Identifier
stringlengths 37
82
| Paragraph
stringlengths 1.95k
9.23k
| Citation Text
list | Functions Text
list | Functions Label
list | Citation Start End
list | Functions Start End
list |
|---|---|---|---|---|---|---|
2021ApJ...917...24Z__Hall_&_Evans_2019_Instance_1
|
As Advanced LIGO and Virgo will upgrade while additional interferometers (KAGRA and LIGO-India) come online in the future, the detection horizon would be several hundred megaparsecs for BNS and BH–NS mergers with the localization accuracy reaching ∼10 deg2 for a network of these 2nd generation GW detectors (Abbott et al. 2018). In contrast, higher localization accuracy would be obtained by networking of 3rd generation GW detectors (Zhao & Wen 2018; Chan et al.2018; Hall & Evans 2019; Maggiore et al. 2020). In these regimes, present and future wide-field-of-view survey projects will be able to cover the sky localization given by GW detections in a few pointings and achieve deep detection depths with relatively short exposure integration times. The detection of GW170817/AT 2017gfo was made in an i-band filter and subsequently confirmed by larger telescopes (Coulter et al. 2017). Similar to this case, we assume that one can use only one filter to make follow-up searches after GW triggers. With quick response times of follow-up observations after GW events, we also assume that the survey telescopes would not miss the peaks of kilonovae in each band. Therefore, by assuming that all GW detectors in each era can operate normally (i.e., the most optimal situation for detector networks of different generations), the detection rates for target-of-opportunity observations for a given X-band limiting magnitude mX,limit can be expressed as
15
N
X
=
∫
0
z
max
ρ
m
X
,
limit
(
z
)
1
+
z
dV
(
z
)
dz
dz
,
where
z
max
is the maximum detectable distance and
ρ
m
X
,
limit
(
z
)
represents the cosmological event rate densities of the BH–NS mergers that can be triggered by GW detectors and be detected by telescopes in band X. Here, the astronomical magnitude system we adopted is the AB magnitude system, i.e.,
m
ν
=
−
2.5
log
(
F
ν
/
3631
Jy
)
.
|
[
"Hall & Evans 2019"
] |
[
"In contrast, higher localization accuracy would be obtained by networking of 3rd generation GW detectors"
] |
[
"Compare/Contrast"
] |
[
[
470,
487
]
] |
[
[
330,
434
]
] |
2021AandA...646A..96C__Freundlich_et_al._2019_Instance_1
|
Observations of cold molecular gas are more promising since any activity from the AGN (radiation, outflows, or jets) will have an impact on the molecular gas reservoir first, and then on the SFR, which previous studies have focused on. The molecular gas provides an instantaneous measure of the raw fuel from which stars form and can be used as a more direct tracer of potential feedback effects. Over the last decade, large observational efforts have been devoted to map the molecular gas reservoir of galaxies (e.g., Daddi et al. 2010a; García-Burillo et al. 2012; Bauermeister et al. 2013; Bothwell et al. 2013; Tacconi et al. 2013, 2018; Genzel et al. 2015; Silverman et al. 2015; Decarli et al. 2016; Cicone et al. 2017; Saintonge et al. 2017; Fluetsch et al. 2019; Freundlich et al. 2019). These studies are largely based on observations of carbon monoxide (CO) rotational emission lines, used as a tracer of cold molecular hydrogen H2 (the ground-state rotational transition in particular traces the total molecular gas best), but there are also many studies based on dust emission (e.g., Tacconi et al. 2018 and references therein). The main targets of such CO campaigns have primarily been inactive1 galaxies that mostly lie on the main sequence of star-forming galaxies (e.g., Noeske et al. 2007; Schreiber et al. 2015), where the majority of the cosmic star-formation activity occurs. The fundamental relation between SFR and the molecular gas content of galaxies, the Schmidt-Kennicutt relation, provides precious information about how efficiently galaxies turn their gas into stars (Schmidt 1959; Kennicutt 1989). This star-formation law is usually presented in terms of surface densities, therefore requiring spatially resolved measurements of galaxies. However, in high-redshift studies, an integrated form of this relation with global measurements of the SFR and molecular mass is normally used (e.g., Carilli & Walter 2013; Sargent et al. 2014).
|
[
"Freundlich et al. 2019"
] |
[
"Over the last decade, large observational efforts have been devoted to map the molecular gas reservoir of galaxies (e.g.,",
"These studies are largely based on observations of carbon monoxide (CO) rotational emission lines, used as a tracer of cold molecular hydrogen H2 (the ground-state rotational transition in particular traces the total molecular gas best),"
] |
[
"Background",
"Background"
] |
[
[
771,
793
]
] |
[
[
397,
518
],
[
796,
1033
]
] |
2020MNRAS.492.5675C__Pérez-Montero_et_al._2019_Instance_1
|
In regarding AGNs, the Te method tends to underestimate the oxygen abundance by an average value of about 0.6 dex in comparison to estimations based on strong-line methods and it produces subsolar O/H values for most of these objects (Dors et al. 2015; Dors, Freitas-Lemes & Âmores 2020). An alternative method to derive the metallicity or abundances in the nuclear regions of spiral galaxies is the extrapolation of the radial oxygen abundance. Along decades, results based on this indirect method have indicated Z near or slightly above the solar value in nuclear regions (Vila-Costas & Edmunds 1992; Zaritsky, Kennicutt & Huchra 1994; van Zee et al. 1998; Pilyugin, Vílchez & Contini 2004; Gusev et al. 2012; Dors et al. 2015; Zinchenko et al. 2019), in consonance with predictions of chemical evolution models (e.g. Mólla & Díaz 2005) and with the use of strong-line methods (e.g. Groves, Dopita & Sutherland 2004; Groves, Heckman & Kauffmann 2006; Feltre, Charlot & Gutkin 2016; Pérez-Montero et al. 2019; Thomas, Kewley & Dopita 2019; Dors et al. 2020). Therefore, Temethod does not seem to work for AGNs. The origin of the discrepancy between Z values calculated via Te method and via strong-line methods, the so-called Teproblem, could be attributed, in part, to the presence of heating/ionization by gas shock in the narrow-line region (NLR) of AGNs. In fact, Contini (2017) carried out detailed modelling of AGN optical emission lines by using the SUMA code (Contini & Aldrovandi 1983) and suggested the presence of gas shock with low velocity ($v \: \lesssim \: 400 \: \rm km \:s^{-1}$) in a sample of Seyfert 2 nuclei. This result is supported by recent spatially resolved observational studies of Seyfert 2 nuclei, in which the presence of gas outflows with velocity of the order of 100–300 $\rm km \: s^{-1}$ have been found (e.g. Riffel, Storchi-Bergmann & Riffel 2017; Riffel, Hekatelyne & Freitas 2018). Moreover, the Te problem can also be originated due to the use of an unappropriate calculation of the ionization correction factor (ICF) for oxygen in AGNs (Pérez-Montero et al. 2019; Dors et al. 2020).
|
[
"Pérez-Montero et al. 2019"
] |
[
"Along decades, results based on this indirect method have indicated Z near or slightly above the solar value in nuclear regions",
"in consonance with predictions of chemical evolution models",
"and with the use of strong-line methods"
] |
[
"Similarities",
"Similarities",
"Similarities"
] |
[
[
984,
1009
]
] |
[
[
446,
573
],
[
754,
813
],
[
839,
878
]
] |
2017MNRAS.471...80S__Kalugina_et_al._2014_Instance_2
|
The electronic computations were performed using the molpro (Molpro 2015, http://www.molpro.net) package. In a preliminary work, we used the complete active space self-consistent field (Knowles & Werner 1985; Werner & Knowles 1985) to examine the electronic wavefunction of the HNCO–He complex. These computations showed that this wavefunction is dominantly described by a unique electron configuration (with a weight ≥0.93) over the grid used for the generation of this PES. This justifies hence the use of monoconfigurational ab initio methods. Accordingly, we applied the recently established methodology by Hochlaf and co-workers for mapping multidimensional PESs of weakly bound molecular systems with high accuracy and relatively low computational cost (Lique, Klos & Hochlaf 2010; Halvick et al. 2011; Ajili et al. 2013; Mathivon, Linguerri & Hochlaf 2013; Kalugina et al. 2014; Mogren Al Mogren et al. 2014). Briefly, these electronic computations were carried out using the explicitly correlated coupled cluster method with single, double and perturbative treatment of triple excitations (CCSD(T)-F12) (Adler, Knizia & Werner 2007; Knizia, Adler & Werner 2009) in connection with the augmented correlation-consistent aug-cc-pVTZ basis set of Dunning and co-workers (Dunning 1989; Kendall, Dunning & Harrison 1992). In addition, molpro default choices for the density fitting and resolution of identity basis sets have been applied (Yousaf & Peterson 2008). Benchmarks by Hochlaf and co-workers (Lique et al. 2010; Halvick et al. 2011; Ajili et al. 2013; Mathivon et al. 2013; Kalugina et al. 2014; Mogren Al Mogren et al. 2014) showed that results obtained from this highly correlated approach are close to those deduced using standard coupled cluster techniques extrapolated to the complete basis set (CBS) limit, whereas a strong reduction in CPU time and disc occupancy are observed. For illustration, we performed CCSD(T)/aug-cc-pVXZ calculations (X = D, T, Q, 5) on the HNCO—He cluster. Then the energies were extrapolated to the CBS limit. The comparison between CCSD(T)-F12/aug-cc-pVTZ and CBS calculations is given in Table 1. It shows that the CCSD(T)-F12/aug-cc-pVTZ results are off by 4 per cent (at the maximum) with those deduced from CBS extrapolation. We compare also our results with those done using the CCSD(T)/aug-cc-pV5Z approach. We can clearly see that the CCSD(T)-F12/aug-cc-pVTZ approach offers a good agreement with the CCSD(T)/aug-cc-pV5Z calculations with a very reduced computational cost.
|
[
"Kalugina et al. 2014"
] |
[
"Benchmarks by Hochlaf and co-workers",
"showed that results obtained from this highly correlated approach are close to those deduced using standard coupled cluster techniques extrapolated to the complete basis set (CBS) limit, whereas a strong reduction in CPU time and disc occupancy are observed."
] |
[
"Similarities",
"Similarities"
] |
[
[
1585,
1605
]
] |
[
[
1466,
1502
],
[
1637,
1895
]
] |
2021AandA...653A..83V__Krühler_et_al._(2015)_Instance_1
|
The comparisons show that except for GRBs 060707 and 060605, which have a very high vpeak, the LAE-LGRBs fall in the same parameter space as LyC leakers and follow the correlations between the indirect indicators found by Verhamme et al. (2017). Following their study, we can cut these plots into two regions corresponding to strong LyC leakers (fesc(LyC) > 5%, red rectangle) and weak LyC leakers (fesc(LyC) 5%, blue rectangle). All LAE-LGRBs fall in the category of the weak LyC leakers except for GRB 021004, which appears systematically to agree well with the region of the strong leakers. In panel b of Fig. 12, we also superimpose the distribution of [OIII]/[OII] ratio for the GRB sample of Krühler et al. (2015) and GRB 081121 reported here. For the majority of the GRBs, this ratio is about two, with seven cases at O32 > 4. GRB 021004 has a high value of O32 > 10 and is the strongest LAE of our golden sample (with fesc(Lyα) of 60%) in agreement with potential high fesc(LyC). Nevertheless, the Lyα profile of GRB 021004 is a single peak with no residual flux at the Lyα line center. This is not the typical line shape observed for confirmed LyC emitters, which have the tendency to show double- or triple-peak profiles (Verhamme et al. 2017; Rivera-Thorsen et al. 2017; Vanzella et al. 2020). Our shell-model fitting also suggests that the column densities are too high to allow LyC photons to escape. However, this model already predicted a similar HI column density (log(NHI/cm−2) ≈ 19−20) for four green pea galaxies out of the five with detected LyC emission reported in Yang et al. (2017). This suggests that LyC emission can escape through holes in the ISM even if the Lyα photons probe denser neutral gas. The only LAE-LGRB for which LyC leakage has been detected along the LGRB line of sight (
f
esc
(
LyC
)
=
0
.
35
−
0.11
+
0.10
$ f_{\mathrm{esc}}(\rm LyC) = 0.35^{+0.10}_{-0.11} $
) is GRB 191004B (Vielfaure et al. 2020). However, Fig. 12 (panel c) shows that it does not fall in the high escape fraction region, but in the lower left area. The reasons could be that (i) fesc(LyC) is lower at the scale of the galaxy than along the LGRB line of sight, and (ii) the indicators of strong LyC leakage evolves with redshift. As a comparison, the LyC emitters from Fletcher et al. (2019) are found out of the high escape region (red rectangle). They show lower rest-frame EW(Lyα) and higher vpeak than the local LyC emitters, whereas their escape fraction of ionizing photons is significantly higher (fesc(LyC) = 15 − 60%). This could also suggest that strong LyC leakers span a wider parameter space than predicted by the study of local LyC emitters. Overall, it is clear that this type of studies is still limited by the poor statistics, and the current results show the difficulty of characterizing LyC leakage based on these properties alone.
|
[
"Krühler et al. (2015)"
] |
[
"In panel b of Fig. 12, we also superimpose the distribution of [OIII]/[OII] ratio for the GRB sample of",
"and GRB 081121 reported here.",
"For the majority of the GRBs, this ratio is about two, with seven cases at O32 > 4. GRB 021004 has a high value of O32 > 10 and is the strongest LAE of our golden sample (with fesc(Lyα) of 60%) in agreement with potential high fesc(LyC). Nevertheless, the Lyα profile of GRB 021004 is a single peak with no residual flux at the Lyα line center."
] |
[
"Uses",
"Uses",
"Compare/Contrast"
] |
[
[
699,
720
]
] |
[
[
595,
698
],
[
721,
750
],
[
751,
1095
]
] |
2018MNRAS.475.1160H__Werk_et_al._2013_Instance_2
|
Galaxies are surrounded by vast gaseous haloes which extend well beyond the hosts’ stellar components: Early observations of quasar sight lines attributed the presence of absorption at multiple intermittent redshifts to gaseous haloes of intervening galaxies (e.g. Bergeron 1986; Bergeron & Boissé 1991; Lanzetta et al. 1995; Tripp, Savage & Jenkins 2000; Chen, Lanzetta & Webb 2001). In the past decade, owing to the rise of large spectroscopic surveys of galaxies with well-determined physical properties (e.g. SDSS), all sky UV surveys (e.g. GALEX), and improved sensitivity of UV spectrographs (e.g. COS), studies of the gaseous haloes of galaxies could systematically connect gas absorption properties to galaxy properties in statistically meaningful samples (e.g. Cooksey et al. 2010; Prochaska et al. 2011; Tumlinson et al. 2013; Liang & Chen 2014; Lehner, Howk & Wakker 2015). The aforementioned gaseous haloes are commonly referred to as the circum-galactic medium (CGM) and are ubiquitous in galaxies regardless of mass or star formation activity: even sub-L* galaxies (Bordoloi et al. 2014), and passive galaxies host a CGM (Thom et al. 2012). The current model of the CGM suggests the presence of a clumpy multiphase medium which extends beyond the virial radius of the host galaxy, with a declining radial density profile, containing a substantial amount of gas and metals (e.g. Werk et al. 2013, 2014; Liang & Chen 2014; Lehner et al. 2014, 2015; Prochaska et al. 2017). Observational studies targeting the CGM of L* galaxies showed that the CGM gas content is comparable to the mass of the interstellar medium (ISM; e.g. Chen et al. 2010; Tumlinson et al. 2011; Werk et al. 2014; Prochaska et al. 2017) and correlates positively with ISM properties (Borthakur et al. 2015). Additionally, CGM observations infer a significant amount of metals (e.g. Werk et al. 2013; Peeples et al. 2014) where CGM metallicities can extend to supersolar metallicities (Prochaska et al. 2017). The clumpy multiphase CGM consists of a warm gas T ∼ 104 − 5 K (clumpy in nature) embedded within a hot diffuse T ∼ 106 K medium (e.g. Heitsch & Putman 2009; Armillotta et al. 2017; Bordoloi et al. 2017). The multiphase structure of the CGM is corroborated by the variety of observed ionic species which survive at a vast range of temperatures: While the warm gas hosts the low ionization species (e.g. H i, Si ii, Si iii, C ii, C iv), the hot medium is home for the most highly ionized species (e.g. O vi, O vii). Additionally, the spectral line profiles of absorbers in the CGM can be reproduced by invoking a patchy medium (e.g. Stern et al. 2016; Werk et al. 2016), i.e. multiple high density gas clouds contribute to the optical depth along the line of sight thus leaving their kinematic imprint on the absorption line profile. For a review of the CGM, see Putman, Peek & Joung (2012) and Tumlinson, Peeples & Werk (2017).
|
[
"Werk et al. 2013"
] |
[
"Additionally, CGM observations infer a significant amount of metals (e.g."
] |
[
"Background"
] |
[
[
1863,
1879
]
] |
[
[
1789,
1862
]
] |
2019AandA...630A..30L__Hässig_et_al._(2015)_Instance_2
|
The many unexpected surprises of comet 67P/Churyumov-Gerasimenko (hereafter 67P) revealed by the historic Rosetta mission highlight the importance of observing the evolution of comets throughout their orbits. One of the surprises was the drastic heterogeneity in both the major and minor volatile species in the coma that was observed early on in the mission (Hässig et al. 2015; Luspay-Kuti et al. 2015, hereafter ALK15). When Rosetta first arrived at comet 67P in August 2014, the Rosetta Orbiter Mass Spectrometer for Ion and Neutral Analysis/Double Focusing Mass Spectrometer (ROSINA/DFMS; Balsiger et al. 2007) detected large diurnal variations in the intensity profiles of various species in the coma from distances to the comet as far as 250 km. At this time, 67P was still at a distance of about 3 AU and inbound from the Sun. The intensity variations in the major and minor volatile species were found to be periodic, and were dependent on both the observing sub-spacecraft latitude and longitude (Hässig et al. 2015; Luspay-Kuti et al. (2015)). As reported in Hässig et al. (2015), the intensity of H2O in the coma dominated the overall signal, with maxima in the H2O signal every ~6 h, about twice during a rotation. Interestingly, however, CO2 and CO displayed a separate additional maximum when the H2O signal was near its minimum. This independent maximum in CO2 and CO only occurred at negative observing latitudes that are associated with a particular “view” of Rosetta at 67P, with the larger lobe blocking out the neck and head. At this time, 67P had not yet reached its first equinox (10 May 2015), and the poorly illuminated southern hemisphere was experiencing winter. In addition, the largest H2O activity was localized at the well-illuminated neck region, as also seen by the Microwave Instrument on the Rostta Orbiter (MIRO; Gulkis et al. 2015; Biver et al. 2015; Lee et al. 2015) and by the Visible InfraRed Thermal Imaging Spectrometer (VIRTIS; Bockelée-Morvan et al. 2015; Migliorini et al. 2016). VIRTIS also measured weak H2O production in regions with low solar illumination, while CO2 was outgassing from both illuminated and non-illuminated regions pre-inbound equinox (Bockelée-Morvan et al. 2015; Migliorini et al. 2016; Fink et al. 2016). The observed outgassing pattern of the major cometary species suggested that CO and CO2 may be sublimating from a depth below the diurnal skin depth.
|
[
"Hässig et al. 2015"
] |
[
"The intensity variations in the major and minor volatile species were found to be periodic, and were dependent on both the observing sub-spacecraft latitude and longitude"
] |
[
"Background"
] |
[
[
1007,
1025
]
] |
[
[
835,
1005
]
] |
2022MNRAS.513.5245A__Done_&_Jin_2016_Instance_2
|
We assume the time-scales we observe here are generated in the corona itself (and note that longer time-scale changes will be driven by the disc outside of the corona) and are made visible by a changing electron temperature and density as a result of local turbulence and coupling to mass accretion rate propagations through the flow from rout to rin. We note that we can discount variations in the seed photon population as the driver for changes in the power spectrum, as the UV emission from the disc is established to be considerably less variable than the corona in NLS1s (Leighly 1999; Smith & Vaughan 2007; Ai et al. 2013; Alston, Vaughan & Uttley 2013; Done & Jin 2016). We assume that the variability generated locally at each radius (rν) is at the viscous frequency (see Churazov et al. 2001) such that
(9)$$\begin{eqnarray}
r_{\nu } = \left[ \frac{2\pi \nu }{\alpha }\left(\frac{H}{R}\right)^{-2}\right]^{-2/3} ,
\end{eqnarray}$$where the frequency is in units of c/Rg (see e.g. Kato, Fukue & Mineshige 1998; Arévalo & Uttley 2006). In the above, α and $\frac{H}{R}$ are the dimensionless viscosity parameter of Shakura & Sunyaev (1973) and scale height of the accretion disc, respectively. We assume that our frequency range of interest, 0.01−1 mHz (i.e. the range over which we can practically fit to the data) corresponds to radii between the ISCO (rin) and some radius within the true outer edge of the corona (i.e. rout ≤ rcorona). The actual frequencies generated in our model therefore depend on the SMBH spin and the combination $\left(\frac{H}{R}\right)^2 \alpha$ (and somewhat on the SMBH mass – although here the range is small). Given the reported high spin values for these bright AGNs (Ogle et al. 2004; Fabian et al. 2013; Done & Jin 2016; Kara et al. 2017; Buisson et al. 2018b), we expect the ISCO to sit at ∼1.25Rg. For the corona at the ISCO to produce variability above our upper frequency limit of 1 mHz requires $\left(\frac{H}{R}\right)^2 \alpha \gtrsim 0.01$. We note that for the mass range subtended by our AGN sample (from 106.00 to 106.63 M⊙, see Table 2), should we instead assume zero spin (rin = 6Rg), the viscous frequency at rin is lower and we have strong curvature in our observed bandpass.
|
[
"Done & Jin 2016"
] |
[
"Given the reported high spin values for these bright AGNs",
"we expect the ISCO to sit at ∼1.25Rg."
] |
[
"Uses",
"Uses"
] |
[
[
1751,
1766
]
] |
[
[
1654,
1711
],
[
1809,
1846
]
] |
2015MNRAS.450.3458C__Cichowolski_et_al._2001_Instance_4
|
The kinetic energy stored in the CO shell can be estimated as $E_{\rm kin} = 0.5\, M_{\rm shell}\, V^2_{\rm exp}$, where Vexp is the expansion velocity of the shell and Mshell is the total (molecular, atomic, and ionized) shell mass. Adopting an expansion velocity equal to half the velocity interval where the structure is observed, Vexp = 7.0 ± 1.3 km s− 1 , the molecular mass given in Table 1 and the atomic and ionized masses estimated by Cichowolski et al. (2001), 1450 and 3000 M⊙, respectively, we obtain Ekin = (2.5 ± 1.0) × 1049 erg, assuming a 40 per cent error for the masses.. Although Cichowolski et al. (2001) concluded that WR 130 could have alone created the observed structure, it is important to note that they did not take into account the molecular mass present in the shell, which considerably increases the kinetic shell energy. Thus, we can compare now the new value obtained for Ekin with the mechanical energy deposited in the ISM by the wind of the WR star, Ew = (0.7–2.2) × 1050 erg (Cichowolski et al. 2001). We obtain ϵ = Ekin/Ew = 0.007–0.5. The ratio ϵ measures the energy conversion efficiency in the shell, and according to evolutionary models ϵ ≤ 0.2 (Koo & McKee 1992). Thus, not all the possible values of ϵ are compatible with the scenario where the energy injected during the WR phase is enough to create the structure. In this case, the contribution of the energy injected during the O-star phase and/or other massive stars, should be considered. As mentioned in the Introduction, WR 130 is a WNH star, and according to Smith & Conti (2008) its age would be of about 2–3 Myr and its initial mass of at least 60 M⊙. A rough estimation of the energy injected by such a star during its main sequence yields Ew = (2.5–3.5) × 1050 erg (de Jager, Nieuwenhuijzen & van der Hucht 1988), which would be enough to create the observed structure. We have nevertheless looked for the presence of other massive stars in the region. We queried the available catalogues such as the Galactic O-Star Catalog (Maíz Apellániz et al. 2013), the Early-Type Emission-Line Stars Catalogue (Wackerling 1970), the Catalogue of Be stars (Jaschek & Egret 1982), the H-alpha Stars in the Northern Milky Way Catalogue (Kohoutek & Wehmeyer 1997), and the Catalog of Galactic OB Stars (Reed 2003), for early-type and emission stars. No stars were found in any catalogue. The only massive star located nearby is, as mentioned by Cichowolski et al. (2001), an OB star, which has an uncertain spectral type and no distance estimate (Stock, Nassau & Stephenson 1960). It is located in projection not in the centre of the structure but on to the shell (there is a second OB star mentioned by Cichowolski et al. 2001 but its location is actually outside the structure, see fig. 1 of Cichowolski et al. 2001). Although we cannot completely rule out the possibility that the OB star may be playing a role in creating the shell structure, we think that the action of WR 130 is sufficient and most likely dominant in the region.
|
[
"Cichowolski et al. 2001"
] |
[
"Thus, we can compare now the new value obtained for Ekin with the mechanical energy deposited in the ISM by the wind of the WR star, Ew = (0.7–2.2) × 1050 erg"
] |
[
"Compare/Contrast"
] |
[
[
1012,
1035
]
] |
[
[
852,
1010
]
] |
2020MNRAS.491..560D__Straten_&_Manchester_2004_Instance_1
|
As MKT J170456.2–482100 is coincident with a star in a spectroscopic binary, it is possible that the companion star may be a pulsar or pulsating white dwarf, similar to AR Scorpii (Marsh et al. 2016). To investigate this possibility, we performed high-time resolution observations to search for pulsations over a range of periods spanning milliseconds to minutes. We observed the position of MKT J170456.2–482100 on utc 2019 January 30 with the ultra-wide-bandwidth (UWL) receiver deployed on the 64-m Parkes radio telescope in Australia. The output of the receiver is processed by the Medusa Graphics Processing Unit (GPU) cluster which produces 8-bit, full-Stokes filter banks with spectral and temporal resolutions of 1 MHz and 128 μs respectively over the $3328\, \mathrm{MHz}$ band from 0.740 to $4.03\, \mathrm{GHz}$. The data were recorded in psrfits format and processed using PSRCHIVE tools (Hotan, van Straten & Manchester 2004). Since our data were strongly affected by radio frequency interference (RFI) we used the clfd8 package described in (Morello et al. 2019) to perform more sophisticated RFI mitigation. Since radio waves are dispersed by free electrons in the interstellar medium along the line of sight, the data need to be corrected for this dispersion delay before any analyses can be made. This dispersion delay can be quantified by the dispersion measure (DM), which is the integrated free electron column density along the line of sight. We dedispersed the data over a range of trial DMs, $0.0 \le \mathrm{DM} \le 30.0$ pc cm$^{-3}$ in steps of 1 pc cm$^{-3}$ to account for the uncertainty in the distance to TYC 8332-2529-1. The DM range was estimated using various distances to the source (Bailer-Jones et al. 2018), and the NE2001 model (Cordes & Lazio 2002). For each trial DM, the resulting dedispersed time series was searched for long and short-period pulsations using both a Fast Folding Algorithm (FFA) and a Fast Fourier Transform (FFT), respectively.
|
[
"Hotan, van Straten & Manchester 2004"
] |
[
"The data were recorded in psrfits format and processed using PSRCHIVE tools"
] |
[
"Uses"
] |
[
[
901,
937
]
] |
[
[
824,
899
]
] |
2016AandA...585A..76W__Neau_et_al._2000_Instance_1
|
The median value of the column density of OH in the first quadrant is 3.9 × 1014 cm-3 and for OH+ it is 0.68 × 1014 cm-3. In the fourth quadrant, the median column densities amount to 1.7 × 1014 cm-3 and 0.55 × 1014 cm-3, respectively. Column densities in excess of 1015 cm-2 (for OH) and above ~1014 cm-2 (for OH+) are rather the exception. The median N(OH) /N(OH+) ratio over all sightlines and velocity components is 3.3. While the formation of OH+ results from cosmic-ray induced reactions involving atomic and molecular hydrogen and atomic oxygen, the bottleneck expected for the formation of OH via ion-neutral chemistry is the availability of H2. The reaction of OH+ with H2 yields H3O+, via the two hydrogen abstraction reactions OH+(H2,H)H2O+(H2,H)H3O+ (see Appendix C). Then the dissociative recombination of H3O+ yields OH and H2O, with a branching ratio of ~74% to 83% in favor of OH (determined by ion storage ring experiments, Jensen et al. 2000; Neau et al. 2000), while less than ~1% forms O i . In the following, we attempt to confirm these predictions, that is, the bottleneck reaction OH+(H2,H)H2O+, and the N(OH) /N(H2O) ratio. As for the former, a strong anticorrespondence between the column densities of OH and OH+might naively be expected, where the availability of H2 tips the scales in favor of OH, while OH+ traces predominantly atomic gas (Hollenbach et al. 2012, further references therein). But even if a clear anticorrelation between OH and OH+ existed, it would be impossible to observe it. On a given sightline several clouds with high and low molecular hydrogen fractions \hbox{$f^{\rm N}_\HH = N(\HH)/(2N(\HH)+N($}fH2N=N(H2)/(2N(H2)+N(H i )) line up. Even across a single diffuse cloud, the N(OH) /N(OH+) ratio is expected to vary substantially, depending on the degree of self-shielding of H2 against the interstellar UV radiation field. To quantify the anticorrelation, we normalized the velocity-specific OH column density with the total OH and OH+ reservoir and obtained an abundance ratio r = Nv(OH)/(Nv(OH) + Nv(OH+) varying from zero (only OH+, no OH) to one, where all the OH+ abundance is exhausted owing to the formation of OH and (see below) water. (The normalization chosen here avoids the divergence of the distribution if OH has no spectral counterpart in OH+.) The resulting distribution (Fig. 12) indeed shows that these extremes are present in the data, although the second case is by an order of magnitude more frequent. We suggest two explanations for this. One reason is that if \hbox{$f^{\rm N}_\HH$}fH2N is too small, OH+ can be efficiently destroyed by the dissociative recombination with free electrons (Appendix C), while the formation of OH+ by the reaction chain H+(O,H)O+(H2,H)OH+ and the secondary, less important path H2+(H2,H)H3+(O,H2)OH+ become less efficient (see Appendix D) because less H2 is available. Another reason is that the fraction \hbox{$f^{\rm N}_\HH$}fH2N is larger in denser gas (cf. Table 3) where column densities are higher and absorption features easier to observe.
|
[
"Neau et al. 2000"
] |
[
"determined by ion storage ring experiments,"
] |
[
"Background"
] |
[
[
961,
977
]
] |
[
[
897,
940
]
] |
2017AandA...605A..20C__Momjian_&_Sarma_2017_Instance_1
|
In principle, when ΔνZ>δν, with δν being the observed full width at half maximun (FWHM) of the line, the complete set of information concerning the magnetic field B can be derived. However, for most of the Zeeman detections, the Zeeman splitting ΔνZ turns out to be significantly smaller than δν. Indeed, even if we neglect the non-thermal line broadening due to typical processes occurring in star forming regions such as jets (with velocities up to hundreds of km s-1), outflows, and accreting/rotating disks (typically ≤10 km s-1), already the thermal broadening (even at kinetical temperatures lower than 10 K) is definitely larger than the Zeeman splitting (e.g., Frank et al. 2014). For instance, a typical linewidth of the coldest starless cores is ~0.1 km s-1 (e.g., di Francesco et al. 2007), indeed too broad to allow an observer to unveil the Zeeman effect. The expected B values in star forming regions range from ~100 μG, for low-mass objects (Li et al. 2014), to ~1 mG, as measured in regions hosting high-mass young stars (e.g., Pillai et al. 2016; Momjian & Sarma 2017). If we consider the N,J = 1, 1 ← 0, 1 transition (286.3 GHz), for which the largest Zeeman shift was obtained (Z = 0.87 Hz/μG), and we assume that B varies in the 0.1–1 mG range, we derive ΔνZ≃ values in the 87–870 Hz range, that is, 0.3–3 m s-1. Moving to higher frequencies and considering the SO N,J = 2, 2 ← 1, 2 transition at 309.5 GHz, we derive ΔνZ ≃ 61−610 Hz, that is, 0.06−0.59 m s-1. In conclusion, in such cases, only information about the LoS component of B, B∥, is obtained. In fact, the Stokes parameter Q and U spectra are usually too weak to be detected. The analysis is therefore limited to the Stokes V spectrum, which has the shape of the first derivative of the Stokes I spectrum and is proportional to (ΔνZ/δν)2 × B∥. According to Crutcher et al. (1993), Crutcher (2012), and Uchida et al. (2001), the standard procedure is to fit dI/dν to the observed V spectrum, with B∥ being the free parameter to be determined from the strength of the V spectrum. Furthermore, the direction (toward or away from the observer) of B∥ is obtained. Therefore, accurate laboratory (either experimental or theoretical) determination of the g factors might provide the opportunity to complete the missing information.
|
[
"Momjian & Sarma 2017"
] |
[
"The expected B values in star forming regions range from ~100 μG, for low-mass objects",
", to ~1 mG, as measured in regions hosting high-mass young stars (e.g.,"
] |
[
"Uses",
"Uses"
] |
[
[
1065,
1085
]
] |
[
[
870,
956
],
[
973,
1044
]
] |
2020ApJ...898...92C__Lawson_et_al._2012_Instance_1
|
Neutral MF occurs as syn and anti conformers (also denoted Z and E rotamers) with Cs symmetry, with the syn rotamer being more stable (by ∼20 kJ mol−1) and well separated from the anti rotamer by a significant energy barrier of ∼35 kJ mol−1 for anti
syn isomerization (Wilmshurst 1957; Curl 1959; Blom & Günthard 1981; Müller et al. 1983; Bagno & Scorrano 1996; Neill et al. 2011, 2012; Zeegers-Huyskens & Kryachko 2011; Ferro-Costas & Mosquera 2013). The zero-kinetic-energy photoelectron spectrum of MF provides an accurate estimate for its adiabatic ionization energy and vibrational frequencies of the radical cation ground state (Waterstradt et al. 1994). The electronic excitation spectrum of MF is very similar to that of acetic acid featuring three valence transitions (Nunes et al. 2010). In contrast to MF and its radical cation (MF+), only very limited information is available for H+MF. Theoretical studies indicate the preference for protonation at the carbonyl oxygen (CO) over the methyl ester (OMe) by around 80 kJ mol−1, and the CO-protonated tautomer has four rotamers within 20 kJ mol−1 (Zeegers-Huyskens & Kryachko 2011; Ferro-Costas & Mosquera 2013). Numerous mass spectrometric studies have investigated the production of H+MF via various ion–molecule reactions and its fragmentation by metastable decay and collisions (Harrison & Tsang 1976; Benoit & Harrison 1977; Hopkinson et al. 1979; van Baar et al. 1986; Horn et al. 2004; Lawson et al. 2012), and the measured proton affinity (PA) is reported as PA = 782.5 kJ mol−1 (Hunter & Lias 1998). However, the site of protonation and the rotational conformation of H+MF could not be established from these mass spectrometric measurements. The dominant fragment channel of H+MF is decarbonylation (loss of CO), whereby the C and O atoms come from the CO group. Significantly, there are no spectroscopic data available for H+MF and its clusters. To this end, our combined spectroscopic and DFT approach provides the first reliable experimental evidence for the preferred protonation site in the isolated molecule and the isomer assignment along with the intermolecular interaction of this prototypical protonated aliphatic ester with nonpolar ligands. There are a few experimental data for H+MF in the condensed phase. However, it is well known that solvation and counter ions can have a drastic effect on both the position and energetics of protonation in the various possible tautomers and rotamers. Previous 1H-NMR spectra of H+MF revealed three OH+ resonances in a super acid solution, demonstrating protonation at the CO group with three rotamers (although their configurations could not be established) (Fraenkel 1961; Olah et al. 1967). IR, Raman, and 1H/13C-NMR spectra of single crystal salts of H+MF with
and
show only a single isomer (the most stable (ts) conformer), which forms a strong ionic hydrogen bond (H band) between the OH+ group and the anion (Minkwitz et al. 2000). The characteristic OH stretch frequency is thus largely redshifted and thus not identified in both the IR and Raman spectra of the salts. Also, the IR intensities of the CH stretches are too weak to be observed.
|
[
"Lawson et al. 2012"
] |
[
"Numerous mass spectrometric studies have investigated the production of H+MF via various ion–molecule reactions and its fragmentation by metastable decay and collisions"
] |
[
"Background"
] |
[
[
1459,
1477
]
] |
[
[
1179,
1347
]
] |
2022MNRAS.514.5192L__Borsa_et_al._2021_Instance_1
|
We introduce the quantity ξ to represent a scaled product of equilibrium temperature (Teq) and surface gravity (g):
(14)$$\begin{eqnarray}
\xi = \left(\frac{T_{\text{eq}}}{1000~\text{K}}\right)\left(\frac{g}{g_{\rm J}}\right) ~,
\end{eqnarray}$$where gJ is the surface gravity of Jupiter. The calculated ξ values are listed in Table 4, along with the equilibrium temperature, surface gravity, and scale height for each planet. Fig. 6 shows the relative height of sodium for all 10 planets against ξ. The weight-combined results are shown in dark-blue, together with the unweighted results (light-blue) and literature values using the same HARPS/HARPS-N data (grey). The salmon-coloured points show recent results from other high-resolution spectrographs: WASP-69b with CARMENES (Khalafinejad et al. 2021), WASP-76b with ESPRESSO (Tabernero et al. 2021) and GRACES (Deibert et al. 2021), and WASP-121b with ESPRESSO (Borsa et al. 2021). We find that hNa is well described by an exponential trend of the form
(15)$$\begin{eqnarray}
h_{\lambda } = a\text{e}^{-b\xi } + c ~.
