fluxes soft (0.3-2 keV). The spectra with highest and lowest soft are selected for generating the stacked spectra at high and low flux states, respectively. In addition, considering that both the accretion disc and jet may contribute to the X-ray flux, the dominant contributor of the X-ray emission is possibly different at different epochs even though they are in similar overall flux state, but the main contributing mechanism is unlikely to change in a short time or a stationary phase. Therefore, we also stack the spectra within a continuous period of time when soft is near a peak (T1) as the high flux state and when soft is in a valley (T2) as the low flux state (shaded areas in Figure 1 and 2).
To compare the spectral properties of high and low flux state, the number of spectra for stacking is chosen such that the stacked spectra have roughly equal statistics. The more spectra are selected for stacking in each state, the higher statistics do the stacked spectra have. On the other hand, the more spectra are selected we will more possibly get misleading results as we may stack the spectra with different spectral shapes. We select the spectra for stacking so that the stacked spectra have photon counts of ∼3000 for 1H 0323+342, and ∼900 for PMN J0948+0022.
According to the SED decomposition of these sources in literature, their optical/UV emission is likely dominated by the accretion disc most of the time, while the X-rays are more complicated and the decomposition is model dependent. Study of the spectral evolution when the relative fraction of X-ray emission increases (flatter UV/Xray slope, i.e. smaller ox ) or decreases (steeper UV/X-ray slope, i.e. larger ox ) is important to explore whether the accretion disc or the jet makes more contribution to the observed X-rays at different states. Thus, the spectra for stacking are also selected based on the broad band spectral slope ox : the X-ray spectra from epochs with largest ox and smallest ox are stacked, respectively. Again, the stacked spectra with large ox and small ox have roughly equal statistics as mentioned above.
The spectra are stacked using the task addspec, which automatically generated the corresponding response files. Then the spectra are grouped so that they have at least 25 counts in each bin. A blackbody + power-law model is adopted to fit all the stacked spectra of 1H 0323+342, while a single power-law model is adequate to well fit all the stacked spectra of PMN J0948+0022. The Galactic neutral hydrogen absorption is included during the fitting. The best-fit parameters are summarized for 1H0323+342 and PMN J0948+0022 in Table 4 and 5, respectively. As an example we show the stacked spectra and the best-fit models for 1H 0323+342 at peak (T1) and valley (T2) in Figure 10.
For PKS 1502+036, the X-ray light curve does not show any visually obvious 'peak' or 'valley'. So we do not choose such continuous periods as the high or low flux state. The spectra for stacking are selected only based on soft and ox , and each stacked spectrum has photon counts of ∼ 200. For PKS 2004-447, there is neither a visually obvious 'peak'/'valley' period nor a correlation between ox and the X-ray flux. The spectra are only selected based on soft and each stacked spectrum has photon counts of ∼ 300. A single power-law model with a Galactic neutral hydrogen absorption is adopted to fit the stacked spectra of PKS 1502+036 and PKS 2004-447, and the best-fit parameters are summarized in Table 6 and 7, respectively. For the other three sources in the sample, no spectral stacking is performed, since they were observed only a few times.

Notes on individual objects
Here, we provide a short overview of the sources of our sample, and also highlight similarities and differences in their multiwavelength properties, their black hole masses, and redshifts.

1H 0323+342.
This galaxy is one of the early identified -ray NLS1 galaxies and the nearest one. 1H 0323+342 has a black hole mass of only a few 10 7 M ⊙ estimated by several different methods including the reverberation mapping method (e.g. Zhou et al. 2007 Figure 9. SEDs of the -ray NLS1 galaxies in our sample, built using simultaneous Swift UVOT and XRT data in the observer's frame. Simultaneous UVOT/XRT data are marked in the same colour. The UVOT data are corrected for Galactic extinction. The X-ray spectra are plotted using the best-fit model parameters. The black points represent radio data collected from NED, including measurements by Bennett et al. (1986), Wright & Otrupcek (1990), Becker et al. (1991), Gregory & Condon (1991), White & Becker (1992), Laurent-Muehleisen et al. (1997), Condon et al. (1998), Healey et al. (2007), Richards et al. (2011) and Richards et al. (2014). The gray dotted line represents the SED model of PMN J0948+0022 taken from Foschini et al. (2012) and shifted vertically for comparison. In the case of 1H 0323+342 the large number of available UVOT spectra overlap strongly in this plot.     16.4/7 Notes. A Galactic absorption of column density of Gal H = 3.47 × 10 20 cm −2 is always included during the fitting. Column 2: Photon index. Column 3: 2 and the degrees of freedom. the narrow emission lines are all weak (e.g. Zhou et al. 2007;León Tavares et al. 2014;Landt et al. 2017;Wang et al. 2016). Its host galaxy image has revealed a Sersic bulge and a possible disc component (Olguín-Iglesias et al. 2020). A one-armed spiral or ring-like structure is present in both near-infrared and HST optical images, suggesting a recent galaxy interaction (Zhou et al. 2007;Antón et al. 2008;León Tavares et al. 2014;Olguín-Iglesias et al. 2020) and making 1H 0323+342 an interesting case to study the triggering of AGN activity. 1H 0323+342 is highly variable across the electromagnetic spectrum (e.g., Yao et al. 2015a; D'Ammando 2020a) and shows X-ray spectral complexity (e.g. Yao et al. 2015a;Mundo et al. 2020).
