Datasets:

SouthernCrossAI
/

Arxiv

Dataset card Data Studio Files Files and versions Community

Split (1)

train · 132k rows

Filename stringlengths 22 64	Paragraph stringlengths 8 5.57k
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	mode. X-ray pulsations have only been detected in the high mode (Papitto et al. 2015; Archibald et al. 2015). PSR J1023+0038 has also been known to enter an ex- tended ﬂare mode occasionally. These extended ﬂaring 10 hours in X-rays (Tendulkar episodes can last up to et al. 2014; Li et al. 2014) and 14 hours in optical (Pa- pitto et al. 2018). Optical pulsations were also detected when PSR J1023+0038 was in the ﬂare mode (Papitto et al. 2019), suggesting that the ﬂares are happening near the neutron star. Recently, Coti Zelati et al. (2019) found another tMSP candidate, CXOU J110926.4-650224, to ﬂare in X-rays for up to 4.5 hours. The tMSP identity, however, remains questionable because of the insigniﬁ- cant GeV γ-ray counterpart (The Fermi-LAT collabora- tion 2019; Hui & Li 2019).
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	In the radio band, the steep spectrum emission and radio pulsations that were clearly detected in the pulsar state disappear in the sub-luminous disk state (Archibald et al. 2009; Papitto et al. 2013; Stappers et al. 2014). Instead, ﬂat-spectrum radio emission is detected (Hill et al. 2011; Deller et al. 2015). Recently, Bogdanov et al. (2018) found a strong anti-correlation between the radio and X-ray emission in the simultaneous radio and X- ray observations of PSR J1023+0038—the radio source is bright in the low mode, and faint in the higher mode. These data suggest that a synchrotron-emitting outﬂow is launched during the low mode.
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	Fig. 2.— The X-ray light curve (0.2–10 keV; blue) of 3FGL J0427.9−6704 observed by XMM-Newton in May 2017. The left y-axis shows the corrected count rate measured by pn and MOS 1/2 (the high ﬂaring background intervals are also included), while the right axis refers to the X-ray luminosity inferred from the best-ﬁt partially-absorbed power-law model in Average1 (see Table 1 for the best-ﬁt parameters) with d = 2.3 kpc. The three 2200-sec eclipses are all indicated by the gray shaded regions based on the solution in S16 while the two vertical dashed lines indicate the inferior conjunction of the binary. The red line is the best-ﬁt Gaussian model for the X-ray ﬂares (see text for details). Every ﬂare detected is marked with a vertical bar on the top. Red and thicker bars are used to indicate the 16 representative ﬂares (see Figure 4), and the rest are black in color. The gray narrow curve is the OM light curve scaled to ﬁt the ﬁgure (see Figure 3 for the original). The three horizontal green dashed lines represent the X-ray luminosities (0.3–10 keV) of PSR J1023+0038 in the ﬂare (≈ 1034 erg s−1), high (≈ 3 × 1033 erg s−1), and low (≈ 5 × 1032 erg s−1) modes, respectively, which are obtained from Bogdanov et al. (2015).
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	cluded with a 6-pixel radius circular region. All times are presented in the frame of Barycentric Dynamical Time (TDB), converted by the SAS task barycen with the JPL solar system ephemeris DE200 (Standish 1990).
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	We arranged strictly simultaneous radio observations of 3FGL J0427.9 6704 using the Australia Telescope Compact Array (ATCA), observing for as long as the source was above the horizon, from 18:10 UT on 2017 May 2 to 13:07 UT on 2017 May 3, under project codes CX364 and C3170. The array was in the extended 6A conﬁguration, with all six antennas aligned east-west, with a maximum baseline of 6 km. We used the Com- pact Array Broadband Backend (CABB; Wilson et al. 2011) to observe simultaneously at central frequencies of 5.5 and 9.0 GHz, with 2048 MHz of bandwidth in each of the two frequency bands.
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	Fig. 3.— The same as in Figure 2, but for the OM optical/UV observations alone.
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	with a robust parameter of 0.5, as a compromise be- tween sensitivity and minimizing sidelobe levels. The source was detected at high signiﬁcance in both fre- quency bands.
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	We imaged several subsets of data, to determine whether the radio emission changed when in eclipse, and how it varied with the changing X-ray count rates de- ﬁned in Section 3.2. Finally, we made time-resolved light curves. Since the array was in an east-west conﬁgura- tion, the instantaneous uv-coverage was linear, and not suitable for imaging. We therefore subtracted out all other sources in the ﬁeld, using our best image made from the entire data set at each frequency, leaving only 6704 in the visibility data, which we ﬁt 3FGL J0427.9 in the uv-plane using a point source model ﬁxed at the known source position. We found that ten-minute bins provided the best compromise between time resolution and sensitivity.
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	of 3FGL J0427.9 6704 from 22:29 UT on 2016 Aug 27 to 04:38 UT on 2016 Aug 28, under project code CX365 and in the extended 6C conﬁguration. The receiver setup, calibrators, and imaging were the same as for the 2017 data. A single image was made at each of the central frequencies of 5.5 and 9.0 GHz. These radio data were not simultaneous with any X-ray observations.
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	cts s−1 (pn + MOS 1/2) and 6 cts s−1, respectively (Fig- ure 2 & 3). Both light curves are dominated by strong ﬂares on time-scales of 10–100 sec. In addition, each light curve shows three eclipses, which perfectly match the orbital ephemeris presented in S16 (Figure 2 & 3). During the eclipse phases, the X-ray and optical ﬂuxes are lower, and the ﬂaring phenomenon is not observed. We tested both light curves for periodic signals longer than 1 sec using powspec, with no clear periodicity found in the frequency range 10−5 Hz to 1 Hz.
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	Fig. 4.— The EPIC X-ray (blue and thinner) and optical/UV (red and thicker) light curves of 16 arbitrarily chosen ﬂares to show the range of relationships between the X-ray and optical ﬂares. Each zoom-in view is centred at the peak of the chosen ﬂare. The optical/UV light curves shown are scaled linearly to ﬁt the X-ray ones for easier comparison, i.e., Cs = R(Co − O), where Cs is the scaled count rate, Co is the original count rate, R is the X-ray-to-optical/UV ﬂare amplitude ratio, and O is the minimum count rate of the original optica/UV light curve in the interval. R and O can be found in the upper right corner of each plot. 4 cts s−1 were adopted. A higher detection threshold of 4.5σ was also used to avoid over-ﬁtting. Otherwise, we found that all residual features were ﬁt with low- amplitude “ﬂares”. We emphasize that this technique is not designed to achieve a perfect statistical model of the data, but instead to give a ﬁrst-order sense of the frequency and amplitude of the ﬂares.
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	As shown in Figures 2 and 3, the X-ray and opti- cal/UV ﬂares that are visually obvious are all iden- tiﬁed by our technique. Outside of eclipse, on aver- age we detect an X-ray (optical/UV) ﬂare every 540 (470) sec. The occurrence rate is high compared with PSRs J1023+0038 and J1227 4853, of which the ﬂare occurrence rates are just up to a few tens of events per day on average (de Martino et al. 2013; Bogdanov et al. 2015; Papitto et al. 2018; Kennedy et al. 2018). The ﬂaring state of PSR J1023+0038 observed in opti- cal occupied about 15.6–22% of the time during the Ke- pler K2 observation (Papitto et al. 2018; Kennedy et al. 2018). Using the FWHM durations of the detected OM ﬂares, we found that the ﬂaring fraction is 43% for 3FGL J0427.9 6704, which is double the fraction seen in PSR J1023+0038. We show a selection of 16 pairs of ﬂares in Figure 4.
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	Fig. 5.— From top to bottom, the ﬁgures show the total number (a and b), the best-ﬁt Gaussian amplitudes (c and d), and the best-ﬁt Gaussian sigma (e and f) of the ﬂares detected by EPIC (left) and OM (right) at diﬀerent orbital phases. In the panels (a) and (b), the step function shows the distribution of the number of ﬂares normalized by the exposure of each phase bin w.r.t. that of the ﬁrst bin. Note that the gaps around phase 0.75 are due to the X-ray/optical eclipses. The vertical bars at the upper right corners show the typical estimates of the uncertainties. mine the uncertainty in this value, we simulated 104 pairs of light curves based on the EPIC and OM data, and repeated the crosscor calculations with the simulated data. The distribution of the simulated cross-correlation peaks (Figure 6) gives ∆td = 4.5 6.8 sec (90% conﬁ- dence interval). Hence there is no strong evidence for a time delay of the optical/UV emission compared to the X-ray emission. To put this in context, the light curve modeling of S16 found that the outer edge of the ac- 2.3 light-seconds) from the cretion disk is neutron star.