\end{eqnarray}$$As shown in Fig. 6, we fit this curve to the weight-combined results using the optimize.curve_fit function from scipy. The best-fitting values for the variables are: a = 1.70 ± 1.04, b = 5.04 ± 1.63, and c = 0.113 ± 0.013. The reduced chi-square of the fit across the full sample is $\chi ^2_\nu = 3.7$, much of which is skewed by two planets that deviate from the fit by more than 3σ (WASP-79b and HD 189733 b). The reduced chi-square when excluding these two planets is $\chi ^2_\nu = 1.2$. Since the exponential curve is asymptotic to the value of c, our results suggest that planets with ξ ≳ 1.25 are likely to have an upper limit on hNa of ∼0.113. The sodium features could potentially be muted due to other atmospheric effects, such as high-altitude clouds and hazes in lower temperature planets, or ionization of most of the sodium in extremely irradiated environments. Therefore, results lower than those given by equation (15) are also possible. The underlying physical processes behind this trend are not discussed in this current work, but these results motivate further observations to confirm the trend and theoretical studies to investigate possible physical mechanisms. Further refinement of this curve will be possible with more observations of low-ξ planets.
|
[
"Borsa et al. 2021)"
] |
[
"The salmon-coloured points show recent results from other high-resolution spectrographs:",
"WASP-121b with ESPRESSO"
] |
[
"Uses",
"Uses"
] |
[
[
918,
936
]
] |
[
[
668,
756
],
[
893,
916
]
] |
2017MNRAS.471.3856T__Tempel_et_al._2011_Instance_1
|
It is important to note that we have not classified our simulated galaxies by morphology itself in this paper, and more detailed analysis of the morphology, including kinematical decomposition, will be made in future work. In Fig. 4, we show the distributions of our morphology indicators, ETGΔs, LTGΔs, ETGΔt, and LTGΔt, with stellar mass. The top panel shows that LTGΔt dominate at low mass, and ETGΔt, which have experienced a major merger within 1 Gyr, are more numerous above M* ∼ 5 × 1010M⊙; the ETGΔt fraction is ∼45 per cent and ∼85 per cent at M* ∼ 5 × 1010M⊙ and ∼1011M⊙, respectively. Since mergers are strong drivers of morphological change, this implies that our high-mass galaxies are mostly elliptical, which is in good agreement with observational results for ellipticals and S0 galaxies (e.g. Bernardi et al. 2010; Tempel et al. 2011). There is not a perfect correlation between galaxy morphology and tmerge, however, and the details of each individual merger are important. In the lower panel of Fig. 4, galaxies are separated by ΔSFMS. Galaxies that lie close to the SFMS (LTGΔs) are most numerous at lower masses, and constitute a declining fraction of the galaxy population with increasing mass, but the fraction of ETGΔs is not perfectly consistent with observations (e.g. Renzini & Peng 2015, ∼50 per cent and ∼75 per cent at log M* = 10.5 and 11). This means that our AGN feedback is probably not strong enough (see TK15a for more discussion). Our predicted distribution of gradients could be affected by these effects. Our galaxies that are classified as both LTGΔs and ETGΔt should be ellipticals, and their metallicity gradients may be too steep because of the lack of quenching of star formation. Therefore, we think our sample of ETGΔs is better for the comparison with observations of SAURON and ATLAS3D. For low-mass galaxies, the lack of resolution could result in underestimating gradients in general. However, our low-mass LTGΔs and LTGΔt galaxies, it could also overestimate the gradients in the case that the gradients would be measured along small star-forming discs, which are not resolved in our simulations.
|
[
"Tempel et al. 2011"
] |
[
"Since mergers are strong drivers of morphological change, this implies that our high-mass galaxies are mostly elliptical, which is in good agreement with observational results for ellipticals and S0 galaxies (e.g."
] |
[
"Similarities"
] |
[
[
833,
851
]
] |
[
[
597,
810
]
] |
2018MNRAS.479.3254V___2000_Instance_1
|
The lifetime of molecular clouds (MCs) remains an active research topic in the study of the interstellar medium and star formation, and most recent studies, both observational and theoretical, place this lifetime at a few times 107 yr for clouds in the 105–106M⊙ mass range (e.g. Blitz & Shu 1980; Kawamura et al. 2009; Zamora-Avilés, Vázquez-Semadeni & Colín 2012; Zamora-Avilés & Vázquez-Semadeni 2014; Lee, Miville-Deschênes & Murray 2016). In addition, several observational studies have suggested that the star formation rate (SFR) of the clouds appears to increase over their lifetimes. For example, studies of young clusters embedded in moderate-mass MCs (∼104M⊙) (e.g. Palla & Stahler 1999, 2000; Da Rio et al. 2010) have shown that their age histograms contain a large majority of young (1–2 Myr) objects, but also a tail of older (up to several Myr) ones suggesting an accelerating star formation activity, sometimes followed by a subsequent decline (see also Povich et al. 2016; Schneider et al. 2018). In addition, Kawamura et al. (2009) reported a clear evolutionary process over the lifetime of giant molecular clouds (GMCs; of masses ∼105–106M⊙) in the Large Magellanic Cloud, evidenced by the increasing number of massive stars across the sequence of GMC ‘classes’ proposed by those authors. Finally, on the basis of the large scatter in the observed star formation efficiency (SFE) in Milky Way GMCs, Lee et al. (2016) have concluded that the SFR in those clouds must also be time variable. Numerical simulations of MC formation and evolution also exhibit time varying, increasing SFRs during their early stages (e.g. Vázquez-Semadeni et al. 2007; Hartmann, Ballesteros-Paredes & Heitsch 2012). Also, in the presence of stellar feedback, at late times the SFRs reach a maximum and begin to decrease again (e.g. Vázquez-Semadeni et al. 2010; Colín, Vázquez-Semadeni & Gómez 2013). Vázquez-Semadeni, González-Samaniego & Colín (2017) have recently shown that the simulations of Colín et al. (2013) in fact produce stellar age histograms highly resemblant of the observed ones (Palla & Stahler 1999, 2000; Da Rio et al. 2010), and reproduce observed radial age gradients in clusters (Getman et al. 2014) as well as bottom-heavy stellar initial mass functions (IMFs) in scattered regions of massive star formation (Povich et al. 2016).
|
[
"Palla & Stahler",
"2000"
] |
[
"For example, studies of young clusters embedded in moderate-mass MCs (∼104M⊙) (e.g.",
"have shown that their age histograms contain a large majority of young (1–2 Myr) objects, but also a tail of older (up to several Myr) ones suggesting an accelerating star formation activity, sometimes followed by a subsequent decline"
] |
[
"Background",
"Background"
] |
[
[
677,
692
],
[
699,
703
]
] |
[
[
593,
676
],
[
725,
959
]
] |
2021MNRAS.503.6016K__Martocchia_et_al._2018a_Instance_1
|
Observations of splits in the MS and at the eMSTO of young and (at least) moderately massive Magellanic Cloud star clusters have demonstrated the presence of complex stellar populations, suggestive of variations in their stellar rotation properties (Milone et al. 2018a). The effects of stellar rotation on stellar populations have been observed in Galactic open clusters as well (Marino et al. 2018b). Nearly all of these young star clusters are sufficiently massive to host MPs among their MS stars. For example, NGC 1755, a young cluster with a mass of ∼104 M⊙ (Milone et al. 2016) or NGC 419, another young massive cluster with an age of 1.72 Gyr and a mass of ∼104 M⊙, are comparably massive with respect to other star clusters that show clear evidence of MPs (Martocchia et al. 2018a). Thus, clusters with minimum masses of order 104 M⊙ tend to show evidence of MPs. However, at present such a mass threshold has yet to be explained theoretically. If we assume that mass is an important driver of MP formation, then NGC 411 should also have shown clear evidence of chemical abundance variations among its RGB stars. However, we have found no such evidence in NGC 411, which challenges the idea that mass may be the only or even the main driver of MP formation. The similarity of the observed RGB width in NGC 411 with that expected for an SSP suggests an absence of significant chemical abundance variations. In fact, for NGC 411, NGC 1718, and NGC 2213 we derive maximum possible helium-abundance variations of δY = 0.003 ± 0.001(Y = 0.300), 0.002 ± 0.001(Y = 0.350), and 0.004 ± 0.002(Y = 0.300), respectively. We determined an upper limit to the nitrogen-abundance variation in NGC 411 of Δ[N/Fe] = 0.3 dex, although the available data did not allow us to determine useful upper limits for our other sample clusters. Combined with similar results for NGC 419, NGC 1806, and NGC 1846 (Martocchia et al. 2017), our result is indeed inconsistent with mass being the primary driver of MP formation, although it seems likely that a sizeable minimum mass is still required.
|
[
"Martocchia et al. 2018a"
] |
[
"NGC 419, another young massive cluster with an age of 1.72 Gyr and a mass of ∼104 M⊙, are comparably massive with respect to other star clusters that show clear evidence of MPs"
] |
[
"Differences"
] |
[
[
766,
789
]
] |
[
[
588,
764
]
] |
2021AandA...656A..44R__Duc_2012_Instance_1
|
In this work we aim to explore the presence and properties of LSBGs in the environment of the NGC 1052 group of galaxies. This region is of particular interest in light of the “exotic” properties that recent works claim for some LSBGs found in this region. For instance, van Dokkum et al. (2018a) in [KKS2000] 04 (more commonly known as NGC 1052-DF2) and van Dokkum et al. (2019a) in NGC 1052-DF4 claimed the similarity between the observed baryonic matter and the dynamic matter obtained by the radial velocity dispersion of their GCs and also the stellar component (Emsellem et al. 2019; Danieli et al. 2019). This observational evidence would imply an extreme and unexpected deficit (even lack) of dark matter in these LSBGs. Taking into account that their metallicity content follows the usual stellar mass–metallicity relation for dwarf galaxies (Fensch et al. 2019; Ruiz-Lara et al. 2019), the possibility that they are tidal dwarf galaxies – galaxies formed by strong interactions with recycled material from massive host galaxies, high in metals, and intrinsically born with a lack of dark matter (see e.g., Duc 2012; Román et al. 2021) – is ruled out. The properties of these LSBGs have been the subject of much debate (see e.g., Martin et al. 2018; Ogiya 2018; Famaey et al. 2018; Kroupa et al. 2018; Lewis et al. 2020; Haslbauer et al. 2019; Müller et al. 2019; Nusser 2019; Montes et al. 2020, 2021). One hypothesis put forward by Trujillo et al. (2019) is that these LSBGs “lacking dark matter” could be at a closer distance than initially estimated by van Dokkum et al. (2018a, 2019a). This would not only make the stellar mass smaller (leaving room for a higher M/L fraction and therefore recovering a certain amount of dark matter, alleviating its strong deficit) but would also resolve the other anomalies that these LSBGs have related to their globular cluster luminosity function (GCLF) (e.g., Shen et al. 2021a), something considered a universal property of any galactic population, including LSBGs (e.g., Jordán et al. 2007; Villegas et al. 2010; Rejkuba 2012; Amorisco et al. 2018; Prole et al. 2019b). However, the distance to these galaxies is still the subject of intense debate (van Dokkum et al. 2018b, 2019b; Blakeslee & Cantiello 2018; Monelli & Trujillo 2019; Danieli et al. 2020; Shen et al. 2021b; Zonoozi et al. 2021) and there is currently no broad consensus on these distance values, with the remaining debated range of distances being approximately 12–22 Mpc.
|
[
"Duc 2012"
] |
[
"the possibility that they are tidal dwarf galaxies – galaxies formed by strong interactions with recycled material from massive host galaxies, high in metals, and intrinsically born with a lack of dark matter (see e.g.,",
"– is ruled out."
] |
[
"Differences",
"Differences"
] |
[
[
1116,
1124
]
] |
[
[
896,
1115
],
[
1145,
1160
]
] |
2016MNRAS.461.1719C__Fu_et_al._2012_Instance_1
|
HATLAS12-00 had already been identified as a candidate gravitationally lensed galaxy as a result of its high submm flux (i.e. F500 > 100 mJy), red Herschel colours and the lack of a bright optical or radio counterpart (see e.g. Negrello et al. 2010 for a discussion of the selection of lens candidates in H-ATLAS and other Herschel surveys). This source was therefore observed spectroscopically in the submm. A CO spectroscopic redshift of 3.26 was first suggested by Z-spec (Bradford et al. 2004) observations, then subsequently confirmed by observations by the CARMA interferometer (Leeuw et al., in preparation) and the Zpectrometer instrument (Harris et al. 2007) on the Greenbank Telescope (Harris et al. 2012; see also Fu et al. 2012). Additional followup observations in the optical, near-IR, submm and other wavelengths were targeted at the lensed z = 3.26 source and the foreground objects responsible for the lensing, resulting in detailed analyses of this lensing system by Fu et al. (2012) and Bussmann et al. (2013). Their conclusions are that the z = 3.26 source HATLAS12-00 is subject to gravitational lensing, with a magnification of 9.6 ± 0.5 in both the submm continuum and CO, and 16.7 ± 0.8 in the K′ band, by two foreground galaxies, one at a spectroscopically determined redshift of 1.22, and another with photometry suggesting that it lies at a similar redshift. The submm photometry of HATLAS12-00 at 890 μm acquired with the Submillimeter Array (SMA) as part of this programme (Fu et al. 2012; Bussmann et al. 2013) is fully consistent with the 870 μm and 850 μm fluxes derived for this source from the LABOCA and SCUBA2 data to be presented here. The spectral energy distribution (SED) of the lensed source, after correcting for the lensing amplification, is well matched by the optically thick SED model for Arp220 from Rangwala et al. (2011), with a lensing-corrected far-IR luminosity of 1.2 ± 0.2 × 1013 L⊙, and an implied star formation rate of 1400 ± 300 M⊙ yr−1. In many ways the unlensed properties of this object match those of the broader population of bright submm selected galaxies first discovered by the SCUBA submm imager (see e.g. Chapman et al. 2005; Clements et al. 2008; Michałowski, Hjorth & Watson 2010). The unlensed 870 μm flux of this object would be ∼7.7 mJy.
|
[
"Fu et al. 2012"
] |
[
"A CO spectroscopic redshift of 3.26 was first suggested by Z-spec",
"observations, then subsequently confirmed by observations by",
"the Zpectrometer instrument",
"on the Greenbank Telescope"
] |
[
"Compare/Contrast",
"Compare/Contrast",
"Compare/Contrast",
"Compare/Contrast"
] |
[
[
725,
739
]
] |
[
[
409,
474
],
[
498,
558
],
[
619,
646
],
[
668,
694
]
] |
2019MNRAS.490.3875L__Onaka_et_al._1996_Instance_1
|
Similar to Chen et al. (2017), we also follow Li & Draine (2001a) to derive upper limits on the abundance of the graphene C24 in the Galactic cirrus and in the diffuse ISM towards l = 44°20′, b = −0°20′ based on comparison of the observed IR emission with the calculated emission spectrum of C24. For the Galactic cirrus, the average emission per H has been measured by COBE/DIRBE (Arendt et al. 1998), COBE/FIRAS (Finkbeiner, Davis & Schlegel 1999), and Planck (Planck Collaboration XVII 2014). The diffuse ISM towards l = 44°20′, b = −0°20′ has been observed by COBE/DIRBE (Hauser et al. 1998). The Mid-Infrared Spectrograph (MIRS) aboard the Infrared Telescope in Space (IRTS) has also obtained the 4.7–11.7$\, {\rm \mu m}$ spectrum for the diffuse ISM towards l = 44°20′, b = −0°20′ (Onaka et al. 1996). As shown in Fig. 5(a), even locking up 20 $\, {\rm ppm}$ of C/H – the upper limit of graphene derived from the interstellar extinction – all in the specific graphene species C24, the 6.6, 9.8, and 20 $\, {\rm \mu m}$ emission features of C24 would still be hidden by the PAH features at 6.2, 7.7, 8.6, and 11.3 $\, {\rm\, \mu m}$ and would remain undetected by Spitzer or by the Short Wavelength Spectrometer (SWS) aboard the Infrared Space Observatory (ISO). This is also true for the diffuse ISM towards l = 44°20′, b = −0°20′. As illustrated in Fig. 5(b), as much as $\sim \, 20\, {\rm ppm}$ of C/H could also be tied up in the C24 graphene while the characteristic 6.6, 9.8, and 20 $\, {\rm \mu m}$ emission features of C24 are still not strong enough to be detected by IRTS. Therefore, for both the Galactic cirrus and the line of sight towards l = 44°20′, b = −0°20′, a upper limit of C/H ≲ 20 $\, {\rm ppm}$ is imposed by the COBE/DIRBE photometric data and the IRTS spectrum. Nevertheless, the actual abundance of C24 could be much lower than 20 $\, {\rm ppm}$ since, if graphene is indeed present in the ISM, it could span a wide range of sizes and charging states.
|
[
"Onaka et al. 1996"
] |
[
"The Mid-Infrared Spectrograph (MIRS) aboard the Infrared Telescope in Space (IRTS) has also obtained the 4.7–11.7$\\, {\\rm \\mu m}$ spectrum for the diffuse ISM towards l = 44°20′, b = −0°20′"
] |
[
"Uses"
] |
[
[
788,
805
]
] |
[
[
597,
786
]
] |
2020ApJ...901...50E__Dame_et_al._2001_Instance_1
|
Molecular hydrogen H2 is conventionally believed to make up 17% of the interstellar medium (ISM) by mass in the Galaxy (e.g., Draine 2011, Table 1.2, not including He), but, unlike atomic hydrogen H i (at 60% by mass), H2 is not generally detectable in the cold, diffuse conditions that predominate the volume of the ISM. In order to determine the mass and structure of interstellar H2, emission from other “trace” molecules must be observed instead. The primary tracer for H2 in current use is the J = (1−0) ground-state rotational transition of 12CO at a wavelength of 3 mm. While this tracer has become widely adopted over the years (see, e.g., Heyer et al. 1998; Dame et al. 2001) and has been often defended as a reliable large-scale tracer for molecules in the ISM, it also remains the primary such tracer in common use. There are several reasons to be concerned about this. First, the critical volume density of the 12CO (1–0) transition at low optical depths is
; this means that gas at lower volume density will be less excited by collisions, thus emitting fewer photons, and may even effectively disappear from sensitivity-limited surveys. Second, in sufficient quantities to avoid the “excitation” problem, the 12CO signal is optically thick, so that direct calculation of column densities from line profile integrals is not possible. Instead, indirect methods of inferring the quantity of H2 must be used, such as the “X-factor” method (see, e.g., Bolatto et al. 2013, for a review). However, recent evidence from gamma rays and IR surveys has suggested that there are significant regions of the ISM apparently containing molecular gas that is not adequately traced by CO emission (see, e.g., Grenier et al. 2005). The “dark gas” appears not to be hidden optically thick H i (see Murray et al. 2018), and as such is most likely molecular. While several other molecular gas tracers have been studied (see, e.g., Jacob et al. 2019, on CH), these tracers may not be ideal for detecting low-density, diffuse portions of the large-scale ISM in emission owing to their high critical densities. A reliable alternative molecular tracer for diffuse gas is needed to confirm the identity of this “CO-dark” gas, as well as to provide information on its nature, kinematics, column density, and structure.
|
[
"Dame et al. 2001"
] |
[
"The primary tracer for H2 in current use is the J = (1−0) ground-state rotational transition of 12CO at a wavelength of 3 mm. While this tracer has become widely adopted over the years (see, e.g.,",
"and has been often defended as a reliable large-scale tracer for molecules in the ISM, it also remains the primary such tracer in common use. There are several reasons to be concerned about this."
] |
[
"Background",
"Compare/Contrast"
] |
[
[
667,
683
]
] |
[
[
451,
647
],
[
685,
880
]
] |
2020MNRAS.495.2949F__Stauffer_et_al._2016_Instance_1
|
Stellar rotation is known to be related to the activity level; e.g. faster rotators tend to have larger star-spot coverages (e.g. Paper I), and stronger chromospheric emissions (e.g. Stauffer et al. 1997; Douglas et al. 2014; Newton et al. 2017; Paper II) and X-ray emissions (e.g. Pizzolato et al. 2003; Mamajek & Hillenbrand 2008); thus, a correlation between rotation and variation level should also be expected. In fact, Fig. 7 shows that there exists a large scatter in variation amplitude even for stars in the same cluster. It is reasonable to connect the scatter to the stellar rotation rate diversity, considering the fact that there exist noticeable differences in rotation periods between stars of the same temperature in young open clusters (e.g. Stauffer et al. 2016). To understand further on this scenario, we provided variation amplitudes as a function of temperature in Fig. 8, wherein shown are stars with known rotation periods. For comparison purpose, we also displayed the corresponding rotation periods in the upper panels. The periods were collected from the literature (Pleiades: Hartman et al. 2010; Covey et al. 2016; Rebull et al. 2016a; Praesepe: Agüeros et al. 2011; Delorme et al. 2011; Kovács et al. 2014; Rebull et al. 2017; Hyades: Hartman et al. 2011; Armstrong et al. 2015; Douglas et al. 2016); for a few stars in Praesepe, such as EPIC 211852399 and EPIC 211875458, the periods were obtained by us based on K2 data (see Appendix A). The colour contrast in Fig. 8 denotes the Rossby number, Ro, the normalized rotation period by the convective turnover time (the convective turnover time was estimated by using the correlation reported by Wright et al. 2011; see Paper I/II for details). It is known that the open cluster members mainly locate in two rotation sequences in the rotation–colour diagram (Barnes 2003), i.e. I (interface) and C (convective) sequence, corresponding to slow and fast rotators, respectively, as shown in the top panels in Fig. 8. The figure shows a tendency that GK-type C sequence stars in Pleiades have larger variation amplitude of EWH α compared to I sequence stars of the same temperature. We found no evident difference between different rotators among M-type Pleiades stars, partially because most of them are very fast rotators that reside in activity saturation regime (when Ro ≲ 0.1; e.g. Paper II). For Praesepe/Hyades sample, almost all GK-type stars rotate slowly (e.g. Ro ∼ 0.5), located in the I sequence (those stars with blue colours); there is no clear trend among these stars. Among M-type stars, some stars are fast rotators in C sequence (red or green colours), while others are slow rotators (in I sequence), as shown in the upper right panel of Fig. 8. Thus, there is a tendency that faster rotators have larger variation amplitude of EWH α. Unlike EWH α, TiO2n shows no clear difference among all sample stars with different rotation periods, probably due to its small intrinsic variation amplitude, which could be dominated by observational noise.
|
[
"Stauffer et al. 2016"
] |
[
"It is reasonable to connect the scatter to the stellar rotation rate diversity, considering the fact that there exist noticeable differences in rotation periods between stars of the same temperature in young open clusters (e.g."
] |
[
"Uses"
] |
[
[
759,
779
]
] |
[
[
531,
758
]
] |
2022AandA...666L...5G__Ramos_et_al._2014_Instance_1
|
More recently, García-Bernete et al. (2022) found that the PAH molecules responsible for the 11.3 μm PAH emission band are more resilient in the hard environments often present in AGN. In particular, the authors found larger 11.3/7.7 μm and 11.3/6.2 μm PAH ratios in AGN-dominated systems compared to SF galaxies, indicating a larger fraction of neutral PAH molecules (as noted by Smith et al. 2007 using a sample of relatively weak AGN). However, these studies were limited by the spatial resolution (∼4″) and the low spectral resolution (R ∼ 60–130) of Spitzer/InfraRed Spectrograph (IRS). Previous sub-arcsecond angular resolution N-band (∼8–13 μm) ground-based spectroscopic studies investigated the 11.3 μm PAH feature in the nuclear and circumnuclear regions of AGN (e.g., Hönig et al. 2010; González-Martín et al. 2013; Alonso-Herrero et al. 2014, 2016; Ramos et al. 2014; Esquej et al. 2014; García-Bernete et al. 2015; Jensen et al. 2017; Esparza-Arredondo et al. 2018). However, these works were unable to provide definitive details regarding the effect of the AGN on the PAH molecules due to limited wavelength coverage and sensitivity. The changes in the PAH properties due to the presence of the AGN might be more prominent in their innermost regions of galaxies. Therefore, the unprecedented combination of high angular and spectral resolution (R ∼ 1500 − 3500) in the entire mid-IR range (4.9–28.1 μm) afforded by the James Webb Space Telescope (JWST)/Mid-Infrared Instrument (MIRI; Rieke et al. 2015; Wells et al. 2015; Wright et al. 2015) is key to investigating PAH properties. In this Letter we report on the first investigation of PAH emission in the nuclear regions of three luminous Seyfert (Sy) galaxies and compare them with emission from SF regions using JWST/MIRI Medium Resolution Spectrograph (MRS) data. This enables us, for the first time, to characterise the PAH properties of local luminous Sy galaxies (log (Lbol)> 44.46 erg s−1)1 at sub-arcsecond scales (∼0.45″, ∼142–245 pc).
|
[
"Ramos et al. 2014"
] |
[
"Previous sub-arcsecond angular resolution N-band (∼8–13 μm) ground-based spectroscopic studies investigated the 11.3 μm PAH feature in the nuclear and circumnuclear regions of AGN (e.g.,",
"However, these works were unable to provide definitive details regarding the effect of the AGN on the PAH molecules due to limited wavelength coverage and sensitivity."
] |
[
"Background",
"Motivation"
] |
[
[
861,
878
]
] |
[
[
592,
778
],
[
980,
1147
]
] |
2020AandA...633A.147B__Xu_et_al._(2013)_Instance_1
|
We used the position and distance of the MCs in the FQS catalogue to describe the structure of the Milky Way in this poorly known portion of the third quadrant. In Fig. 16 we show the distribution of the clouds projected onto the Milky Way plane. At any Galactic longitude the clouds are grouped in three well-separated structures that trace the position of the spiral arms. The dispersion of the positions in each structure is given not only by the width of the arms but also by the uncertainty of the distance determination. Indeed, kinematic distances are affected by significant uncertainties due not only to the assumed Galactic rotation model but also to the unknown possible peculiar motions of the clouds themselves. In Fig. 16 we compare the positions of the clouds derived from FQS data with the expected locations of the spiral arms, adopting a log-periodic spiral form of the Galaxy (Reid et al. 2014). The pitch angle and the distance at a reference position of the three arms are taken from Xu et al. (2013) for the Local arm, Choi et al. (2014) for the Perseus arm, and Hachisuka et al. (2015) for the Outer arm. These authors used the positions of water masers derived from parallax measurements as part of the Bar and Spiral Structure Legacy (BeSSeL) survey to fit the parameters of the spiral arms. We found good agreement between the position of the FQS clouds and the modelled position of the Local and the Perseus arms, while the modelled Outer arm lies further than the observed clouds. We note that while the parameters of the Local and Perseus arms are derived by fitting about 20 water masers located in both the second and third quadrants, covering also the Galactic longitude of the FQS survey, the parameters of the Outer arm are derived from only 5 water masers, 4 of which are located in the second quadrant. Therefore it is not surprising that they might not be fully appropriate to describe the location of the Outer arm in the longitude range of the FQS survey of 220° l 240°. In fact, an alternative model of the Outer arm from Hou & Han (2014), also shown in Fig. 16, seems in better agreement with the FQS results.
|
[
"Xu et al. (2013)"
] |
[
"The pitch angle and the distance at a reference position of the three arms are taken from",
"for the Local arm",
"These authors used the positions of water masers derived from parallax measurements as part of the Bar and Spiral Structure Legacy (BeSSeL) survey to fit the parameters of the spiral arms.",
"We found good agreement between the position of the FQS clouds and the modelled position of the Local and the Perseus arms,"
] |
[
"Uses",
"Uses",
"Background",
"Similarities"
] |
[
[
1005,
1021
]
] |
[
[
915,
1004
],
[
1022,
1039
],
[
1128,
1316
],
[
1317,
1440
]
] |
2015AandA...584A..75V__Essen_et_al._(2014)_Instance_1
|
The data presented here comprise quasi-simultaneous observations during secondary eclipse of WASP-33 b around the V and Y bands. The predicted planet-star flux ratio in the V-band is 0.2 ppt, four times lower than the accuracy of our measurements. Therefore, we can neglect the planet imprint and use this band to measure the stellar pulsations, and most specifically to tune their current phases (see phase shifts in von Essen et al. 2014). Particularly, our model for the light contribution of the stellar pulsations consists of eight sinusoidal pulsation frequencies with corresponding amplitudes and phases. Hence, to reduce the number of 24 free parameters and the values they can take, we use prior knowledge about the pulsation spectrum of the star that was acquired during von Essen et al. (2014). As the frequency resolution is 1/ΔT (Kurtz 1983), 3.5 h of data are not sufficient to determine the pulsations frequencies. Therefore, during our fitting procedure we use the frequencies determined in von Essen et al. (2014) as starting values plus their derived errors as Gaussian priors. As pointed out in von Essen et al. (2014), we found clear evidences of pulsation phase variability with a maximum change of 2 × 10-3 c/d. In other words, as an example after one year time a phase-constant model would appear to have the correct shape with respect to the pulsation pattern of the star, but shifted several minutes in time. To account for this, the eight phases were considered as fitting parameters. The von Essen et al. (2014) photometric follow-up started in August, 2010, and ended in October, 2012, coinciding with these LBT data. We then used the phases determined in von Essen et al. (2014) during our last observing season as starting values, and we restricted them to the limiting cases derived in Sect. 3.5 of von Essen et al. (2014), rather than allowing them to take arbitrary values. The pulsation amplitudes in δ Scuti stars are expected to be wavelength-dependent (see e.g. Daszyńska-Daszkiewicz 2008). Our follow-up campaign and these data were acquired in the blue wavelength range. Therefore the amplitudes estimated in von Essen et al. (2014), listed in Table 1, are used in this work as fixed values. This approach would be incorrect if the pulsation amplitudes would be significantly variable (see e.g., Breger et al. 2005). Nonetheless, the short time span of LBT data, and the achieved photometric precision compared to the intrinsically low values of WASP-33’s amplitudes, make the detection of any amplitude variability impossible.
|
[
"von Essen et al. 2014"
] |
[
"Therefore, we can neglect the planet imprint and use this band to measure the stellar pulsations, and most specifically to tune their current phases (see phase shifts in"
] |
[
"Uses"
] |
[
[
418,
439
]
] |
[
[
248,
417
]
] |
2021AandA...649A.126T__Luck_(2018b)_Instance_2
|
Studies of the radial n-capture-to-iron abundance gradients are very scarce so far. We can only search for a broad agreement of our results with several studies of abundance gradients with galactocentric distances (Rgc). da Silva et al. (2016) studied n-capture elements across the Galactic thin disc based on Cepheid variables. Because the Cepheids are young stars, their Rgc may be rather close to their birthplaces and Rmean. da Silva et al. (2016) supplemented their sample of 111 Cepheids with 324 more stars from other studies and found that the [Y/Fe] distribution is flat throughout the entire disc. In our study, we confirm this finding not only based on the whole thin-disc sample of stars and on a subsample of younger ≤4 Gyr stars, but also add another light s-process dominated element strontium. Like in our study, da Silva et al. (2016) also obtained positive [El/Fe] radial gradients for La, Ce, Nd, and Eu. The slopes are rather similar. For [Eu/Fe], they differ just by 0.002 dex kpc−1. More recently, Luck (2018b) also investigated the gradients of n-capture element abundance-to-iron ratios with respect to Rgc for a sample of 435 Cepheids. It is interesting to note that the [Ba/Fe] versus Rgc slope according to this Cepheid sample is also negative, as in our study. [Ba/Fe] is the only n-capture element-to-iron ratio with a negative radial gradient in our sample of stars and in Luck (2018b). Overbeek et al. (2016) investigated trends of Pr, Nd, and Eu to Fe abundance ratios with respect to Rgc using 23 open clusters. As in our study, they found that these elements have positive linear trends with galactocentric radius (the linear regression slopes are of about +0.04 dex kpc−1). They also suggested that the [El/Fe] relation of Pr and Nd, but not Eu, with the galactocentric radius may not be linear because the [El/Fe] of these elements appears to be enhanced around 10 kpc and drop around 12 kpc. Because only a small number of stars lie at these large radial distances, we cannot address this question. For the thick-disc stars, the radial abundance-to-iron slopes are negligible, as was found for α-process elements by Li et al. (2018), even though the production sites of α-elements and s-processes dominated elements are quite different.