In 2019-2021, we detect a broad flare, more pronounced in the UV and with an amplitude of ∼ 1 magnitude. As can be seen from the individual and stacked X-ray spectra and the broad-band SED in Figure 9 and 10, we find a significant excess of soft X-rays over a single powerlaw. The spectrum of 1H 0323+342 is well fitted by a two-component model, where the soft X-ray emission likely represents emission related to an accretion disc, or is otherwise mimicked by the presence of ionized absorption.

PMN J0948+0022.
Another prototype of the -ray emitting NLS1 galaxies is PMN J0948+0022 (Abdo et al. 2009a). Similar to 1H 0323+342, the optical spectrum of PMN 0948+0022 shows strong Fe ii multiplets and very weak narrow emission lines (Zhou et al. 2003). By using width of H and the optical luminosity, and the empirical scaling relations, the black hole mass of PMN J0948+0022 is estimated to be a few 10 7 M ⊙ (Zhou et al. 2003;Xin et al. 2022). The modeling of PMN J0948+0022's host galaxy image has shown a brightness profile with Sersic index of < 2, suggesting a disclike morphology. But unlike 1H 0323+342, no interaction feature is detected in this source (Olguín-Iglesias et al. 2020). Berton et al. (2019) have reported that the XMM-Newton and NuSTAR joint spectra are characterized by a soft excess on top of a hard power law with a hard photon index, indicating a combination of emission from the accretion disk and the jet.
We find, that a single power law (absorbed with Galactic absorption) is adequate to provide a good fit to the Swift X-ray spectra discussed here. In short exposures, multicomponent models cannot be distinguished with Swift. However, the majority of X-ray spectral indices we measure for PNM J0948+0022 are < 2, with Γ X as low as 1.1, indicating that its X-ray spectrum is hard and the X-ray emission is dominated by the jet most of the time. Only in very few cases we find steeper spectral indices up to 2.3, making a disc contribution likely at those spectral states. The X-ray flux and the ox of PMN J0948+0022 have a similar variability trend as seen for 1H0323+342 (upper panel of Figure 8).

PKS 1502+036
. This is the NLS1 galaxy with the lowest reported SMBH mass of the sample discussed here. A black hole mass of 4 × 10 6 M ⊙ was obtained when employing single-epoch optical spectroscopy and empirical scaling relations (Yuan et al. 2008). Its host galaxy image is well fitted by a disc-like galaxy (Olguín-Iglesias et al. 2020), however, an elliptical host was favored by D' Ammando et al. (2018) instead, and a much higher black hole mass of ∼ 7 × 10 8 M ⊙ was obtained from bulge luminosity in that study. There is a companion close to PKS 1502+036 in the band image but the source does not show any signs of interaction (Olguín-Iglesias et al. 2020).