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	Fig. 6.— The cross-correlation function (5-sec resolution) be- tween the EPIC and the OM light curves computed by crosscor. The cross-correlations were normalized by dividing by N1N2, where N1 and N2 are the numbers of bins in the two light curves, respectively. The three eclipse phases were excluded in the cal- culation. In the function, a positive time delay (∆td) refers an optical/UV emission delay w.r.t. the X-ray emission. No evidence of time delay is found in the analysis with ∆td = 4.5 ± 6.8 sec (in- dicated by the vertical lines in the zoomed-in version in the inset box).
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	larger than 0.8 with the X-ray time-scales just (cid:46) 10 sec longer. Therefore, their time-scale ratios could be consis- tent, considering the per-ﬁt uncertainties. The other two outlier ﬂares have rather low time-scale ratios ((cid:46) 0.7), and their zoomed-in light curves can be found in Figure 4o and 4e. The X-ray and optical/UV ﬂare amplitudes are roughly correlated (larger X-ray ﬂares generally lead to larger optical/UV ﬂares), but with a large scatter (Fig- ure 7b). The amplitudes themselves arguably change on short time-scales (e.g., Figure 4f), which could explain part of this scatter.
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	Average1 (MOS 1/2): the out-of-eclipse spectrum, selected based on the orbital timing solution in S16. High1 (MOS 1/2): a subset of Average1, with EPIC net count rates > 0.6 cts s−1 per 30-sec bin (Figure 2). Medium1 (MOS 1/2): similar with High1, but with count rates between 0.3 and 0.6 cts s−1. Low1 (MOS 1/2): similar with High1, but with count rates (cid:54) 0.3 cts s−1. Eclipse (pn and MOS 1/2): data taken during the three eclipse phases.
Processed_The_Flare-Dominated_Accretion_Mode_of_a_Radio-Brig.txt	Fig. 7.— The ﬁgures show the comparisons between the ﬂare characteristics observed in X-ray and optical/UV. The upper panel (a) refers to the ﬂare time-scale comparison, while the lower panel (b) is for the ﬂare amplitudes. In general, the ﬂares last signiﬁ- cantly longer in optical/UV than in X-rays (see the dashed line, which indicates the time-scale ratio as one). No clear relation can be found between the optical/UV and X-ray amplitudes. The 16 representative ﬂares that are shown in Figure 4 are indicated by empty circles and labelled accordingly from (a) through (p). The crosses at the upper right corners show the typical estimates of the uncertainties.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	ABSTRACT We present results from two statistical analyses applied to an neutral hydrogen (HI) dataset of the nearby tidal bridge in the Magellanic System. Primarily, analyses of the Spatial Power Spectrum suggest that the Magellanic Bridge, historically considered to be a single contiguous feature, may in fact be a projection of two kinematically and morphologically distinct structures. The southern and more obviously turbulent parts appear to show structure organized similarly to the adjacent Small Magellanic Cloud (SMC), while the northern regions are shown to be relatively deﬁcient in large scale power. The extent of modiﬁcation to the spatial power index by the velocity ﬂuctuations is also highly variant across these parts of the Bridge. We ﬁnd again that the northern part appears distinct from the southern part and from the SMC, in that the power spectrum is signiﬁcantly more aﬀected by slower velocity perturbations.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	The Magellanic Bridge (MB) is an elongated, pre- dominantly neutral hydrogen (Hi) structure, located be- tween and joining the Large and Small Magellanic Clouds (Kerr, Hindman, & Robinson, 1954). Numerical simula- tions (Ruˇziˇcka, 2003; Gardiner, Sawa, & Fujimoto, 1994; Gardiner & Noguchi, 1996) and the relative geometrical arrangement of the components of the Magellanic Sys- tem (e.g. Mathewson, Schwarz & Murray, 1977) indi- cate that the MB was formed directly from the Magel- lanic Clouds, following a particularly close mutual pass ∼200 Myr ago. Recent high resolution observations have shown that the Hi in the MB has a complex and chaotic ﬁlamentary structure (Muller et al. 2003). Apparently coherent Hi features are identiﬁable across the entire range of observed angular scales, from ∼98′′ up to ∼7◦, although the ﬁlamentary component is clearly unevenly distributed throughout the entire MB region.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	theoretical framework. While there are many statistical techniques that have been used to analyze diﬀerent data sets, this paper focuses on two complimentary methods: the Spatial Power Spectrum (SPS) and the Spectral Cor- relation Function (Rosolowsky et al, 1999).
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	The SPS and the SCF were chosen because they high- light the importance of structure across a wide range of spatial scales. The organisation of spatial scales has a direct inﬂuence on the eﬃciency of energy transport throughout the interstellar medium (ISM) of the subject system and also vice versa: the transport of energy is an important operator on the arrangement and distribution of spatial scales.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	The relationship between the observed SPS and the characteristics of the underlying turbulent ﬂows has been the work of several theoretical studies (Lazarian & Pogosyan 2000; Goldman 2000; Miville-Deschenes et In contrast with the SPS which measures al. 2003). structure of data sets on a more global scale, the SCF measures local variations as a function of spatial scale.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	No studies have yet made a direct and dedicated com- parison of the outcomes and predictions of the SPS and the SCF. Measurements of the SCF on the Hi of the LMC (Padoan et al. 2001) were shown to be in rough, though not particularly good agreement with earlier studies us- ing the SPS (Elmegreen et al. 2001) and a more detailed discussion of the predictions of the SCF in the context of the turbulent and small-scale structure of the LMC Hi were not entered into.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	To date, studies of the SPS have been made of Hi in the SMC and the LMC (Stanimirovic et al. 1999a; Stan- imirovic & Lazarian 2001; Elmegreen et al. 2001). SCF studies have been made only of the LMC (Padoan et al. 2001). These results have highlighted the existence of important active processes and structure trends within these systems and have been used as a quantiﬁer for the organisation of structure within these systems indepen- dently. The MB forms a physical link between the SMC and the LMC, and represents a feature which has evolved purely in response to tidal forces. The MB is therefore an ideal system to test and compare the response of the SCF to a) a system which has developed in response to large-scale forces and b) in the context of the host sys- tems, i.e. the SMC and LMC. Such an analysis may characterize the turbulence in tidal features in general and in addition, a comparison of the SPS results derived for diﬀerent components in the Magellanic System can provide insights into diﬀerent evolutionary paths for or- ganization of Hi structures in an galactic versus an in- tergalactic medium.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	The structure of this paper is organized as follows. In Section 2 we review the SPS and SCF. In Section 3 we summarize the observations of the subject Hi MB dataset. Sections 4 and 5 detail the methodology and results of the calculation of the Hi SPS and SCF on the MB Hi dataset. Sections 6 presents the discussion and interpretation of our results. Finally, we summarize our results in Section 7.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	P (k) ∝ kγ or P (k) ∝ C1kγ1 + C2kγ2. The composite power law is observed when the dynam- ics of a system are shaped by a process (or processes) which generate excess power over a speciﬁc range of scales. For example, supernovae are capable of inject- ing power across scales of tens of parsecs. The constants C1 and C2 are dependent on the relative importance of the two terms (see also Elmegreen & Scalo, 2004; Scalo & Elmegreen, 2004, for a detailed review of turbulent processes).