|
[
"Luck (2018b)"
] |
[
"It is interesting to note that the [Ba/Fe] versus Rgc slope according to this Cepheid sample is also negative, as in our study. [Ba/Fe] is the only n-capture element-to-iron ratio with a negative radial gradient in our sample of stars and in"
] |
[
"Similarities"
] |
[
[
1403,
1415
]
] |
[
[
1161,
1402
]
] |
2018MNRAS.474.2277D__Oliveira,_Dottori_&_Bica_1998_Instance_1
|
There are three possible explanations for the origin of these systems: (1) they formed from the fragmentation of the same molecular cloud (Elmegreen & Elmegreen 1983), (2) they were generated in distinct molecular clouds and then became bound systems after a close encounter leading to a tidal capture (Vallenari, Bettoni & Chiosi 1998; Leon, Bergond & Vallenari 1999), or (3) they are the result of division of a single star-forming region (Goodwin & Whitworth 2004; Arnold et al. 2017). Their subsequent evolution may also have different outcomes. Dynamical models and N-body simulations (see, e.g., Barnes & Hut 1986; de Oliveira, Dottori & Bica 1998, and references therein) have shown that, depending on the initial conditions, a bound pair of clusters may either become unbound, because of significant mass-loss in the early phases of stellar evolution, or merge into a single and more massive cluster on a short time-scale (≈60 Myr) due to loss of angular momentum from escaping stars (see Portegies Zwart & Rusli 2007). The final product of a merger may be characterized by a variable degree of kinematic and morphologic complexity, mostly depending on the values of the impact parameter of the pre-merger binary system (de Oliveira, Bica & Dottori 2000; Priyatikanto et al. 2016). In some cases, the stellar system resulting from the merger event may show significant internal rotation (in fact, for many years this has been the preferred dynamical route to form rotating star clusters; see Sugimoto & Makino 1989; Makino, Akiyama & Sugimoto 1991; Okumura, Ebisuzaki & Makino 1991; de Oliveira, Dottori & Bica 1998). Merger of cluster pairs has been sometimes invoked to interpret the properties of particularly massive and dynamically complex clusters (e.g. see the study of ω Centauri by Lee et al. 1999, G1 by Baumgardt et al. 2003 and NGC 2419 by Brüns & Kroupa 2011), and, more in general, as an avenue to form clusters with multiple populations with different chemical abundances both in terms of iron and light elements (e.g. van den Bergh 1996; Catelan 1997; Amaro-Seoane et al. 2013; Gavagnin, Mapelli & Lake 2016; Hong et al. 2017).
|
[
"de Oliveira, Dottori & Bica 1998"
] |
[
"Dynamical models and N-body simulations (see, e.g.,",
"and references therein) have shown that, depending on the initial conditions, a bound pair of clusters may either become unbound, because of significant mass-loss in the early phases of stellar evolution, or merge into a single and more massive cluster on a short time-scale (≈60 Myr) due to loss of angular momentum from escaping stars"
] |
[
"Background",
"Background"
] |
[
[
621,
653
]
] |
[
[
550,
601
],
[
655,
991
]
] |
2022AandA...667A..82D__Lyne_et_al._(1985)_Instance_1
|
Much of the progress in PPS has come from thorough Monte Carlo simulations that generate pulsars and test whether they fulfil the criteria for detection according to geometrical factors and sensitivity issues. It is then possible to develop and optimize a model for the underlying pulsar population, which informs us about the important intrinsic neutron star parameters and distribution, enabling predictions for future surveys. Usually, PPS studies follow two simple approaches. The first one is to take a ‘snapshot’ of the Galaxy as it appears today, where no assumptions are made regarding the prior evolution of the pulsar population. Instead, this population is generated assuming various distribution functions (typically spatial distribution, spin period P and,
E
˙
$ \dot{E} $
- or Ṗ), which are optimized to match the observations. Inspired by earlier studies from Taylor & Manchester (1977) and Lyne et al. (1985), Lorimer et al. (2006) applied the snapshot approach to the canonical3 pulsar population to determine best-fitting probability density functions in Galactocentric radius (R), luminosity (L), height with respect to the Galactic plane (z), and the period P for the currently observed population of pulsars. Alternatively, one may consider ‘evolution’ strategies where the pulsars are evolved from birth up to the present era, starting from an initial spatial distribution, and an initial period and magnetic field distribution. A fine example of the latter genre is the comprehensive study of Faucher-Giguère & Kaspi (2006), which quite successfully reproduced the properties of the main part of the radio pulsar population using a model in which the luminosity has a power-law dependence on P and Ṗ. PPS studies can be also extended to the population of neutron stars observed in other bands, such as X-rays (see for instance Popov et al. 2010) and γ-rays. Indeed, with the broad increase in γ-ray pulsar numbers, a statistical treatment of the γ-ray population in combination with deep radio surveys of the Galactic plane is now feasible. Early works on radio-loud gamma-ray pulsar populations carried out before the Fermi era include Gonthier et al. (2002), (2004), (2007a,b). With the advent of Fermi/LAT, new studies emerged (see for instance Gonthier et al. 2018; Ravi et al. 2010; Takata et al. 2011; Watters & Romani 2011; Pierbattista et al. 2012), trying to constrain the geometry and the location of the gamma-ray emission sites. Watters & Romani (2011) showed that an initial spin period of P0 = 50 ms and a birth rate of one neutron star per 59 yr were required to reproduce the observed γ-ray population. They made the prediction that after ten years of operations, Fermi should detect ∼120 young γ-ray pulsars, of which about one half would be radio-quiet. Gonthier et al. (2004) included an exponentially decaying magnetic field with a 2.8 Myr timescale and displaying a satisfactory agreement with the P−Ṗ distribution at that time. Later, more accurate magnetic field decay models were elaborated, especially for magnetars, as presented in Viganò et al. (2013).
|
[
"Lyne et al. (1985)"
] |
[
"Inspired by earlier studies from Taylor & Manchester (1977) and",
"Lorimer et al. (2006) applied the snapshot approach to the canonical3 pulsar population to determine best-fitting probability density functions in Galactocentric radius (R), luminosity (L), height with respect to the Galactic plane (z), and the period P for the currently observed population of pulsars."
] |
[
"Background",
"Background"
] |
[
[
914,
932
]
] |
[
[
850,
913
],
[
934,
1237
]
] |
2020MNRAS.495L..27H__Cao_et_al._2018_Instance_1
|
Independent measurements of distances and redshifts, z, allow astronomers to constrain cosmological models, since they both define the distance–redshift relation. When it was determined that type Ia supernovae (SNIa) could be standardized and therefore used to measure distances, this led to the discovery of the accelerated expansion of our Universe (Riess et al. 1998; Perlmutter et al. 1999). The combination of SNIa (Betoule et al. 2014), baryonic acoustic oscillations (Eisenstein et al. 2005; Alam et al. 2017), and the cosmic microwave background (CMB; Komatsu et al. 2011; Planck Collaboration et al. 2018) led to the emergence of the concordance Λ cold dark matter (ΛCDM) model, in which the energy density is dominated by dark energy as a cosmological constant Λ. In a flat Friedmann–Lemaître–Robertson–Walker universe, the comoving distance is defined as
(1)$$\begin{eqnarray*}
D(z) = \int _0^z \frac{c\mathrm{d}z}{H(z)},
\end{eqnarray*}$$ where, in the ΛCDM model,
(2)$$\begin{eqnarray*}
H(z) = H_0 \sqrt{\Omega _\mathrm{m} (1+z)^3 + 1-\Omega _\mathrm{m}}
\end{eqnarray*}$$is the Hubble parameter, H0 is the Hubble–Lemaître constant, and Ωm is the matter energy density at the current epoch. The luminosity and angular diameter distances are defined as
(3)$$\begin{eqnarray*}
D_\mathrm{L}(z) = (1+z) D(z)
\end{eqnarray*}$$(4)$$\begin{eqnarray*}
{\mathrm{ and}} \,D_\mathrm{A}(z) = \frac{R}{\theta }= \frac{D(z)}{(1+z)}.
\end{eqnarray*}$$Type Ia supernovae can only be used up to redshift of around 2 (Jones et al. 2013), and there are tensions between direct local measurements of the Hubble–Lemaître constant and model-dependent estimates using CMB observations (Planck Collaboration et al. 2018). Therefore, independent distance measurements to extragalactic objects are desired. We should emphasize here, that model-independent distance indicators can have various important applications in physical cosmology. In particular, to test different aspects of cosmological models and theories of gravity. For instance, we can use these model-independent measurements to test the FLRW metric (Clarkson, Bassett & Lu 2008; Wiltshire 2009; Shafieloo & Clarkson 2010; L’Huillier & Shafieloo 2017; Shafieloo, L’Huillier & Starobinsky 2018; Cao et al. 2019a; Qi et al. 2019a), to test general relativity and some modified gravity models (Cao et al. 2012; Shafieloo, Kim & Linder 2013a; Cao et al. 2015, 2017b; Qi et al. 2017; L’Huillier, Shafieloo & Kim 2018; Shafieloo et al. 2018; Xu et al. 2018; Chen, Sesana & Conselice 2019; L’Huillier et al. 2020), to test natural constants such as the speed of light (Cao et al. 2018), to test cosmic duality relationships, and to measure cosmic curvature (Shafieloo et al. 2013b; Cao et al. 2019b; Qi et al. 2019a, b; Zheng et al. 2020). Using a combination of such model-independent distance indicators can also be used to measure some key cosmological parameters such as the Hubble Constant (Liao et al. 2015; Suyu et al. 2017; Jee et al. 2019; Liao et al. 2019, 2020; Wong et al. 2019). Amongst the most energetic objects in our Universe are active Galactic nuclei (AGNs). AGNs are the nuclei of massive galaxies that sometimes produce relativistic jets of material launched from near a central supermassive black hole (SMBH). When these jets are not aligned close to our line of sight, AGNs are observed as radio galaxies, whereas if the jet is aligned to within a small angle to our line of sight, they are observed as blazars (Urry & Padovani 1995). Blazars and quasars are amongst the most consistently bright objects in our Universe and can be observed at redshifts as high as ∼7 (Mortlock et al. 2011). Attempts have been made to measure distances to AGNs in various ways, with some claiming deviations from the expected cosmology at high redshifts (Risaliti & Lusso 2017, 2019; Turner & Shabala 2019). Very long baseline interferometry (VLBI) has also been used to attempt to measure cosmological distances. The approach pioneered by Gurvits, Kellermann & Frey (1999) attempted to measure cosmological parameters by assuming that AGN could be used as a standardizable rod. Vishwakarma (2001) used this data set and compared it with supernovae data, and found it was not possible to differentiate different cosmological models with the VLBI data of Gurvits et al. (1999). Cao et al. (2015) revisited this technique and investigated the evolution of the standard rod by assuming a Planck cosmology. Cao et al. (2017a, c) then introduced a cosmology independent method for calibrating the standard rod and was able to provide reasonable constraints on cosmological parameters. Our approach differs from this by using the speed of light to normalize the rod. This approach was first attempted by Wiik & Valtaoja (2001), which found that that the apparent angular sizes of AGNs maximized at z ∼ 2. In this paper, we demonstrate the method on the famous nearby radio galaxy 3C 84 (NGC 1275) and discuss possible systematic errors. The source is known to exhibit extremely high energy emission despite not exhibiting strong relativistic effects (Jorstad et al. 2017; Liodakis et al. 2018), and has multiple independent measures of distance (Theureau et al. 2007; Hicken et al. 2009), thus making it an ideal source to test our methodology.
|
[
"Cao et al. 2018"
] |
[
"For instance, we can use these model-independent measurements",
"to test natural constants such as the speed of light"
] |
[
"Background",
"Background"
] |
[
[
2624,
2639
]
] |
[
[
2026,
2087
],
[
2570,
2622
]
] |
2021ApJ...919...30D__Staguhn_et_al._2014_Instance_2
|
The first SMGs were detected using SCUBA at 850 μm (Smail et al. 1997; Barger et al. 1998; Hughes et al. 1998), which remains one of the prime wavelengths to detect these galaxies (e.g., Geach et al. 2017), thanks to a combination of available instruments, spectral window, and the negative k-correction at that wavelength. Other single-dish samples of SMGs have also been obtained at 1.1–1.3 mm using MAMBO (e.g., Eales et al. 2003; Bertoldi et al. 2007; Greve et al. 2008) and AzTEC (e.g., Aretxaga et al. 2011; Yun et al. 2012), at 1.4 mm/2 mm with the SPT (Vieira et al. 2010), and at 2 mm with GISMO (Staguhn et al. 2014; Magnelli et al. 2019). Selecting SMGs from observations at longer wavelengths is thought to favor galaxies at higher redshifts (e.g., Smolčić et al. 2012; Vieira et al. 2013; Staguhn et al. 2014; Magnelli et al. 2019; Hodge & da Cunha 2020), although it is difficult to compare the redshift distributions in an unbiased way (see, e.g., Zavala et al. 2014 for a discussion), and account for intrinsic variations of galaxy far-IR spectral energy distributions (SEDs). Nevertheless, the 2 mm band has been put forth as a potential candidate to detect high-redshift (z > 3) galaxies (e.g., Casey et al. 2018a, 2018b, 2019; Zavala et al. 2021). The negative k-correction is stronger at 2 mm than at 850 μm; thus, for a fixed SED, the 2 mm band should pick up more high-redshift galaxies than at 870 μm. In addition, better atmospheric transmission and larger fields of view can be achieved at 2 mm (but corresponding poorer resolution). Such an effort is currently ongoing (see Zavala et al. 2021 for first results). To understand the relationship between the populations detected at 850 μm and at 2 mm, we require a detailed characterization of the (sub)millimeter SEDs of these sources. Multiwavelength submillimeter observations are still rare, with most observations focusing on a single wavelength. Only a handful of sources observed at 2 mm have complementary shorter-wavelength detections (Staguhn et al. 2014; Magnelli et al. 2019). Thus, a more systematic multiwavelength dust continuum investigation is warranted in order to reveal the dust properties of (sub)millimeter-detected sources.
|
[
"Staguhn et al. 2014"
] |
[
"Selecting SMGs from observations at longer wavelengths is thought to favor galaxies at higher redshifts (e.g.,",
"although it is difficult to compare the redshift distributions in an unbiased way",
"and account for intrinsic variations of galaxy far-IR spectral energy distributions (SEDs)."
] |
[
"Compare/Contrast",
"Compare/Contrast",
"Compare/Contrast"
] |
[
[
802,
821
]
] |
[
[
650,
760
],
[
869,
950
],
[
1001,
1092
]
] |
2022ApJ...931...70B__Gabrielse_et_al._2012_Instance_2
|
RFs can propagate from the magnetotail to Earth over a long distance more than 10 R
E together with BBFs behind them (Runov et al. 2009; Cao et al. 2010). Studies have suggested that RFs are crucial regions for particle acceleration, pitch-angle evolution, wave–particle interactions, and electromagnetic energy conversion during their Earthward propagation. For instance, rapid increases in energy fluxes of electrons and ions from tens to hundreds of keV are a typical feature of RF events (Khotyaintsev et al. 2011; Liu et al. 2013, 2018c, 2021a, 2022b; Zhou et al. 2018; Liu & Fu 2019; Gabrielse et al. 2021), pitch-angle distribution of suprathermal electrons can evolve dramatically around RFs (Runov et al. 2013; Liu et al. 2020), strong particle and wave activity can occur in the vicinity of RFs (Ono et al. 2009; Zhou et al. 2009, 2014; Fu et al. 2014; Breuillard et al. 2016; Greco et al. 2017; Yang et al. 2017), and RFs are associated with energy conversion from electromagnetic fields to particles (Sitnov et al. 2009; Huang et al. 2015; Khotyaintsev et al. 2017; Liu et al. 2018a, 2022a). The energetic plasma in the vicinity of RFs plays a key role in connecting the magnetotail with the inner magnetosphere because they carry a large amount of energy and can be injected into the inner magnetosphere to affect the ring current and radiation belt (Gabrielse et al. 2012; Duan et al. 2014; Turner et al. 2014). Possible mechanisms responsible for the energization of particles around RFs have been widely investigated based on both spacecraft observations and numerical simulations during the past decade. The strong convection electric field induced by the strong magnetic field gradient of RFs provides significant adiabatic acceleration of the ambient particles (Birn et al. 2004, 2013, 2015; Gabrielse et al. 2012, 2014, 2016; Ganushkina et al. 2013; Liu et al. 2016; Turner et al. 2016). Nonadiabatic effects, caused by particle reflection ahead of the RFs (Zhou et al. 2018), resonance with RFs (Ukhorskiy et al. 2013, 2017), and scattering by wave emissions (Zhou et al. 2009; Greco et al. 2017), are also significant for particle energization. These above studies usually assumed that the RF surface has a planar boundary at a typical thickness comparable to the ion gyroradius and below (Nakamura et al. 2002; Sergeev et al. 2009; Zhou et al. 2009; Schmid et al. 2011; Liu et al. 2013; Vapirev et al. 2013). Divin et al. (2015b) revealed that the RF surface is unstable to instabilities ranging from electron scales to ion scales. Simulation studies found that RFs can be unstable to interchange instability and that finger-like structures on ion–electron hybrid scales can develop at the RF (Vapirev et al. 2013). Such finger-like structures are found to play a role in modulating the electron acceleration process (Wu et al. 2018). Bai et al. (2022) also reported significant ion trapping acceleration at the RF with ion-scale ripples. Unlike these surface structures with ion or ion–electron hybrid scales, Liu et al. (2018b) recently reported that the RF layer has electron-scale density gradients, currents, and electric fields, based on the MMS mission, which consists of four spacecraft separated by 30 km. Such electron-scale ripple structure can be generated by lower hybrid drift instability (Divin et al. 2015b; Pan et al. 2018). Liu et al. (2021c) presented a detailed investigation of energy flux densities at two RFs with/without the electron-scale surface ripples and indicated that surface ripples may play an important role in the particle dynamics. But how such electron-scale RF structure impacts the electron energization and transport still remains unknown. In this paper, with the aid of observation-based test-particle simulation, we aim to investigate in detail the effect of the front surface ripples on the local electron dynamics.
|
[
"Gabrielse et al. 2012"
] |
[
"Possible mechanisms responsible for the energization of particles around RFs have been widely investigated based on both spacecraft observations and numerical simulations during the past decade. The strong convection electric field induced by the strong magnetic field gradient of RFs provides significant adiabatic acceleration of the ambient particles"
] |
[
"Background"
] |
[
[
1811,
1832
]
] |
[
[
1426,
1779
]
] |
2017ApJ...835...25E__Rutten_1984_Instance_1
|
We compare our results with a new reduction of observations from the Lowell Observatory SSS, which is running a long-term stellar activity survey complementary to the MWO HK Project. The SSS observes solar and stellar light with the same spectrograph, with the solar telescope consisting of an exposed optical fiber that observes the Sun as an unresolved source (Hall & Lockwood 1995; Hall et al. 2007). The basic measurement of SSS is the integrated flux in 1 Å bandpasses centered on the Ca ii H & K cores from continuum-normalized spectra, ϕHK, which can then be transformed to the S-index using a combination of empirical relationships derived from stellar observations:
7
where
is the continuum flux scale for the Ca ii H & K wavelength region, which converts ϕHK to physical flux (erg cm−2 s−1).
is a function of Strömgren
and is taken from Hall (1996).
(simply K in other works) is the conversion factor from the MWO HKP-2 H & K flux (numerator of Equation (1)) to physical flux (Rutten 1984). Ccf is a factor that removes the color term from S and is a function of Johnson
(Rutten 1984). Finally, Teff is the effective temperature. See Hall et al. (2007) and Hall & Lockwood (1995) for details on the extensive work leading to this formulation. What is important to realize about this method of obtaining S is that it requires three measurements of solar properties,
,
, and
, along with the determination of one constant,
. The solar properties are taken from best estimates in the literature, which vary widely depending on the source used, and can dramatically affect the resulting SSSS for the Sun. Hall et al. (2007) used
,
, and
. The constant
was empirically determined to be 0.97 ± 0.11 erg cm−2 s−1 in Hall et al. (2007) as the value that provides the best agreement between SSSS and
from Baliunas et al. (1995) for an ensemble of stars and the Sun. This combination of parameters resulted in a mean SSSS of 0.170 for the Sun using observations covering cycle 23. A slightly different calibration of SSS data in Hall & Lockwood (2004) used a flux scale
based on Johnson
, set to 0.65 for the Sun, and
K. In Table 1 we estimated that this calibration resulted in a mean S = 0.168 for cycle 23. Hall et al. (2009), which included a revised reduction procedure and one year of data with the upgraded camera (see below), found
.
|
[
"Rutten 1984"
] |
[
"(simply K in other works) is the conversion factor from the MWO HKP-2 H & K flux (numerator of Equation (1)) to physical flux"
] |
[
"Uses"
] |
[
[
1020,
1031
]
] |
[
[
893,
1018
]
] |
2019MNRAS.485L..78V__Chatterjee_et_al._2017_Instance_2
|
The properties of the persistent radio source associated with FRB 121102 may be constrained independently of the Faraday-rotating medium. We assume equipartition between the relativistic gas and magnetic field as is common in synchrotron sources3 (Readhead 1994). The source becomes self-absorbed at $1.5$ GHz for radius $R_{\rm per} < 0.05$ pc; this is thus the lower bound on the source size. European Very Long Baseline Interferometry (VLBI) Network observations of the source at 5 GHz set an upper bound on the source radius of Rper ≲ 0.35 pc (Marcote et al. 2017). This is consistent with the ${\approx } 30\, {{\rm per\, cent}}$ amplitude modulations observed in the source at 3 GHz (Chatterjee et al. 2017) being caused by refractive interstellar scintillation in the Milky Way interstellar medium (ISM; Walker 1998). For any radius within the allowed range (0.05 Rper/pc 0.35), we can determine the equipartition magnetic field, Beq, and the column of relativistic electrons, Nrel, using the standard expressions for synchrotron emissivity and absorption coefficients (Rybicki & Lightman 1979, their equations 6.36 and 6.53). We assume a power-law energy distribution of radiating electrons with somewhat shallow index of b = −1.5 that can account for the relatively flat spectrum of the source (Chatterjee et al. 2017). The peak Lorentz factor of the distribution, γmax, is chosen to correspond to the observed spectral break frequency of $\nu _{\rm max}=10$ GHz. If the lower Lorentz factor cut-off corresponds to emission at $\nu _{\rm min}=1$ GHz,4 then the equipartition magnetic field and electron column thus determined for minimum and maximum source sizes are $B_{\rm eq}\approx 140$ mG, $\gamma _{ \rm min}\approx 50$, γmax ≈ 160, $N_{\rm rel} \approx 0.95\, {\rm pc}\, {\rm cm}^{-3}$ for $R_{\rm per}=0.05$ pc, and $B_{\rm eq}\approx 27$ mG, $\gamma _{ \rm min}\approx 120$, γmax ≈ 370, $N_{\rm rel} \approx 0.1\, {\rm pc}\, {\rm cm}^{-3}$ for $R_{\rm per}=0.35$ pc. The reader can scale the equipartition field to other source sizes using Beq(R) ∝ R−6/7. The total energy contained in the relativistic electrons and the magnetic field (‘equipartition energy’) is ∼1049.1 and ∼1050.2 erg, respectively. If the relativistic electrons were injected in a one-off event, the synchrotron cooling rates at γmax yield source ages of $14$ yr for R = 0.05 pc and 60 yr for $R=0.35$ pc. The corresponding expansion velocities are $0.011\, c$ and $0.02\, c$, respectively.
|
[
"Chatterjee et al. 2017"
] |
[
"We assume a power-law energy distribution of radiating electrons with somewhat shallow index of b = −1.5 that can account for the relatively flat spectrum of the source"
] |
[
"Uses"
] |
[
[
1306,
1328
]
] |
[
[
1136,
1304
]
] |
2020MNRAS.497.2941S__Loeb_&_Barkana_2001_Instance_1
|
After the cosmological recombination, the universe went into the dark ages during which the density fluctuations in the matter distribution grew, and after reaching a threshold, the matter collapsed to make the first bound objects. The nature of dark matter sets the timeline and characteristics of these first bound objects, which were the hosts for the first sources of light, so it is essential to see the impact of different dark matter models on the observables from the Cosmic Dawn and EoR. Then, the natural question that arises is whether one can use the differences in these observables, estimated for different dark matter models, in order to constrain the nature of dark matter. The present observational probes that allow us to have a peak in this epoch are the absorption spectra of high-redshift quasars (Loeb & Barkana 2001; White et al. 2003; Boera et al. 2019) and the Thomson scattering optical depth of the cosmic microwave background (CMB) radiation photons (Kaplinghat et al. 2003; Komatsu et al. 2011). However, these indirect probes provide very limited and weak constraints on the CD-EoR. The H i 21-cm line, which arises due to the hyperfine splitting of the ground state of the neutral hydrogen, is a direct and most promising probe to study this period. Motivated by this, a large number of radio interferometers, including the GMRT (Paciga et al. 2013), LOFAR (Ghara et al. 2020; Mertens et al. 2020), MWA (Barry et al. 2019; Li et al. 2019), and PAPER (Kolopanis et al. 2019), are attempting a statistical detection of this signal using the power spectrum statistic. In parallel, there is a complementary approach to detect the sky-averaged global 21-cm signal from the CD-EoR using experiments, e.g. the EDGES (Bowman et al. 2018), DARE (Burns et al. 2017), and SARAS (Singh et al. 2018). The next-generation interferometers like the SKA (Koopmans et al. 2015; Mellema et al. 2015) are expected to see a giant leap in the sensitivity, which will enable them to make tomographic images of the H i distribution across cosmic time.
|
[
"Loeb & Barkana 2001"
] |
[
"The present observational probes that allow us to have a peak in this epoch are the absorption spectra of high-redshift quasars"
] |
[
"Background"
] |
[
[
819,
838
]
] |
[
[
690,
817
]
] |
2022MNRAS.509.1959S__Ezzeddine_et_al._2019_Instance_1
|
However, the transition between the two extremes of modern (metal-rich) and primordial (metal-poor) star formation, and in particular the role of dust coupling and stellar radiation feedback at low metallicity, has thus far received limited exploration. Krumholz (2011) present analytical models for radiation feedback and predict a weak scaling of IMF peak mass with metallicity, while Myers et al. (2011) and Bate (2014, 2019) carry out radiation-hydrodynamic simulations of star formation over a metallicity range from $0.01{-}3\, Z_{\rm {\odot }}$ and find negligible effects on gas fragmentation. However, these studies do not explore lower metallicities, despite available evidence for the existence of a low-metallicity ISM in the past through the discovery of stars with metallicities as low as $10^{-4}\, \rm {Z_{\odot }}$ (Caffau et al. 2011; Starkenburg et al. 2018), as well as several others with $\rm {[Fe/H]} \lt -5$ (Christlieb et al. 2004; Keller et al. 2014; Frebel et al. 2015; Aguado et al. 2017, 2018; Ezzeddine et al. 2019; Nordlander et al. 2019). Coming from the opposite direction, Bromm et al. (2001), Omukai et al. (2005), and Omukai, Hosokawa & Yoshida (2010) consider the thermodynamics of gas of increasing metallicity, and find that dust and metal line cooling permits fragmentation to reach masses ≲1 M⊙ only once the metallicity exceeds ∼10−3.5 Z⊙. Dust is a more efficient coolant than metal lines, and permits fragmentation to lower masses at lower metallicity (e.g. Meece, Smith & O’Shea 2014; Chiaki & Yoshida 2020; Shima & Hosokawa 2021), but exactly by how much depends on the poorly known distribution of dust grain sizes in the early Universe (Schneider et al. 2006, 2012; Omukai et al. 2010; Schneider & Omukai 2010; Chiaki et al. 2015). However, the early Universe star formation models are fundamentally misanalogous to the modern ones that consider decreasing metallicity, in that the early Universe models consider dust solely as a coolant that enables fragmentation, whereas the modern ones assign it a more nuanced role, as both a source of cooling and later, once stellar feedback begins, a source of heating – a changeover that seems crucial to explaining why the IMF in the present-day Universe peaks at ${\sim}0.2\, \rm {M_{\odot }}$ rather than ${\sim}10^{-2}\, \rm {M_{\odot }}$ (Kroupa 2001; Chabrier 2003, 2005).
|
[
"Ezzeddine et al. 2019"
] |
[
"However, these studies do not explore lower metallicities, despite available evidence for the existence of a low-metallicity ISM in the past through the discovery of stars with metallicities as low as $10^{-4}\\, \\rm {Z_{\\odot }}$",
"as well as several others with $\\rm {[Fe/H]} \\lt -5$"
] |
[
"Compare/Contrast",
"Compare/Contrast"
] |
[
[
1023,
1044
]
] |
[
[
602,
831
],
[
879,
931
]
] |
2022MNRAS.509..903N__Murguia-Berthier_et_al._2014_Instance_1
|
Binary neutron star (BNS) mergers have long been suspected to produce the central engines of short gamma-ray bursts (sGRBs) (Eichler et al. 1989). The link was firmly established in August 2017, after the combined detection of gravitational waves and an sGRB from the same BNS merger (Abbott et al. 2017a,b,c; Goldstein et al. 2017; Savchenko et al. 2017). The actual origin of the gamma-ray signal is still debated, coming from either an off-axis jet or from the shock breakout of a relativistic cocoon inflated by the jet itself (Kasliwal et al. 2017; Nakar & Piran 2017; Gottlieb, Nakar & Piran 2018a; Gottlieb et al. 2018b; Mooley et al. 2018a; Lundman & Beloborodov 2021). Nevertheless, the multiband observations of a rising afterglow (Alexander et al. 2017; Hallinan et al. 2017; Margutti et al. 2017, 2018; Troja et al. 2017, 2018, 2019; D’Avanzo P. et al. 2018; Lyman et al. 2018; Lamb et al. 2019) together with the detection of superluminal motion (Mooley et al. 2018b; Ghirlanda et al. 2019; Hotokezaka et al. 2019) have settled the presence of a jet that successfully broke out from the surrounding ejecta, observed off-axis with a viewing angle θobs ≈ 19° (Murguia-Berthier et al. 2017; Lamb, Mandel & Resmi 2018; Lazzati et al. 2018; Mooley et al. 2018a; Margutti & Chornock 2020). The information obtainable from afterglow observations is strongly dependent on the angular structure of the emerging jet. This structure might be mainly determined by the launching process (Kathirgamaraju et al. 2019) or it may arise as a consequence of the interaction with the surrounding environment during propagation. Therefore, an understanding of the processes that shape the jet could in principle provide insights into both jet formation, and the post-merger environment. Recent relativistic (magneto-) hydrodynamics simulations (Murguia-Berthier et al. 2014, 2017, 2021b; Geng et al. 2019; Beniamini et al. 2020b; Gottlieb, Levinson & Nakar 2020; Gottlieb & Nakar 2021; Gottlieb et al. 2021a; Gottlieb, Nakar & Bromberg 2021b; Hamidani & Ioka 2021; Lazzati et al. 2021; Nathanail et al. 2021; Pavan et al. 2021; Urrutia et al. 2021) have illustrated the importance of the ambient medium in shaping the jet, thereby highlighting the importance of understanding the remnant structure and ejecta properties. The joint events GW170817 and GRB170817A were followed by an additional electromagnetic transient spanning the spectral bands from UV to optical and IR on time-scales from days to weeks (Abbott et al. 2017b; Arcavi et al. 2017; Cowperthwaite et al. 2017; Drout et al. 2017; Evans et al. 2017; Pian et al. 2017; Smartt et al. 2017; Soares-Santos et al. 2017; Tanvir et al. 2017; Utsumi et al. 2017). The observed properties were consistent with the expectations for a thermal transient powered by the radioactive decay of freshly synthesized r-process elements (a so-called ‘macronova’ or ‘kilonova’ e.g. Li & Paczyński 1998; Kulkarni 2005; Rosswog 2005; Metzger et al. 2010; Roberts et al. 2011; Kasen, Badnell & Barnes 2013; Yu, Zhang & Gao 2013; Kasen, Fernández & Metzger 2015; Kasen et al. 2017; Metzger 2017; Perego, Radice & Bernuzzi 2017; Tanaka et al. 2017; Rosswog et al. 2018). Understanding the properties of this signal requires in-depth investigation of all the processes that can unbind material during and after a BNS merger, together with the available amount of free neutrons provided by each ejection channel. By now, several mass-ejection channels have been identified and they differ in terms of launch time, mass, electron fraction Ye, and velocity. During the merger ∼10−3 − 10−2 M⊙ of material are ejected dynamically (Rosswog et al. 1998, 1999; Rosswog & Davies 2002; Oechslin, Janka & Marek 2007; Bauswein, Goriely & Janka 2013; Radice et al. 2018a). On longer time-scales (∼1 s) a few 10−2 M⊙ of material can be unbound from the torus surrounding the remnant by the action of nuclear heating (Metzger, Piro & Quataert 2008; Fernandez & Metzger 2013; Fernandez et al. 2015; Just et al. 2015), magnetic (Siegel & Ciolfi 2015; Ciolfi et al. 2017; Siegel & Metzger 2017, 2018; Ciolfi & Kalinani 2020; Murguia-Berthier et al. 2021a), and viscous effects (Shibata, Kiuchi & Sekiguchi 2017; Fujibayashi et al. 2018, 2020; Radice et al. 2018b; Shibata & Hotokezaka 2019). Weak interactions play a key role since they can change the initially extremely low electron fraction and they are therefore of paramount importance for nucleosynthesis and the electromagnetic appearance of a BNS merger (Ruffert et al. 1997; Rosswog & Liebendörfer 2003; Dessart et al. 2009; Perego et al. 2014, 2017; Martin et al. 2015, 2018; Miller et al. 2019; Murguia-Berthier et al. 2021a).
|
[
"Murguia-Berthier et al. 2014"
] |
[
"Recent relativistic (magneto-) hydrodynamics simulations",
"have illustrated the importance of the ambient medium in shaping the jet, thereby highlighting the importance of understanding the remnant structure and ejecta properties"
] |
[
"Motivation",
"Background"
] |
[
[
1837,
1865
]
] |
[
[
1779,
1835
],
[
2141,
2311
]
] |
2016MNRAS.463.2716M__Cho_&_Lazarian_2007_Instance_2
|
Returning to the case of HL Tau, where the possible contribution of an infalling envelope is not an issue, how can one reconcile the strong indication of a dominant radial field component in the polarization map with the expectation that the bulk of the mm-wavelength emission originates near the disc mid-plane, where the azimuthal field component dominates? One possibility is that the non-negligible optical depth inferred in the bright emission rings of HL Tau at mm-wavelengths (Jin et al. 2016; Pinte et al. 2016) shifts the emission centroid to finite disc elevations where the magnetic field already has a measurable radial component. However, in view of the very small scale height inferred for the mm-emitting dust in this source, this is unlikely to be the main explanation. Perhaps a more likely possibility is that, even in this comparatively young source, the grains near the mid-plane, which dominate the total intensity, have already grown to sizes that exceed the maximum size $a_\mathrm{max} = \lambda /2\pi$ for producing polarized emission at wavelength λ (e.g. Cho & Lazarian 2007; for λ = 1.25 mm, amax = 0.2 mm), while the smaller grains (with sizes a amax), which contribute efficiently to the polarized flux, remain suspended at high elevations (where the field is predominantly radial). Another effect that could lower the polarized emission from grains that have settled to the mid-pane is the likelihood that grains become less elongated as they grow (e.g. Hughes et al. 2009), which would tend to reduce the value of the coefficient C in equation (2) (C → 0 as the grain axis ratio → 1).7 This interpretation is supported by the finding in the high-resolution observations of IRAS 4A (Cox et al. 2015) of an average polarization of 15 per cent at 8 mm and 10 per cent at 10 mm, with a peak fractional polarization of ∼20 per cent. If the intrinsic degree of mm-wavelength polarization in HL Tau is also of the order of 20 per cent, then it may be possible to explain the factor of ∼10 lower value of P measured in this source at 1.25 mm8 in terms of a dilution of the polarized emission from a ≲ 0.1 mm grains at high disc elevations by weakly polarized emission of larger grains residing near the mid-plane. In this scenario, most of the grains that are responsible for the mm-wavelength flux have settled to the mid-plane and grown to sizes a ≳ 1 mm.9 Although a fraction of these grains may have sizes in excess of 1 mm and would therefore emit less efficiently at that wavelength than a ≲ 1 mm grains (e.g. Miyake & Nakagawa 1993), the mid-plane region should still dominate the total mm-wavelength flux if most of the a ≳ 1 mm grains are concentrated there. Grains of size a ≲ 0.1 mm may be kept at high elevations by turbulent motions that can persist below the wind-driving surface layers (e.g. Simon et al. 2013, 2015; Bai 2015) as well as by the emerging outflows within these layers (see Safier 1993), with porosity effects (e.g. Ormel, Spaans & Tielens 2007) possibly also helping to mitigate gravity's pull towards the mid-plane. This scenario of course needs to be backed by detailed calculations and observational tests. One such test would be to obtain a polarization map of HL Tau at longer ( ≳ 1 cm) wavelengths: if the above picture is correct and the grains in the mid-plane region are aligned, such a map could reveal a stronger (or even dominant) contribution from the azimuthal and (especially if Λ0 ≪ 1) vertical field components.10 It is, however, conceivable that the large grains in this source are not well aligned because the radiative torque mechanism does not operate efficiently on them: this could happen if the characteristic wavelength of the anisotropic component of the local radiation field were much smaller than the mid-plane grain sizes (e.g. Cho & Lazarian 2007) or if the anisotropic radiation component inside the dust disc were weak due to finite optical depth effects. Note that the possibility of the mid-plane grains not being well aligned provides another reason for why the polarized mm-wavelength emission from this region could be weak.
|
[
"Cho & Lazarian 2007"
] |
[
"It is, however, conceivable that the large grains in this source are not well aligned because the radiative torque mechanism does not operate efficiently on them: this could happen if the characteristic wavelength of the anisotropic component of the local radiation field were much smaller than the mid-plane grain sizes (e.g."