The optical/UV is likely dominated by the emission from the accretion disc in most cases but could also by jet during flares when the optical/UV spectrum becomes apparently redder (Figure 9). Compared to other sources in the sample, PKS 1502+036 is relatively fainter in X-rays. This can be seen from its ox ( Figure 8) and its broad-band SED (Figure 9). The best fit to the X-ray spectrum of PKS 1502+036 is provided by a single power law with Galactic absorption either for individual spectra or stacked ones, consistent with the results reported by Berton et al. (2019), and no intrinsic absorption is detected. With respect of the broad band spectral variation, the ox and the X-ray flux in the band 2-10 keV still follow a similar trend as in 1H 0323+342 and PMN J0948+0022, but this trend is not seen for X-rays below 2 keV. More simultaneous optical/UV and X-ray monitoring with higher S/N on PKS 1502+036 is needed to explore the broad-band spectral evolution.  (Gokus et al. 2021;Yang et al. 2018). Although they were also identified as NLS1 galaxies, their spectra are characterized by much stronger Balmer emission lines from the narrow-line region (NLR) compared with other sources in the sample (see optical spectrum presented in Oshlack et al. 2001and Berton et al. 2021bfor PKS 2004−447, and in Yang et al. 2018and Järvelä et al. 2020 for SDSS J2118=0732). PKS 2004−447 is a compact steep-spectrum (CSS) radio source (Gallo et al. 2006). Its host galaxy appears to be a barred disc galaxy from the best-fit model to its infrared images (Kotilainen et al. 2016), while the SDSS image of the host galaxy of SDSS J2118−0732 is well resolved and shows the very characteristic structures of an ongoing galaxy interaction or merger (Yang et al. 2018;Komossa 2018). Based on the analysis of the SED and/or the X-ray spectra at past epochs, the X-rays of these two sources are likely dominated by emission from the jet and the disc only has little contribution if not absent (Kreikenbohm et al. 2016;Yang et al. 2018;Berton et al. 2019).
As shown in our SED of PKS 2004−447 and SDSS J2118−0732 (Figure 9), their X-ray spectra are very flat indicating the dominance of jet contribution, consistent with previous findings. In particular, the optical/UV spectra are very red, indicating that the emission from accretion disc contributes little to the optical/UV continuum if any. Instead, the optical/UV continuum is likely dominated by synchrotron emission from the jet for most of the time. There is no obvious trend for the broad-band spectral shape evolution for PKS 2004−447 based on its ox vs flux, and the data points are too few to perform this study for SDSS J2118−0732. More intense simultaneous multiband monitoring is required to explore the interplay between accretion and jet.
3.4.0.5 SDSS J0946+1017 and SDSS J1222+0413. These two NLS1 galaxies have spectral features similar to PMN J0948+0022 or PKS 1502+036, strong FeII and weak narrow emission lines, but they have black hole masses of a few 10 8 M ⊙ (Yao et al. 2015b(Yao et al. , 2019, significantly higher than the other sources. They are of particular interest, because they represent the high-redshift population of our sample, and are among the highest-redshift -ray emitting NLS1 galaxies identified so far (Yao et al. 2015b(Yao et al. , 2019Rakshit et al. 2021). Their host galaxies are not yet detected due to large distances ( ≈ 1) (Olguín-Iglesias et al. 2020). Based on the analysis of the SED and/or the X-ray spectra at past epochs, the X-rays of these two galaxies are likely dominated by emission from the jet and the disc only has little contribution if not absent (Yao et al. 2015b(Yao et al. , 2019Berton et al. 2019). Due to their high redshift, any soft X-ray accretion-disc contribution, even if intrinsically present, would be likely missed since shifted out of the observable band.
As shown in Figure 9, the Swift/UVOT has caught strong optical/UV spectral variability of SDSS J1222+0413, from a very blue spectrum at high flux state indicating a strong contribution from the accretion disc, to a red spectrum. This indicates that the accretion disc has faded strongly. Similarly, the UV spectra of SDSS J0946+1017 are very red during Swift observations although there is a lack of simultaneous optical detections.

DISCUSSION
Only a small number of NLS1 galaxies, about 20, has been detected in the -ray band so far (see Komossa 2018, for a review). These sources provide us with important new laboratories of the disc-jet connection in a previously unexplored parameter space. Swift in particular delivers simultaneous SEDs from the optical to the X-rays (e.g., Grupe et al. 2010;Yao et al. 2015a;D'Ammando 2020a). Since all -ray NLS1 galaxies are highly variable across the electromagnetic spectrum, important questions are if their SEDs show the same spectral components at all times, or if there is systematic trends, either with brightness state, or with time, and also if there are significant dependencies on black hole mass or on accretion rate / Edd where Edd is the Eddington accretion rate. An important question in this respect regards the contribution of thermal (accretion disc) versus non-thermal (jet-related) processes to their SEDs.