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	cascade of energy over intermediate scales and dissipa- tion of energy only on smallest scales) the SPS of both 3-d density and velocity ﬂuctuations is a power law with a slope equal to −11/3.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	However, a spectral index of γ ∼ −3.7 does not nec- essarily imply Kolmogorov-like turbulence. Recent work using simulated datasets have shown that other turbu- lent regimes are capable of producing spectral indices which are approximately equal to this value. For ex- ample, Miville-Deschˆenes, Levrier & Falgarone (2003) have developed synthetic fractal Brownian motion (fBm) datasets which can yield spatial spectral indices of −3.7. Furthermore, synthetic datasets containing randomly distributed, supernovae-driven expanding shells also ap- pear to produce a Kolmogorov spectrum (Hodge & Desh- pande, in preparation). Both of these studies are in need of further clariﬁcation however, speciﬁcally in that the fBm set does not represent coherent turbulent structures (i.e. it is constructed from Brownian noise), and that the Hodge & Deshpande analysis involves unphysical forma- tion processes (i.e. they do not solve the hydrodynamic equations in the development of the set). To distinguish and disentangle the nature and extent of velocity ﬂuc- tuations from static hierarchical density structure, tech- niques which measure the velocity statistics of the set must be used.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	An important advantage the SPS has over other sta- tistical techniques is in relating statistics of the observed intensity ﬂuctuations with the statistics of the underly- ing 3-d density and velocity ﬁelds. A recent development in SPS studies involves the averaging of data across a number of observed velocity channels. By integrating a spectral-line data cube over varying ranges of velocity, Lazarian & Pogosyan (2000) showed that it is possible to disentangle the inﬂuences of density and velocity ﬂuc- tuations in the turbulent ﬂow, i.e. the power-law index is a function of both the velocity, and the velocity width of the observations: γ = γ(v, ∆v). When an image is generated from a small range in velocities (small ∆v), a signiﬁcant fraction of the observed structure results from the velocities of the turbulent ﬂow moving emis- sion into diﬀerent velocity channels. Since an SPS is usually inferred to characterize the density ﬂuctuations in a turbulent ﬂow, the velocity ﬂuctuations would cor- rupt the results. However, when integrated over a large velocity range (large ∆v), most of the velocity ﬂuctu- ations average out, leaving only static density ﬂuctua- tions. By measuring the power law spectrum over a range of widths in the velocity window, Lazarian & Pogosyan (2000) showed that it was possible to separate the power- law spectrum of the density ﬂuctuations from those of the velocity ﬂuctuations. In their Velocity Channel Analy- sis (VCA), they refer to the regime where the velocity thickness of the integrated slice averages out the velocity ﬂuctuations as the “thick slice regime” and the opposite regime where the velocity ﬂuctuations are important as the “thin slice regime”.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	By contrast, the SCF does not transform the data and operates on entire spectra in a data cube instead of en- tire images. The SCF parameterizes the ‘similarity’, S0, between two spectra as a function of their spatial sep- aration, ∆r and often shows a power-law relationship S0 ∝ ∆rα over a range of separations (e.g. Padoan, Rosolowsky & Goodman, 2001). The SCF is normalized so that a perfect match between two spectra will yield a value of S0 = 1, and a value of zero is obtained for no similarity, or for an anti-correlation. Since this technique does not use a Fourier transform it does not require the data to be edge tapered and thereby reduces problems arising from the ﬁnite sampling and the associated edge eﬀects. Therefore, the minimum size of the region that may be sampled with this technique is much smaller than that available to the power spectrum analysis. However, the SCF is sensitive to the noise in the data cube and its application is limited to high signal-to-noise data.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	The early forms of the SCF technique were used to de- rive an estimate of spectral similarity over a lag range which was deﬁned by the velocity dispersion of the spec- tra of interest. It was therefore limited to data cubes that were described by a simple, single velocity compo- nent. Later modiﬁcations to the SCF made by Padoan, Rosolowsky & Goodman (2001) included a parameter which weights the function by the ratio of the RMS of the signal and noise. These modiﬁcations improved the ability of the algorithm to handle noise and expanded its application to more complex and multi-component spectra and across any range of spatial lags (Padoan, Rosolowsky & Goodman 2001). An interesting applica- tion of the SCF was shown in the study by Padoan et al. (2001) where a break in the power-law behavior of the SCF was interpreted as a signature of the LMC’s disk thickness.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	As the SCF is a rather recent development, the re- lationship between the SCF results and the proper- ties of the observed turbulence are not understood as completely as for the SPS. We have attempted here a more dedicated study and comparison of the the SCF in the context of predictions from the relatively well- parameterized SPS.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	sponding to angular scale of ∼7◦), assuming a distance of 60 kpc to the MB. An important note in the context of this work, is that the ATCA interferometer, being com- posed of discrete elements is not capable of uniformly sampling the entire UV plane. The missing data is ex- trapolated during the deconvolution and inversion pro- cess which also homogonises the sensitivities somewhat (refer to Muller et al. 2003 and also Stanimirovi´c, 1999 for explicit information on the data deconvolution and inversion process).
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	Fig. 1 shows an RA-Dec. peak intensity image of the Hi spectral line data cube used in this study. The diverse range of morphologies in the Hi of the MB is striking in this image: some regions are bright and ﬁlamentary, particularly in the southern parts, while others are more tenuous and smooth. Fig. 2 shows a peak pixel image of the same dataset viewed in the Dec.-Velocity projec- tion. The distribution of Hi in this ﬁgure appears well- organised into at least two components. Notably, these two components appear to be distinct also in declina- tion. The velocity separation of these components is ∼ 40 km s−1.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	The rapid spatial and velocity variability of the Hi in the MB shown in Figs. 1 & 2, indicates a turbulent and highly inhomogeneous morphology over the entire range of observed spatial and velocity scales. Consequently, we have applied a separate SPS analysis on four subregions within the data set, as shown in Fig. 1 with the overlaid grid. Each region spans ∼ 2.2×2.2 kpc (256×256 pixels). We will refer to these regions as (clockwise starting from the bottom left on Fig. 1): the South East (SE), the North East (NE), the North West (NW) and the South West (SW).
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	The SW region is dominated by bright, ﬁlamentary features across its entire extent. The Hi in the SE region is similarly ﬁlamentary, although the spatial coverage of the brightest feature is less complete. The NE region deﬁnes an area of less diverse structure where the Hi ap- pears relatively tenuous and smooth. The NW region appears to be dominated by part of a large loop-like fea- ture. Fig. 2 shows that the northern two sub-regions (NE and NW) encompass Hi which occurs at lower Heliocen- tric velocities. The SPS analyses was applied on each region separately.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	Due to the unpredictable response of the SCF to noisy data, we have restricted the SCF analysis to a the bright- est region of the Magellanic Bridge, shown in Fig. 3. Much of this area has a spectral quality of Q > 6 (see Eq. 2 in Section 5.1) . Restricting the tested data to those with a quality of this level limits the eﬀects of noise on the SCF to < 20%. In contrast, the northern regions of the data set is more dominated by noise, having areas where Q < 6. Although binning can increase the spec- tral quality, the degree of averaging necessary to eﬀect a substantial increase in Q ultimately obscures the velocity structure, thereby compromising the utility of the SCF. The 10×3 grid overlaid on Fig. 3 shows the eight areas over which the SCF was averaged (see §5. Each small square deﬁnes a 500 ×500 pc area (63 × 63 pixels). As the telescope beam subtends approximately three image pixels, a new data cube was generated that sampled ev- ery third pixel in the x and y directions to ensure that the SCF samples independent spectra.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	function. The median Hi column density for the ﬁnal Hi data cube is quite low (∼4×1020 cm−2), so any correc- tions for self-absorption are unnecessary (see discussion by Minter 2002).
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	We emphasise that this dataset is a combination of observations obtained with the Parkes single-dish tele- scope and with the ATCA interferometer. Therefore, a complete range of spatial frequencies was sampled from zero spacing, up to ∼ 600λ, corresponding to a minimum spatial scale of ∼ 100 pc (note that this is the contigu- ously sampled range of spatial frequencies; although spa- tial frequencies up to ∼ 1600λ have been sampled, these are not part of a complete contiguous spatial frequency distribution).
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	A sample SPS from the MB dataset, typical for velocity channels with signiﬁcant Hi emission, is shown in Fig. 4. Data points calculated from a continuous range of spatial frequencies < 600λ were used for ﬁtting and are denoted by squares. The error bars in the ﬁgure correspond to 1- σ statistical uncertainties, which were taken into account during the ﬁtting procedure.
Processed_A_Statistical_Investigation_of_HI_in_the_Magellani.txt	Fig. 2.— Integrated intensity map of the Declination-Velocity projection of the combined ATCA-Parkes data cube. The northern part of the Hi in the Bridge is clearly oﬀset to higher velocities with respect to the southern part by ∼40-50 km s−1. We see a clear bimodal distribution of Hi in the more southern part in this ﬁgure. This apparent two-fold structure originates in the body of the SMC and extends approximately 4 kpc towards the LMC (Muller et al. 2003). It has clearly been signiﬁcant in the evolution of the Bridge and is the subject of a future work. The grey scale is a linear transfer function as shown in the wedge and units are K.arcsec . As this map is integrated along the x image axis, the labeled Declination axis is correct only for RA=2h08m8s.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	We present 10 µm, ISO-SWS and Australia Telescope Compact Array observations of the region in the cluster Wd1 in Ara centred on the B[e] star Ara C. An ISO- SWS spectrum reveals emission from highly ionised species in the vicinity of the star, suggesting a secondary source of excitation in the region. We ﬁnd strong radio emission at both 3.5cm and 6.3cm, with a total spatial extent of over 20 arcsec. The emission is found to be concentrated in two discrete structures, separated by ∼14”. The westerly source is resolved, with a spectral index indicative of thermal emission. The easterly source is clearly extended and nonthermal (synchrotron) in nature. Positionally, the B[e] star is found to coincide with the more compact radio source, while the southerly lobe of the extended source is coincident with Ara A, an M2 I star. Observation of the region at 10µm reveals strong emission with an almost identical spatial distribution to the radio emission. Ara C is found to have an extreme radio luminosity in comparison to prior radio observations of hot stars such as O and B supergiants and Wolf-Rayet stars, given the estimated distance to the cluster. An origin in a detatched shell of material around the central star is therefore suggested; however given the spatial extent of the emission, such a shell must be relatively young (τ ∼103yrs). The extended non thermal emission associated with the M star Ara A is unexpected; to the best of our knowledge this is a unique phenomenon. SAX (2-10keV) observations show no evidence of X-ray emission, which might be expected if a compact companion were present.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	et al. (1986) to propose a two component wind to explain the spectra, with a slowly expanding cool equatorial com- ponent responsible for the optical spectral features, and a hotter, high velocity polar component producing the high excitation features in the UV spectra. Near IR observations of B[e] stars by McGregor et al. (1988) and subsequently Morris et al. (1996) reveal Na i and CO bandhead emission, conﬁrming the presence of cooler circumstellar regions as indicated by the dust emission.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	the morphological similarities between the spectra they sug- gest a close evolutionary link between the three types of star. If conﬁrmed this would be an important result, as there is still some confusion as to the exact evolutionary status of B[e] stars (see for example Lamers et al. 1998). In order to investigate the evolutionary status of these objects, obser- vations of the galactic B[e] star Ara C were made as it is the only known galactic B[e] star in a cluster, permitting an accurate measurement of its luminosity, and hence age (since very luminous young B[e] stars are not expected to be seen).