] |
[
"Future Work"
] |
[
[
3812,
3831
]
] |
[
[
3485,
3811
]
] |
2021ApJ...910..124X__Leslie_et_al._2016_Instance_1
|
What is the dominant mode of star formation for AGNs in general, and for quasars in particular? The main sequence gives a useful framework for discussing the evolutionary status of AGN host galaxies and their relation to the galaxy population at large. The existing literature in this field, however, is complicated enormously by the diverse strategies of AGN sample selection, the accuracy of the SFR and M* tracers, and the myriad choices of main-sequence prescription. While there is almost unanimous agreement that AGNs of low to moderate luminosity (Lbol ≲ 1045 erg s−1) at z ≈ 0–3 lie on or below the main sequence (e.g., Shao et al. 2010; Mullaney et al. 2012, 2015; Santini et al. 2012; Rosario et al. 2013a; Shimizu et al. 2015; Ellison et al. 2016; Leslie et al. 2016; Suh et al. 2017; Bernhard et al. 2019; Grimmett et al. 2020; Jackson et al. 2020), no consensus exists for AGNs with Lbol > 1045 erg s−1. The situation is particularly contentious at redshifts higher than ∼0.5, where luminous AGNs have been reported to be above (e.g., Rovilos et al. 2012; Florez et al. 2020; Kirkpatrick et al. 2020), on (e.g., Harrison et al. 2012; Xu et al. 2015; Stanley et al. 2017; Schulze et al. 2019), and below (e.g., Scholtz et al. 2018; Stemo et al. 2020) the main sequence. Fortunately, a better consensus of opinion can be found for luminous (Lbol ≳ 1045 erg s−1) AGNs at z ≲ 0.5. Most agree that low-redshift quasars are located largely on and above the main sequence (Husemann et al. 2014; Xu et al. 2015; Zhang et al. 2016; Stanley et al. 2017; Jarvis et al. 2020). Regardless of redshift, it appears that the magnitude of an AGN’s offset from the main sequence correlates positively with its luminosity (Bernhard et al. 2019; Grimmett et al. 2020). For their large sample of z ≈ 0.3 type 1 AGNs with uniform SFRs based on extinction-corrected [O II] λ3727 emission, Zhuang & Ho (2020) demonstrated that SFR systematically rises with increasing Lbol at fixed M*.
|
[
"Leslie et al. 2016"
] |
[
"While there is almost unanimous agreement that AGNs of low to moderate luminosity (Lbol ≲ 1045 erg s−1) at z ≈ 0–3 lie on or below the main sequence (e.g.,",
", no consensus exists for AGNs with Lbol > 1045 erg s−1."
] |
[
"Similarities",
"Differences"
] |
[
[
759,
777
]
] |
[
[
472,
627
],
[
860,
916
]
] |
2017MNRAS.464.2545C__López-Corredoira_&_Molgó_2014_Instance_1
|
One of the fundamental tasks of the Galactic studies is to estimate the structure parameters of the major structure components. Bahcall & Soneira (1980) fit the observations with two structure components, namely a disc and a halo. Gilmore & Reid (1983) introduce a third component, namely a thick disc, confirmed in the earliest Besancon Galaxy Model (Crézé & Robin 1983). Since then, various methods and observations have been adopted to estimate parameters of the thin and thick discs and of the halo of our Galaxy. As the quantity and quality of data available continue to improve over the years, the model parameters derived have become more precise, numerically. Ironically, those numerically more precise results do not converge (see table 1 of Chang, Ko & Peng 2011, table 2 of López-Corredoira & Molgó 2014 and sections 5 and 6 of Bland-Hawthorn & Gerhard 2016, for a review). The scatters in density law parameters, such as scale lengths, scale heights and local densities of these Galactic components, as reported in the literature, are rather large. At least parts of the discrepancies are caused by degeneracy of model parameters, which in turn can be traced back to the different data sets adopted in the analyses. Those differing data sets either probe different sky areas (Bilir et al. 2006a; Du et al. 2006; Cabrera-Lavers et al. 2007; Ak et al. 2007; Yaz & Karaali 2010; Yaz Gökçe et al. 2015), are of different completeness magnitudes and therefore refer to different limiting distances (Karaali et al. 2007), or consist of stars of different populations of different absolute magnitudes (Karaali, Bilir & Hamzaolu 2004; Bilir et al. 2006b; Jurić et al. 2008; Jia et al. 2014). It should be noted that the analysis of Bovy et al. (2012), using the SEGUE spectroscopic survey, has given a new insight on the thin and thick disc structural parameters. This analysis provides estimate of their scale height and scale height as a function of metallicity and alpha abundance ratio. However, it relies on incomplete data (since it is spectroscopic) with relatively low range of Galactocentric radius as for the thin disc is concerned.
|
[
"López-Corredoira & Molgó 2014"
] |
[
"As the quantity and quality of data available continue to improve over the years, the model parameters derived have become more precise, numerically. Ironically, those numerically more precise results do not converge",
"table 2 of",
"The scatters in density law parameters, such as scale lengths, scale heights and local densities of these Galactic components, as reported in the literature, are rather large."
] |
[
"Motivation",
"Motivation",
"Motivation"
] |
[
[
785,
814
]
] |
[
[
518,
734
],
[
774,
784
],
[
885,
1060
]
] |
2015AandA...579A.132P__Simha_et_al._(2009)_Instance_1
|
A common feature of all previous models is that the relation between the central galaxy stellar mass and the halo mass reaches a maximum at halo masses ~1012 M⊙. According to Yang et al. (2012), below this threshold the mass accretion of the central galaxy is dominated by star formation. Thus, when the halo mass reaches ~1012 M⊙ a process takes place that quenches the star formation. Interestingly, this mass scale is very similar to the cold-mode to hot-mode transition scale (Birnboim & Dekel 2003; Kereš et al. 2005) in the theory of gas accretion, as derived in hydrodynamic simulations, whereas large halos primarily accrete hot gas and low mass halos cold gas. This would suggest that the quenching of central galaxies coincides with the formation of a hot gaseous halo, and thus with a lack of cold gas supply. What would be the fate of satellites? According to Simha et al. (2009), the subhalos also retain their identity for quite some time after accreting a larger halo, so satellites in subhalos less massive than ~1012 M⊙ do not immediately see the effect of the hot gas in the larger halo and accrete in cold mode. Thus, consistent with the results of Yang et al. (2012) and Béthermin et al. (2013), satellite galaxies continue to accrete gas and convert it to stars over a rather long period, which according to Simha et al. (2009) is about of 0.5−1 Gyr after the merger. The gas accretion declines steadily over this period. Since star formation follows mass accretion with a short delay, satellites should experience quenching in a similar amount of time. This scenario would be consistent with our observations. Indeed, at z ~ 1 when massive halos are just forming via merger, the SF activity in the accreted subhalos is still high. At later epochs, instead, the transition to the hot mode accretion of the satellites and the consequent progressive quenching of their SF activity would lead to the faster decline of their contribution to the CSFH with respect to lower mass halos, which evolve in a cold mode accretion phase maintaining a high SFR.
|
[
"Simha et al. (2009)"
] |
[
"According to",
"the subhalos also retain their identity for quite some time after accreting a larger halo, so satellites in subhalos less massive than ~1012 M⊙ do not immediately see the effect of the hot gas in the larger halo and accrete in cold mode.",
"Thus, consistent with the results of Yang et al. (2012) and Béthermin et al. (2013), satellite galaxies continue to accrete gas and convert it to stars over a rather long period,"
] |
[
"Uses",
"Uses",
"Similarities"
] |
[
[
872,
891
]
] |
[
[
859,
871
],
[
893,
1130
],
[
1131,
1309
]
] |
2016MNRAS.457.2480C___2008_Instance_1
|
A number of ideas have been put forward to explain the formation and early evolution of the compact Kepler and radial velocity systems, which in cases such as Gliese 581 and HD 69830 appear to contain in excess of ∼30 M⊕ of solid material within a few tenths of an au (Lovis et al. 2006; Udry et al. 2007). This concentration of solids close to the star led to classical core accretion models combined with disc-driven migration being developed using population synthesis codes (Alibert et al. 2006). More recent population synthesis calculations that also include prescriptions for planet–planet interactions have also been presented (Ida & Lin 2010). N-body simulations, combined with either hydrodynamic simulations or analytic prescriptions for migration and eccentricity/inclination damping of planetary growth, have also been used to examine the origins of such systems (Cresswell & Nelson 2006, 2008; Terquem & Papaloizou 2007; McNeil & Nelson 2009, 2010; Hellary & Nelson 2012; Cossou, Raymond & Pierens 2013; Coleman & Nelson 2014; Hands, Alexander & Dehnen 2014). A common outcome of these N-body simulations is the formation of resonant convoys of planets in the presence of convergent migration, an outcome that is not reflected in the Kepler systems. Various ideas have been put forward to explain why the resonances may be unstable, including tidal eccentricity damping followed by separation of the resonance for short-period systems (Terquem & Papaloizou 2007), stochastic migration due to local turbulence (Adams, Laughlin & Bloch 2008; Rein & Papaloizou 2009; Rein 2012) – a process that is likely to only operate close to the star where the disc can be thermally ionized (Umebayashi & Nakano 1988; Desch & Turner 2015), resonance breaking due to overstable librations (Goldreich & Schlichting 2014), orbital repulsion due to non-linear spiral wave damping in planet coorbital regions (Podlewska-Gaca, Papaloizou & Szuszkiewicz 2012; Baruteau & Papaloizou 2013).
|
[
"Cresswell & Nelson",
"2008"
] |
[
"N-body simulations, combined with either hydrodynamic simulations or analytic prescriptions for migration and eccentricity/inclination damping of planetary growth, have also been used to examine the origins of such systems"
] |
[
"Background"
] |
[
[
877,
895
],
[
902,
906
]
] |
[
[
653,
875
]
] |
2017MNRAS.470..612F__Feng_etal._2016_Instance_2
|
The millimetre bump in M87 as recently observed by the Atacama Large Millimeter/submillimeter Array can be naturally modelled by the synchrotron emission of the thermal electrons in the ADAF, which is different from the prediction of the jet model. Therefore, it provides an opportunity to explore the accretion process near the BH horizon. In particular, Feng etal. (2016) and Li etal. (2016) both found that the rotation measure predicted from the ADAF is roughly consistent with the observational values. It is still difficult to constrain the BH spin parameter from the modelling of the SED of M87 due to some degeneracy in model parameters (wind parameter, s, magnetic parameter ; Feng etal. 2016). The spin parameter can be better constrained from the jet model if the relativistic jet is indeed powered by the rotating BHs as suggested by MHD simulations and some observations. We find that the dimensionless BH spin parameter should be larger than 0.96 for the lower limit of jet power derived from the X-ray cavities (e.g. Rafferty et al. 2006; Russell etal. 2013b). In this work, we adopt several typical values of parameters (e.g. 0.1, 0.3 and 0.5). The larger value of will lead to a lower accretion rate near the horizon to explain the observed millimetre bump, and the BHs need to rotate faster to reproduce the observed jet power. The peak of synchrotron emission from the thermal electrons of ADAF will move to the submillimetre waveband if is too small, which is different from the observed millimetre bump. We adopt the equipartition case of 0.5 in our calculations, where magnetic energy will become dominant if the BH is fast spinning, considering the possible amplification of the magnetic field by the frame dragging effect. For the weaker magnetic case (e.g. 0.5), the BH needs to rotate faster to explain the observed SED and jet power. We find that our results are not sensitive to the viscosity parameter . Therefore, we suggest that the BH should be fast rotating in M87 even after considering the possible uncertainties.
|
[
"Feng etal. 2016)"
] |
[
"It is still difficult to constrain the BH spin parameter from the modelling of the SED of M87 due to some degeneracy in model parameters (wind parameter, s, magnetic parameter ;"
] |
[
"Compare/Contrast"
] |
[
[
686,
702
]
] |
[
[
508,
685
]
] |
2018MNRAS.473.1879R__Freudling_et_al._2011_Instance_1
|
Cosmic H i gas density ($\Omega _{\rm {H}\,\small {I}}$) as a function of redshift (bottom axis) and lookback time (top axis). All measurements are corrected to the same cosmological parameters. Some DLA measurements adopted a different definition of the cosmic H i density taking into account neutral gas abundance (Ωgas) including helium or contributions from Lyman-α absorbers with lower column density ($\log N({\rm H\,{\small {I}}}) < 20.3$). We have corrected all measurements to a consistent definition of $\Omega _{\rm {H}\,\small {I}}$. The large red star shows the $\Omega _{\rm {H}\,\small {I}}$ measurement from this work. The small black square and triangle at z ∼ 0 are the HIPASS and ALFALFA 21-cm emission measurements by Zwaan et al. (2005) and Martin et al. (2010), respectively. The red diamonds are from the Parkes telescope with an H i stacking technique (Delhaize et al. 2013). The open circle and square are the results from the Arecibo Ultra Deep Survey (AUDS) (Freudling et al. 2011; Hoppmann et al. 2015). Two right-pointing triangles are measured using the WSRT and H i stacking technique by Rhee et al. (2013). The pink triangle is measured by Lah et al. (2007) applying the GMRT 21-cm emission stacking. The green hexagon denotes the H i stacking measurement for the COSMOS field using the GMRT (Rhee et al. 2016). The black arrow line is the upper limit constrained using the GMRT H i stacking (Kanekar et al. 2016). The closed and open diamonds, left-pointing triangle, big circle, open triangles, open diamonds and small circles are damped Lyman-α measurements from the HST and the SDSS by Rao et al. (2006, 2017), Neeleman et al. (2016), Prochaska et al. (2005), Noterdaeme et al. (2009), Noterdaeme et al. (2012) and Bird et al. (2017), respectively. The downward triangles and big square at high redshift of z > 2 are ESO UVES and Gemini GMOS measurements of DLAs by Zafar et al. (2013), Crighton et al. (2015), respectively. The black line with a shaded area shows a weighted fit of all $\Omega _{\rm {H}\,\small {I}}$ measurements and its 95 per cent confidence interval. The solid blue line shows the semi-analytic model prediction of Kim et al. (2015), and the dashed red line shows the mufasa (Davé et al. 2017) simulation.
|
[
"Freudling et al. 2011"
] |
[
"The open circle and square are the results from the Arecibo Ultra Deep Survey (AUDS)"
] |
[
"Uses"
] |
[
[
986,
1007
]
] |
[
[
900,
984
]
] |
2018ApJ...863..194L__Kraft_et_al._1991_Instance_1
|
Five XRT observations got only low exposure (i.e., much less than 1 ks) and the X-ray source was therefore undetected in these data sets. Surprisingly, we also found that the source was undetected in a “deep” observation taken on 2015 August 19 with an exposure time of about 1.6 ks (Table 2). Within a 47″ radius circular region centered at the source position (corresponding to 90% of the encircled energy fraction of XRT at 1.5 keV; Moretti et al. 2005), only one photon (which is located near the edge of the region) was detected in this 1.6 ks observation. Even assuming that only this event is from the source, the inferred count rate is much lower than the measurements in 2010 and 2015–2017; for example, seven source counts would have been detected in a 1.6 ks observation with the count rate of 4.4 × 10−3 cts s−1 measured five days later (Table 2). Using a Bayesian approach (Kraft et al. 1991), we computed 95% upper limits for all the nondetections. As expected, the upper limits for data with 1 ks are not very much constraining (i.e., a few ×10−2 cts s−1, while the average count rate of the four individual detections is about 4 × 10−3 cts s−1). The upper limit for the 1.6 ks data is deeper (i.e., 8.7 × 10−3 cts s−1), but still insufficient to clarify whether the low-count-rate measurement is physically or statistically based. For a deeper constraint, we combined all the six XRT observations, and the X-ray source can be marginally detected in the stacked image with
cts s−1. Although this marginal detection shows a ∼50% decrease on flux in the period from 2015 February 04 through August 19, the variability is not statistically significant (i.e., less than 2σ). To check whether this variability was seen at other frequencies, we performed a Fermi-LAT analysis with the data collected between 2015 February 04 and August 19, and the γ-ray flux (100 MeV–100 GeV) did not vary significantly. In UV, there are some Ultraviolet/Optical Telescope (UVOT) images taken simultaneously with the XRT observations. Although the UVOT magnitudes (obtained by aperture photometry using the uvotsource task in HEAsoft v6.22) significantly changed over time (Table 2), this was due to the orbital modulation (Figure 3; will be discussed in the coming sections).
|
[
"Kraft et al. 1991"
] |
[
"Using a Bayesian approach",
"we computed 95% upper limits for all the nondetections."
] |
[
"Uses",
"Uses"
] |
[
[
887,
904
]
] |
[
[
860,
885
],
[
907,
962
]
] |
2019MNRAS.487.2412G__Weinberg_1993_Instance_1
|
There are only three exceptions from the picture discussed above. In Fig. 2 we see that the model with the smallest number of objects, $N=40\, 000$, does not show abrupt dissolution. This model is characterized with the shortest initial relaxation time (among models considered in this paper), about 1.6 Gyr, and the smallest number of retained BHs, about 30. In such a situation, according to Breen & Heggie (2013), BHs are quickly kicked out from the system and BHS cannot survive for a long time. So, the cluster evolution is not governed by strong energy generation by BHs in BHS. In Fig. 3 we can see that the model with W0 = 3 dissolves extremely fast and the model with W0 = 9 does not show the abrupt dissolution feature. The very fast dissolution of tidally filling models with low King model concentration was already extensively discussed in the literature (e.g. Weinberg 1993; Fukushige & Heggie 1995; Whitehead et al. 2013; Contenta et al. 2015). The dissolution of such clusters is controlled by very strong initial mass-loss powered by stellar/binary evolution. Relaxation process is not important at all. The situation is much different for the model with W0 = 9. As it can be seen from Fig. 5, the cluster enters the core-collapse phase, so it has to be controlled by the relaxation process. The difference between the model with W0 = 6 and the model with W0 = 9 is connected with the rate of BHS evolution. Model W0 = 9 is initially much denser, so its half-mass radius and half-mass relaxation time are shorter than for the W0 = 6 model. Nevertheless, for both models the BH mass segregation ends nearly at the same time, about 4 Gyr. At that time, the models contain about 50 BHs and 160 BHs for W0 = 9 and W0 = 6, respectively. According to Breen & Heggie (2013) the evolution of BHs is controlled by the energy flow through the cluster half-mass radius, which is proportional to Eb/Trh ≈ GM/Rh/Trh, where Eb is the cluster binding energy. Models with W0 = 9 have smaller half-mass radius and half-mass relaxation time than models with W0 = 6, so the energy demand is larger for this model and leads to a much faster burning out of BHs – larger number of dynamical interactions leading to BH removal from the system. After the time of BH mass segregation (about 4 Gyr) the model W0 = 6 contains enough BHs to form a BHS and enters the phase of balanced evolution (Breen & Heggie 2013). Contrarily, the model with W0 = 9 has too small a number of BHs and continues to remove BHs quickly to support the needed energy flow. Finally, it enters the phase when other energy sources connected with ordinary binaries take over and the cluster enters the core-collapse and then core-bounce phases. It is important to note that the tidal field plays an important role in the above picture. The less concentrated model has larger Rh and loses more mass, so it is easier for the BHS to provide the needed energy to support the cluster structure. This phase of evolution ends when the cluster suddenly loses its virial equilibrium.
|
[
"Weinberg 1993"
] |
[
"The very fast dissolution of tidally filling models with low King model concentration was already extensively discussed in the literature (e.g."
] |
[
"Background"
] |
[
[
874,
887
]
] |
[
[
730,
873
]
] |
2015ApJ...815..127W__Borucki_et_al._2010_Instance_1
|
Since its launch in March of 2009, the NASA Kepler mission has been monitoring ∼160,000 stars in order to detect transiting extrasolar planets with high relative photometric precision (∼20 ppm in 6.5 hr, Jenkins et al. 2010). In 2013 May, the Kepler main mission ended with the failure of a second reaction wheel; however, the first four years of Kepler data have led to a wealth of planetary discoveries with a total of 4706 announced planet candidates13
13
http://exoplanetarchive.ipac.caltech.edu/ as of 2015 November 11.
(Borucki et al. 2010, 2011; Batalha et al. 2013; Burke et al. 2014). The confirmed and candidate exoplanets typically have orbital periods shorter than 1000 days because at least three detected transits are needed for identification by the automated Transit Planet Search algorithm. Therefore, transiting exoplanets with periods longer than ∼1000 days are easily missed. The detection of short-period planets is further favored because the transit probability decreases linearly with increasing orbital distance. For these reasons, estimates of the statistical occurrence rate of exoplanets tend to focus on orbital periods shorter than a few hundred days (e.g., Dong & Zhu 2013; Fressin et al. 2013; Petigura et al. 2013). Radial velocity (RV) techniques also favor the detection of shorter period orbits. While gas giant planets have been discovered with orbital periods longer than a decade, their smaller reflex velocity restricts detection of sub-Neptune mass planets to orbital radii less than ∼1 AU (Lovis et al. 2011). In principle, astrometric observations favor longer period orbits; however, high precision needs to be maintained over the correspondingly longer time baselines. For shorter periods, the planets need to be massive enough to introduce a detectable astrometric wobble in the star and Gaia should begin to contribute here (Perryman et al. 2001). Microlensing offers sensitivity to planets in wider orbits and has contributed to our statistical knowledge about occurrence rates of longer period planets (e.g., Gaudi 2010; Cassan et al. 2012) and direct imaging of planets in wide orbits is also beginning to contribute important information (Oppenheimer & Hinkley 2009).
|
[
"Borucki et al. 2010"
] |
[
"In 2013 May, the Kepler main mission ended with the failure of a second reaction wheel; however, the first four years of Kepler data have led to a wealth of planetary discoveries with a total of 4706 announced planet candidates"
] |
[
"Background"
] |
[
[
529,
548
]
] |
[
[
226,
453
]
] |
2019MNRAS.484.1912R__Iyyani_et_al._2016_Instance_1
|
Even though most of the energy released by a gamma-ray burst (GRB) is emitted during the prompt emission phase, the emission mechanism is still not understood. In order to determine the emission process typically, the low-energy spectral index, α, of the GRB spectrum is analysed. This analysis has not given a conclusive answer since the α-distribution is broad and has not been uniquely explained by a single emission process. For instance, the peak of the distribution has been used as an argument for synchrotron emission since it is close to the expected value. However, a large fraction (${\sim } 28{{\ \rm per\ cent}}$) was found to be inconsistent with the theoretical limit of −2/3 (‘line of death’; Preece et al. 1998; Ghirlanda, Celotti & Ghisellini 2002; Guiriec et al. 2015; Goldstein et al. 2016; Yu et al. 2016); and only specific physical scenarios remain plausible (large emitting radii and Lorentz factors of the flow, Beniamini & Piran 2013; Iyyani et al. 2016; Beniamini, Barniol Duran & Giannios 2018; Burgess et al. 2018). Alternatively, the spectral width or sharpness angle has been used as a tool to characterize the spectrum, which also take into account the high-energy spectral slope (Axelsson & Borgonovo 2015; Yu et al. 2015b). Around 80 per cent of the bursts were found to have fitted Band spectra that are narrower than what was expected for synchrotron emission. However, direct fitting with a synchrotron model decreases this fraction (Burgess 2017). Therefore, a firm conclusion based on the spectral width or sharpness angle alone cannot either be reached. Yet another approach has been to study the correlation between spectral parameters. A well-studied and strong correlation is the Golenetskii correlation (Golenetskii et al. 1983) which relates the instantaneous flux, F, and peak of the spectrum Epk (Kargatis et al. 1994; Borgonovo & Ryde 2001; Lu et al. 2012). Another prominent correlation is between Epk and α which is valid in a fraction of bursts (Crider et al. 1997; Kaneko et al. 2006). Both of these correlations show a variety of behaviours. In some individual pulses, the spectral parameters track each other, while in others the correlation is different during the rise and decay phases of the pulses. The variety of behaviours have complicated the search for physical explanations.
|
[
"Iyyani et al. 2016"
] |
[
"However, a large fraction (${\\sim } 28{{\\ \\rm per\\ cent}}$) was found to be inconsistent with the theoretical limit of −2/3 (‘line of death’;",
"and only specific physical scenarios remain plausible (large emitting radii and Lorentz factors of the flow,"
] |
[
"Differences",
"Differences"
] |
[
[
961,
979
]
] |
[
[
567,
708
],
[
828,
936
]
] |
2018MNRAS.480.1639D__cent,_Ricci_et_al._2015_Instance_1
|
Six out of the eight sources (75 per cent of the sample) show evidence for high obscuration by cold gas (column density NH≳ 1023 cm−2). The number of AGN with NH≳ 1024 cm−2 is in the range 2–4 when 90 per cent confidence errors on the column density are considered (see Table 4). Although our sample is definitely too small to make any strong conclusion in terms of statistical incidence of obscured AGN in dual systems, we note that the fraction of CT AGN in our sample of dual AGN is 25–50 per cent, i.e. higher than the fraction of hard X-ray selected CT AGN in isolated systems in a similar luminosity interval (BAT: 27 ± 4 per cent, Ricci et al. 2015, NuSTAR: ∼4 per cent, Marchesi et al. 2018). Recently, it has been suggested (Ricci et al. 2017) that AGN in multiple systems in a late stage of merging (with a projected distance d ≤ 10 kpc) are often type 2 AGN due to the huge amount of gas and dust obscuration channelled towards the nuclear region by the close galaxy encounter. Furthermore, in a sample of 44 UltraLuminous IR Galaxies (ULIRG) from the GOALS survey (Sanders et al. 2003) in the local universe, it has been demonstrated that the fraction of CT AGN in late mergers is higher (65$^{+12}_{-13}$ per cent) than in local isolated hard X-ray selected AGN (27$^{+4}_{-4}$ per cent, Ricci et al. 2017), while it is marginally higher than isolated AGN in early mergers (35$^{+13}_{-12}$ per cent). The increasing of the obscuration with the disturbed or interacting morphology of the galaxy host has been also observed up to higher redshift (up to z = 1.5) using X-ray Chandra data from deep surveys (Kocevski et al. 2015). These authors have analysed 154 heavily obscured AGN and compared their morphologies with control samples composed of moderately obscured and unobscured AGN. They found that the fraction of galaxies undergoing mergers is higher in AGN with NH above 3 × 1023 cm−2 with respect to unobscured AGN with NH 1022 cm−2 (7.8$^{+1.9}_{-1.3}$ versus 21.5$^{+4.2}_{-3.3}$ per cent). X-rays (Kocevski et al. 2015; Ricci et al. 2017) as well as more recent mid-IR colour selection (Satyapal et al. 2017) strongly suggest that AGN in late state of merging (10 kpc) are highly absorbed (but see also Ellison et al. 2017), with NH above 1024 cm−2. All these observational results suggest that CT AGN in dual systems correspond to a phase where the supermassive black hole is still accreting gas during an early stage of an interaction or merging.
|
[
"Ricci et al. 2015"
] |
[
"Although our sample is definitely too small to make any strong conclusion in terms of statistical incidence of obscured AGN in dual systems, we note that the fraction of CT AGN in our sample of dual AGN is 25–50 per cent, i.e. higher than the fraction of hard X-ray selected CT AGN in isolated systems in a similar luminosity interval (BAT: 27 ± 4 per cent,"
] |
[
"Compare/Contrast"
] |
[
[
638,
655
]
] |
[
[
280,
637
]
] |
2016ApJ...827...58S__Stern_et_al._2014_Instance_1
|
The multiwavength UV to mid-IR SED of J1224+5555 supports the scenario of a highly obscured AGN, based on the empirically derived templates of Assef et al. (2010). These templates consist of a set of three galaxy templates and one AGN template. The galaxy templates, E, Sbc, and Im, are based on templates from Coleman et al. (1980) with wavelength ranges extended using stellar models from Bruzual & Charlot (2003) and added dust and polycyclic aromatic hydrocarbon components from Devriendt et al. (1999). The AGN template is based on the average type 1 AGN template from Richards et al. (2006) with efforts to remove host galaxy contamination and create a template for an unobscured AGN. This method has since been used to successfully uncover highly obscured and even Compton-thick AGNs in several other works thus far (Chung et al. 2014; Hainline et al. 2014; Lansbury et al. 2014; Stern et al. 2014; Assef et al. 2015; Chen et al. 2015; Lansbury et al. 2015). Figure 7 shows the best-fit linear combination of the three galaxy templates to the observed SDSS, Two Micron All Sky Survey (2MASS), and WISE photometry of the target galaxy, where the best fit is obtained through a
minimization. The multiwavelength fluxes were obtained from the NASA/IPAC Extragalactic Database (NED).17
17
https://ned.ipac.caltech.edu/
The maximum wavelength of the Assef et al. (2010) templates is 30 μm; therefore, we cannot use the IRAS 60 μm flux in our fits. As discussed in Section 5, the 60 μm flux for J1224+5555 is likely overestimated, and its use is in any case questionable. While the SDSS
, 2MASS
, and WISE W1 bands are well fit by a combination of the Sbc and Im templates, the combined model flux is much lower than the observed fluxes in the W2–W4 bands. In fact, Table 2 of Assef et al. (2010) shows that at z = 0 and z = 0.1 no galaxy template has a W1–W2 color greater than 0.2 (Vega). However, when we add in the AGN template with a variable level of obscuration (red line) in Figure 7, we see that the observed multiwavelength photometry of the galaxy is better fit with a combination of a young and old stellar population and highly obscured AGNs (red line) dominating the WISE bands, with a
value almost a factor of 7 times lower than the best fit obtained using the galaxy templates alone. As an additional test, we investigated the effect of adding variable additional extinction to the galaxy templates alone to determine if a comparable fit can be obtained without invoking the presence of an obscured AGN contribution. However, as can be seen from Figure 7, the best-fit model cannot adequately fit the SED in the WISE bands and results in a
value that is significantly worse than the model that includes an AGN (Figure 7). We note that the improvement in fit when the AGN template is added cannot be attributed simply to having additional model parameters in the fit. In the first model without an AGN, we achieve a
of 1323.29, with three model parameters. While the addition of an AGN template more than doubles the number of model parameters to 8,
is reduced by a factor of
, indicating that a model that includes an AGN template produces a significant improvement to the fit of the data.
|
[
"Stern et al. 2014"
] |
[
"This method has since been used to successfully uncover highly obscured and even Compton-thick AGNs in several other works thus far"
] |
[
"Background"
] |
[
[
887,
904
]
] |
[
[
691,
822
]
] |
2017ApJ...845..160P__Camenzind_1986a_Instance_1
|
It is a pressing question as to how the radiation that is observed in relativistic jets in active galactic nuclei (AGNs) is generated (e.g., Blandford & Königl 1979; Marscher 1980; Zensus 1997; Laing & Bridle 2002; Honda 2010; Levinson & Rieger 2011; Mościbrodzka et al. 2011; Ito et al. 2013; Mason et al. 2013; Potter & Cotter 2013; Hovatta et al. 2014; Scott & Stewart 2014; Shih & Stockton 2014; Wang et al. 2014; Turner & Shabala 2015; Asada et al. 2016; Hirotani et al. 2016; Koay et al. 2016; Khabibullin et al. 2016; Prieto et al. 2016). Although there is a common consensus that the emitters are energetic particles, how these particles are accelerated to such high energies, how they dissipate their energy, and how they are transported with the jets themselves are still the subject of feverish investigation (e.g., Blandford & Eichler 1987). It is argued that relativistic outflows from black holes are associated with accretion flows (Blandford 1976; Fender et al. 2004; Meier 2005; Ferreira et al. 2006; Trump et al. 2011; Pu et al. 2012; Wu et al. 2013; Ishibashi et al. 2014; Sbarrato et al. 2014). In the case of collimated relativistic jets, magnetic fields must play an important role (Camenzind 1986a, 1986b, 1987; Fendt & Greiner 2001; Vlahakis & Königl 2004; Komissarov et al. 2007; Lyubarsky 2009; Nakamura & Asada 2013; Homan et al. 2015), and it is argued that jets are powered at the expense of the black hole, wherein energy is extracted from a reservoir of rotational energy from the black hole itself, either by electromagnetic means (Blandford & Znajek 1977; Komissarov 2004, 2005; Toma & Takahara 2014) or through magnetohydrodynamical processes (Phinney 1983; Takahashi et al. 1990; Koide et al. 2002; McKinney & Gammie 2004; Hawley & Krolik 2006). Models have been proposed for both of these cases, and they both in principle possess certain testable predictions. In particular, for the latter, numerical GRMHD simulations (e.g., McKinney 2006) and analytical GRMHD studies (e.g., Takahashi et al. 1990; Pu et al. 2016) consistently show the presence of a stagnation or separation surface (a separatrix). This surface separates the (inner) inflow region from the (outer) outflow region, both of which follow the same global, black-hole-threading magnetic field lines. The relatively slow radial velocities near the stagnation surface imply a high concentration of fluid particles. If energetic particles are injected in the vicinity of the stagnation surface or near the black hole event horizon, they must accumulate in high concentrations near the stagnation surface, provided that the cooling timescale is not significantly shorter than the dynamical timescale of the jet fluid flow. This surface is a unique feature of relativistic GRMHD jets and in contrast to an ideal force-free magnetic jet (e.g., McKinney & Narayan 2007; Tchekhovskoy et al. 2008; Broderick & Loeb 2009).
|
[
"Camenzind 1986a"
] |
[
"In the case of collimated relativistic jets, magnetic fields must play an important role"
] |
[
"Background"
] |
[
[
1205,
1220
]
] |
[
[
1115,
1203
]
] |
2016AandA...586A..81H__Massey_(2002)_Instance_1
|
Table 2
Catalogue description.