Of particular interest is the identification of NLS1 galaxies, that show both, accretion disc and jet components in their SEDs. Because NLS1 galaxies have, on average, lower SMBH masses and therefore faster accretion-disc variability timescales than their classical blazar counterparts, they are particularly well suited in studying the jetdisc symbiosis across short timescales (Komossa 2018). We have therefore analyzed all available Swift data of recent years for a small sample of -ray emitting NLS1 galaxies, and have also added their long-term Swift light curve evolution, and used these multiyear data for various spectral analyses.
Most of the NLS1 galaxies of our study have shown significant variability in optical/UV and X-rays, including events of rapid flaring. An exception is J0946+1017 which may be due to the sparsity of observations. In addition to the flux variability, there is also strong UV-optical and X-ray spectral variability in most of the sources. X-ray spectral index measurements reveal exceptionally flat X-ray spectra for most of the galaxies most of the time, in marked contrast to radio-quiet NLS1 galaxies that are characterized by ultra-steep soft X-ray spectra (e.g., Puchnarewicz et al. 1992;Grupe et al. 1998), interpreted as signs of reprocessing from an accretion disc and/or ionized absorption of the intrinsic emission (see Komossa 2008, for a review). The exceptionally flat X-ray spectra therefore already indicate a strong contribution of non-thermal jet emission in the X-ray band in the majority of sources.
The different patterns of the soft and hard X-ray flux variability of the NLS1 galaxies studied here, and their SEDs at different epochs, are evident in Figure 9. However, there is no homogeneous trend throughout the sample. The optical/UV spectra become bluer when the flux increases in 1H 0323+342 and SDSS J1222+0413. The Xray spectral slope of 1H 0323+342 becomes slightly steeper when flux increases. Considering that the optical/UV emission of these two sources is dominated by the contribution from a strong accretion disc component based on the modeling and decomposition of their broad band SEDs (e.g. Abdo et al. 2009b;Yao et al. 2015bYao et al. , 2019, the spectral variability behavior in these two sources is likely due to the variable emission from the accretion disc. For PMN J0948+0022 and PKS 1502+036, the decomposition of their SEDs has revealed the dominance of strong thermal emission in the optical/UV from the accretion disc by previous work, whereas non-thermal emission from the jet component likely dominates their X-ray spectra (Abdo et al. 2009b;Foschini et al. 2012;Paliya & Stalin 2016). At peak luminosities the optical/UV spectra of PMN J0948+0022 and PKS 1502+036 become obviously redder, suggesting that the brightening is mainly due to the increasing contribution from emission peaked in the radio/infrared band, i.e., synchrotron emission from the jet (e.g. D' Ammando et al. 2015Ammando et al. , 2016Paliya & Stalin 2016). This is in stark contrast to the variability behavior reported above of 1H 0323+342 and SDSS J1222+0413.
At selected epochs, a decomposition of the broad band SED of PKS 2004−447 (e.g. Abdo et al. 2009b;Gokus et al. 2021) and SDSS J2118−0732 (Yang et al. 2018) has shown that the contribution from an accretion disc is very weak at those epochs. Here, we find that PKS 2004−447 and SDSS J2118−0732 have red optical/UV spectra and very hard X-ray spectra, indicating that the emission peaks at infrared and -rays, respectively, and that the emission from the jet dominates at all observed frequencies.

SUMMARY AND CONCLUSIONS
In this work we have studied the Swift monitoring data up to 2021 December of seven bright -ray detected NLS1 galaxies. The simultaneous optical/UV and X-ray observations by Swift provide good data sets to investigate the variability of the emission from the coupled accretion and jet process. Among the sample, the accretion disc dominates the emission in the optical/UV and makes significant contributions to the X-rays of 1H 0323+342 as derived from its X-ray spectra and broad-band SEDs. The jet contribution may only be significant in hard X-rays. In other sources, the optical/UV emission is mainly from the accretion disc but could also be dominated by synchrotron emission during flares, while their X-ray spectra are always dominated by the emission from the jet and the accretion disc makes very little contribution even though some of the galaxies have similar black hole masses and host galaxies and some do not. A decreasing trend of ox with increasing X-ray flux can be seen in 1H 0323+342, PMN J0948+0022 and PKS 1502+036 although the main contributors to X-rays in these source are different. However, this trend is not seen in PKS 2004-447 (Figure 8). There is no adequate number of simultaneous UV and X-ray observations for the other three sources to explore this trend. More simultaneous multifrequency monitoring observations of these -ray detected NLS1 galaxies are needed to investigate their optical/UV and X-ray emission variability in order to study the accretion disk/jet evolution in this rare species.