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	The diﬀuse object in Ara found at 2.2µm by Price (1968) was identiﬁed by Westerlund (1968) as a heavily reddened cluster. Further work by Borgman et al. (1970), Koornneef (1977) and most recently Westerlund (1987) have done much to clarify the properties of the cluster. The cluster is thought to be young, with an age of only ∼7 million years (Wester- lund 1987). The distance is estimated to be ∼5.2kpc, which gives a cluster size of some 3pc, within which photometry is available for some ∼258 stars (Westerlund 1987). The clus- ter is heavily reddened, with AV ∼ 11, of which AV = 7 is thought to be due to dust in the region of the cluster. West- erlund (1987) estimates a minimum mass of dust of ∼20M⊙, which implies the presence of ∼2000M⊙ of gas compared to 1200M⊙ to be found in 82 of the brightest cluster stars (lower limiting luminosity of log L/L⊙=3.5). The diﬀuse Hα emission seen in images of the cluster is attributed to exci- tation by the stellar radiation ﬁelds.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	The intrinsically brightest stars of the cluster form a continuous sequence from B2 Ia to M2 Ia. Below the su- pergiant sequence Westerlund (1987) ﬁnds that most of the stars are luminosity class II-III, with the faintest stars on the main sequence. Ara C, identiﬁed as star 9 in Westerlund (1987), was of immediate interest due to a remarkable emis- sion line spectrum, consisting of very strong Balmer emis- sion (EWHα∼250˚A), HeI emission, as well as FeII, [NII], [OII] and [ArIII]. Westerlund classiﬁed the star as Be, but the Hα equivalent width is at least a factor 5 higher than the largest values reported for classical Be stars, and compari- son with the spectra of other massive emission line objects suggests that star C may be a B[e] star. Due to uncertainty in the intrinsic colours of B[e] stars, the values of Mv ∼6 and Av ∼8.70 derived for Ara C in Westerlund (1987) are the subject of some uncertainty. Likewise, the estimate of log(L∗/L⊙)=4.86 for Ara C was derived under the assump- tion of Tef f ∼20000K, and should be treated with caution. Many B[e] stars are known in our Galaxy, but their nature is unclear, since their spectral characteristics coincide closely with those of massive YSO’s. The situation would improve if B[e] stars could be identiﬁed in clusters, as is proposed for Ara C. This implies that Ara C is a post-main- sequence object since the age of the cluster is such that massive stars can no longer be YSO’s. This is, to the best of our knowledge, the ﬁrst case in which a galactic B[e] star can deﬁnitively be classiﬁed as a post-main-sequence object, which demonstrates that Galactic counterparts to the LMC B[e] supergiants, as discussed by Zickgraf et al. (1986), do exist.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	ISOCAM images at ∼3-10µm of the Ara C region were made on 1996 September 10. The ISOCAM data were taken with the 1.5 pixel ﬁeld of view (FOV) setting. The array is 32x32 pixels and so the total FOV is about 47x47 arcsec (the PSF for the wavelengths used can be calculated from Cesarsky et al 1996). The 3µm image shows Ara C to be an unresolved point source (several other bright stars in the cluster are also visible as point sources). Ara C is also visible in the 10µm image as a point source, as is Ara A (nomenclature by Borgman et al 1970; denoted star 26 by Westerlund 1987). low resolution optical spectroscopy and photometry Westerlund (1987) classi- ﬁes this as an M2 I star, with a bolometric luminos- ity of log(L∗/L⊙)=5.26 and an eﬀective temperature of log(Tef f )=3.54. Further extended emission is visible to the north of this source in the 10µm image. A 10µm TIMMI image is shown in Figure 2, and conﬁrms the the identiﬁca- tion of extended emission to the north of Ara A (TIMMI is fully described in K¨auﬂ et al 1992; the broad band N ﬁlter was used in the observations). Due to its absence from the 3µm image this is most likely due to emission from cold dust (compared to the warmer dust thought to be responsible for the unresolved emission centred on Ara C and Ara A).
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	A fast SWS spectrum centred on Ara C was also ob- tained at this time (ISO-SWS is fully described in De Graauw et al. 1996). The spectral range employed was 2.4- 45µm; gaussian ﬁts to the strongest forbidden lines are pre- sented in Figure 3. This was expected to yield information on the presence and composition of circumstellar dust, and forbidden line emission (we note that based on the presence of both emission lines in the spectrum, and hot dust associ- ated with the star, the classiﬁcation of Ara C as a B[e] star is conﬁrmed). There is no evidence in the spectrum for the presence of the Unidentiﬁed Infrared (UIR) emission bands 3.29, 6.2, 7.6/7.7, 8.6, 11.3, and 12.7µm that tend to occur together. These features are referred to by Tokunaga (1996) as ‘carbonaceous emission bands’, and are sometimes at- tributed to polycyclic aromatic hydrocarbons (PAHs), which might be expected to arise in the dusty circumstellar enve- lope. However, a strong narrow emission feature is visible at 20.59µm, possibly due to a simple oxide (although an association with crystaline silicates cannot be ruled out at this time). The spectrum also revealed the presence of [O iv]. These observations are deeply puzzling, as this observa- tion implies temperatures of ∼80,000K, which cannot be due to the stellar radiation ﬁeld (∼20,000K), and clearly imply the presence of another source of ionising radiation in the system. It should be noted that there is likely to be some contamination from the M2 I star Ara A in the spectrum, and so the exact location of the highly ionised gas cannot be conclusively identiﬁed with the circumstellar environment of Ara C from this observation alone.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	Observations of the Ara C region at 3.6 and 6.3cm were obtained with the Australian Telescope Compact Ar- ray (ATCA) on 1997 April 30. These observations were in- terleaved with those of a nearby phase referenced source (1646-50), with a total on source integration of 93 minutes. The ﬂux scale was determined by observations of J1934-638. After determining initial antennae gain solutions by phase referencing, the gain solutions were improved by three iter- ations of self-calibration.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	Strong radio emission at both 3.5 and 6.3 cm was de- tected at the expected position of Ara C, and in an extended structure ∼ 10 arcsec to the east (Figure 1). The eastern emission is concentrated into two lobes which we label Ara A (North) and Ara A (South) respectively. We ﬁnd that the emission associated with Ara C is thermal in nature (positive spectral index) while that of Ara A(N & S) is nonthermal in origin (negative spectral index). Examination of the radio images suggests that Ara C can be closely represented by a 2D gaussian. To estimate the size of Ara C we have ex- amined the visibility data directly. After locating the peak of Ara C at the phase-tracking centre of the image, we re- moved the ﬂux contributed from Ara A(S) and Ara A(N) using the AIPS routine UVSUB. The remaining visibilities were then binned into a 15 × 15 grid to improve the signal- to-noise ratio. These visibilities clearly show that Ara C is resolved by the array. A two-dimensional Gaussian ﬁt to this gridded data then gives an estimate of the source size, the total ﬂux of the source and the position angle of the emis- sion distribution (see Tables 1 and 2). For comparison, we also estimated the source size by ﬁtting two-dimensional- Gaussian functions to the image of Ara C (AIPS routine JMFIT). These give the same results as the visibility ﬁtting within the uncertainties.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	The coordinates of the Gaussian ﬁt to the western com- pact radio condensation are RA 16 47 04.2 Dec -45 50 31.6 (J2000). An estimate for the position of Ara C of RA 16 47 04.2 Dec -45 50 30.7 was derived from a combination of the ISO-CAM image and an STScI DSS image (ie within 1 arcsec of the radio co-ordinates). In Figure 2 we overplot the 10µm TIMMI image on the 6.3cm map, centering both im- age and map on Ara C (ISOCAM ﬂuxes for Ara C and Ara A of ∼50Jy and ∼190Jy respectively). We ﬁnd an excellent coincidence between emission at both wavelengths in both lobes of Ara A. Optical co-ords for Ara A(S) are RA 16 47 05.25 Dec -45 50 39. Radio co-ords for Ara A(S) and Ara A(N) are RA 16 47 05.42 Dec -45 50 36.7 and RA 16 47 05.85 Dec -45 50 29.66 respectively. We further plot the positions of other cluster stars on this image (notation by Borgman et al 1970). We cite the excellent agreement of these positions with the TIMMI image as evidence for the identiﬁcation of Ara C with the westerly radio source. We further note that our co-ordinates are consistent, within the quoted errors, to all prior published positions.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	Figure 1. 3.5 cm (upper) and 6.3 cm (lower) maps of the region surrounding Ara A and Ara C (contours are at -1, 1, 2, 4, 8, 16, 32 ,64, 128 and 256 times the r.m.s. noise level of 1.1mJy and 1.5mJy respectively). Beam sizes are indicated in the lower right corner of each panel (1.8×1.2 arcsecs at 14 degrees and 3.7×2.3 arcsecs at 10 degrees for 3.5cm and 6.3cm respectively).