Column
Description
1
Source number
2−3
X-ray coordinates, right ascension and declination (epoch 2000.0)
4
Uncertainty of X-ray position [′′]. For XMM-Newton positions taken from Sturm et al. (2013c) the 1σ error includes a systematic uncertainty of 0.5′′.
5
Origin of the X-ray coordinate (A: ASCA, C: Chandra, E: Einstein, I: Integral, N: XMM-Newton, R: ROSAT, S: Swift, X: RXTE). When no reliable position could be determined from the non-imaging RXTE collimator-instruments, a radius of 30′ for the position error indicates the size of the field of view.
6
Reference for source discovery.
7
Identification of optical counterpart with emission-line star from Meyssonnier & Azzopardi (1993). The negative number indicates a star found in the catalogue of Murphy & Bessell (2000).
8−15
Flags indicating different source properties. For their description see Table 3.
16
Confidence class (values 1−6, see Table 4).
17−18
Optical coordinates, right ascension and declination (epoch 2000.0) for the identified counterpart from Zaritsky et al. (2002), or – when not available there – from Massey (2002).
19−26
The Magellanic Clouds Photometric Survey (MCPS): U, error(U), B, error(B), V, error(V), I, error(I) [mag] from Zaritsky et al. (2002).
27−32
Colour indices U−B, error(U−B), B−V, error(B−V), V−I, error(V−I) [mag] derived from MCPS photometry.
33−34
Reddening-free Q-value (Q = U−B−0.72 × (B−V)) and error(Q) [mag].
35
Near-IR counterpart to the optical star from the Two Micron All Sky Survey (2MASS, Skrutskie et al. 2006).
36−41
Near-IR magnitudes with corresponding errors: J, error(J), H, error(H), K, error(K) [mag].
42−45
Near-IR colour indices, J−H, error(J−H), H−K, error(H−K) [mag].
46−53
Spitzer IRAC fluxes at 3.6, 4.5, 5.8 and 8.0 μm [mag] with respective errors (from the SAGE project, for a description see Meixner et al. 2006).
54
Angular distance between X-ray and optical position [′′].
55
Angular distance between optical and near-IR position [′′].
56
Neutron star spin period [s] inferred from X-rays.
57
Orbital period [days] (see flags for origin).
58−59
Maximum and minimum X-ray flux [erg cm-2 s-1] when available in the 0.2−10 keV band. Fluxes in the SMC XMM-Newton catalogue of Sturm et al. (2013c) are given for the 0.2 to 4.5 keV band. To convert them into the 0.2−10 keV band, we multiplied them by a factor of 2.6 assuming a standard power law with photon index 0.9 (Haberl et al. 2008) and a column density of 1021 cm-2 (solar abundance). For Swift XRT count rates we used a flux conversion factor of 1.1 × 10-10 erg cm-2 cts-1
60
Flag for minimum flux: 1 for a non-detection with an upper limit; −1 when unknown; 0 for detection.
61−62
References for maximum and minimum X-ray flux.
63
X-ray variability factor (ratio of maximum to minimum flux).
64
Equivalent width of the Hα line [Å] (minimum value if more than one measurement is available).
65
Maximum equivalent width of the Hα line [Å].
66
References for the Hα measurements.
67
Comments with key references.
|
[
"Massey (2002)"
] |
[
"Optical coordinates, right ascension and declination (epoch 2000.0) for the identified counterpart from Zaritsky et al. (2002), or – when not available there – from"
] |
[
"Uses"
] |
[
[
1148,
1161
]
] |
[
[
983,
1147
]
] |
2019ApJ...875...90L__Velli_et_al._2015_Instance_1
|
When energy flows from the interior of the Sun outward into the solar atmosphere, why is the Sun’s outer atmosphere, the corona, much hotter than the inner atmosphere, the underlying chromosphere and photosphere? This is the long-standing problem of the coronal heating, which is one of the eight key mysteries in modern astronomy (Kerr 2012). For about 80 yr since the discovery of the extremely hot corona around the late 1930s (Grotian 1939; Edlen 1945), people have worked hard on addressing this issue, and great advances have been made in observation and theoretical studies (Parnell & De Moortel 2012; Amari et al. 2015; Arregui 2015; Cargill et al. 2015; De Moortel & Browning 2015; Jess et al. 2015; Klimchuk 2015; Longcope & Tarr 2015; Peter 2015; Schmelz & Winebarger 2015; Velli et al. 2015; Wilmot-Smith 2015). Especially during recent decades, high-resolution observations of solar super-fine structures indicate that small spicules, minor hot jets along small-scale magnetic channels from the low atmosphere upwards to the corona, petty tornados and cyclones, and small explosive phenomena such as mini-filament eruptions and micro- and nano-flares—all of these small-scale magnetic activities contribute greatly to coronal heating (De Pontieu et al. 2011; 2018; Zhang & Liu 2011; Parnell & De Moortel 2012; Klimchuk 2015; Peter 2015; Schmelz & Winebarger 2015; Henriques et al. 2016; Li et al. 2018a). Additionally, contributions of MHD waves to heating the corona have been observationally illustrated (van Ballegooijen et al. 2011; Jess et al. 2015; Kubo et al. 2016; Morton et al. 2016; Soler et al. 2017; Morgan & Hutton 2018). Meanwhile, with the progress of observational studies, two groups of theoretical models, magnetic reconnection models and magnetohydrodynamic wave models, have traditionally attempted to explain coronal heating, but so far no models can address the key mystery perfectly (van Ballegooijen et al. 2011; Arregui 2015; Cargill et al. 2015; Peter 2015; Velli et al. 2015; Wilmot-Smith 2015). Maybe we do not need to intentionally take to heart such the classical dichotomy, because waves and reconnections may interact with each other (De Moortel & Browning 2015; Velli et al. 2015). Additionally, statistical studies may look at coronal heating from a comprehensive perspective. Li et al. (2018b) found that the long-term variation of the heated corona, which is represented by coronal spectral irradiances, and that of small-scale magnetic activity are in lockstep, indicating that the corona should statistically be effectively heated by small-scale magnetic activity. Observational and theoretical model studies through heating channels and modes, and statistical studies by means of heating effect (performance of the heated corona), both suggest that coronal heating originates from small-scale magnetic activity.
|
[
"Velli et al. 2015"
] |
[
"For about 80 yr since the discovery of the extremely hot corona around the late 1930s (Grotian 1939; Edlen 1945), people have worked hard on addressing this issue, and great advances have been made in observation and theoretical studies"
] |
[
"Background"
] |
[
[
785,
802
]
] |
[
[
344,
580
]
] |
2021AandA...653A..85S__However,_Genovali_et_al._(2014)_Instance_1
|
Figure 6 shows the orbital eccentricities as a function of [M/H] for the metal-rich disc sample. The solid lines correspond to the required eccentricity (see Eq. (2)) for different values of ISM radial metallicity gradients: −0.10 dex kpc−1 (black), −0.07 dex kpc−1 (our measured gradient for young stars in Table 1; see also Minchev et al. 2018, red), −0.04 dex kpc−1 (orange), and −0.06 dex kpc−1 (Cepheids analysis from Genovali et al. 2014, green). For the three first cases, we assumed ISM[M/H](R⊙) = 0.0 to estimate Rbirth from the stellar metallicity. However, Genovali et al. (2014) have their own zero point, defined as: [Fe/H] = −0.06 * Rg + 0.57, with a clear shift in the relation compared to the other ones assumed in this work. The impact of the ISM gradient value and the zero-point assumption on the derived Rbirth, and therefore on the required eccentricity to reach the solar vicinity without the need for churning, is clearly observed. As described in Hayden et al. (2020), given the measured [M/H] and eccentricity, stars lying to the left are able to reach the solar neighbourhood through blurring, while the stars to the right of the line are possible candidates to have migrated through churning. This is the case for most of the SMR stars (70% of the SMR stars lie below the line that corresponds to the Cepheids analysis); they are therefore likely to have been brought to the solar neighbourhood by churning, which is in close agreement with previous studies (e.g., Kordopatis et al. 2015a; Wojno et al. 2016). However, it is worth noting that the observed metallicity distribution function in Fig. 2 peaks around 0.2 dex, which is higher than previous reported solar vicinity MDFs (see e.g., Fuhrmann et al. 2017). A possible ignored bias towards more metal-rich objects in the sample selection could be pulling the percentage of possible migrators to higher values. Among the entire distribution, our churned candidates with [M/H] > + 0.1 comprise around 17% of the sample. If we constrain the number of migrators to only stars with [M/H] > + 0.25, the global percentage decreases to 8% of the sample.
|
[
"Genovali et al. 2014"
] |
[
"Cepheids analysis from",
"green"
] |
[
"Uses",
"Uses"
] |
[
[
423,
443
]
] |
[
[
400,
422
],
[
445,
450
]
] |
2016ApJ...830...28P__Zank_et_al._2015_Instance_1
|
Our basic model for acceleration is shown in Figure 1. The left panel shows a cartoon of the model for acceleration, transport, and radiation at the X-reconnection site in the corona. Acceleration can be by the second order Fermi (or stochastic) acceleration by turbulence (see, e.g., Petrosian & Liu 2004), by a standing shock produced by the downflow from the X-reconnection site (see, e.g., Guo & Giacalone 2012), or in the merging of islands shown to arise in PIC simulations during reconnection (Drake et al. 2006, 2013; Le et al. 2012; Oka et al. 2010). The latter model has been invoked as a possible mechanism in the downstream of the CME shock (Le Roux et al. 2015; Zank et al. 2015). The right panel shows a similar cartoon for acceleration in the environment of the CME. The (red) rectangles show the cross section of the box within which particles are accelerated. For our purposes here the relevant parameters of the acceleration and transport of particles are the momentum (p) and pitch angle cosine (μ) diffusion rates
and
, and direct energy gain
and loss
rates, where E is the particle energy.1
1
Here and in what follows we neglect the effects of the third diffusion coefficient
, which are generally small (see Schlickeiser 1989; Petrosian & Liu 2004).
For an isotropic pitch angle distribution the evolution of the volume-integrated energy spectrum,
, is governed by the so-called leaky-box model kinetic equation (see Petrosian 2012)
1
where
is a source term,
, and the direct acceleration coefficients for stochastic and shock accelerations are given as
2
Here
is the shock velocity,
is the coefficient of spatial diffusion (along the field lines),
,
is the Lorentz factor, and ζ depends on the shock compression ratio and other factors (see, e.g., Steinacker et al. 1988). The energy-loss rate at low energies (for both electrons and ions) is dominated by Coulomb collisions with background particles (mainly electrons).2
2
Coulomb collisions also cause pitch angle scattering, and therefore spatial diffusion along the field lines, with a rate that is comparable to the energy-loss rate at nonrelativistic energies but decreases as
. They also cause energy diffusion, which is negligible in the cold target case (
), but can be comparable to the energy-loss rate for a hot target as E approaches kT (see Petrosian & Kang 2015), where T is the background plasma temperature.
At higher energies (not relevant for the discussion here) inelastic interactions (synchrotron, inverse Compton, and bremsstrahlung) for electrons and (nuclear line excitation, neutron and pion productions) for ions become important.
|
[
"Zank et al. 2015"
] |
[
"The latter model has been invoked as a possible mechanism in the downstream of the CME shock"
] |
[
"Uses"
] |
[
[
675,
691
]
] |
[
[
560,
652
]
] |
2017ApJ...850..197P__Langer_2012_Instance_1
|
In order to explode as an ECSN, several ingredients need to be in place. Nomoto (1984) argued that stars with helium cores between 2.0 and
(which corresponds roughly to initial masses between 8 and
) would explode as an ECSN. His models, however, did not develop a second dredge-up, which can significantly reduce the mass of the helium core and diminishes the predictive power of this criterion (see Podsiadlowski et al. 2004; Poelarends et al. 2008). Since then, several authors (Siess 2007; Poelarends et al. 2008; Doherty et al. 2010, 2015; Jones et al. 2013) have established precise initial mass ranges for ECSNe to occur, although these mass ranges are highly sensitive to the adopted convection criteria, overshooting, and mass-loss prescriptions (Poelarends et al. 2008; Doherty et al. 2010; Langer 2012). Jones et al. (2013) produced several detailed models, computed all the way to electron captures on 24Mg and 20Ne and found that CO cores with masses over
are able to reach densities high enough for this to occur (
g cm−3). If the CO core is massive enough, neon will ignite off-center (Jones et al. 2013; Schwab et al. 2016), but Jones et al. (2014) also found that the upper boundary for ECSNe is affected by uncertainties regarding the progression or stalling of the neon flame. There seems to be consensus, however, that the mass of the CO core is a reliable indicator for the final fate of stars in this mass range. How this translates into the initial mass of the star depends on the adopted convection criterion, with the Schwarzschild criterion producing more massive cores than the Ledoux criterion for the same initial mass. Inclusion of overshooting will also lead to larger cores (see Siess 2007). However, although the initial mass range for ECSNe is therefore quite sensitive to the adopted convection criteria, this does not seem to be the case for the final MCO, as most authors find similar values for MCO at which neon ignites, even though they treat convection differently. In the context of binary evolution, this provides an additional reason to adopt MCO as an indicator for whether the star explodes as an ECSN or not, as the CO core is generally not eroded by mass transfer, in contrast to the He core.
|
[
"Langer 2012"
] |
[
"Since then, several authors",
"have established precise initial mass ranges for ECSNe to occur, although these mass ranges are highly sensitive to the adopted convection criteria, overshooting, and mass-loss prescriptions"
] |
[
"Background",
"Background"
] |
[
[
814,
825
]
] |
[
[
466,
493
],
[
577,
767
]
] |
2022AandA...666A.112L__Cormier_et_al._2015_Instance_2
|
Local dwarf galaxies were the focus of large Herschel and Spitzer surveys (e.g., The Dwarf Galaxy Survey, DGS; Madden et al. 2006). Studies on both resolved and integrated-galaxy scales have highlighted some distinctively unique observational signatures of star-forming low-metallicity dwarf galaxies. A non-linear relation of the dust-to-gas mass (D/G) with metallicity is observed, with extremely low dust masses observed for the lowest metallicity galaxies (Z ≤ 0.1 Z⊙; Herrera-Camus et al. 2012; Fisher et al. 2014; Rémy-Ruyer et al. 2015; Galliano et al. 2018, 2021; Cigan et al. 2021). Furthermore, the hard radiation fields in star-forming dwarf galaxies, along with their lower dust abundance, result in extended ionized gas regions prominent on global galaxy scales (Hunter et al. 2011; Cormier et al. 2015, 2019). The consequence is often a largely photodissociated molecular phase, existing in clumps which are difficult to observe with the usual molecular gas tracer, CO (1-0) (Cormier et al. 2014; Hunt et al. 2015; Accurso et al. 2017b), beckoning the presence of a CO-dark molecular gas phase (Grenier et al. 2005; Röllig et al. 2006; Wolfire et al. 2010; Glover & Clark 2012; Bolatto et al. 2013; Accurso et al. 2017a; Madden et al. 2020). Other emission lines, however, such as the far-infrared [C ii]λl58 µm line, are strikingly enhanced on global scales in dwarf galaxies (e.g., Cormier et al. 2015, 2019; Cigan et al. 2016; Lebouteiller et al. 2017; Jameson et al. 2018), making the [C ii]λl58 µm line a potential tool for tracing star formation activity (Malhotra et al. 2001; Papadopoulos et al. 2007; Pineda et al. 2014; De Looze et al. 2014; Olsen et al. 2015; Herrera-Camus et al. 2015, Herrera-Camus et al. 2018; Carniani et al. 2018; Matthee et al. 2019; Izumi et al. 2021; Fujimoto et al. 2021) and a tracer of the total H2 in galaxies, near and far (Poglitsch et al. 1995; Wolfire et al. 2010; Pineda et al. 2013; Nordon & Sternberg 2016; Fahrion et al. 2017;
Accurso et al. 2017b; Zanella et al. 2018; Madden et al. 2020; Schaerer et al. 2020; Tacconi et al. 2020).
|
[
"Cormier et al. 2015"
] |
[
"Other emission lines, however, such as the far-infrared [C ii]λl58 µm line, are strikingly enhanced on global scales in dwarf galaxies (e.g.,"
] |
[
"Compare/Contrast"
] |
[
[
1398,
1417
]
] |
[
[
1256,
1397
]
] |
2022AandA...660A.135V__Nissen_2016_Instance_1
|
The first study, to our knowledge, to notice the net increase in the abundance of slow (s) neutron capture elements in young stellar populations is D’Orazi et al. (2009), in which the abundance of barium in young star clusters was seen to be higher than in the older ones. Maiorca et al. (2011, 2012) added a few more elements with important s-process contributions (yttrium, zirconium, lanthanum, and cerium), confirming the increasing trend towards younger ages. Subsequently, a number of works have attempted to both clarify the origin of this increase (see, e.g., Bisterzo et al. 2014; Mishenina et al. 2015; Trippella et al. 2016; Magrini et al. 2018; Spina et al. 2018; Busso et al. 2021) and to use their abundances to estimate the ages of stars, often using neutron capture s-process elements in combination with other elements with opposite behaviours, such as α elements – that we indicate as chemical clocks – and thus maximising the dependence of the relationship with age (see, e.g., Tucci Maia et al. 2016; Nissen 2016; Feltzing et al. 2017; Fuhrmann et al. 2017; Slumstrup et al. 2017; Titarenko et al. 2019). Once the existence of a relationship between age and chemical clocks was established (see, e.g., Spina et al. 2016; Delgado Mena et al. 2019; Jofré et al. 2020), the next steps were the following: (i) to clarify the applicability of these relationships with luminosity class (dwarf or giant) (see, e.g., Tucci Maia et al. 2016; Slumstrup et al. 2017; Casamiquela et al. 2021), metallicity (see, e.g., Feltzing et al. 2017; Delgado Mena et al. 2019; Casali et al. 2020), and population type (thin disc, thick disc, halo) (see, e.g., Titarenko et al. 2019; Nissen et al. 2020; Tautvaišienė et al. 2021), or even in dwarf galaxies (Skúladóttir et al. 2019; ii) to calibrate them with a sample of stars with reliable age determination, which are usually open star clusters (OCs), solar twins, or targets with asteroseismic observations. Finally, it is essential to understand whether these relationships are valid throughout the Galactic disc, or whether they are necessarily position-dependent. For the first time, Casali et al. (2020) applied the relations derived from a large sample of solar-like stars located in the solar neighbourhood and noted that they fail to reproduce the ages of star clusters in the inner disc. They concluded that the relationship between age and chemical clocks is not universal and that it varies with galactocentric position. Later, Magrini et al. (2021b) suggested that the differences in the relationships between age and chemical clocks in different parts of the Galactic disc are due to the strong dependence on the metallicity of the yields of low-mass stars, which produce s-process elements during the final stages of their evolution. Casamiquela et al. (2021) used red clump stars in open clusters to investigate the age dependence of several abundance ratios, including those that contain s-process and α elements. They found that the relationship between [Y/Mg] and ages outlined by open clusters is similar to the one found using solar twins in the solar neighbourhood. They also found that the abundance ratios involving Ba are those with the highest correlation with age. However, they also note that as one moves away from the solar neighbourhood, the dispersion increases and is in agreement with the findings of Casali et al. (2020), which attributed this to the spatial variation of the star formation history along the galactocentric radius.
|
[
"Nissen 2016"
] |
[
"Subsequently, a number of works have",
"and to use their abundances to estimate the ages of stars, often using neutron capture s-process elements in combination with other elements with opposite behaviours, such as α elements – that we indicate as chemical clocks – and thus maximising the dependence of the relationship with age (see, e.g.,"
] |
[
"Background",
"Background"
] |
[
[
1021,
1032
]
] |
[
[
465,
501
],
[
695,
996
]
] |
2018AandA...616A..43S__Breeveld_&_Puchnarewicz_1998_Instance_1
|
Within the same scenario, Smith et al. (2004, 2005) have theoretically predicted that, when observed in polarized light, NLS1s should have a low polarization fraction, the polarization angle should change monotonically from one wing of the BELs to the other, and the BELs should be significantly broader than those observed in direct light (Smith et al. 2004, 2005). We assume that polarized BELs are scattered into the line of sight by material that is close to coplanar with the BLR (i.e. the torus). This prediction has been searched for in RQ-NLS1s, but many objects have not been detected in polarized light, and none of the detected show any significant broadening of the BELs (e.g. Goodrich 1989; Breeveld & Puchnarewicz 1998; Kay et al. 1999). On the other hand, only one RL-NLS1 (PKS 2004-447) has been the object of a spectropolarimetry study (Baldi et al. 2016), and the three predicted features have been observed that have a FWHM in the polarized Hα of ≈ 9000 km s−1. We propose for the first time a simple scenario that can resolve such apparent tension between the two classes of NLS1s. We speculate that in those NLS1s that are intrinsically without a jet (a fraction of the RQ-NLS1) a significant amount of polarized light could be due to polar scattering of the BELs, by material above the BLR, as observed for type II AGN (Antonucci 1983; Miller & Antonucci 1983; Antonucci & Miller 1985). A polar-scattered component could easily overwhelm the planar-scattering contribution, predicted to result in low polarization fractions for face-on objects because of the axi-symmetry of the scattering material. In RL-NLS1s, on the contrary, the jet could evacuate the intervening material, resulting in a lower degree of polarization in which only the broadened component of the BEL is left. The evacuation of the polar material by a jet closely aligned withthe line of sight has been already proposed by Ghisellini & Sbarrato (2016) to account for the lack of high-z blazar parent population. This simple speculative scenario can be tested through spectropolarimetry of a larger sample of RL-NLS1s.
|
[
"Breeveld & Puchnarewicz 1998"
] |
[
"We assume that polarized BELs are scattered into the line of sight by material that is close to coplanar with the BLR (i.e. the torus). This prediction has been searched for in RQ-NLS1s, but many objects have not been detected in polarized light, and none of the detected show any significant broadening of the BELs (e.g."
] |
[
"Compare/Contrast"
] |
[
[
704,
732
]
] |
[
[
367,
688
]
] |
2019MNRAS.490.4975R__Neilson_et_al._2012_Instance_1
|
Fig. 6 displays the MLR for the investigated CCs whose intrinsic stellar parameters were derived from the best-fitting models listed in Table 1. These data are compared with the predicted canonical (no overshooting, no mass-loss) MLR (Bono et al. 2000; the solid lines) and with the relations obtained by increasing the zero-point of the canonical MLR by 0.25 dex (the dashed lines) and 0.5 dex (the dotted lines) to reproduce the effect of mild and full overshooting,5 respectively (see Chiosi et al. 1993; Bono et al. 1999, for details). Inclusion of mass-loss and/or rotation would produce a similar increase in the Cepheid luminosity level at fixed mass (see Neilson et al. 2012, for details). As the light curves of $OGLE\_CEP\_LMC\_2019$ are best reproduced adopting a different value of the helium content (see above), in Fig. 6 we also show the MLR for Y = 0.30, Z = 0.008 (the green lines). Note that this relation is slightly more luminous than those calculated for the standard LMC elemental composition (Y = 0.25, Z = 0.008). According to the location of the variables in the ML plane, the canonical MLR is not strictly satisfied, as the points are spread between the canonical and full overshooting predictions. Even if at this stage we cannot disentangle the role of overshooting, mass-loss, and rotation in producing the quoted excess luminosity, at fixed mass, the detected dispersion might indicate a combination of these different non-canonical phenomena. Indeed, if only overshooting were efficient, one would in principle expect the same amount of excess luminosity for all stellar masses (within small uncertainties). Rotation produces similar effects as overshooting because it implies a larger He burning core and a brighter luminosity at fixed mass (see e.g. Anderson et al. 2016) On the other hand, if the mass-loss process were efficient, this could be inferred from the predicted deviation of the best-fitting stellar mass from the value corresponding to the canonical MLR. Such a deviation is represented in Fig. 7 as a function of the canonical mass (top) and of the pulsation period (bottom) for the CCs in our sample. We note that the expected mass differences range from $0{{\ \rm per\ cent}}$ to almost $\sim 50{{\ \rm per\ cent}}$ and are not clearly correlated with the pulsation period or the stellar mass.
|
[
"Neilson et al. 2012"
] |
[
"Inclusion of mass-loss and/or rotation would produce a similar increase in the Cepheid luminosity level at fixed mass (see",
"for details)."
] |
[
"Uses",
"Uses"
] |
[
[
663,
682
]
] |
[
[
540,
662
],
[
684,
697
]
] |
2017MNRAS.472..876O__Rodono_et_al._1985_Instance_1
|
Leitzinger et al. (2014) briefly compared the mass estimates of stellar mass ejection events and the estimated X-ray energies of their associated flares with the solar scaling. We extend their comparison here. Specifically, we consider the events observed as blueshifted extra emissions in Balmer lines on the active M dwarfs AD Leo (Houdebine et al. 1990), AT Mic (Gunn et al. 1994), V374 Peg (Vida et al. 2016) and the pre-main-sequence star DZ Cha (Guenther & Emerson 1997). Published estimates of their minimum masses are 7.7 × 1017, ∼1015, ∼1016 and 1.4 × 1018…7.8 × 1019 g, respectively. Since no simultaneous X-ray observations of the associated flares are available, we estimate the corresponding X-ray flare energies based on the assumption that the published U-band energies are of comparable magnitude (cf. Hawley & Pettersen 1991). Doing so, we find EX ≈ EU ∼ 2 × 1032 erg for AD Leo (Rodono et al. 1985; Hawley & Pettersen 1991), 3 × 1031 erg for AT Mic (Gunn et al. 1994) and 1.2…2.5 × 1035 erg for DZ Cha (Guenther & Emerson 1997), respectively. The recently published event at V374 Peg has an Hγ flare energy of ∼4 × 1030 erg, which corresponds to ∼1032 erg in X-rays (Butler, Rodono & Foing 1988). We add two other events, which are, however, more uncertain in their interpretation as real CME events. Doyle et al. (1988) observed a strong increase in neutral hydrogen column density during a flare on the active M dwarf YZ CMi, which may be interpreted as a rising filament obscuring parts of the flare region. The estimated mass of this event is 3 × 1017 g and the estimated X-ray flare energy EX ≈ EU ∼ 8 × 1030 erg, since the observed X-ray flare emission was likely partly absorbed by the neutral material (Doyle et al. 1988). The second one is a long-decay flare on the young, active M dwarf AU Mic, which was interpreted as an eruptive event with a CME mass in the order of 1020 g by Cully et al. (1994). However, there is a different model of this strong flare (∼3 × 1035 erg) including post-eruptive energy release, which does not involve a CME (Katsova, Drake & Livshits 1999). Other studies that observed stellar CME events (cf. Section 1) did not provide mass estimates and had to be excluded.
|
[
"Rodono et al. 1985"
] |
[
"Doing so, we find EX ≈ EU ∼ 2 × 1032 erg for AD Leo"
] |
[
"Uses"
] |
[
[
897,
915
]
] |
[
[
844,
895
]
] |
2022MNRAS.510.5088B__Faisst_et_al._2017_Instance_2
|
In the expectation of a relationship between the observed IRX and the rest-frame UV slope there is the assumption that the stars and dust are well mixed, which leads to the coupling of any observed reddening in the UV to the FIR emission detected (e.g. Meurer et al. 1999; Charlot & Fall 2000; Calzetti 2001). If instead the galaxy consists of regions of significantly different obscuration, then the relationship will break down for the galaxy as a whole. Indeed, geometric effects have been put forward as an explanation of the discrepant results at z > 5 (Faisst et al. 2017; Popping, Somerville & Galametz 2017). In local starburst galaxies, the existence of an IRX–β relation and a clear morphological similarity, indicates that the rest-frame UV emission from young stars is being attenuated from dust that is tracing broadly the same star-forming regions of the galaxy (e.g. in the spiral arms; Kennicutt et al. 2003; Gil de Paz et al. 2007). In the high-redshift Universe, however, where galaxies become more turbulent and irregular (e.g. Förster Schreiber et al. 2011; Buitrago et al. 2013; Guo et al. 2015) the expected morphology of the dust relative to the observed UV emission is not clear. Evidence for offset dust continuum emission relative to the rest-UV has been identified in several high-redshift Lyman-break galaxies (LBGs; Koprowski et al. 2016; Faisst et al. 2017; Laporte et al. 2017; Bowler et al. 2018), and similar trends have been found when comparing the [C ii] FIR line and the rest-UV continuum (e.g. Maiolino et al. 2015; Carniani et al. 2017). Although some of these offsets have been attributed to astrometric systematics (e.g. Dunlop et al. 2017) there is a growing consensus that FIR continuum and line emission are frequently physically offset as compared to the observed rest-UV emission (see Carniani et al. 2018). Whether high-redshift galaxies show large and distinct regions of obscured and unobscured star formation has implications for the use of the IRX–β relation in deriving the cosmic SFR density (e.g. Bouwens et al. 2016b), as the assumed energy balance will break down (Buat et al. 2019) and the global β measurement will not be representative of the full source.
|
[
"Faisst et al. 2017"
] |
[
"Evidence for offset dust continuum emission relative to the rest-UV has been identified in several high-redshift Lyman-break galaxies (LBGs;"
] |
[
"Compare/Contrast"
] |
[
[
1368,
1386
]
] |
[
[
1204,
1344
]
] |
2017ApJ...844...17P__Xie_et_al._2004_Instance_1
|
Observations by a single coronagraph allow us to obtain CME 2D images projected on the plane of the sky. Using the image sequences of CME expansions, we can determine CME 2D parameters such as speed and angular width. Such observations do have some limitations, mostly the projection effect. Kwon et al. (2015) showed that the apparent width of CMEs does not represent an accurate measure of the CME ejecta size. In previous studies, geometrical models based on the assumption of different cone shapes were developed to obtain CME 3D parameters from single spacecraft observations (Xie et al. 2004; Xue et al. 2005; Michalek 2006). Na et al. (2013) compared the 3D parameters of halo CMEs obtained from the cone shapes (elliptical-cone, ice-cream-cone, and asymmetric-cone models) and found that their 3D speeds are comparable to one another but their angular widths and inclination angles are quite different. We note that stereoscopic observations by twin spacecraft STEREO make it possible to obtain the CME 3D parameters using geometrical triangulation methods (Mierla et al. 2008, 2009; Liewer et al. 2009; Temmer et al. 2009; Liu et al. 2010a, 2010b) and a flux rope method (Chen et al. 2000; Thernisien et al. 2009). Recently, the STEREO CME Analysis Tool (StereoCAT), based on a triangulation method from multiple spacecraft observations, was developed by the Community Coordinated Modeling Center (CCMC) for real-time CME analysis carried out by the CCMC/SWRC forecasting team. CME 3D parameters can be determined by selecting two coronagraphs from SOHO/LASCO C3 and STEREO/SECCHI COR2 images using the model. The triangulation algorithm and the performance of the CME measurements from StereoCAT are described in Sections 3.1 and 3.2 of Mays et al. (2015). Lee et al. (2015) showed, using 44 halo CMEs from 2010 to 2011, that the CME 3D parameters (speed, angular width, and source location) obtained from StereoCAT are similar to those obtained from the graduated cylindrical shape flux rope model (Thernisien et al. 2009; Thernisien 2011). Jang et al. (2016) showed, using 306 LASCO front-side halo CMEs from 2009 to 2013, that around 20% of the CME 2D speeds in the LASCO CME catalog are underestimated compared with the 3D ones obtained from StereoCAT.
|
[
"Xie et al. 2004"
] |
[
"In previous studies, geometrical models based on the assumption of different cone shapes were developed to obtain CME 3D parameters from single spacecraft observations"
] |
[
"Background"
] |
[
[
582,
597
]
] |
[
[
413,
580
]
] |
2017ApJ...850...75S__Bugaev_et_al._2016_Instance_2
|
A realistic EoS that is able to reproduce the properties of compact astrophysical objects has to fulfill several requirements. The possibility of including many particle species, which is known as multicomponent character, is of crucial importance for modeling the NS interiors, which in even the simplest treatment include neutrons, protons, and electrons, while more advanced descriptions have to account for the presence of hyperons (Schaffner-Bielich et al. 2002). Therefore, the grand canonical ensemble is the natural choice for the formulation of such an EoS. Another element of the realistic phenomenological hadronic EoS corresponds to the short-range repulsive interaction of the hard core nature between particles (Andronic et al. 2006; Bugaev et al. 2016). Analysis of the particle yields produced in relativistic A+A collisions within statistical (thermal) models, i.e., the Hadron Resonance Gas (HRG) model (Bugaev et al. 2016), shows the importance of the particle hard core repulsion. In this approach every particle species is defined as a rigid sphere with a fixed radius estimated from experimental data analysis. These radii do not exceed 0.5 fm (Andronic et al. 2017; Sagun et al. 2017a). Note that the hard core of hadrons in phenomenological EoSs is important in order to suppress thermal excitations of the hadronic spectrum and provide deconfinement of the color degrees of freedom expected at high temperatures/densities (Satz 2012). Another requirement to the phenomenological EoS is related to its causal behavior when the speed of sound cannot exceed the speed of light. At sufficiently high densities this condition is violated by the hard core repulsion. As was shown by Sagun et al. (2017a), introducing the induced surface tension (IST) of particles to the model with the hard core repulsion between an arbitrary number of hadron species makes the EoS significantly softer and extends its causality range up to 7.5 normal nuclear densities, where formation of the quark-gluon plasma is expected. The IST is the key element of this approach (Sagun et al. 2014), as it allows us to account for the hard core repulsion between constituents in the most accurate way, and to properly reproduce the virial expansion of the multicomponent EoS. Recently, the IST EoS was used to describe the experimental data of hadron multiplicities measured at AGS, SPS, RHIC, and LHC energies of nuclear collisions (Sagun et al. 2017b), as well as the nuclear matter properties (Sagun et al. 2014). In this work the focus is on the application of IST EoS to the study of NS properties.
|
[
"Bugaev et al. 2016"
] |
[
"Analysis of the particle yields produced in relativistic A+A collisions within statistical (thermal) models, i.e., the Hadron Resonance Gas (HRG) model",
"shows the importance of the particle hard core repulsion."