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	Table 1. Flux data for the source coincident with Ara C and the 2 components of the emission coincident with Ara A.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	Figure 2. Upper Panel: TIMMI 10µm image of the ﬁeld (ﬂuxes derived from ISOCAM observations of ∼50Jy and ∼190Jy for Ara C and Ara A respectively). Lower Panel: overlay of the TIMMI (greyscale) and 6.3cm (contour) plots. The positions of other clus- ter stars (notation by Borgman et al 1970) are also shown.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	Figure 3. Line proﬁles for the strongest forbidden lines in the ISO-SWS spectrum. Data points from the up and down scans are shown separately. The nominal (solid line) and best ﬁt (broken line) wavelengths are also indicated.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	model consisting of a (partially) optically thick stellar wind, and an extended optically thin component produced by rera- diation. This could be the result of emission from ambient intracluster material (although the previous main sequence wind regime might have cleared a cavity around the star), or from a detached shell of material surrounding the star. A similar structure has been observed before in radio and IRAS observations of G79.29+0.46, with an optically thin detached shell of material centred on a stellar source with a partially optically thick radio spectrum (Higgs et al. 1994, Waters et al. 1996). Waters et al. (1996) conclude that the present day wind characteristics of this star suggest that it could be a member of the class of Luminous Blue Vari- ables (LBVs), while the detatched shell is either the result of a Red Supergiant (RSG) phase, or an extended period of LBV-like mass loss.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	underestimate of Av (possibly due to local inhomogeneities in the intra cluster medium) or uncertainties in the spectral type of the star (and hence bolometric correction) could re- sult in an increase in the bolometric luminosity of Ara C to log(L∗/L⊙)=5.0. Such a luminosity would then place it in the region of the HR diagram occupied by LBVs (and would also allow it to have evolved through a RSG phase). Unfortunately due to lack of high resolution spectral or ra- dio observations it is not possible to extract information on the present wind characteristics of Ara C, and so it is not possible to determine a present day mass loss rate or ter- minal wind velocity, both of which would help to determine the evolutionary status of the star. However, it is possible to estimate the age of the detached shell from the its angular extent (given in Table 2). If the distance to the cluster is taken to be 5.2kpc, and a value of 30kms−1 is adopted for the terminal wind velocity (appropriate for a RSG; Waters et al 1996), then the age of the shell is found to be ∼600yrs. Since Maeder & Meynet (1988) estimate the time taken for massive stars to evolve from the red to blue region of the HR diagram to be 1-3 104yr, this would tend to argue against the shell being formed during a RSG phase (although it should be noted that the higher densities of the ISM expected in this cluster could aﬀect the shell expansion velocity).
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	An explanation for the nonthermal emission position- ally coincident with Ara A is more problematic. Previous observations of M-type giants and supergiants (for exam- ple Reid & Menton 1996, Skinner et al. 1996) have revealed them to be radio sources. The emission has been attributed to free-free emission in a radio photosphere extending a few AU. However, the nonthermal nature of the emission observed clearly eliminates this possibility. Radio emission from OH masers has also been observed in some systems, with individual masers lying at up to ∼500AU from the star. However, at an estimated distance of ∼5.2kpc, the extended emission along the NE/SW axis is ∼105AU long, ∼103 times larger than the distance at which OH masers are seen. The low resolution optical spectrum of Westerlund (1987) shows low level H α emission, which possibly originates from ex- cited gas associated with the dust seen in the TIMMI image. No other lines are seen in emission in this spectrum, ruling out the identiﬁcation of the M2 I star as a component in a symbiotic system, since symbiotic systems typically show a wealth of emission lines in their spectra. Given the linear structure of the emission, it is tempting to attribute it to jet emission from accretion onto a compact object. Such a sce- nario appears to be excluded by SAX 2-10keV observations, which provide an upper limit of 10 milliCrab to any emission (ie a lower luminosity than expected for a neutron star ac- creting at the Eddington limit; J. Heise priv. comm.), but we note that possible X-ray emission might be softer than that detectable by SAX, or such a source might be a transient source of X-rays. Likewise, given the current observations we are unable to excluded the possibility of a chance alignment between Ara A and a background radio source.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	source coincident with the optical coordinates of Ara C, and a second extended source, the southernmost lobe of which is coincident with Ara A, an M2 I star. An analysis of the spec- tral indexes of these two detections reveals that the emission associated with Ara C is most likely thermal in origin, while that associated with Ara A is non thermal. Comparison to 10 µm images of the region shows a remarkable coincidence between radio and IR continuum emission (thought to arise from hot dust).
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	The physical processes giving rise to the emission are not clear. Despite the thermal nature of the emission associ- ated with Ara C, it would appear unlikely that the emission is the result of a steady stellar wind, with both luminosity and its resolved nature arguing against such an interpreta- tion. A more plausible scenario would appear to be emission from a combination of a compact, partially optically thick stellar wind, and an extended optically thin source, possi- bly a detected shell of material formed in a possible RSG or LBV phase. The relationship between the eastern emis- sion component and Ara A and C is as yet unclear. Despite the positional coincidence between the southernmost lobe of both radio and 10µm emission and Ara A it is uncer- tain whether the emission is associated with this star or the result of a chance alignment. We note that if emission is as- sociated with Ara A then it would be the ﬁrst observation of non thermal radio emission from an M supergiant.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	Reid M. J., Menton K. M., in Radio Emission from the Stars and the Sun, ASP Conf. Series, Vol 93, 1996, A. R. Taylor and J. M. Paredes (eds), 1996, 87.
Processed_Discovery_of_extended_radio_emission_in_the_young_.txt	Tokungaa A. T., 1996, Paper presented at ‘Diﬀuse Infrared Ra- diation and the IRTS’, Tokyo, November 1996 (preprint IfA- 97/13: ‘A Summary of the ‘UIR’ Bands’.
Processed_On_the_contribution_of_unresolved_blazars_to_the_e.txt	Abstract. We present a model for estimating the blazar contribution to the extragalactic γ-ray background (EGRB) in the EGRET energy range. It is based on inverse-Compton scattering as the dominant γ-ray production process in the jets of ﬂat spectrum radio quasars (FSRQs) and BL Lac objects, and on the uniﬁcation scheme of radio-loud AGN. Depending on the evolutionary behaviour of radio-loud AGN, our esti- mate for the point source contribution to the extragalactic background emission ranges from about 40% for a redshift drop oﬀ at 20% ∼ 85% for a cutoﬀ at Zmax = 5. Roughly 70%- Zmax = 3 to around 90% of the AGN contribution to the EGRB would be due to BL Lacs/FR I objects. However, in case of BL Lacs our model parameters are not en- tirely in agreement with the results from studies at radio wavelengths indicating systematical uncertainties whereas our estimate for the FSRQ contribution seems to be consistent. Thus we end up with two possibilities depending on whether we underpredict or overpredict the BL Lac contri- bution: either unresolved AGN can not account for the entire EGRB or unresolved BL Lacs produce the observed background.
Processed_On_the_contribution_of_unresolved_blazars_to_the_e.txt	One of the most important results of the Energetic Gamma Ray Experiment Telescope (EGRET) in the ﬁeld of extragalactic astronomy is the measurement of an isotropic diﬀuse γ-ray background over three decades of photon energy (Sreekumar et al. 1998). The diﬀerential spectrum can be described by a power 0.03. The intensity of the EGRB law with photon spectral index α = 2.10 10−5 photons cm−2 s−1 sr−1. integrated above 100 MeV is I = (1.45 ± In this paper we follow a new ansatz for modeling the blazar contribution to the EGRB, which is based on the uniﬁcation scheme of radio-loud AGN and a particular γ-ray emission model, namely external inverse-Compton scattering of accretion disk photons (Dermer & Schlickeiser 1993).
Processed_On_the_contribution_of_unresolved_blazars_to_the_e.txt	According to the uniﬁcation scheme of radio-loud AGN blazars are the beamed sub-class of Fanaroﬀ-Riley (FR) radio galaxies (i.e. Urry & Padovani 1995) with the most probable parent population of FSRQs being the Fanaroﬀ-Riley II (=FR II) radio galaxies and BL Lacs being the beamed population of FR I radio galaxies (i.e. Padovani & Urry 1992; Urry, Padovani & Stickel 1991). For a homogeneous distribution of viewing angles and an isotropic distribution of these objects throughout the universe, the γ-ray emission from radio-loud AGN observed under a large viewing angle and/or at a large distance is expected to fall below the EGRET-sensitivity limit. The integration of the γ-ray intensity emitted by those objects yields the point-source contribution to the EGRB. We base the calculation of the non-thermal γ-ray emission of blazars on the leptonic model of Dermer & Schlickeiser (1993) and extend it by considering In particular not only instantaneous but also continuous electron injection. we consider a power law with index s=2 for the injected electron spectrum and a ﬁxed luminosity for the Shakura-Sunyaev-accretion disk luminosity for all sources. The calculation of the template source spectrum is described in detail in M¨ucke & Pohl (1998). Using a power law distribution for the energy injected into the jets of AGN, ρF R(Ein) = ρ0 Ein,1), we calculate the time and energy integrated ﬂux (from 100 MeV to 10 GeV) from BL Lac/FR I and FSRQ/FR II objects. The low energy cut-oﬀ Ein,1 and the normalization ρ0 of the distribution ρF R(Ein) are the free parameters of the model. We use a Monte-Carlo method to model the Log N-Log S and the redshift distribution of resolved BL Lacs and FSRQs which are our model constraints. Objects which are brighter than the sensitivity limit determine the resolved Log N-Log S, while sources with ﬂux values below that limit contribute to the EGRB. The unresolved blazar contribution to the EGRB then follows by sum- ming up all unresolved AGN per solid angle of the whole survey. Our calculations consider a redshift drop oﬀ at Zmax = 3 and Zmax = 5. Note that the γ-ray emission in the EGRET energy range is moderated by pair production on the IR-background if it originates at redshifts Z > 3 (Stecker et al. 1992). The observed power law spectrum of the EGRB up to 100 GeV therefore excludes signiﬁcant contributions from sources at redshifts of Z > 3.
Processed_On_the_contribution_of_unresolved_blazars_to_the_e.txt	model we can only explain 40% of the observed extragalactic γ-ray background as caused by unresolved radio-loud AGN. If the redshift cut-oﬀ is assumed to occur at Zmax = 5 this contribution increases to 85% of which 90% would be BL Lacs/FR I. The typical unresolved source in our model is either weakly beamed or has a weak intrinsic luminosity of the blazar jet. In Figure 1 we show the predicted diﬀerential power distribution S2dN/dS of FSRQs/FR II and BL Lacs/FR I for the case of instantaneous electron injection and a source density cut-oﬀ at Zmax = 3. It indicates that the main contribution 10−8 ph cm−2 s−1. to the EGRB stems from sources with a ﬂux around S Our model predicts a signiﬁcant ﬂattening of the Log N-Log S distribution in the next two decades of source ﬂux below the EGRET threshold and number 0.1 deg−2. counts of roughly of BL Lacs/FR I and FSRQs/FR II from our Comparing the total number model ﬁt with the observations in the radio band (see Padovani & Urry (1992), Urry, Padovani & Stickel (1991)) reveals a signiﬁcant underestimate of the to- tal number of BL Lac/FR I-objects while for the FSRQ/FR II population our model is in agreement with the radio observations. We note that the γ-ray emission model used in our study may not be ideally suited for BL Lacs anyway, since it is based on the assumption that the accretion disk provides the target photons for the dominant inverse-Compton process (EIC). However, BL Lacs are thought to have rather weak accretion disks which may imply that the SSC mechanism is the dominant γ-ray radiation process. We therefore conclude that, depending on whether we underpredict or overpredict the BL Lac contribution, either unresolved AGN can not account for the entire EGRB or unresolved BL 103 Lacs/FR I produce the observed background. The detection of roughly 102 10 100MeV - 10GeV with a telescope more blazars in the energy range of times more sensitive compared to EGRET is expected.
Processed_On_the_contribution_of_unresolved_blazars_to_the_e.txt	Fig.1: Predicted Log (S2 (dN/dS))-Log S-distribution of our background model for BL Lacs/FR I and FS- RQs/FR II for the case of a ﬁnite Ein-distribution and instantaneous electron injec- tion assuming a redshift cut oﬀ at Z = 3. The statisti- cal uncertainty in each bin lies at about 10%. The main contribution to the EGRB comes from sources at S 10−7...−8 phot cm−2 s−1.
Processed_Several_recent_developments_in_estimation_and_robu.txt	Abstract— This paper summarizes several recent develop- ments in the area of estimation and robust control of quantum systems and outlines several directions for future research. Quantum state tomography via linear regression estimation and adaptive quantum state estimation are introduced and a Hamiltonian identiﬁcation algorithm is outlined. Two quantum robust control approaches including sliding mode control and sampling-based learning control are illustrated.
Processed_Several_recent_developments_in_estimation_and_robu.txt	III presents the identiﬁcation problem of quantum processes and system Hamiltonian. Section IV outlines two quantum robust control methods including sliding mode control and sampling-based learning control. Conclusions are presented in Section V.
Processed_Several_recent_developments_in_estimation_and_robu.txt	Index Terms— Quantum system, quantum control, quantum state tomography, quantum system identiﬁcation, quantum robust control.
Processed_Several_recent_developments_in_estimation_and_robu.txt	Quantum technology has shown powerful potential for developing future technology [1]. Practical applications of quantum technology include secure quantum communication, powerful quantum computation and high-precision quantum metrology. These potential applications have attracted many mathematicians, physicists, computer scientists and control engineers to this booming ﬁeld.
Processed_Several_recent_developments_in_estimation_and_robu.txt	A fundamental task in quantum technology is to character- ize the state of a quantum system and identify the parameters in the system. The estimation procedure of a static quantum state is often referred as quantum state tomography [1]. For estimating a dynamical state, quantum ﬁltering theory [2], [3] has been developed, which is especially useful for addressing the measurement-based quantum feedback control problem [4]. The area of identifying key parameters in quantum systems can be referred as quantum system identiﬁcation [5]. Another important problem is robustness of quantum systems in developing practical quantum technology since real quantum systems are often subject to noises, incomplete knowledge or uncertainties [6]. In this paper, we introduce some recent developments in the areas of estimation and robust control of quantum systems. We do not intend to survey the main progress in these areas. Instead, we present several examples and methods that were recently developed by the authors and their collaborators, and aim to illustrate several classes of signiﬁcant quantum estimation and con- trol problems as well as outline open questions for future research.
Processed_Several_recent_developments_in_estimation_and_robu.txt	Quantum state tomography provides a framework to re- construct quantum states. The state of a quantum system can be described by a density matrix ρ which is a Hermitian, positive semideﬁnite matrix satisfying Tr(ρ) = 1. A pure state can also be described by a unit complex vector ψ i \| [1]. A mixed state is linear combination with ρ = of independent pure states. To estimate a quantum state, we usually need to make measurements on many copies of the state. For quantum measurement, a set of positive operator valued measurement (POVM) elements is prepared 0 and ∑i Pi = I (e.g., mutually unbiased basis [7]), where Pi with I being the identity matrix. The occurrence probability of the ith outcome can be calculated as pi = Tr(ρPi) ac- cording to the Born Rule, where Tr(A) returns the trace of the matrix A. For a given unknown quantum state, we need to design POVM measurement and develop an estimation algorithm to reconstruct the quantum state from measurement data. Various quantum state tomography methods have been developed such as maximum likelihood estimation (MLE) method [8]-[10], Bayesian mean estimation approach [11], [12] and linear regression estimation (LRE) [13]. For more details of MLE and Bayesian mean estimation, please refer to [8], [11]. Here we only introduce the LRE method recently developed for quantum state tomography.
Processed_Several_recent_developments_in_estimation_and_robu.txt	An advantage of LRE for quantum state tomography is that its computational complexity can be characterized and the theoretical error upper bound may be obtained. In [13], its computational complexity O(d4) for estimating a d- dimensional quantum state has been proven and numerical results showed that the LRE algorithm is around 10000 times faster than the MLE approach for quantum state estimation.