] |
[
"Background",
"Background"
] |
[
[
923,
941
]
] |
[
[
770,
921
],
[
944,
1001
]
] |
2022MNRAS.509.3599T__Du_et_al._2015_Instance_3
|
Here we report the X-ray spectral and timing analysis of the joint XMM–Newton and NuSTAR observations of an IRAS 04416+1215, a nearby (z = 0.0889; Boller et al. 1992) hyper-Eddington AGN. The source is part of a XMM–Newton/NuSTAR campaign that aims to constrain the broad-band X-ray properties of eight super-Eddington AGN from the best sample of bona fide super-Eddington sources available, i.e. super-Eddington accreting massive black holes (SEAMBHs; Du et al. 2014, 2015; Wang et al. 2014) that contains exclusively objects with black hole masses estimated from reverberation mapping. In this campaign we are carrying out to study the broad-band X-ray properties of super-Eddington AGN, all the sources have new NuSTAR observations performed simultaneously with XMM–Newton or Swift-X-ray Telescope (XRT). IRAS 04416+1215 has bolometric luminosity $\log (L_{\rm bol}/\rm erg\, s^{-1})=47.55$, according to Castelló-Mor, Netzer & Kaspi (2016), and $\log (L_{\rm bol}/\rm erg\, s^{-1})=45.52$, according to Liu et al. (2021). The former estimate is computed using, for the SED fitting procedure, the Slone & Netzer (2012) code, including the comparison of the observed SED with various combinations of disc SEDs covering the range of mass, accretion rate, spins, and taking into account the correction for intrinsic reddening and host galaxy contribution. In the latter estimate, the SED fitting is done using the more semplicistic templates from Krawczyk et al. (2013). The dimensionless accretion rate (Du et al. 2014) and black hole mass of the source are $\log (\dot{\mathscr {M}})$ = $2.63^{+0.16}_{-0.67}$ and log (MBH/M⊙) = $6.78^{+0.31}_{-0.06}$ with the reverberation mapping technique (Du et al. 2015), respectively, where $\dot{\mathscr {M}}\equiv \dot{M}_{\bullet }c^2/L_{\rm Edd}$, $\dot{M}_{\bullet }$ is mass accretion rates, c is speed of light, and LEdd is the Eddington luminosity. The dimensionless accretion rate is estimated by $\dot{\mathscr {M}}=20.1\, \ell _{44}^{3/2}M_7^{-2}$ from the Shakura–Sunyaev disc model (Du et al. 2015), where ℓ44 is the 5100 Å luminosity in units of $10^{44}\, {\rm erg\, s^{-1}}$ and $M_7=M_{\bullet }/10^7\, \mathrm{M}_{\odot }$. This approximation is valid for $\dot{\mathscr {M}}\lesssim 10^3$. To compute the Eddington ratio we assumed the bolometric luminosity value from Castelló-Mor et al. (2016), which is a better and more trustable estimate of the bolometric luminosity of the source, obtaining λEdd ∼ 472. This value is in perfect agreement with the dimensionless accretion rate from Du et al. (2014). However even assuming the luminosity from Liu et al. (2021), with which the value of the accretion rate would be λEdd ∼ 4.40, the source would remain a super-Eddington accreting AGN. IRAS 04416+1215 turned out to be the most peculiar of our sample, it is classified as NLS1 galaxy, showing narrow Hβ line [full width at half-maximum (FWHM) = $1670 \, \rm km \, \rm s^{-1}$; Moran, Halpern & Helfand 1996] and very broad [O iii] (FWHM = $1150 \, \rm km \, \rm s^{-1}$; Véron-Cetty, Véron & Gonçalves 2001) lines, which is typically found in sources accreting at such high Eddington accretion rates (Greene & Ho 2005; Ho 2009). The source shows a photon index in the Roentgen Satellite (ROSAT) (0.1–2.4 keV) energy band, of Γ = 2.96 ± 0.50 (Boller et al. 1992) and of $\Gamma =2.46^{+0.27}_{-0.26}$ for the rest-frame >2 keV spectrum, according to Liu et al. (2021).
|
[
"Du et al. 2015"
] |
[
"The dimensionless accretion rate is estimated by $\\dot{\\mathscr {M}}=20.1\\, \\ell _{44}^{3/2}M_7^{-2}$ from the Shakura–Sunyaev disc model",
"where ℓ44 is the 5100 Å luminosity in units of $10^{44}\\, {\\rm erg\\, s^{-1}}$ and $M_7=M_{\\bullet }/10^7\\, \\mathrm{M}_{\\odot }$."
] |
[
"Uses",
"Uses"
] |
[
[
2039,
2053
]
] |
[
[
1900,
2037
],
[
2056,
2184
]
] |
2016AandA...592A..74S__Sobolewska_&_Papadakis_(2009)_Instance_2
|
In Fig. B.1 we plot the soft X-ray light curves for our candidate highly variable AGN using available X-ray data taken by the satellite missions Einstein, ROSAT, XMM, Suzaku and Swift. The count rates were obtained from different archives including HEASARC, the XMM Science Archive, the Swift UKSSDC and from our own Swift XRT data analysis, and for upper limits the 1SXPS catalogue (Evans et al. 2014) and the XMM upper limit server2 were queried. The count rates of the different satellites were converted to fluxes between 0.2–2.0 keV using PIMMS3 assuming a power law with a photon index of 1.7 as a spectral shape taking into account Galactic extinction as given by Willingale et al. (2013). Sobolewska & Papadakis (2009) found a positive correlation between flux and spectral slope for a sample of bright RXTE AGN in the 2–10 keV band. This could affect the relative fluxes seen in our sample which are plotted in Fig. B.1. We have attempted to quantify this for the different detectors used in the creation of our light curves. The sample of Sobolewska & Papadakis (2009) showed spectral changes with observed power-law slope varying between 1.0 and 2.0 (see their Fig. 7). For a typical Galactic absorption of 3 × 1020 cm-2 the change from slope of 1.0 to 2.0 would alter our estimated fluxes by −14% (ROSAT), −13% (XMM-Newton), +7% (Swift-XRT), +76% (Suzaku), +25% (Einstein-IPC). The change is large for Suzaku observations since in this case we use the count rate between 2.0–10.0 keV and extrapolate it to the soft band. All other satellites are sensitive in the soft band and hence the fluxes are less dependent upon the assumed spectral index. Six sources within our sample (XMMSL1 J024916.6-041244, J034555.1-355959, J045740.0-503053, J051935.5-323928, J070841.3-493305, and J193439.3+490922) display variation in flux of a factor of ten or greater between at least one pair of XMM and Swift observations, on timescales of months to years. The ratio between the soft X-ray flux observed with Swift and that observed with XMM for the remaining sources is typically a factor of a few. We observed the two TDE candidates with XRT, and found that both had faded significantly, following expectations from previous and later fluxes and upper limits (Figs. 1p and h).
|
[
"Sobolewska & Papadakis (2009)"
] |
[
"The sample of",
"showed spectral changes with observed power-law slope varying between 1.0 and 2.0 (see their Fig. 7). For a typical Galactic absorption of 3 × 1020 cm-2 the change from slope of 1.0 to 2.0 would alter our estimated fluxes by −14% (ROSAT), −13% (XMM-Newton), +7% (Swift-XRT), +76% (Suzaku), +25% (Einstein-IPC). The change is large for Suzaku observations since in this case we use the count rate between 2.0–10.0 keV and extrapolate it to the soft band. All other satellites are sensitive in the soft band and hence the fluxes are less dependent upon the assumed spectral index."
] |
[
"Uses",
"Uses"
] |
[
[
1049,
1078
]
] |
[
[
1035,
1048
],
[
1079,
1657
]
] |
2015ApJ...805...44S__Tanaka_&_Haiman_2009_Instance_1
|
Pair-instability supernovae (PI SNe) are the most energetic thermonuclear explosions known and can be detected near the edge of the observable universe. They have now been studied by several groups for their potential to probe the properties of the first stars and galaxies (Greif et al. 2008, 2010, 2011, 2012; Johnson et al. 2009, 2014; Turk et al. 2009; Stacy et al. 2010, 2012; Clark et al. 2011; Hosokawa et al. 2011; Smith et al. 2011; Glover 2013; Jeon et al. 2012; Susa 2013; Pawlik et al. 2011, 2013; Wise et al. 2012; Whalen 2013; Hirano et al. 2014). They can also shed light on the origins of supermassive black holes and early cosmological reionization and chemical enrichment (Mackey et al. 2003; Whalen et al. 2004, 2008a, 2010; Abel et al. 2007; Smith & Sigurdsson 2007; Wise & Abel 2008; Alvarez et al. 2009; Tanaka & Haiman 2009; Smith et al. 2009; Park & Ricotti 2011, 2012, 2013; Volonteri 2012; Agarwal et al. 2012; Ritter et al. 2012; Whalen & Fryer 2012; Choi et al. 2013; Latif et al. 2013a, 2013b; Reisswig et al. 2013; Schleicher et al. 2013; Johnson et al. 2014; Chiaki et al. 2013; Safranek-Shrader et al. 2014). For example, detections of both PI and core-collapse (CC) SNe at high redshift could be roughly binned by mass, thereby building up a simple Pop III IMF over time as enough events are discovered (de Souza et al. 2013, 2014). If the Pop III initial mass function (IMF) proves to be top heavy, this could account for the origins of SMBHs because enough 200–300
seed black holes might be formed at
20 for a few to reach 109
by
7. If not, alternatives for SMBH seeds must be found, such as BHs forming by direct collapse in atomically cooled halos at slightly later epochs, z ∼ 10–15 (Johnson et al. 2012, 2013b). PI SN candidates such as SN 2007bi (Gal-Yam et al. 2009; Kozyreva et al. 2014) and SN 2213–1745 (Cooke et al. 2012) have now been discovered at z = 0.126 and 2.05, respectively.
|
[
"Tanaka & Haiman 2009"
] |
[
"They can also shed light on the origins of supermassive black holes and early cosmological reionization and chemical enrichment"
] |
[
"Motivation"
] |
[
[
826,
846
]
] |
[
[
562,
689
]
] |
2022MNRAS.511.4946N__Orr_&_Browne_1982_Instance_1
|
In the standard AGN unification theory, the anisotropic radio emission produced by relativistic jets points to a description where properties like radio morphology and radio spectral index depend on orientation. Two widely used orientation indicators in the radio regime are the (a) core-to-lobe flux density ratio (R), which is the ratio of the flux densities of the core and lobe in radio wavelengths, and (b) radio-to-optical ratio of the quasar core (RI), which is the ratio of flux densities of the core in radio and optical wavelengths (See equation 5 and 6 for the exact definition of R and RI parameters). A high R-value suggests a small viewing angle to the jet axis, while a low R-value indicates more of an edge-on view to the quasar (Kapahi & Saikia 1982; Orr & Browne 1982; Morisawa & Takahara 1987; Morganti et al. 1997). Studies have shown that R correlates with core radio luminosity (Hardcastle & Worrall 2000) as well with the core optical luminosity (Kharb & Shastri 2004). Factors like age and environment are believed to introduce the scatter in the correlation involving the R parameter suggesting that R might not be the best measure for orientation. The R parameter depends on both the core’s radio flux as well as the radio flux of the lobes arising from the quasar jets. These jets are suggested to be affected by environmental factors. Wills & Brotherton (1995) and Kharb, Lister & Cooper (2010) showed that the radio-to-optical ratio (RI) of the quasar core might be a better measure of studying the orientation. The RI parameter measures the core-boosting factor by normalizing the core’s radio flux by the core’s optical flux. Although with some scatter in the correlation, Kimball et al. (2011) demonstrated a strong correlation between R and RI parameter suggesting that the two parameters are good indicators of quasar orientation. The scatter in the correlation supports the idea that other factors like age, size, environment, and luminosity also influence these two measurements.
|
[
"Orr & Browne 1982"
] |
[
"A high R-value suggests a small viewing angle to the jet axis, while a low R-value indicates more of an edge-on view to the quasar"
] |
[
"Background"
] |
[
[
768,
785
]
] |
[
[
614,
744
]
] |
2020AandA...639A..46B__Štverák_et_al._(2009)_Instance_5
|
The linear relationship that we observe between breakpoint energy and core temperature is in line with previous measurements (e.g. McComas et al. 1992; Štverák et al. 2009), for both the halo and strahl. According to Scudder & Olbert (1979), a linear trend in the halo relation also follows under the assumption that binary Coulomb collisions dominate electron dynamics in the solar wind. However, in order to align with available experimental data, Scudder & Olbert (1979) set a scaling factor of Ebp/kBTc = 7, which differs from our scaling factor of Ebp/kBTc = 5.5 ± 0.1. With a scaling factor of Ebp/kBTc = 7, Scudder & Olbert (1979) predict that a transformation of thermal electrons into the suprathermal population occurs as the solar wind flows out from the Sun. Findings by Štverák et al. (2009), on the other hand, show that the (nh + ns)/nc ratio remains roughly constant with heliocentric distance in the slow wind, suggesting a lack of interchange between the thermal and suprathermal populations. However Štverák et al. (2009) observes some variability in the (nh + ns)/nc ratio in the fast wind, which they attribute to either statistical effects due to a lack of samples or a possible “interplay” between thermal and suprathermal electrons. Scudder & Olbert (1979) also predict that the halo Ebp/kBTc ratio remains constant with heliocentric distance, whereas Štverák et al. (2009) find that the halo Ebp/kBTc ratio decreases with heliocentric distance. These findings by Štverák et al. (2009), along with the discrepancy between our calculated ratio of Ebp/kBTc = 5.5 ± 0.1 and the prediction of Ebp/kBTc = 7, suggest that the model of Scudder & Olbert (1979) requires a minor update to either the theory or to the input parameters. The discrepancy, however, may also be indicative of other processes, such as wave-particle scattering (e.g. Gary et al. 1994), that possibly modifies the ratio between breakpoint energy and core temperature while preserving its linear relationship.
|
[
"Štverák et al. (2009)"
] |
[
"These findings by",
", along with the discrepancy between our calculated ratio of Ebp/kBTc = 5.5 ± 0.1 and the prediction of Ebp/kBTc = 7, suggest that the model of Scudder & Olbert (1979) requires a minor update to either the theory or to the input parameters."
] |
[
"Compare/Contrast",
"Compare/Contrast"
] |
[
[
1488,
1509
]
] |
[
[
1470,
1487
],
[
1509,
1749
]
] |
2021AandA...654L...5P__Beaugé_&_Nesvorný_2013_Instance_1
|
Armstrong et al. (2020) recently announced the discovery of TOI-849b, a planet having a size comparable to the one of Neptune (
R
pl
=
3
.
45
−
0.12
+
0.16
R
⊕
$ R_{\mathrm{pl}} = 3.45^{+0.16}_{-0.12}\,R_{\oplus} $
), but an anomalously larger mass (
M
pl
=
40
.
8
−
2.5
+
2.4
M
⊕
$ M_{\mathrm{pl}} = 40.8^{+2.4}_{-2.5}\,M_{\oplus} $
) and a density similar to the one of the Earth (ρ = 5.5 ± 0.8 g cm−3). This is the densest Neptune-sized planet discovered to date. Interior structure models suggest that for this kind of planet, any H/He envelope would consist of no more than 3.9% of the total mass (Armstrong et al. 2020). TOI-849b orbits around a late G-type star, with an orbital period of P = 18.4 h. Its equilibrium temperature is Teq = 1800 K (for a Bond Albedo 0.3). With such properties, TOI-849b represents one of the few planets populating the hot Neptune desert, a region on the radius-orbital distance plane characterised by a surprising deficit of Neptune-sized planets on very short orbits (e.g. Lecavelier Des Etangs 2007; Beaugé & Nesvorný 2013; Mazeh et al. 2016). Growing evidence suggests that the evaporation of hot Neptunes due to stellar irradiation represents a major process in shaping the desert (Owen 2019). Some of the planets within the desert or at its lower-radius border could thus be the remnant cores of larger gas-rich progenitors that lost most of their atmosphere (e.g. Lecavelier des Etangs et al. 2004). Alternatively, orbital migration may have also played a part in sculpting the desert, with different classes of planets forming differently (e.g. Batygin et al. 2016), or following different dynamical tracks (e.g. Matsakos & Königl 2016). Even so, the origin of this key feature in the demographics of close-in planets remains debated (Zahnle & Catling 2017; Owen & Lai 2018), and investigating the past history of planets such as TOI-849b contributes to the global effort towards disentangling the important mechanisms at the root of the desert. Since the mass of the planet is larger than the threshold value for runaway gas accretion (roughly 10 − 20 M⊕, Mizuno et al. 1978; Rafikov 2006; Piso et al. 2015), TOI-849b might have been a gas giant before undergoing extreme mass loss via thermal self-disruption, collisions with other giant planets, or it was prevented from accreting gas because of gap openings in the protoplanetary disc, or because of late formation (Armstrong et al. 2020). In their work, Armstrong et al. (2020) found that their estimated photoevaporation rates cannot provide the mass-loss rates needed to remove a roughly 280 M⊕ envelope from a Jupiter-like gas giant. Nevertheless, in their estimations the planetary atmosphere is not self-consistently monitored, and neither is the luminosity emitted by the host star relative to its rotational history. This is critical, considering how recent works emphasise the interplay between all these elements in driving the evolution of close-in worlds (e.g. Owen & Alvarez 2016; Kubyshkina et al. 2018; King & Wheatley 2021; Pezzotti et al. 2021).
|
[
"Beaugé & Nesvorný 2013"
] |
[
"With such properties, TOI-849b represents one of the few planets populating the hot Neptune desert, a region on the radius-orbital distance plane characterised by a surprising deficit of Neptune-sized planets on very short orbits (e.g."
] |
[
"Background"
] |
[
[
1077,
1099
]
] |
[
[
813,
1048
]
] |
2022AandA...657A..50G__Qian_&_Wasserburg_2003_Instance_1
|
Carbon-enhanced metal-poor (CEMP) stars form an important class of metal-poor giants, sub-giants, and dwarfs, with a large fraction of them showing enhanced abundances of heavy elements (see Beers & Christlieb 2005; Frebel 2018 for a general review). Among the different types of CEMP stars, the CEMP-s stars are enriched with products of s-process nucleosynthesis, the CEMP-r stars are enriched with the products of r-process nucleosynthesis, and CEMP-r/s stars are enriched with products of i-process nucleosynthesis. Understanding the diverse abundance patterns exhibited by different groups of CEMP stars that are believed to be associated with different formation mechanisms has been a challenge. In Goswami et al. (2021), we present a detailed analysis and discussion on the classification criteria of CEMP stars, as well as the formation scenarios of CEMP stars put forward by different authors (Cowan & Rose 1977; Hill et al. 2000; Qian & Wasserburg 2003; Cohen et al. 2003; Jonsell et al. 2006; Campbell & Lattanzio 2008; Campbell et al. 2010; Stancliffe et al. 2011; Herwig et al. 2011; Doherty et al. 2015; Abate et al. 2016; Jones et al. 2016; Banerjee et al. 2018; Clarkson et al. 2018; Denissenkov et al. 2017; Côté et al. 2018). In this paper, we report an extremely metal-poor carbon-enhanced star, HE 1005–1439, whose surface chemical composition is found to be enriched with both s-process and i-process nucleosynthesis that forms a new class of object with a distinct abundance pattern. The peculiar abundance pattern, observed for the first time in a CEMP star, was investigated based on a parametric-model-based analysis that revealed almost equal contributions from both the s-process and the i-process to its surface chemical composition. We examined various production mechanisms and formation scenarios for this object. A formation scenario involving effective proton ingestion episodes (PIEs) triggering i-process nucleosynthesis followed by s-process asymptotic giant branch (AGB) nucleosynthesis with limited third-dredge-up (TDU) episodes seems to be most promising for this type of object.
|
[
"Qian & Wasserburg 2003"
] |
[
"In Goswami et al. (2021), we present a detailed analysis and discussion on the classification criteria of CEMP stars, as well as the formation scenarios of CEMP stars put forward by different authors"
] |
[
"Background"
] |
[
[
940,
962
]
] |
[
[
702,
901
]
] |
2022ApJ...935....7L__Takeuchi_&_Kono_2020_Instance_1
|
Among the GRB empirical correlations, the Amati correlation is a very popular one, which connects the spectral peak energy in the GRB cosmological rest frame and the isotropic equivalent radiated energy (E
p
− E
iso; Amati et al. 2002; Amati 2006a, 2006b; Amati et al. 2008, 2009; Amati & Della Valle 2013). Recently, we proposed an improved Amati correlation (Liu et al. 2022) by using the Gaussian copula, which is a powerful statistical tool capable of describing the dependence structures between multivariate random variables, and has been applied to various fields by the astronomical community (Benabed et al. 2009; Jiang et al. 2009; Koen 2009; Scherrer et al. 2010; Takeuchi 2010; Yuan et al. 2018; Qin et al. 2020; Takeuchi & Kono 2020). In Liu et al. (2022), by choosing the spatially flat ΛCDM model with Ωm0 = 0.30 and H
0 = 70 km s−1 Mpc−1 as the fiducial model, we utilize the low-redshift (z 1.4) GRB data to calibrate the standard and improved Amati correlations, and then extrapolate the results to the high-redshift GRB data to achieve the GRB Hubble diagram, where Ωm0 is the present dimensionless matter density parameter. Using these calibrated GRBs to constrain the flat ΛCDM model, we found that the improved Amati correlation can give results well consistent with the fiducial model, while the standard one cannot. Thus, in Liu et al. (2022), the reliability of the improved Amati correlation was ascertained with a fiducial model, but its cosmological application was not carried out. In this work, we will fill this gap. In order to obtain the Hubble diagram of the latest A220 GRB samples (Khadka et al. 2021) model-independently, we use the Pantheon SN Ia data (Scolnic et al. 2018) to calibrate the standard and improved Amati correlations, and then use these calibrated GRB data to constrain the ΛCDM and wCDM models. Besides the GRB data, the H(z) data set is also added in our analysis to obtain a tight constraint on model parameters. The rest of the paper is organized as follows: Section 2 introduces the improved Amati correlation briefly, and standardizes the GRB samples by using the method of the low-redshift calibration. Section 3 studies the constraints on the ΛCDM and wCDM models from the GRB data and the GRB + H(z) data. Our conclusions are summarized in Section 4.
|
[
"Takeuchi & Kono 2020"
] |
[
"Recently, we proposed an improved Amati correlation",
"by using the Gaussian copula, which is a powerful statistical tool capable of describing the dependence structures between multivariate random variables, and has been applied to various fields by the astronomical community"
] |
[
"Background",
"Background"
] |
[
[
727,
747
]
] |
[
[
310,
361
],
[
380,
602
]
] |
2021ApJ...919..140S__Bartos_et_al._2017_Instance_2
|
Resonant dynamical friction may have applications beyond the relaxation of IMBHs examined in this paper. It may affect all objects in stellar clusters much more massive than the individual constituents of the disk, if present, including massive stars, stellar mass black holes (BHs), or the center of mass of massive binaries. Furthermore, it is also expected to operate in any type of disk with a high number of particles, including active galactic nucleus (AGN) accretion disks. Previously, it has been argued that stars and BHs crossing the disk on low-inclination orbits get captured by Chandrasekhar dynamical friction into the disk (Bartos et al. 2017; Panamarev et al. 2018; Tagawa et al. 2020). An interesting implication is that, if BHs settle into the disk, they interact dynamically and form BH–BH binaries efficiently, and frequent dynamical interactions and gas effects drive the BHs to merger, producing gravitational waves (GWs) detectable by LIGO, VIRGO, and KAGRA (McKernan et al. 2014, 2018; Bartos et al. 2017; Leigh et al. 2018; Yang et al. 2019; Tagawa et al. 2020, 2021; Samsing et al. 2020). Mergers are also facilitated by Lidov–Kozai oscillations in anisotropic systems (Heisler & Tremaine 1986; Petrovich & Antonini 2017; Hamilton & Rafikov 2019). The results in this paper show that resonant dynamical friction may accelerate the capture of objects in the accretion disks by a factor proportional to the SMBH mass over the local disk mass for large orbital inclinations. Pressure and viscosity in a gaseous disk do not inhibit the orbit-averaged torque from the IMBH, which leads to realignment and the warping of the disk (Bregman & Alexander 2012). Thus, RDF may efficiently catalyze the alignment of the orbital planes of BHs even in low-luminosity AGN or Seyfert galaxies with relatively small disk masses, which may not be possible for Chandrasekhar dynamical friction. In fact, this mechanism extends the scope of the “AGN merger channel” for GW source populations even beyond low-luminosity AGN and Seyfert galaxies, as it may organize BHs into disks also in nonactive galaxies with nuclear stellar disks.
|
[
"Bartos et al. 2017"
] |
[
"An interesting implication is that, if BHs settle into the disk, they interact dynamically and form BH–BH binaries efficiently, and frequent dynamical interactions and gas effects drive the BHs to merger, producing gravitational waves (GWs) detectable by LIGO, VIRGO, and KAGRA"
] |
[
"Background"
] |
[
[
1010,
1028
]
] |
[
[
703,
980
]
] |
2019AandA...631A.167D__George_et_al._2013_Instance_1
|
The applicability of mid-IR and far-IR FSL as diagnostic tools of the ISM has received a boost thanks to the publication of samples of nearby galaxies observed with the infrared space observatory (ISO) and Herschel (e.g. Brauher et al. 2008; Farrah et al. 2013; Sargsyan et al. 2014; Kamenetzky et al. 2014; Cormier et al. 2015; Cigan et al. 2016; Herrera-Camus et al. 2016; Fernández-Ontiveros et al. 2016; Zhao et al. 2016; Díaz-Santos et al. 2017; Zhang et al. 2018). At high redshift (z ≳ 1), the far-IR FSL conveniently shift into the (sub)millimetre atmospheric windows. The most popular line is clearly [CII] 158 μm, followed by the [CI] 370, 609 μm lines (e.g. Walter et al. 2011; Bothwell et al. 2017). Both single dish submillimetre telescopes and interferometers have also detected the [NII] 122 μm and 205 μm lines at high redshift (Ferkinhoff et al. 2011, 2015; Combes et al. 2012; Nagao et al. 2012; Decarli et al. 2012, 2014; Béthermin et al. 2016; Pavesi et al. 2016; Lu et al. 2017; Tadaki et al. 2019; Novak et al. 2019). After a slow start, the [OIII] 88 μm line is quickly becoming a popular line to confirm redshifts of galaxies in the epoch of reionization (z ≳ 6), where it shifts into the submillimetre atmospheric windows below 500 GHz (Ferkinhoff et al. 2010; Inoue et al. 2016; Carniani et al. 2017; Marrone et al. 2018; Vishwas et al. 2018; Hashimoto et al. 2018, 2019; Walter et al. 2018; Tamura et al. 2019; Tadaki et al. 2019; Novak et al. 2019). Deep Herschel/SPIRE spectroscopy has also revealed a number of FSL detections, either in individual objects (Valtchanov et al. 2011; Coppin et al. 2012; George et al. 2013; Uzgil et al. 2016; Rigopoulou et al. 2018; Zhang et al. 2018), or in stacked spectra (Wardlow et al. 2017; Wilson et al. 2017; Zhang et al. 2018). These include the only detections of the [OI] 63 μm line at high redshift reported thus far. This [OI]3P2−3P1 line is arguably the best tracer for the star-forming gas, as it traces the very dense (
n
crit
H
$ n^{\mathrm{H}}_{\mathrm{crit}} $
= 5×105 cm−3) neutral gas (one caveat being the frequeny presence of self-absorption observed in local ultra-luminous IR galaxies, Rosenberg et al. 2015). Like the [CII] line, the [OI] 63 μm line also shows a “deficit” in the most luminous far-IR sources, though with a higher scatter (Graciá-Carpio et al. 2011; Cormier et al. 2015; Díaz-Santos et al. 2017). Surprisingly, this bright FSL has not been frequently observed with ALMA, probably because it is only observable in the highest frequency bands. At least as surprising is that the fainter, but more accessible [OI]3P1−3P0 transition at λrest = 145 μm (
n
crit
H
$ n^{H}_{\mathrm{crit}} $
= 9.5×104 cm−3) has only recently been detected at high redshifts (Novak et al. 2019 report a tentative detection in a z = 7.5 quasar). Also in nearby galaxies, this [OI] 145 μm line has not been observed very frequently as in most cases, it is fainter than the nearby [CII] 158 μm line. After initial detections with ISO (Malhotra et al. 2001; Brauher et al. 2008), Herschel has now detected [OI] 145 μm in significant samples of nearby galaxies (Spinoglio et al. 2015; Cormier et al. 2015; Fernández-Ontiveros et al. 2016; Herrera-Camus et al. 2018), and recently in a z = 6.5 lensed quasar (Yang et al. 2019).
|
[
"George et al. 2013"
] |
[
"Deep Herschel/SPIRE spectroscopy has also revealed a number of FSL detections, either in individual objects"
] |
[
"Background"
] |
[
[
1631,
1649
]
] |
[
[
1478,
1585
]
] |
2019ApJ...886...34F__Sahijpal_&_Goswami_1998_Instance_1
|
If the variation in 10Be/9Be ratios of CAIs reflects those episodic accretion events, 10Be/9Be ratios of CH–CB CAIs observed in this study would give important constraints on the evolution of the solar protoplanetary disk. Astronomical observations suggest that FUori-type outbursts are confined to the first few hundreds of thousands of years, which correspond to the class I stage of the protoplanetary disk evolution (e.g., Schulz 2012). We propose that the high and variable 10Be/9Be ratios recorded in CH–CB CAIs reflect episodic cosmic-ray fluxes caused by FUori-type outbursts. On the other hand, relatively low and less variable 10Be/9Be ratios recorded in CV CAIs may reflect less intensive episodic accretion events, possibly the EXori-type outbursts, which are confined to the evolutional stage of a few million years after the formation of the protoplanetary disk (=class II). Note that CH–CB CAIs studied here show no (or very low) signs of 26Al-derived 26Mg excesses, while most CV CAIs show clear evidence for the past presence of 26Al. If 26Al was introduced into the solar system at the earliest stage of the disk evolution (e.g., Sahijpal & Goswami 1998), differences in Be–B and Al–Mg systematics between CH–CB and CV CAIs imply that the injection of 26Al have occurred between the evolutionary stages class I and class II of the solar protoplanetary disk. This scenario is in agreement with arguments by other authors that the 26Al-free CAIs formed prior to injection and homogenization of 26Al in the early solar system (Sahijpal & Goswami 1998; Sahijpal et al. 2000; Krot et al. 2008a see more discussion in Krot et al. 2012a). Importantly, as mentioned in the introduction, CH–CB chondrites may have accreted a significant amount of outer solar system materials (Murty et al. 2007; Ivanova et al. 2008; Briani et al. 2009; Bonal et al. 2010; Olsen et al. 2016; Van Kooten et al. 2016), suggesting that CH–CB chondrites formed at outer parts of the solar protoplanetary disk relative to CV chondrites. In this case, our new Be–B and Al–Mg data set implies that the earliest formed CAIs tend to be transported into the outer part of the solar protoplanetary disk, where the parent bodies of CH–CB chondrites likely accreted. Yang & Ciesla (2012) modeled the evolution of the protoplanetary disk and material transport in the protoplanetary disk. Interestingly, Yang & Ciesla (2012) showed that outward radial transport in class I would have been greater than that of later stages of YSO evolution, suggesting that the earliest formed CAIs could be preserved in primitive bodies that accreted in the outer part of the disk. This model is consistent with our interpretation for the Be–B and Al–Mg systematics on CH–CB CAIs. It should be noted, however, that it is possible that 26Al were heterogeneously distributed in the CAI-forming regions at the earliest stage of the solar system evolution (e.g., Krot et al. 2008a; Holst et al. 2013; Park et al. 2017 and reference therein). Because no Pb–Pb ages of CH–CB CAIs are available at present, we cannot discard that possibility. Very recently, Kööp et al. (2018) found helium and neon excesses in the 26Al-free hibonite-rich CAIs, which can be attributed to in situ irradiation by energetic particles. Because 26Al-rich CAIs in CV chondrites lack comparable noble gas irradiation records (Vogel et al. 2004), Kööp et al. (2018) concluded that 26Al-free hibonite-rich CAIs experienced intense energetic particle irradiation at the earliest stage of solar protoplanetary disk evolution. This conclusion seems to be consistent with our above scenario for 26Al-free CH–CB CAIs. Note, however, that 10Be/9Be ratios of 26Al-free hibonite-rich CAIs in CM chondrites tend to be in the range of those for 26Al-rich CV CAIs (Liu et al. 2009, 2010), which is inconsistent with the above scenario. Therefore, the relationship between 10Be and 26Al in the early solar system would be more complicated than we thought.
|
[
"Sahijpal & Goswami 1998"
] |
[
"If 26Al was introduced into the solar system at the earliest stage of the disk evolution (e.g.,",
"differences in Be–B and Al–Mg systematics between CH–CB and CV CAIs imply that the injection of 26Al have occurred between the evolutionary stages class I and class II of the solar protoplanetary disk."