Processed_Several_recent_developments_in_estimation_and_robu.txt	The LRE algorithm can be further optimized and imple- mented on GPU, and such improvement has demonstrated the realization of reconstructing a 14-qubit state using only 3.35 hours [16].
Processed_Several_recent_developments_in_estimation_and_robu.txt	Another advantage of LRE is that it is suitable for devel- oping adaptive quantum state estimation method. Adaptive protocols [17]-[21] have been proven to have the capability to improve the quantum estimation precision. In adaptive state estimation as shown in Figure 1, we ﬁrst make measurements on part of copies and get a rough estimate of the quantum state. Then we ﬁnd optimal measurement bases to make measurements on some other copies. It may involve multiple steps of adaptivity according to practical tasks.
Processed_Several_recent_developments_in_estimation_and_robu.txt	Using (6) and (7), one can recursively incorporate new measurement data into historical measurement data, which provides a convenient way for adaptive estimation of ρ. In this sense, the LRE method is more suitable for adaptive reconstruction of quantum states due to its recursive pro- cedure than traditional MLE or Bayesian mean method. In the LRE framework, instead of repeatedly calculating all the historical data when new data arrive, we only need to add the new data into historical information matrix and vector, which signiﬁcantly reduces the calculation cost.
Processed_Several_recent_developments_in_estimation_and_robu.txt	Suppose after s steps, we have Qs and ˆΘs. We deﬁne gs+1 ≡ TrQs+1. According to the experimental ca- pability, if the candidate measurement basis set is ﬁnite, one can calculate gs+1 for each candidate basis and pick up the one with the largest gs+1 as the measurement basis at (s + 1)-th step. When is an inﬁnite set, one { needs to either analytically ﬁnd the optimal basis or try to obtain an approximate optimal basis. In [22], an analytical optimal solution has been provided for estimating a single- qubit state. For two-qubit states, a heuristic deduction to search for the optimal measurement bases was presented in [14], where numerical and experimental results showed that the adaptive quantum state tomography can improve the estimation precision.
Processed_Several_recent_developments_in_estimation_and_robu.txt	Many problems remain open in the area of quantum state tomography. For example, the efﬁciency of the estimation al- gorithms may be further enhanced if there is prior knowledge on the quantum state to be reconstructed. Various variants of LRE could be developed for quantum state tomography and different adaptivity criteria can be explored for adaptive estimation of quantum states. The capability of parallel processing for quantum state estimation and the precision limit of adaptive estimation are worth further exploring. Moreover, new approaches such as coherent observers [23], [24] and machine learning methods may provide different angles for estimating the state of a quantum system.
Processed_Several_recent_developments_in_estimation_and_robu.txt	Here B is determined once the bases are cho- sen, and Λ is obtained from experimental data. In practice, direct inversion or pseudo-inversion of B may fail to generate a physical estimation ˆX due to noise or uncertainty. A central issue in quantum process tomography is to design an algorithm to ﬁnd a physical estimation ˆX such that Bvec( ˆX) is close enough to vec( ˆΛ). MLE [28], [29] aims to ﬁnd the most likely process that can generate the current data. Bayesian mean deduction [18], [30], [31] uses Bayes formula to generate a posterior probability distribution, and the ex- pectation of this probability distribution is taken as the ﬁnal estimation result. We will discuss Hamiltonian identiﬁcation within the framework of quantum process tomography in the next subsection.
Processed_Several_recent_developments_in_estimation_and_robu.txt	where i = √ 1, ¯h is reduced Planck’s constant (we set ¯h = 1 in the following) and H is the system Hamiltonian. It task in quantum systems. Some approaches have been developed for Hamiltonian identiﬁcation. For example, a Hamiltonian identiﬁcation method using measurement time traces has been proposed based on classical system identiﬁcation theory [32] and it has also been used to experimentally identify the Hamiltonian in spin systems [33]. In [34], dynamical decoupling was employed for identifying parameters in the Hamiltonian.
Processed_Several_recent_developments_in_estimation_and_robu.txt	of rank one. Now the semideﬁnite requirement is naturally satisﬁed. Let X = gg† and g = vec(G). We thus know G . must be unitary if we choose natural bases } Furthermore, from Theorem 2 in [26], B is now unitary. 1(B†vec( ˆΛ)). Hence we need to ﬁnd a Denote ˆD = vec− unitary ˆG to minimize \|\| · \|\| is the matrix Frobenius norm. This problem can be solved using a two-step optimization approach. We ﬁrst ﬁnd an ˆS and then ﬁnd a unitary ˆG vec( ˆS)vec( ˆS)† to minimize − vec( ˆG)vec( ˆG)† to minimize . Then Schur − decomposition can be used to obtain the ˆH through the i ˆHt = ˆGT . Further, the computational com- relationship e− plexity O(d6) is given and an upper bound of estimation error is also established in [26]. Numerical results show that the Hamiltonian identiﬁcation algorithm has much lower computational complexity than the approach in [32]. In [35], it is shown that a more efﬁcient algorithm with computational complexity O(d3) can be developed if only input pure states are used.
Processed_Several_recent_developments_in_estimation_and_robu.txt	Quantum system identiﬁcation has been growing quickly in the last several years. A large number of challenging problems is waiting for exploring. For example, although several results on identiﬁability of quantum systems [36] have been presented, the identiﬁability of more general quantum systems was not investigated. Adaptive approaches were only applied to the identiﬁcation problems of several simple quantum systems such as estimating the Hamiltonian parameter of a two-level system [37] and more adaptive algorithms could be developed to enhance the identiﬁcation precision for quantum systems. Since the computational complexity of quantum system identiﬁcation algorithm usu- ally exponentially increases with the number of qubits, it is expected to develop more efﬁcient identiﬁcation algorithms for quantum parameter identiﬁcation. Other new directions for future research include the mechanism identiﬁcation of physical process [38] and the identiﬁcation of quantum networks [39], [40] where the network topology could be taken advantaged of.
Processed_Several_recent_developments_in_estimation_and_robu.txt	In recent years, robust control approaches have been devel- oped to enhance the robustness performance in quantum sys- tems. In particular, several robust control design approaches including H∞ control [41]-[44], small gain theorem and Popov method [45] have been extended to linear quantum systems. More detailed results on control of linear quantum systems can refer to [46], [47]. In this section, we brieﬂy introduce two robust control design methods of sliding mode control (SMC) and sampling-based learning control (SLC) for quantum systems.
Processed_Several_recent_developments_in_estimation_and_robu.txt	The goal of the training step is to ﬁnd a control ﬁeld u∗ that maximizes the performance function in (15). The gradient ﬂow algorithm has been developed for achieving this goal for several classes of quantum robust control problems. Then in the testing step we apply the optimal control u∗ obtained in the training step to additional samples to evaluate the control performance of each sample. If the performance for all the tested samples is satisfactory, we the designed control law. Otherwise, we need to accept improve the algorithm to achieve acceptable performance. The SLC method has been successfully applied to many quantum control tasks including control and classiﬁcation of inhomogeneous quantum ensembles [55], [57], robust control of quantum superconducting systems [58], learning robust pulses for generating universal quantum gates [59] and synchronizing collision of molecules with shaped laser pulses [60]. Other machine learning algorithms [61], [62], [63] can be easily integrated into the SLC method. Recently, the SLC method has been integrated into an improved differential evolution algorithm for control fragmentation of halomethane molecules CH2BrI using femtosecond laser pulses [64].
Processed_Several_recent_developments_in_estimation_and_robu.txt	We introduced some recent progress in the areas of quan- tum state estimation, quantum Hamiltonian identiﬁcation and quantum robust control. A large number of open questions remain in these emerging areas, and systems control theory may make more contributions to address these challenging issues by integrating it with the unique characteristics of quantum systems.
Processed_Several_recent_developments_in_estimation_and_robu.txt	quantum systems,” Automatica, vol. 48, no. 5, pp. 725-735, 2012. [51] D. Dong and I. R. Petersen, “Notes on sliding mode control of two- level quantum systems,” Automatica, vol. 48, no. 12, pp. 3089-3097, 2012.
Processed_Several_recent_developments_in_estimation_and_robu.txt	robust control of a single qubit,” IEEE Transactions on Automatic Control, vol. 58, no. 10, pp. 2654-2659, 2013.
Processed_A_Search_for_Annihilating_Dark_Matter_in_47_Tucana.txt	A Search for Annihilating Dark Matter in 47 Tucanae and Omega Centauri Lister Staveley-Smith,1,2 Emma Bond,1 Kenji Bekki,1 and Tobias Westmeier1,2 1International Centre for Radio Astronomy Research (ICRAR), University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia 2ARC Centre of Excellence for All Sky Astrophysics (ASTRO 3D), Australia Author for correspondence: Lister Staveley-Smith, Email: [email protected].
Processed_A_Search_for_Annihilating_Dark_Matter_in_47_Tucana.txt	Abstract A plausible formation scenario for the Galactic globular clusters 47 Tucanae (47 Tuc) and Omega Centauri (ω Cen) is that they are tidally stripped remnants of dwarf galaxies, in which case they are likely to have retained a fraction of their dark matter cores. In this study, we have used the ultra-wide band receiver on the Parkes telescope (Murriyang) to place upper limits on the annihilation rate of exotic Light Dark Matter particles (χ) via the χχ → e+e– channel using measurements of the recombination rate of positronium (Ps). This is an extension of a technique previously used to search for Ps in the Galactic Centre. However, by stacking of spectral data at multiple line frequencies, we have been able to improve sensitivity. Our measurements have resulted in 3-σ ﬂux density (recombination rate) upper limits of 1.7 mJy (1.4 × 1043 s–1) and 0.8 mJy (1.1 × 1043 s–1) for 47 Tuc and ω Cen, respectively. Within the Parkes beam at the cluster distances, which varies from 10–23 pc depending on the frequency of the recombination line, and for an assumed annihilation cross section (cid:104)σv(cid:105) = 3 × 10–29 cm3 s–1, (mχ/MeV c–2) M(cid:12) and (cid:46) 48 – 54f –0.5 we calculate upper limits to the dark matter mass and rms dark matter density of (cid:46) 1.2 – 1.3 × 105f –0.5 (mχ/MeV c–2) M(cid:12) pc–3 for the clusters, where fn = Rn/Rp is the ratio of Ps recombination transitions to annihilations, estimated to be ∼ 0.01. The radio limits for ω Cen suggest that, for a ﬁducial dark/luminous mass ratio of ∼ 0.05, any contribution from Light Dark Matter is small unless (cid:104)σv(cid:105) < 7.9 × 10–28 (mχ/MeV c–2)2 cm3 s–1. Owing to the compactness and proximity of the clusters, archival 511-keV measurements suggest even tighter limits than permitted by CMB anisotropies, (cid:104)σv(cid:105) < 8.6 × 10–31 (mχ/MeV c–2)2 cm3 s–1. Due to the very low synchrotron radiation background, our recombination rate limits substantially improve on previous radio limits for the Milky Way.
Processed_A_Search_for_Annihilating_Dark_Matter_in_47_Tucana.txt	reviews of the dark matter problem, particle candidates and de- tection schemes, see Bergström (1998), Garrett & D ¯uda (2011) and Bertone & Hooper (2018).
Processed_A_Search_for_Annihilating_Dark_Matter_in_47_Tucana.txt	Astronomical searches for the electromagnetic signature of dark matter through their decay or annihilation products (particularly WIMPs, LDM and axions) have been conducted at many wavelengths from γ-rays to the radio regime, and over size scales from the magnetospheres of individual pulsars (a few 10’s of km) to the largest virialised structures in the Universe – clusters of galaxies.
Processed_A_Search_for_Annihilating_Dark_Matter_in_47_Tucana.txt	the Galactic Centre, annihilation cross section upper limits in the range (cid:104)σv(cid:105) < 10–26 to 10–25 cm3 s–1 in the mass range 10 GeV < mχc2 < 50 TeV are typical. The Cherenkov Tele- scope Array should improve on these limits in certain mass ranges (Silverwood et al., 2015) .
Processed_A_Search_for_Annihilating_Dark_Matter_in_47_Tucana.txt	Radio observations are also sensitive to WIMP GeV an- nihilation products due to the synchrotron radiation emitted by the acceleration of charged particles (e.g. protons, elec- trons and their anti-particles) in pre-existing magnetic ﬁelds (Borriello et al., 2009). Observations of nearby galaxies have resulted in similarly useful annihilation cross section limits of around (cid:104)σv(cid:105) < 10–26 to 10–25 cm3 s–1 in the mass range 1- 1000 GeV c–2 (Regis et al., 2017; Cook et al., 2020; Basu et al., 2021), though with a dependence on other parameters such as magnetic ﬁeld strength, diﬀusion coeﬃcients and equipar- tition (Colafrancesco et al., 2006). Future radio observations, including with the Square Kilometre Array, are expected to substantially improve on these limits (Storm et al., 2017; Chen et al., 2021b).
Processed_A_Search_for_Annihilating_Dark_Matter_in_47_Tucana.txt	Self-annihilating LDM particles in the 1-1000 MeV c–2 mass range can produce lower-energy γ-rays, including from the e+e– annihilation line at 511 keV. Although this line is strongly detected in the Galactic Centre (Johnson & Haymes, 1973; Leventhal et al., 1978; Siegert et al., 2016a), its origin is unclear as other e+ production channels exist, including pair production around pulsars (Jones, 1979) and β+ decay in supernovae (Crocker et al., 2017; Siegert et al., 2022b). Observations of nearby galaxies have not yet revealed any signiﬁcant 511 keV detections, but have nevertheless resulted in excellent cross-section limits, especially for low mass LDM particles. For example, for mχ ∼ 1 MeV c–2, Siegert et al. (2022a) deduce a limit of (cid:104)σv(cid:105) < 5 × 10–28 cm3 s–1 for the dwarf galaxy Reticulum II (Ret II).
Processed_A_Search_for_Annihilating_Dark_Matter_in_47_Tucana.txt	example, the strong detection of the 511 keV annihilation line in the Milky Way may have a nucleosynthetic origin, rather than a dark matter origin (Siegert et al., 2022b). However, a disadvantage is that their distances (typically 20 to 500 kpc) are mostly larger than for the Milky Way or LMC, leading to weaker signal strength predictions.
Processed_A_Search_for_Annihilating_Dark_Matter_in_47_Tucana.txt	Another avenue are the much closer giant globular clus- ters 47 Tucanae (47 Tuc) and Omega Centauri (ω Cen), also known as NGC 5139. These objects have distances of 4.4 kpc and 5.5 kpc, respectively (Chen et al., 2018; Del Principe et al., 2006) and may be the nucleated remnants of ancient tidally-stripped dwarf galaxies (Peebles, 1984; Norris et al., 1996; Hilker & Richtler, 2000; Bekki & Norris, 2006; Lee et al., 2009; McKenzie & Bekki, 2021) and therefore in pos- sible possession of a dark matter core. Various authors have suggested the role that dark matter may play in explaining their velocity dispersion proﬁles (e.g. Peñarrubia et al., 2017). Evans et al. (2022) have used GAIA, HST and spectroscopic data to infer strong evidence for a non-luminous component in ω Cen, with a mass contained within the half-light radius of up to 106 M(cid:12). No 511 keV radiation has been detected from these clusters (Knödlseder et al., 2005), though observa- tions with Fermi-LAT have detected strong γ-ray continuum (Abdo et al., 2010). Due to the poor angular resolution of γ-ray telescopes, the source of the emission is not clear – it is consistent with models of the annihilation of massive dark matter particles (Gaskins, 2016; Brown et al., 2019; Wirth et al., 2020), but could also arise from a population of old millisecond pulsars which is known to exist in these clusters (Abdo et al., 2010; Reynoso-Cordova et al., 2019). There are 25 known mil- lisecond pulsars in 47 Tuc (Freire et al., 2017) and 5 in ω Cen (Dai et al., 2020), plus a number of unidentiﬁed X-ray sources with similar properties to millisecond pulsars (Bhattacharya et al., 2017; Henleywillis et al., 2018).
Processed_A_Search_for_Annihilating_Dark_Matter_in_47_Tucana.txt	Wirth et al. (2020) used high-resolution computer simula- tions to model the evolution of nucleated dwarf galaxies near the Milky Way and found that they can transform into glob- ular clusters similar to ω Cen. Their simulated nuclei were found to have central dark matter densities of 0.1 to several M(cid:12) pc–3. Wirth et al. (2020) were able to match the predicted γ-ray spectrum (assuming dark matter annihilation via the b¯b channel) of one of their simulated clusters with that observed by Fermi-LAT for ω Cen using an annihilation cross section (cid:104)σv(cid:105) = 5 × 10–28 cm3 s–1.
Processed_A_Search_for_Annihilating_Dark_Matter_in_47_Tucana.txt	Whether or not the γ-ray detections of 47 Tuc and ω Cen represent the signature of dark matter annihilation, they at least give an upper limit to the dark matter annihilation rate, under the assumption that their cores really do contain dark matter. This applies to scenarios of WIMP annihilation, as explored by Wirth et al. (2020), as well as annihilation of LDM and other candidate dark matter particles.
Processed_A_Search_for_Annihilating_Dark_Matter_in_47_Tucana.txt	inate between diﬀerent astrophysical origins. Furthermore, spectral-line observations also have the potential to reduce con- fusion with diﬀuse and/or compact sources of emission in the foreground and background.

Subsets and Splits

No saved queries yet

Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.