] |
[
"Uses",
"Uses"
] |
[
[
1148,
1171
]
] |
[
[
1052,
1147
],
[
1174,
1375
]
] |
2021MNRAS.504.5575K__Cohen_et_al._2018_Instance_1
|
First detected as a radio source during the Vermilion River Observatory Sky Survey (Dickel et al. 1967) and the Ohio Sky Survey (Kraus 1977), OJ 287 has been studied extensively in the radio regime. Its relativistic jet is pointing at us with an average viewing angle of ∼2° (Jorstad et al. 2005; Agudo et al. 2012) and shows remarkable short-time-scale variability interpreted as a turbulent injection process and/or a clumpy accretion disc structure (Agudo et al. 2012) or as a binary-induced wobble (Valtonen & Pihajoki 2013; Dey et al. 2021). The inner jet is the source of highly variable γ-rays (e.g. Agudo et al. 2011; Hodgson et al. 2017) and displays strong and variable radio polarization (e.g. Aller et al. 2014; Cohen et al. 2018; Myserlis et al. 2018). Though only faintly detected in the very-high-energy (VHE) band (>100 GeV; Mukherjee et al. 2017; O’Brien 2017), OJ 287 is a well-known X-ray emitter and has been detected with most major X-ray observatories, including Einstein (Madejski & Schwartz 1988), EXOSAT (Sambruna et al. 1994), ROSAT (Comastri, Molendi & Ghisellini 1995), BeppoSAX (Massaro et al. 2003), ASCA (Idesawa et al. 1997), the Neil Gehrels Swift observatory (Swift hereafter; Massaro et al. 2008), Suzaku (Seta et al. 2009), XMM–Newton (Ciprini et al. 2007), and most recently with NuSTAR (Komossa et al. 2020a). These observations established OJ 287 as a bright and variable X-ray source, and allowed single-component X-ray spectral fits. Its (0.5–10) keV X-ray spectrum was interpreted as a mix of synchrotron and inverse-Compton (IC) emission, with the former more variable (Urry et al. 1996). OJ 287 is found most of the time in a rather flat spectral state with a photon index Γx ≈ 1.5–1.9, though Γx is as steep as 2.6 during one ROSAT observation (Urry et al. 1996) in the band (0.1–2.4) keV. Its steepest state, with an equivalent Γx = 2.8, (better fit with a logarithmic parabolic model) was detected with XMM–Newton in the band (0.3–10) keV, which caught the source right at the peak of one of the brightest X-ray outbursts measured so far (Komossa et al. 2020a). Imaging with the Chandra X-ray observatory has revealed a long, curved X-ray jet consisting of multiple knots and extending out to 20 arcsec or a de-projected scale of >1 Mpc, and bright central emission (Marscher & Jorstad 2011).
|
[
"Cohen et al. 2018"
] |
[
"The inner jet",
"and displays strong and variable radio polarization (e.g."
] |
[
"Background",
"Background"
] |
[
[
724,
741
]
] |
[
[
547,
560
],
[
647,
704
]
] |
2017ApJ...849...52N__Merloni_&_Fabian_2002_Instance_1
|
The bolometric luminosity of 3C 84 is about 0.4% of the Eddington luminosity. Thus, the accretion flow of 3C 84 is likely to be a radiatively inefficient accretion flow (RIAF: Narayan & Yi 1995) rather than a standard disk (Shakura & Sunyaev 1973). However, we note that 3C 84 has a cold (
T
e
∼
10
4
K) disk-like accretion flow, as identified by FFA of the emission from the counter jet in the parsec scale (Walker et al. 2000) and inhomogeneous gas distribution around the black hole (Fujita et al. 2016). A number of theoretical studies predicted that the accretion flow components of hot geometrically thick (RIAF-like) and cold geometrically thin can coexist in either horizontal or vertical stratification (e.g., Miller & Stone 2000; Merloni & Fabian 2002; Liu et al. 2007; Ho 2008; Liu & Taam 2013). The measured Faraday rotation can be caused by such an RIAF-like component. We thus estimate the accretion rate of the RIAF-like component using the measured RM. For a simplicity, we assume that the RIAF-like component is a quasi-spherical Bondi accretion flow with a power-law density profile. We can calculate the accretion rate, following the formulation as follows (Quataert & Gruzinov 2000; Marrone et al. 2006; Kuo et al. 2014).
M
˙
=
1.3
×
10
−
10
[
1
−
(
r
out
/
r
in
)
−
(
3
β
−
1
)
/
2
]
−
2
/
3
×
M
BH
8.0
×
10
8
M
⊙
4
/
3
2
3
β
−
1
−
2
/
3
r
in
7
/
6
RM
rad
m
−
2
2
/
3
.
For an inner effective radius
r
in
of 1 pc (
1.3
×
10
4
R
s
), where the hotspot is located, the observed RM implies an accretion rate of
∼
4.3
×
10
−
2
M
⊙
yr−1 and
∼
8.6
×
10
−
2
M
⊙
yr−1 for
β
=
0.5
and
β
=
1.5
, which are corresponding to convection-dominated accretion flow (CDAF: Narayan et al. 2000; Quataert & Gruzinov 2000) and advection-dominated accretion flow (ADAF: Ichimaru 1977; Narayan & Yi 1995), respectively. Here we assumed the outer effective radius
r
out
of
10
5
R
s
(∼8 pc), which is approximately the same with the Bondi radius of 8.6 pc (Fujita et al. 2016). The derived accretion rate is roughly consistent with that estimated from the bolometric luminosity with a black hole mass of
8
×
10
8
M
⊙
and a radiative efficiency of 10% (
M
˙
∼
L
bol
/
(
0.1
c
2
)
≃
7.1
×
10
−
2
M
⊙
yr−1).
|
[
"Merloni & Fabian 2002"
] |
[
"A number of theoretical studies predicted that the accretion flow components of hot geometrically thick (RIAF-like) and cold geometrically thin can coexist in either horizontal or vertical stratification (e.g.,"
] |
[
"Background"
] |
[
[
759,
780
]
] |
[
[
527,
737
]
] |
2022MNRAS.513.1623C__Koyama,_Taruya_&_Hiramatsu_2009_Instance_1
|
Two-point statistics are central to many of the leading cosmological analyses of the large-scale structure searching for deviations from ΛCDM (Simpson et al. 2013; Song et al. 2015; Amon et al. 2018; Abbott et al. 2019a; Chudaykin, Dolgikh & Ivanov 2021; Lee et al. 2022; Muir et al. 2021; Tröster et al. 2021; Vazsonyi et al. 2021), and a great deal of effort has gone into accurately modelling the non-linear matter power spectrum in modified gravity and dark energy cosmologies – a theoretical ingredient essential to extract the cosmological information locked in small scales (e.g. Koyama, Taruya & Hiramatsu 2009; Brax & Valageas 2012; Takahashi et al. 2012; Heitmann et al. 2014; Zhao 2014; Casarini et al. 2016; Mead et al. 2016; Cusin, Lewandowski & Vernizzi 2018; Cataneo et al. 2019; Winther et al. 2019; Euclid Collaboration 2021; Ramachandra et al. 2021). However, non-linear gravitational clustering converts the nearly Gaussian initial density field (Planck Collaboration VI 2020) to a late-time density field with significant non-Gaussian features that these standard analyses are unable to access (Bernardeau et al. 2002). Non-Gaussian statistics, such as the bispectrum (Brax & Valageas 2012; Munshi 2017; Yamauchi, Yokoyama & Tashiro 2017; Crisostomi, Lewandowski & Vernizzi 2020; Bose et al. 2020b), higher order weak lensing spectra (Munshi & McEwen 2020), the halo mass function (Lam & Li 2012; Cataneo et al. 2016; Hagstotz et al. 2019; McClintock et al. 2019; Bocquet et al. 2020), the void size function (Perico et al. 2019; Verza et al. 2019; Contarini et al. 2021) and Minkowski functionals (Kratochvil et al. 2012; Fang, Li & Zhao 2017), respond strongly to modified gravity and dark energy through the induced changes in the higher moments of the cosmic density field, and their remarkable complementarity to traditional two-point functions leads to tighter joint constraints on the extra non-standard parameters (Shirasaki et al. 2017; Peel et al. 2018; Sahlén 2019; Liu et al. 2021).
|
[
"Koyama, Taruya & Hiramatsu 2009"
] |
[
"Two-point statistics are central to many of the leading cosmological analyses of the large-scale structure searching for deviations from ΛCDM",
"and a great deal of effort has gone into accurately modelling the non-linear matter power spectrum in modified gravity and dark energy cosmologies – a theoretical ingredient essential to extract the cosmological information locked in small scales (e.g."
] |
[
"Background",
"Background"
] |
[
[
587,
618
]
] |
[
[
0,
141
],
[
334,
586
]
] |
2019MNRAS.482.5597A__Bergeron_&_Stasińska_1986_Instance_1
|
There have been many efforts in the past that have used the optical emission lines to understand the AGN and the host galaxy properties (e.g. Baldwin, Wampler & Burbidge 1981; Tadhunter et al. 1998). While broad and high ionization lines tend to arise from the immediate vicinity of the central massive black hole, low ionization lines arise further out (e.g. Richardson et al. 2014), tracing the host-galaxy dynamics, chemical composition and gas distribution. It is known that low ionization gases trace the H i distribution in the galactic discs (e.g. Lin & Zou 2001). Indeed, Mg ii is known as a good tracer of high column density H i gas in damped Lyman α absorbers (e.g. Bergeron & Stasińska 1986; Péroux et al. 2004). [O ii] λ3727 is widely used to trace star formation in galaxy surveys, particularly for redshifts $z$ ≳ 0.4 (e.g. Lilly et al. 1996; Hippelein et al. 2003). In optically selected quasars, [O ii] emission in excess of the base level that is expected from non-stellar photoionization indicates star formation accompanying quasar activity (e.g. Ho 2005). These lines will hence allow a better understanding of the distribution and kinematical properties of the atomic gas. Recently, Shen & Ménard (2012) and Khare et al. (2014) studied the properties of [O ii] λ3727 emission from a large sample of associated Mg ii absorbers, in order to understand the origin of associated Mg ii absorption systems relative to the AGN. Khare et al. (2014) find a high excess [O ii] emission in the composite quasar spectrum (constructed in the quasar rest frame) for the sample of quasars that show outflows, i.e. the systems for which the Mg ii absorption is blueshifted relative to the AGN. Further, the authors find that the presence of an associated Mg ii absorption enhances [O ii] emission in the composite quasar spectrum. However the excess emission flux in the [O ii] line does not depend on the strength of the Mg ii absorption line (Khare et al. 2014). The excess [O ii] line flux could be originating either from the host galaxy, or the parent AGN; in the former case, the excess [O ii] emission could indicate higher star formation in the host galaxy. The association of high [O ii] line emission with the occurrence of outflows suggests either strong stellar outbursts in the host galaxy, or strong AGN jet–gas interactions.
|
[
"Bergeron & Stasińska 1986"
] |
[
"Indeed, Mg ii is known as a good tracer of high column density H i gas in damped Lyman α absorbers (e.g."
] |
[
"Background"
] |
[
[
677,
702
]
] |
[
[
572,
676
]
] |
2020AandA...638A..34J__Hardcastle_(2018)_Instance_1
|
We stress that the age values discussed here should be taken with care as they are based on simulations which rely on a number of assumptions. One of these is the magnetic field which is typically computed using the often unrealistic equipartition conditions. For example, a factor two difference in the magnetic field value translates directly into about a factor two age difference. Another important parameter is the assumed total source age distribution which is assumed to be different in the various simulations. For example, in the case of the modelling presented in Brienza et al. (2017) and Godfrey et al. (2017), the distribution of ton was taken to be a truncated log-normal distribution between 20 and 200 Myr, with a median of 30 Myr. Hardcastle (2018) on the other hand explored two types of models, with ages being either uniformly distributed between 0 and 1000 Myr, or linearly distributed in log space between 1 and 1000 Myr. These latter authors found that the uniform-age models efficiently explain the size and luminosity statistics of bright sources observed by LOFAR, while the log-uniform models were a better representation of the faint population. Similarly, Shabala et al. (2020) showed that power-law age-distribution models (i.e. models where the radio-source population is dominated by short-lived sources) can explain the observed properties of both active and remnant and/or restarted LOFAR sources. These findings are consistent with the modelling assumptions of Brienza et al. (2017) and Godfrey et al. (2017), as the majority of short-lived sources will simply never grow large or luminous enough to make it into the observable LOFAR sample1. Modelling by both Shabala et al. (2020) and Hardcastle (2018) confirms the picture in which the remnant lobes fade quickly below the LOFAR detection limit. This fading is more rapid for older sources. Future studies, including modelling of the spectral index and follow-up observations, will help us to put tighter constraints on the age of the candidate restarted radio galaxies and the time that passed between the two bursts of activity.
|
[
"Hardcastle (2018)",
"Hardcastle (2018)"
] |
[
"on the other hand explored two types of models, with ages being either uniformly distributed between 0 and 1000 Myr, or linearly distributed in log space between 1 and 1000 Myr.",
"These latter authors found that the uniform-age models efficiently explain the size and luminosity statistics of bright sources observed by LOFAR, while the log-uniform models were a better representation of the faint population.",
"Modelling by both Shabala et al. (2020) and",
"confirms the picture in which the remnant lobes fade quickly below the LOFAR detection limit."
] |
[
"Differences",
"Compare/Contrast",
"Similarities",
"Similarities"
] |
[
[
748,
765
],
[
1722,
1739
]
] |
[
[
766,
943
],
[
944,
1173
],
[
1678,
1721
],
[
1740,
1833
]
] |
2022MNRAS.511.2105K__Husemann_et_al._2019_Instance_2
|
The calculation of mass outflow rates, especially in the ionized gas phase, have often come from measurements using integrated fibre or long-slit spectra, where several assumptions are invoked in the outflow modelling. These assumptions, briefly described here, result in ‘time-averaged global mass outflow rate’ with large systematic uncertainties. First, due to the limitations of the current instruments even on large telescopes, an accurate modelling of the outflow geometry is not possible. This is especially true for high redshift galaxies (z∼2) where, with currently available adaptive optics (AO) technology, one can at best achieve a spatial resolution of ∼2 kpc where the bulk of the outflow might reside (e.g. Brusa et al. 2016; Davies et al. 2020b). Therefore, the outflow geometry is either assumed to be a uniformly filled conical, bi-conical, or spherical thin shells (e.g. Veilleux, Shopbell & Miller 2001; Fischer et al. 2013; Riffel, Storchi-Bergmann & Winge 2013; Ishibashi & Fabian 2015; Thompson et al. 2015; Bae & Woo 2016; Husemann et al. 2019; Mingozzi et al. 2019). Secondly, if the data are obtained from fibre and single-slit spectroscopy, the size of the outflow is largely unconstrained. For long-slit observations, as an example, the outflow size depends on whether the slit is oriented along the outflow direction. This can be mitigated by using integral field spectroscopy (IFS) which is being increasingly used for extragalactic studies (e.g. Liu et al. 2013; Rupke & Veilleux 2013; Harrison et al. 2014; Maiolino et al. 2017; Husemann et al. 2019; Schönell et al. 2019; Rupke, Thomas & Dopita 2021), although there could still be projection effects with the IFS data. Thirdly, accurate determination of electron density and electron temperature is required for the ionized mass outflow rate calculations. Electron density is usually derived from emission lines that arise out of two closely spaced ‘metastable’ energy levels such as [S ii]λλ6716,6731 ([S ii] doublet hereafter). Density measured from the [S ii] doublet is sensitive to values between ∼10 and 5000 cm−3, typical in the Narrow Line Region (NLR) of AGN host galaxies (e.g. Osterbrock & Ferland 2006; Perna et al. 2017; Baron & Netzer 2019; Davies et al. 2020a). The [S ii] doublet is significantly weaker than the lines used to trace ionized outflows such as the [O iii]λ5007 and H α. In high redshift galaxies, it is extremely challenging to detect these doublet lines, despite hours of observations on a single target. Therefore, nominal density values are often assumed in mass outflow rate calculations, resulting in systematic uncertainties of up to 2–3 orders of magnitude. Furthermore, the density structure within the outflowing medium is often non-uniform, when resolved in low redshift galaxies (e.g. Kakkad et al. 2018). Therefore assuming a constant density within the outflowing medium often leads to inaccurate outflow rate and kinetic energy values. Collectively, these assumptions result in a systematic uncertainty of approximately 3–4 orders of magnitude (e.g. Harrison et al. 2018). This implies that the quoted values of coupling efficiency in the literature have a wide range, with the actual efficiency still an unknown in most of the studies.
|
[
"Husemann et al. 2019"
] |
[
"This can be mitigated by using integral field spectroscopy (IFS) which is being increasingly used for extragalactic studies (e.g.",
"although there could still be projection effects with the IFS data."
] |
[
"Background",
"Background"
] |
[
[
1561,
1581
]
] |
[
[
1347,
1476
],
[
1635,
1702
]
] |
2017AandA...600A..47D__Becker_2015_Instance_1
|
To check for the validity of this idea, the case of some PACWBs already investigated in X-rays deserves to be examined. Some systems, such as WR 140 (Williams et al. 1990; Pollock et al. 2005) or Cyg OB2 #8A (Harnden et al. 1979; De Becker et al. 2006), are indeed known to be both well-studied PACWBs and bright thermal X-ray emitters, due to their colliding winds. However, other examples should be commented. The first is HD 167971. It consists of a hierarchical triple system with a 3.3-day O-type binary and a third later-type companion evolving on a 21-yr orbit (De Becker et al. 2012; Ibanoglu et al. 2013). Recently, it has been demonstrated that its thermal X-ray spectrum is dominated by the colliding-wind region in the close binary, with only a weak/moderate contribution coming from the wind-wind interaction in the wide orbit (De Becker 2015). However, it has also been demonstrated that HD 167971 presents a non-thermal radio emission modulated with a period of 21 yr, thus providing compelling evidence for a colliding-wind region origin (Blomme et al. 2007). In addition, HD 167971 is the brightest O-type synchrotron radio emitter included to date in the catalogue of PACWB. Another relevant example is HD 93129A. It is known as an early O-type wide binary with undetermined period (probably more than a century), with non-thermal radio emission recently imaged in Long Baseline Array observations (Benaglia et al. 2015). The direct imaging of this emission region points to a clear and significant synchrotron radio emission coincident with the colliding-wind region. However, the investigation in X-rays by Gagné et al. (2011) did not point to any strong over-luminosity attributable to an X-ray bright wind-wind interaction region. These examples demonstrate that the association between bright thermal X-ray emission and bright non-thermal radio emission is not necessarily clear. As a result, the quest for new members of the catalogue should by no means be restricted to objects known to display a bright thermal soft X-ray spectrum.
|
[
"De Becker 2015"
] |
[
"Recently, it has been demonstrated that its thermal X-ray spectrum is dominated by the colliding-wind region in the close binary, with only a weak/moderate contribution coming from the wind-wind interaction in the wide orbit"
] |
[
"Background"
] |
[
[
841,
855
]
] |
[
[
615,
839
]
] |
2019ApJ...883..174X__Lattimer_&_Prakash_2000_Instance_1
|
Both the magnitude and slope of nuclear symmetry energy contribute to the pressure of NS matter. For example, the pressure of npe matter in NSs at β equilibrium at density ρ and isospin asymmetry δ is explicitly
2
The first term is the SNM pressure
, while the last two terms are the isospin-asymmetric pressure
from nucleons and electrons, separately. At the saturation density ρ0, P0 vanishes, and the electron contribution is also negligible, leaving the total pressure completely determined by the slope of the symmetry energy. Both P0 and Pasy increase with density with rates determined separately by the respective density dependences of the SNM EOS and the symmetry energy. In the region around ρ0 ∼ 2.5ρ0, Pasy dominates over P0 using most EOSs available. At higher densities, the SNM pressure P0 dominates, while Pasy also plays an important role, depending on the high-density behaviors of nuclear symmetry energy (Li & Steiner 2006). The exact transition of dominance from Pasy to P0 depends on the stiffnesses of both the SNM EOS and the symmetry energy. It is also well known that the radius R1.4 of canonical NSs is essentially determined by the pressure at densities around ρ0 ∼ 2.5ρ0 (Lattimer & Prakash 2000), while the maximum mass of NSs is determined by the pressure at higher densities reached in the core. Thus, knowledge of the density dependence of nuclear symmetry energy is important for understanding measurements of the masses and especially the radii of NSs. Moreover, the critical densities for forming hyperons (Sumiyoshi & Toki 1994; Lee 1996; Kubis & Kutschera 2003; Providência et al. 2019), Δ(1232) resonances (Drago et al. 2014; Cai et al. 2015; Zhu et al. 2016; Sahoo et al.2018; Ribes et al. 2019), kaon condensation (Odrzywolek & Kutschera 2009), and the quark phase (Di Toro et al. 2010; Wu & Shen 2019) are also known to depend sensitively on the high-density nuclear symmetry energy. Information about the latter is thus a prerequisite for exploring the evolution of the NS matter phase diagram in the isospin dimension. Once Esym(ρ) is better determined and, hopefully, with more astrophysical data, it will be interesting to introduce extra model parameters characterizing the physics associated with the exotic particles and/or new phases predicted to appear in superdense neutron-rich matter. With the very limited data available and the expensive computational costs of simultaneously inferring a lot more than the six EOS parameters we already have in the minimum NS model consisting of only nucleons and two leptons, our goals in this work are conservative and practical. However, inferring new physics parameters associated with the exotic degrees of freedom and new phases in superdense neutron-rich matter from astrophysical data by extending the model used in the present work are high on our working agenda.
|
[
"Lattimer & Prakash 2000"
] |
[
"It is also well known that the radius R1.4 of canonical NSs is essentially determined by the pressure at densities around ρ0 ∼ 2.5ρ0",
"while the maximum mass of NSs is determined by the pressure at higher densities reached in the core."
] |
[
"Uses",
"Uses"
] |
[
[
1220,
1243
]
] |
[
[
1086,
1218
],
[
1246,
1346
]
] |
2018MNRAS.478.4657P__Gadotti_2011_Instance_1
|
We have studied a sample of 263 LSB galaxies observed by Green Bank Telescope (Schneider et al. 1992) which are in overlap with the SDSS footprint. We have performed two-component bulge-disc decomposition of 263 galaxies in the SDSS g, r and, i bands and investigated their structural properties in detail. We have found that $60\hbox{ per cent}$ LSBs in our specific sample are bulgeless, while $40\hbox{ per cent}$ are with bulges. Some of the LSBs are associated with significant bulge component with B/T> 0.1. Since LSBs are known to be dwelling in less-dense environment (Rosenbaum & Bomans 2004), mergers and interactions are unlikely to have led the bulge formation. We also have $15 \hbox{ per cent}$ barred galaxies in our sample. Our findings of bulges and bars suggest a considerable on-going evolution in the local LSB galaxies and the bars might as well be playing a role in the bulge growth (Laurikainen et al. 2007; Gadotti 2011; Cheung et al. 2013). The interesting fact about our sample is that they are not the class of giant LSB galaxies, in fact, most of our LSBs are faint blue and gas-rich and roughly half of them are hosting bars and bulges. Since LSB galaxies are dark matter-dominated, discs are known to be stable against bar formation (Ostriker & Peebles 1973; Efstathiou, Lake & Negroponte 1982; Christodoulou, Shlosman & Tohline 1995; Cervantes Sodi, Li & Park 2015; Algorry et al. 2017), as shown in numerical simulations of stellar discs with dark matter dominance at all radii (Saha 2014). Question arises how these faint blue LSBs are making bars and bulges. One possibility is that LSB discs are embedded in dark matter haloes that are spinning (Jimenez et al. 1998; Vitvitska et al. 2002; Kim & Lee 2013) which might be promoting bar formation provided spin parameter is not too high (Saha & Naab 2013; Cervantes-Sodi et al. 2013; Long, Shlosman & Heller 2014; Collier, Shlosman & Heller 2018). Whether these bars lead to the formation of bulges or other processes such as minor mergers being involved, needs further and detailed investigation.
|
[
"Gadotti 2011"
] |
[
"Our findings of bulges and bars suggest a considerable on-going evolution in the local LSB galaxies and the bars might as well be playing a role in the bulge growth"
] |
[
"Compare/Contrast"
] |
[
[
931,
943
]
] |
[
[
740,
904
]
] |
2021MNRAS.508.4429C__McClintock_et_al._1976_Instance_1
|
Vela X-1 (4U 0900−40) is an eclipsing high-mass X-ray binary (HMXB) discovered during rocket borne X-ray observations in 1967 (Chodil et al. 1967). It is located at a distance of ∼2.0 kpc (Sadakane et al. 1985; Nagase 1989) in the Vela constellation. Recent estimates using Gaia data infer distance of $2.42^{+0.19}_{-0.17}$ kpc (Bailer-Jones et al. 2018). The system consists of a massive B0.5Ib supergiant HD 77581 (Brucato & Kristian 1972; Hiltner, Werner & Osmer 1972; Jones & Liller 1973; Vidal, Wickramasinghe & Peterson 1973) having mass of about ∼23 $\rm {M_\odot }$ and radius of ∼34 $\rm {R_\odot }$ (Van Paradijs et al. 1976; Joss & Rappaport 1984; Nagase 1989; van Kerkwijk et al. 1995) and a neutron star with mass ∼1.8 $\rm {M_\odot }$ (Van Paradijs et al. 1976; Nagase 1989; Barziv et al. 2001; Rawls et al. 2011). The orbital period of the binary system is about 9 d (Hiltner et al. 1972; Forman et al. 1973; Vidal et al. 1973; Watson & Griffiths 1977; van Kerkwijk et al. 1995). Due to the close proximity of about 1.7 $\rm {R_\star }$ (Conti 1978; Quaintrell et al. 2003) between the neutron star and its companion, the neutron star is immersed in the dense stellar wind of the donor star having typical mass-loss rate of about $\dot{M} \mathrm{\sim 10^{-6} ~{M_\odot } \, yr^{-1}}$ (Hutchings 1974; Dupree et al. 1980; Nagase et al. 1986; Sako et al. 1999). A fraction of the stellar wind is captured and channelled along the strong magnetic field (∼2.7 × 1012 G; Kretschmar et al. 1996; Coburn et al. 2002; Kreykenbohm et al. 2002) of the neutron star on to the magnetic poles, producing regular X-ray pulsations caused by the spin period ∼283 s (Rappaport & McClintock 1975; McClintock et al. 1976) of the neutron star. Although Vela X-1 is known to be a persistent source having luminosity of about 4 × 1036 erg s−1 (McCray et al. 1984; Sadakane et al. 1985; Nagase et al. 1986; Kreykenbohm et al. 2002), it shows a plethora of X-ray variabilities such as sudden flares lasting a few minutes to several hours wherein the luminosity increases by several folds within very short time-scales of a few tens of seconds (Lapshov et al. 1992; Staubert et al. 2004; Kreykenbohm et al. 2008). Occurrence of sudden flares in this system are not so well understood and is believed to be due to enhanced accretion rate due to variabilities in the stellar wind from the companion star (Nagase et al. 1983; Haberl & White 1990) or accretion of clumpy stellar wind (Staubert et al. 2004; Ducci et al. 2009; Fürst et al. 2010; Odaka et al. 2013). Some studies suggest that sudden flares might be related to formation of transient accretion disc (Inoue et al. 1984; Taam & Fryxell 1989; Haberl & White 1990; Kreykenbohm et al. 2008). Another bizarre manifestation seen in Vela X-1 is occurrence of abrupt ‘off-states’ wherein X-ray pulsations cessation (within less than the pulse period) is observed for several tens of minutes at a time (Inoue et al. 1984; Lapshov et al. 1992; Kreykenbohm et al. 1999, 2008; Doroshenko, Santangelo & Suleimanov 2011; Sidoli et al. 2015). These states are poorly understood and might be caused by changes in the accretion rate due to variabilities in the stellar wind (Lapshov et al. 1992; Coburn et al. 2002). Some earlier studies also suggest that ‘off-states’ might be associated with formation of transient accretion discs (Inoue et al. 1984) or the accretion is choked due to the sudden onset of propeller effect (Kreykenbohm et al. 2008). It has also been suggested that the onset of these ‘off-states’ can be caused due to transition from the higher luminosity Compton cooling regime to the lower luminosity radiative cooling regime (Shakura, Postnov & Hjalmarsdotter 2013) or due to unstable hydrodynamic flows in the vicinity of the neutron star (Manousakis & Walter 2015a). Recent numerical studies suggest formation of temporary accretion discs in wind-fed X-ray pulsars (Karino, Nakamura & Taani 2019; El Mellah, Sundqvist & Keppens 2019a; El Mellah et al. 2019b) but conclusive evidence of their existence has been elusive. Interestingly, Liao et al. (2020) infer presence of temporary accretion disc in Vela X-1 during an extended low state lasting at least 30 ks that was accompanied by unusual spin-up event and similar Fe Kα fluxes compared to the preceding flaring period.
|
[
"McClintock et al. 1976"
] |
[
"A fraction of the stellar wind is captured and channelled along the strong magnetic field",
"of the neutron star on to the magnetic poles, producing regular X-ray pulsations caused by the spin period ∼283 s",
"of the neutron star."
] |
[
"Background",
"Background",
"Background"
] |
[
[
1696,
1718
]
] |
[
[
1377,
1466
],
[
1552,
1665
],
[
1720,
1740
]
] |
2021ApJ...920..145H__Damone_et_al._2018_Instance_3
|
Over the past decade, many attempts to address this issue have been carried out, such as from the perspective of conventional nuclear physics and even exotic physics beyond the standard BBN framework (Angulo et al. 2005; Cyburt et al. 2008, 2016; Boyd et al. 2010; Pospelov & Pradler 2010; Fields 2011; Kirsebom & Davids 2011; Wang et al. 2011; Broggini et al. 2012; Coc et al. 2012, 2013, 2014; Cyburt & Pospelov 2012; Kang et al. 2012; Voronchev et al. 2012; Bertulani et al. 2013; Hammache et al. 2013; He et al. 2013; Kusakabe et al. 2014; Pizzone et al. 2014; Yamazaki et al. 2014; Hou et al. 2015, 2017; Famiano et al. 2016; Damone et al. 2018; Hartos et al. 2018; Luo et al. 2019; Rijal et al. 2019; Clara & Martins 2020). However, despite the fact some solutions using exotic physics have succeeded in resolving this issue, it appears there is still no universally accepted solution in the academic community since validations of these mysterious exotic physics are beyond the capabilities of current science. Conversely, it seems more worthwhile to exclude any potential possibility of resolving the 7Li discrepancy from the perspective of nuclear physics. It is known that the majority of the primordial 7Li production arises from the decay of 7Be by electron capture during the 2 months after BBN stops. Thus, for the solution of the Li problem, reactions involving 7Be could be more significant than those involving 7Li. Therefore, many reactions that potentially destroy 7Be were investigated to solve this discrepancy over past 10 yr (Kirsebom & Davids 2011; Broggini et al. 2012; Hammache et al. 2013; Hou et al. 2015; Hartos et al. 2018). Meanwhile, enormous efforts have been made to refine the reaction rates of key BBN reactions in the past 20 yr (Smith et al. 1993; Descouvemont et al. 2004; Serpico et al. 2004; Cyburt & Davids 2008; Neff 2011; Pizzone et al. 2014; Tumino et al. 2014; Hou et al. 2015; Barbagallo et al. 2016; Iliadis et al. 2016; Kawabata et al. 2017; Lamia et al. 2017, 2019; Damone et al. 2018; Rijal et al. 2019; Mossa et al. 2020), but the probability of solving or alleviating the 7Li problem by improving our knowledge of relevant nuclear reaction rates still cannot be eliminated. Recent experiments for key nuclear reactions like 7Be(n,p)7Li and 7Be(d,p)24He allow for a reduction of the 7Li production by about 12% (Damone et al. 2018; Rijal et al. 2019) compared to previous calculations. At present, nuclear uncertainties cannot rule out that some of the reactions destroying 7Li are indeed more efficient than those currently used (Boyd et al. 2010; Chakraborty et al. 2011).
|
[
"Damone et al. 2018"
] |
[
"Recent experiments for key nuclear reactions like 7Be(n,p)7Li and 7Be(d,p)24He allow for a reduction of the 7Li production by about 12%",
"compared to previous calculations."
] |
[
"Differences",
"Differences"
] |
[
[
2364,
2382
]
] |
[
[
2227,
2362
],
[
2403,
2437
]
] |
2015AandA...584A.103S__Potekhin_et_al._2013_Instance_3
|
Douchin & Haensel (2001; DH) formulated a unified EoS for NS on the basis of the SLy4 Skyrme nuclear effective force (Chabanat et al. 1998), where some parameters of the Skyrme interaction were adjusted to reproduce the Wiringa et al. calculation of neutron matter (Wiringa et al. 1988) above saturation density. Hence, the DH EoS contains certain microscopic input. In the DH model the inner crust was treated in the CLDM approach. More recently, unified EoSs for NS have been derived by the Brussels-Montreal group (Chamel et al. 2011; Pearson et al. 2012; Fantina et al. 2013; Potekhin et al. 2013). They are based on the BSk family of Skyrme nuclear effective forces (Goriely et al. 2010). Each force is fitted to the known masses of nuclei and adjusted among other constraints to reproduce a different microscopic EoS of neutron matter with different stiffness at high density. The inner crust is treated in the extended Thomas-Fermi approach with trial nucleon density profiles including perturbatively shell corrections for protons via the Strutinsky integral method. Analytical fits of these neutron-star EoSs have been constructed in order to facilitate their inclusion in astrophysical simulations (Potekhin et al. 2013). Quantal Hartree calculations for the NS crust have been systematically performed by (Shen et al. 2011b,a). This approach uses a virial expansion at low density and a RMF effective interaction at intermediate and high densities, and the EoS of the whole NS has been tabulated for different RMF parameter sets. Also recently, a complete EoS for supernova matter has been developed within the statistical model (Hempel & Schaffner-Bielich 2010). We shall adopt here the EoS of the BSk21 model (Chamel et al. 2011; Pearson et al. 2012; Fantina et al. 2013; Potekhin et al. 2013; Goriely et al. 2010) as a representative example of contemporary EoS for the complete NS structure, and a comparison with the other EoSs of the BSk family (Chamel et al. 2011; Pearson et al. 2012; Fantina et al. 2013; Potekhin et al. 2013) and the RMF family (Shen et al. 2011b,a) shall be left for future study.
|
[
"Potekhin et al. 2013"
] |
[
"We shall adopt here the EoS of the BSk21 model",
"as a representative example of contemporary EoS for the complete NS structure,"
] |
[
"Uses",
"Uses"
] |
[
[
1785,
1805
]
] |
[
[
1675,
1721
],
[
1828,
1906
]
] |
2022ApJ...929..186L__Lanzoni_et_al._2013_Instance_1
|
Our group is addressing this problem by combining a variety of complementary perspectives: (i) by constructing a new generation of high-quality star density profiles derived from star counts instead of surface brightness (see Lanzoni et al. 2007a, 2010, 2019; Miocchi et al. 2013; Pallanca et al. 2021); (ii) by investigating the population of stellar exotica (Ferraro et al. 2001, 2003, 2015, 2016; Pallanca et al. 2010, 2013,2014, 2017; Cadelano et al. 2017, 2018, 2020) and their connection with the dynamical evolution of the parent cluster (see Ferraro et al. 2009, 2012, 2018a, 2019; Lanzoni et al. 2016); (iii) by characterizing the three-dimensional (3D) global velocity space through the analysis of the velocity dispersion profile and rotation curve from resolved star spectroscopy (Lanzoni et al. 2013, 2018a, 2018b; Ferraro et al. 2018b) and proper motions (PMs; see Raso et al. 2020). The determination of GGC internal kinematics from resolved star velocities is particularly relevant and challenging. In this context we promoted the ESO-VLT Multi-Instrument Kinematic Survey (hereafter the MIKiS survey; Ferraro et al. 2018b, 2018c), a project specifically designed to characterize the kinematical properties of a sample of GGCs in different dynamical evolutionary stages from the radial velocities (RVs) of hundreds of individual stars distributed over the entire radial range of each stellar system. To this end, the survey fully exploits the spectroscopic capabilities of different instruments currently available at the ESO Very Large Telescope (VLT): originally designed to use the adaptive optics (AO) assisted integral-field spectrograph SINFONI, the multiobject integral-field spectrograph KMOS, and the multiobject fiber-fed spectrograph FLAMES/GIRAFFE, it has been recently complemented with individual projects and an ongoing large program (PI: Ferraro) fully exploiting the remarkable performances of the AO-assisted integral-field spectrograph MUSE.
|
[
"Lanzoni et al. 2013"
] |
[
"Our group is addressing this problem by combining a variety of complementary perspectives:",
"(iii) by characterizing the three-dimensional (3D) global velocity space through the analysis of the velocity dispersion profile and rotation curve from resolved star spectroscopy"
] |
[
"Uses",
"Uses"
] |
[
[
793,
812
]
] |
[
[
0,
90
],
[
612,
791
]
] |
2016MNRAS.455.4426V__Smith_&_Dwek_1998_Instance_1
|
It is interesting that the tracers of atomic and molecular hydrogen become maximum at distances ∼0.85 kpc, where no dust cloud is inferred from the Swift/XRT data (see Fig. 5). A fiducial dust cloud at the distance of ∼0.85 kpc would produce an X-ray ring with angular size ∼9 arcmin on MJD 57205.5. As this would fall well within the Swift/XRT FOV (see Fig. 2), its absence should be related to the dust properties of the respective cloud. We estimated therefore that the maximum grain size should be ≫ 0.2 μm. A dust cloud composed by large grains ( ∼ 1 μm) would suppress the scattered intensity even at 1 keV, while it would significantly attenuate X-rays at ≤0.5 keV (e.g. Smith & Dwek 1998; Corrales & Paerels 2015). Alternatively, the absence of the X-ray ring might indicate a much smaller number of grains per hydrogen atom at ∼0.85 kpc compared to the other clouds. The analysis presented so far, as in most studies, does not take into account possible variations of the radial profiles in the azimuthal direction. We remind that the profiles shown in Figs 3 and B1 were created by summing the counts of each annulus, i.e. integrating over the azimuthal angle. Although this method is sufficient for probing the average properties of the dust clouds, such as position and average column density, it cannot provide information about the spatial inhomogeneity of the dust clouds and/or their inclination with respect to the LOS. The photon statistics of the Swift/XRT observations are sufficient for quantifying the azimuthal variations of the ring intensity (see also Fig. 2). Fig. 12 shows the azimuthal variation of the scattered photons for Swift/XRT observation 2. There are three important features that need to be mentioned: (i) the radial profile of ring 5 is highly variable along the azimuthal direction, with the peak number of counts changing by a factor of ∼5 above the background level (see sub-panels for 180°–216° and 252°–288°); (ii) azimuthal variations are also present in the rings 3 and 4, yet the variability amplitude is lower when compared to the ring 5. The peak number of counts of the fourth ring is for all azimuth angles larger compared to that of the third ring; (iii) the azimuthal behaviour of the rings 1 and 2 is the most intriguing, as the second peak (corresponding to the ring 2) becomes prominent for azimuth angles 0°–36° and 324°–360° while the first peak dominates for azimuth angles of 216°–324°. Obviously, this information is lost from the azimuthal integrated radial profiles shown in Fig. 3. Similar features appear in all Swift/XRT observations with their significance varying according to the photon statistics of the observation.
|
[
"Smith & Dwek 1998"
] |
[
"A dust cloud composed by large grains ( ∼ 1 μm) would suppress the scattered intensity even at 1 keV, while it would significantly attenuate X-rays at ≤0.5 keV (e.g."
] |
[
"Uses"
] |
[
[
678,
695
]
] |
[
[
512,
677
]
] |
2019AandA...624A..60L__Dubernet_et_al._2016_Instance_1
|
One of the main uncertainties in elemental abundances is the quality of the adopted atomic data used in spectral synthesis calculations (Bigot & Thevenin 2008). The Belgian repository of fundamental atomic data and stellar spectra (BRASS) aims to provide astronomers with quality information for the large amount of atomic data available for high-resolution optical spectroscopy, in an attempt to help reduce systematic input errors in quantitative spectroscopy from atomic data and line selection (Lobel et al. 2017). Previously, we retrieved and cross-matched a large quantity of atomic data from several major atomic databases such as the Vienna Atomic Line Database (VALD3; Ryabchikova et al. 2015), the National Institute of Standards and Technology Atomic Spectra Database (NIST ASD; Kramida et al. 2018), and providers within the Virtual Atomic and Molecular Data Centre (VAMDC; Dubernet et al. 2016), in preparation for quality assessment work (Laverick et al. 2018; hereafter Paper 1). In this work the atomic data of seemingly “unblended” spectral lines are quality assessed against several benchmark dwarf stars, including the Sun, spanning late F-type to early K-type stars, for the spectral range 4200–6800 Å1. Unblended spectral lines for stars of ∼G2V spectral type are identified in a homogeneous manner using both the observed benchmark spectra, and the theoretical input line list of Paper I. Astrophysical oscillator strengths2 are derived for these unblended lines, using two commonly utilised methods, to gauge the reliability of the spectral line for quantitative spectroscopy and to produce “benchmark” log(g f) values for quality assessment of atomic data. The literature log(g f) values are then compared against these benchmark log(g f) values to determine which literature values reliably reproduce the stellar spectra of cool dwarf stars, and thus whether the values can be recommended for spectroscopic modelling. This paper presents three main sets of results:
|
[
"Dubernet et al. 2016"
] |
[
"Previously, we retrieved and cross-matched a large quantity of atomic data from several major atomic databases such as the",
"and providers within the Virtual Atomic and Molecular Data Centre (VAMDC;",
"in preparation for quality assessment work"
] |
[
"Uses",
"Uses",
"Uses"
] |
[
[
886,
906
]
] |
[
[
519,
641
],
[
812,
885
],
[
909,
951
]
] |
2020ApJ...897..177P__Lister_et_al._2019_Instance_1
|
Relativistic jets are the manifestation of the extreme processes that occur within the central regions of galaxies (see Blandford et al. 2019 for a review). Active galactic nuclei (AGNs) hosting relativistic jets closely aligned to the line of sight are called blazars. Due to their peculiar orientation, the relativistic amplification of the nonthermal jetted radiation (Doppler boosting; see, e.g., Rybicki & Lightman 1979) leads to the observation of a number of interesting phenomena. A few examples are detection at all accessible frequencies (e.g., Abdo et al. 2011), observation of temporal and spectral variability (Gaidos et al. 1996; Acciari et al. 2011; Fuhrmann et al. 2014; Paliya et al. 2017b), and superluminal motion and high brightness temperature (Scheuer & Readhead 1979; Lister et al. 2019). The optical and radio emissions detected from blazars are found to be significantly polarized (e.g., Fan et al. 2008; Itoh et al. 2016). The flux enhancement also makes blazars a dominating class of γ-ray emitters in the extragalactic high-energy sky (Ajello et al. 2020) and one of the very few astrophysical source classes detected at cosmic distances (e.g., Romani et al. 2004; Sbarrato et al. 2013). Blazars are classified as flat-spectrum radio quasars (FSRQs) and BL Lac objects based on their optical spectroscopic properties. FSRQs are characterized by broad emission lines (rest-frame equivalent width >5 Å), whereas BL Lac sources exhibit weak or no emission lines in their optical spectra, thereby making it challenging to detect their redshift (Stickel et al. 1991). BL Lac objects are known to exhibit a negative or mildly positive evolution compared to the strong positive evolution noticed in FSRQs (Ajello et al. 2012, 2014). Altogether, FSRQs dominate the known population of high-redshift (z ≳ 3) blazars and are found to be much more luminous than the BL Lac population (e.g., Ajello et al. 2009; Ackermann et al. 2017; Paliya et al. 2019d).
|
[
"Lister et al. 2019"
] |
[
"A few examples are",
"and superluminal motion and high brightness temperature"
] |
[
"Background",
"Background"
] |
[
[
791,
809
]
] |
[
[
489,
507
],
[
709,
764
]
] |
2020MNRAS.493.4868L__Lazarian_&_Hoang_2007b_Instance_1
|
Recently, polarized (sub)millimetre emission has been detected in an increasing number of discs by Atacama Large Millimeter/submillimeter Array (ALMA) with its high sensitivity and angular resolution. However, the origin of disc polarization remains uncertain, since grains do not have to be aligned with just the magnetic field (Kataoka et al. 2017; Yang et al. 2019). They may also be aligned in the direction of the radiative anisotropy (Lazarian & Hoang 2007a; Tazaki, Lazarian & Nomura 2017) or the drift velocity of the grains relative to the ambient gas (Gold 1952; Lazarian 1995; Lazarian & Hoang 2007b). Furthermore, even spherical grains can produce polarized emission by self-scattering of large grains in an anisotropic radiation field (Kataoka et al. 2015; Yang et al. 2016, 2017; Stephens et al. 2019). The scattering interpretation of the disc polarization is favoured in several targets (e.g. Stephens et al. 2014, 2017; Kataoka et al. 2016; Bacciotti et al. 2018; Dent et al. 2019; Girart et al. 2018; Harris et al. 2018; Hull et al. 2018; Lee et al. 2018). One way to gauge the effects of scattering and identify polarization from aligned grains would be to observe at multiple wavelengths since the efficiency for scattering for grains of given sizes decreases rapidly with the wavelength in the optically thin and small-particle (or Rayleigh scattering) limit. Indeed, in the disc of Class I protostar BHB 07-11, Alves et al. (2018) detected polarization with ALMA at three wavebands (Bands 3, 6, and 7 or ∼ 3, 1.3, and 0.87 mm, respectively) with consistent polarization orientations across three bands and increasing polarization fraction with wavelength, which is generally not expected for scattering-induced polarization. The rather high mean polarization fractions (∼7.9, 5.3, and 3.5 ${{\ \rm per\ cent}}$ for Bands 3, 6, and 7 respectively) are also higher than those typically produced in models of scattering-induced disc polarization ($\sim \!1 {{\ \rm per\ cent}}$). At least for this well-studied source, scattering is unlikely the main mechanism for producing the observed multiwavelength disc polarization and aligned grains are favoured.
|
[
"Lazarian & Hoang 2007b"
] |
[
"They may also be aligned in the direction of",
"or the drift velocity of the grains relative to the ambient gas"
] |
[
"Background",
"Background"
] |
[
[
588,
610
]
] |
[
[
370,
414
],
[
497,
560
]
] |
2021MNRAS.504.4626K__Kraljic_et_al._2020b_Instance_4
|
Galaxies seem to retain a memory of their spin orientation with respect to the cosmic web filaments and walls, as suggested by the results from large-scale cosmological hydrodynamical simulations (Dubois et al. 2014; Codis et al. 2018; Wang et al. 2018; Ganeshaiah Veena et al. 2019; Kraljic, Davé & Pichon 2020b). The mass dependence of the spin alignment signal is however debated. While some works confirmed the existence of a galaxy spin transition from parallel to perpendicular with respect to the filament’s direction (Dubois et al. 2014; Codis et al. 2018; Kraljic et al. 2020b), and analogously with respect to walls (Codis et al. 2018; Kraljic et al. 2020b), others (Ganeshaiah Veena et al. 2019; Krolewski et al. 2019) found preferential perpendicular orientation with respect to filaments at all masses with no sign of a spin transition. A possible interpretation of this lack of detection of a clear transition is the nature of the filaments, with galaxies in thinner filaments having their spins more likely perpendicular to the filament’s axis, compared to galaxies of similar mass in thicker filaments (Ganeshaiah Veena et al. 2019). This can be in turn understood recalling the multiscale nature of the problem and the conditional TTT (Codis et al. 2015) predicting larger transition mass for denser, thus thicker, filaments. Further support for this interpretation was provided by the findings of stronger impact of large-scale tides on the galaxy spin orientation in denser filaments (Kraljic et al. 2020b, using filament density as a proxy for the thickness of filaments). In addition to the stellar mass, the spin-filament alignment was shown to depend on other internal properties of galaxies. Blue or rotation-supported galaxies were found to dominate the alignment signal at low stellar mass, while red or dispersion-dominated galaxies tend to show a preferential perpendicular alignment (Codis et al. 2018; Wang et al. 2018; Kraljic et al. 2020b).
|
[
"Kraljic et al. 2020b"
] |
[
"In addition to the stellar mass, the spin-filament alignment was shown to depend on other internal properties of galaxies. Blue or rotation-supported galaxies were found to dominate the alignment signal at low stellar mass, while red or dispersion-dominated galaxies tend to show a preferential perpendicular alignment"
] |
[
"Background"
] |
[
[
1950,
1970
]
] |
[
[
1593,
1911
]
] |
2017ApJ...839...26D__Conroy_et_al._2006_Instance_1
|
One might hope that these global observations would strongly constrain the SFHs of individual galaxies, but this connection is not easily established. For example, average SFHs can be inferred by integrating the main sequence SFR over time (e.g., Leitner 2012), but this approach leads to inconsistencies (Leja et al. 2015). Instead, the most successful theoretical models link the growth of stellar mass to the growth of the dark matter halos that galaxies inhabit, for example, via subhalo abundance matching (e.g., Kravtsov et al. 2004; Conroy et al. 2006; Behroozi et al. 2013b; Moster et al. 2013), halo occupation distributions (e.g., Peacock & Smith 2000; Seljak 2000; Hearin et al. 2016), semi-analytic models (Kauffmann et al. 1993; Somerville et al. 2001; Guo et al. 2011), or other assumptions (Bouché et al. 2010; Davé et al. 2012; Lilly et al. 2013; Tacchella et al. 2013; Mitra et al. 2017). One important conclusion from these models is that there has to be significant scatter between halo and galaxy masses (and thus growth histories) in order to explain observations (More et al. 2009; Behroozi et al. 2013b; Reddick et al. 2013; Gu et al. 2016). As a result, even models that agree on global constraints can lead to orthogonal interpretations of the evolution of individual galaxies. A good example for such disagreement is the “rapid quenching”7
7
The term “quenching” is somewhat ambiguous. In this paper, we use it to mean the cessation of star formation, without any presumption as to whether the decrease happens quickly or slowly, and whether it happens due to a diminishing gas supply or other physical processes.
framework where galaxies follow the main sequence until they sharply fall below the main sequence (Peng et al. 2012; Wetzel et al. 2013; Tacchella et al. 2016b; Tinker et al. 2016) and the “stochastic” framework where correlated scatter and the central limit theorem lead to the main sequence (Kelson 2014; Kelson et al. 2016).
|
[
"Conroy et al. 2006"
] |
[
"Instead, the most successful theoretical models link the growth of stellar mass to the growth of the dark matter halos that galaxies inhabit, for example, via subhalo abundance matching (e.g.,"
] |
[
"Background"
] |
[
[
540,
558
]
] |
[
[
325,
517
]
] |
2020ApJ...891..111K__Howe_&_Burrows_2015_Instance_1
|
To explore magma effects on sub-Neptune atmospheres, we assume the atmosphere equilibrates with a well-stirred magma ocean. Our model makes the following simplifications: (a) we consider only the elements Fe, Mg, Si, O, and H. Chemically reduced carbon compounds may also contain H; for simplicity, we omit consideration of them here. We also restrict ourselves to the range of magma elemental compositions for which SiO2 is a major constituent. (b) We set 2000 K ≤ Tmai ≤ 3000 K, so that the magma–atmosphere interface is molten but the vapor pressure of the magma is small relative to the total atmospheric pressure (Fegley et al. 2016; Sossi & Fegley 2018). This Tmai is at the low end of the Tmai output by thermal evolution models for multi-Gyr-old sub-Neptunes (e.g., Bodenheimer & Lissauer 2014; Howe & Burrows 2015; Vazan et al. 2018a). We consider lower Tmai in Section 4.4. (c) We ignore Fe3+ (i.e., Fe2O3). We expect Fe3+ will be a minor constituent of a magma ocean equilibrated with a H2-dominated atmosphere. Equal thermodynamic activities of FeO and Fe2O3 in magma at 3000 K require fO2 = 28 bar, much higher than expected from thermal dissociation of steam at any P and T below 1 kilobar and 3500 K. (d) We assume that metal (if present) is pure Fe for these calculations; in reality, the metal will be an Fe-dominated alloy. We may be (slightly) underestimating fO2 by doing this. (e) We track nonideal behavior of both H2 and H2O (Appendix C), but we assume ideal mixing of H2 and H2O. This is a valid assumption both under mineral-free conditions in the atmosphere for T > 650 K, and also at the magma–atmosphere interface given our model assumptions (Seward & Franck 2019; Bali et al. 2013; Soubiran & Militzer 2015). We ignore joint-solubility effects. (f) We use a single value of gravitational acceleration g, corresponding to 1.2× the bare-rock radius, to convert from bottom-of-atmosphere pressure to atmosphere column mass. (g) We ignore the effect of dissolved volatiles on core mass.
|
[
"Howe & Burrows 2015"
] |
[
"This Tmai is at the low end of the Tmai output by thermal evolution models for multi-Gyr-old sub-Neptunes (e.g.,"
] |
[
"Compare/Contrast"
] |
[
[
803,
822
]
] |
[
[
661,
773
]
] |
2017ApJ...848..126S__Luo_et_al._2017_Instance_1
|
In Figure 2, we show the Chandra 0.3–8 keV images with SDSS contours overlaid for the six mergers in our sample. In Table 4 we list the number of counts detected in the 0.3–8 keV and 2–8 keV bands, and list the detection threshold, PB, associated with each position. There are a number of PB thresholds employed in the literature to define source detection significance. Based on weak sources in Chandra deep field images, the threshold for PB adopted based on a balance between reliability and completeness varies from 0.002 to 0.007 (Xue et al. 2011, 2016; Luo et al. 2017). Based on even the lowest of all these thresholds, all sources are detected. Note that the detection thresholds adopted in the literature for sources at known locations are often PB 0.01, which corresponds to
(e.g., Lansbury et al. 2014). We note that due to insufficient counts, coupled with the fact the exact location of the galaxy nuclei is uncertain in the advanced mergers in our sample (see SDSS images in Figure 1), it is impossible to quantify any possible spatial offsets between the positions of the detected Chandra sources and the actual galactic nuclei. It is therefore not possible to determine using these observations alone if some of the mergers host offset AGNs, another unambiguous signature of a galaxy merger that can probe AGN triggering through galaxy interactions (Barrows et al. 2016, 2017). We note that for sources within the 2′ of the boresight, the absolute accuracy of source locations on the ACIS-S chip has a 90% uncertainty radius of
,18
18
http://cxc.cfa.harvard.edu/proposer/POG/pdf/MPOG.pdf
indicating that the positions of all detected targets are consistent with the locations of the SDSS knots, suggesting that the detected sources likely correspond to the nuclei of the mergers. J1036+0221, J1045+3519 Gal 1, J1221+1137 Gal 1, and J1306+0735 Gal 2 were detected in the hard band using the same detection threshold PB 0.002 using the 2–8 keV source counts. Note that there are insufficient counts to perform a spectral analysis or to obtain reliable HRs for our targets. We therefore list in Table 6 the observed X-ray luminosity, assuming a simple power-law model with
, corrected for Galactic absorption.
|
[
"Luo et al. 2017"
] |
[
"Based on weak sources in Chandra deep field images, the threshold for PB adopted based on a balance between reliability and completeness varies from 0.002 to 0.007"
] |
[
"Uses"
] |
[
[
559,
574
]
] |
[
[
371,
534
]
] |
2021MNRAS.507.5053E__Johnston_et_al._2006_Instance_2
|
Multiwavelength observations of the GC indicate that the number of pulsars in the central few parsecs should be high (Wharton et al. 2012) and conditions are highly favourable for relativistic binaries (Faucher-Giguère & Loeb 2011). The dense nuclear star cluster surrounding Sgr A* (see e.g. Genzel, Eisenhauer & Gillessen 2010, for a review) contains a majority of older late-type stars, but contrary to expectations, massive young main-sequence stars (Ghez et al. 2003) and possible neutron star progenitors such as Wolf–Rayet stars (Paumard et al. 2001). The presence of neutron stars is further indicated by large numbers of X-ray binaries, possible pulsar wind nebulae, X-ray features such as the ‘cannonball’ and compact radio variables (Muno et al. 2005; Wang, Lu & Gotthelf 2006; Zhao, Morris & Goss 2013, 2020). Despite this only six radio pulsars have been discovered within half a degree of Sgr A* (Johnston et al. 2006; Deneva, Cordes & Lazio 2009; Eatough et al. 2013c; Shannon & Johnston 2013) even after many dedicated searches at multiple wavelengths (Kramer et al. 1996a, 2000; Klein et al. 2004; Klein 2005; Deneva 2010; Macquart et al. 2010; Eatough et al. 2013a; Siemion et al. 2013). Hyperstrong scattering of radio waves in the GC has been the principal explanation for the scarcity of detected pulsars (Cordes & Lazio 1997, 2002; Lazio & Cordes 1998a,b), however, scatter broadening measurements of PSR J1745−2900 in Spitler et al. (2014) and Bower et al. (2014) appear to contest this.1 Other authors have noted that the lack of GC pulsars is expected under a certain set of conditions and considering the sensitivity limits of existing pulsar surveys (Chennamangalam & Lorimer 2014; Liu & Eatough 2017; Rajwade, Lorimer & Anderson 2017). Alternatively, the scarcity of detected pulsars might be caused by a more complex scattering structure towards the GC (Cordes & Lazio 1997; Lazio & Cordes 1998a, b; Johnston et al. 2006; Schnitzeler et al. 2016; Dexter et al. 2017).
|
[
"Johnston et al. 2006"
] |
[
"Alternatively, the scarcity of detected pulsars might be caused by a more complex scattering structure towards the GC"
] |
[
"Background"
] |
[
[
1929,
1949
]
] |
[
[
1764,
1881
]
] |
2017ApJ...844...14L__Lim_et_al._2016_Instance_1
|
In order to see whether the CN- and HK′-strong stars in our study are also enhanced in Fe and s-process elements, we have compared our results with high-resolution spectroscopy by M15. In Figure 4, our δCN and δHK′ indices are plotted with [Na/Fe] and [Ca/Fe] abundances, respectively, for 33 common stars. In general, the strength of the CN band is correlated with the N and Na abundances, while the CH band is affected by C abundance (Sneden et al. 1992; Smith et al. 1996; Marino et al. 2008). The upper panel of Figure 4 also shows a strong correlation between [Na/Fe] and the δCN index, which is in good agreement with previous studies (Sneden et al. 1992; Lim et al. 2016).4
4
Careful inspection of the upper panel of Figure 4 also shows the possibility that the s-poor and s-rich groups are probably separated on this diagram, with the s-rich stars more enhanced in both [Na/Fe] and δCN. This would imply that the variations in light elements would be present in each group with different s-process elements abundances, which has already been found in other GCs with s-process element and Fe variations, such as M2 and M22 (Marino et al. 2011; Yong et al. 2014). More samples of stars, however, are needed to confirm this trend in NGC 5286.
The CN-weak (blue) and CN-strong (red) subpopulations are almost identical to the s-poor (triangles) and s-rich (squares) groups, respectively. In addition, the δHK′ index is understandably correlated with the [Ca/Fe] abundance with a few exceptions (see the lower panel of Figure 4). According to this comparison, the difference in δHK′ index between CN-weak and CN-strong stars (∼0.094) is equivalent to 0.09 dex in Δ[Ca/Fe] and 0.15 dex in Δ[Fe/H]. These comparisons confirm that our results from low-resolution spectroscopy are consistent with those from high-resolution spectroscopy by M15. Consequently, the later-generation stars in NGC 5286 show the enhancements not only in light elements (CN) but also in heavy elements (Fe and Ca) and s-process elements, although the presence of Fe spread requires further investigations (see Mucciarelli et al. 2015; Lee 2016).
|
[
"Lim et al. 2016"
] |
[
"The upper panel of Figure 4 also shows a strong correlation between [Na/Fe] and the δCN index, which is in good agreement with previous studies"
] |
[
"Similarities"
] |
[
[
662,
677
]
] |
[
[
497,
640
]
] |
2019MNRAS.485.4841R__Creminelli_et_al._2010_Instance_2
|
Although, the standard form of Press–Schechter mass function with $f(\nu)=\sqrt{{2}/{\pi }} \nu \mathrm{ e}^{-\frac{\nu }{2}}$ which discussed in Press & Schechter (1974) and Bond et al. (1991) can provide a good approximation of the predicted number density of haloes, it fails by predicting approximation too many low-mass haloes and too few high-mass ones (Sheth & Tormen 1999, 2002; Lima & Marassi 2004). Thus, in this study we apply another well-known fitting formula which first proposed in Sheth & Tormen (1999):
(24)
\begin{eqnarray*}
f(\nu)=0.2709\sqrt{\dfrac{2}{\pi }}(1+1.1096\nu ^{0.6})\mathrm{ exp}(-\dfrac{0.707 \nu ^2}{2})\,\, .
\end{eqnarray*}
In a Gaussian density field, σ is given by
(25)
\begin{eqnarray*}
\sigma ^2(R)=\dfrac{1}{2 \pi ^2}{\int _0}^\infty k^2 P(k) W^2(kR) \, {\rm d}k\,\, ,
\end{eqnarray*}
where R = (3M/4πρm0)1/3 is the radius of the spherical overdense region, W(kR) is the Fourier transform of a spherical top-hat profile with radius R and P(k) is the linear power spectrum of density fluctuations (Peebles 1993). To obtain the value of σ, we follow the procedure presented in Abramo et al. (2007a). Following on Ade et al. (2016), we use the normalization of matter power spectrum σ8 = 0.815 for ΛCDM cosmology. The number density of virialized haloes above a certain value of mass M at zc, the collapse redshift obtained by
(26)
\begin{eqnarray*}
N(\: M,z)={\int _0}^\infty \dfrac{{\rm d}n(z)}{{\rm d}M^{\prime }}\, {\rm d}M^{\prime }\,\,.
\end{eqnarray*}
The above limit of integration in equation (26) is $M=10^{18}\, \mathrm{ M}_{\rm \odot}\, \mathrm{ h}^{-1}$ which such gigantic structures could not in practice be observed. Now we can calculate the number density of virialized haloes in both homogeneous and clustered DE scenarios using equations (23) and (26). In this way the total mass of a halo is equal to the mass of pressureless matter perturbations. However, the virialization of dark matter perturbations in the non-linear regime cannot be independent from the properties of DE (Lahav et al. 1991; Maor & Lahav 2005; Creminelli et al. 2010; Basse, Bjlde & Wong 2011). Thus, in clustered DE scenarios, we should consider the contribution of perturbated DE components to the total mass of the haloes (Creminelli et al. 2010; Basse et al. 2011; Batista & Pace 2013). Based on the behaviour of wde(z), DE can reduce or enhance the total mass of the virialized halo. One can obtain ϵ(z), the ratio of DE mass to be taken into account with respect to the mass of dark matter, from:
(27)
\begin{eqnarray*}
\epsilon (z)=\dfrac{m_{\rm DE}}{m_{\rm DM}}\,\, ,
\end{eqnarray*}
where the value of mDE depends on what we consider as the mass of DE component. When one only considers the contribution of the perturbations of DE, the mDE takes the form
(28)
\begin{eqnarray*}
{m_{\rm DE}}^{\mathrm{ Perturbed}}=4 \pi \bar{\rho }_{\rm DE}{\int _0}^{R_{\rm vir}} \, {\rm d}R R^2 \delta _{\rm DE}(1+3{c_{\rm eff}}^2)\,\,.
\end{eqnarray*}
In the other hand, if we assume both DE contributions of perturbation and background level, the total mass of DE in virialized haloes takes this new form
(29)
\begin{eqnarray*}
{m_{\rm DE}}^{\mathrm{ Total}}=4 \pi \bar{\rho }_{\rm DE}{\int _0}^{R_{\rm vir}} {\rm d}R R^2 [(1+3 w_{\rm DE})+ \delta _{\rm DE}(1+3{c_{\rm eff}}^2)].
\end{eqnarray*}
The quantities inside a spherical collapsing region in the framework of the top-hat profile, evolve only with cosmic time. Thus from equation (28) one can find
(30)
\begin{eqnarray*}
\epsilon (z)=\dfrac{\Omega _{\rm DE}}{\Omega _{\rm DM}}\dfrac{\delta _{\rm DE}}{1+\delta _{\rm DM}}\,\,
\end{eqnarray*}
and from equation (29) we can obtain
(31)
\begin{eqnarray*}
\epsilon (z)=\dfrac{\Omega _{\rm DE}}{\Omega _{\rm DM}}\dfrac{1+3 w_{\rm DE}+\delta _{\rm DE}}{1+\delta _{\rm DM}}\,\, .
\end{eqnarray*}
The mass of dark matter also is obtained from (see also Batista & Pace 2013):
(32)
\begin{eqnarray*}
{m_{\rm DM}}=4 \pi \bar{\rho }_{\rm DM}{\int _0}^{R_{\rm vir}} \, {\rm d}R R^2 (1+ \delta _{\rm DM})\,\,.
\end{eqnarray*}
In Fig. 5 we plot the evolution of ϵ(z) using equation (30) as the definition of DE mass. We observe that, at high redshift, where the role of DE is less important, ϵ for both of parametrizations becomes negligible. This parameter has a greater value in the case of parametrization (2).
|
[
"Creminelli et al. 2010"
] |
[
"Thus, in clustered DE scenarios, we should consider the contribution of perturbated DE components to the total mass of the haloes"
] |
[
"Uses"
] |
[
[
2272,
2294
]
] |
[
[
2141,
2270
]
] |
2015AandA...576L..16P__Chandler_&_Sjouwerman_2014_Instance_1
|
To constrain the size of outflows that we could have missed, we performed simple simulations. We placed artificial, unipolar secondary sources next to a primary point source model representing Sgr A* and compared the closure phases obtained from the resulting artificial visibility data with the observations. We considered two geometries: a single-point source and a jet composed of ten equally spaced point sources (knots) with equal fluxes. We probed four orientations for the simulated outflows (see Fig. 2): along the major axis, along the minor axis of the beam, the jet direction claimed by Li et al. (2013), and the jet direction claimed by Yusef-Zadeh et al. (2012). We used total fluxes of 0.2 Jy and 0.55 Jy for the artificial sources; these values ensure that our simulated outflows are sufficiently faint to not violate the constraints given by the known recent brightness evolution of Sgr A* (0.2 Jy from the mean variability of ≈15% from June 2013 to February 2014 at 41 GHz, with 0.55 Jy corresponding to the strongest variation in the same period, Chandler & Sjouwerman 2014). For each simulation setup, we measured the average of absolute values of the closure phases for each triangle. We varied the distances of the model sources (for the jet model: the largest distance) from Sgr A* until we found a critical distance at which the absolute values of the simulated closure phases exceeded those of the observations by more than the 1σ error at all triangles. We summarize our results in Table 1. As expected, the critical distances are smaller for brighter outflows. Jet-like structures lead to larger critical distances than equally luminous single, compact sources. As a consequence of the very elongated beam, the critical distances for sources located along the major axis of the beam are larger by a factor of ≈7 than for those located along the minor axis. In a few cases (denoted “N/A” in Table 1), the absolute values of the simulated closure phases were similar to those of the observations for all distances of the model sources, meaning that we were unable to identify a critical distance. Overall, our observations limit the extension of asymmetric (in the observer frame) jet-like outflows from Sgr A* to projected distances of ≈2.5 mas along the major axis and ≈0.4 mas along the minor axis.
|
[
"Chandler & Sjouwerman 2014"
] |
[
"We used total fluxes of 0.2 Jy and 0.55 Jy for the artificial sources; these values ensure that our simulated outflows are sufficiently faint to not violate the constraints given by the known recent brightness evolution of Sgr A* (0.2 Jy from the mean variability of ≈15% from June 2013 to February 2014 at 41 GHz, with 0.55 Jy corresponding to the strongest variation in the same period,"
] |
[
"Uses"
] |
[
[
1065,
1091
]
] |
[
[
676,
1064
]
] |
2015ApJ...798..104S___2000_Instance_1
|
The Vela Molecular Ridge Cloud-D (hereafter VMR-D) (260° â² â â² 264°; |b| â² 1°) is part of a giant molecular complex located along the Galactic plane (260° â² â â² 272°; |b| â² 3°; Murphy & May 1991) and is then well suited to represent a typical star-forming region (SFR) of our Galaxy. For this reason a subregion of this cloud has been the subject of many previous papers, dealing with different observational aspects of the star formation (SF), such as the presence of outflows (Wouterloot & Brand 1999; Elia et al. 2007), jets (Lorenzetti et al. 2002; Giannini et al. 2005, 2013), and clustering (Massi et al. 2000). The continuum submillimeter emission in the VMR-D cloud was surveyed by Massi et al. (2007), who catalogued 29 resolved dust cores and also obtained a further list of 26 unresolved candidate cores. More recently, thanks to the opportunity offered by the Spitzer Space Telescope, the VMR-D region was observed with the IRAC (λ = 3.6, 4.5, 5.8, 8.0âμm) and MIPS (λ = 24, 70âμm) focal-plane instruments, obtaining in this way six mosaics, covering about 1.2 deg2, which have been analyzed to produce a merged photometric Spitzer-IRAC point-source catalog (hereinafter Spitzer-PSC) complemented with MIPS photometry (Strafella et al. 2010, hereinafter Paper I). Further observational progress was made when the BLAST experiment (Pascale et al. 2008) mapped the whole Vela Molecular Ridge in the far-IR (FIR) spectral region (λ = 250, 350, 500âμm), complementing in this way the Spitzer spectral coverage toward long wavelengths. These observations were discussed by Olmi et al. (2009), who obtained a catalog of dense cores/clumps in the VMR-D cloud. The last important observational progress was made with the Herschel Space Observatory, which, surveying the Galactic plane in the framework of the Hi-GAL key project, partially mapped the VMR-D region in the FIR spectral range (λ = 70, 160, 250, 350, 500âμm) with a spatial resolution and sensitivity almost twice that of BLAST. Here we also analyze these observations for the first time, thanks to the support of the Hi-GAL collaboration that provided us with the corresponding calibrated maps. These have been used to extract five single-band photometries that constitute another important spectral extension of our information about this region.
|
[
"Massi et al. 2000"
] |
[
"For this reason a subregion of this cloud has been the subject of many previous papers, dealing with different observational aspects of the star formation (SF), such as",
"and clustering"
] |
[
"Background",
"Background"
] |
[
[
623,
641
]
] |
[
[
306,
474
],
[
607,
621
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.