Search

Explainability in Graph Neural Networks (GNNs) is a new field growing in the last few years. In this publication we address the problem of determining how important is each neighbor for the GNN when classifying a node and how to measure the performance for this specific task. To do this, various known explainability methods are reformulated to get the neighbor importance and four new metrics are presented. Our results show that there is almost no difference between the explanations provided by gradient-based techniques in the GNN domain. In addition, many explainability techniques failed to identify important neighbors when GNNs without self-loops are used.

Understanding the chemical processes during starless core and prestellar core evolution is an important step in understanding the initial stages of star and disk formation. This project is a study of deuterated ammonia, o-NH$_2$D, in the L1251 star-forming region toward Cepheus. Twenty-two dense cores (twenty of which are starless or prestellar, and two of which have a protostar), previously identified by p-NH$_3$ (1,1) observations, were targeted with the 12m Arizona Radio Observatory telescope on Kitt Peak. o-NH$_2$D J$_{\rm{K_a} \rm{K_c}}^{\pm} =$ $1_{11}^{+} \rightarrow 1_{01}^{-}$ was detected in 13 (59\%) of the NH$_3$-detected cores with a median sensitivity of $\sigma_{T_{mb}} = 17$ mK. All cores detected in o-NH$_2$D at this sensitivity have p-NH$_3$ column densities $> 10^{14}$ cm$^{-2}$. The o-NH$_2$D column densities were calculated using the constant excitation temperature (CTEX) approximation while correcting for the filling fraction of the NH$_3$ source size. The median deuterium fraction was found to be 0.11 (including 3$\sigma$ upper limits). However, there are no strong, discernible trends in plots of deuterium fraction with any physical or evolutionary variables. If the cores in L1251 have similar initial chemical conditions, then this result is evidence of the cores physically evolving at different rates., Comment: 18 pages, 14 figures, MNRAS, in press

Dynamical studies of dense structures within molecular clouds often conclude that the most massive clumps contain too little kinetic energy for virial equilibrium, unless they are magnetized to an unexpected degree. This raises questions about how such a state might arise, and how it might persist long enough to represent the population of massive clumps. In an effort to re-examine the origins of this conclusion, we use ammonia line data from the Green Bank Ammonia Survey and Planck-calibrated dust emission data from Herschel to estimate the masses and kinetic and gravitational energies for dense clumps in the Gould Belt clouds. We show that several types of systematic error can enhance the appearance of low kinetic-to-gravitational energy ratios: insufficient removal of foreground and background material; ignoring the kinetic energy associated with velocity differences across a resolved cloud; and over-correcting for stratification when evaluating the gravitational energy. Using an analysis designed to avoid these errors, we find that the most massive Gould Belt clumps harbor virial motions, rather than sub-virial ones. As a byproduct, we present a catalog of masses, energies, and virial energy ratios for 85 Gould Belt clumps., Comment: Submitted to ApJ

We present Herschel SPIRE and PACS maps of the Cepheus Flare clouds L1157, L1172, L1228, L1241, and L1251, observed by the Herschel Gould Belt Survey (HGBS) of nearby star-forming molecular clouds. Through modified blackbody fits to the SPIRE and PACS data, we determine typical cloud column densities of 0.5-1.0 $\times$ 10$^{21}$ cm$^{-2}$ and typical cloud temperatures of 14-15 K. Using the getsources identification algorithm, we extract 832 dense cores from the SPIRE and PACS data at 160-500 $\mu$m. From placement in a mass vs. size diagram, we consider 303 to be candidate prestellar cores, and 178 of these to be "robust" prestellar cores. From an independent extraction of sources at 70 $\mu$m, we consider 25 of the 832 dense cores to be protostellar. The distribution of background column densities coincident with candidate prestellar cores peaks at 2-4 $\times$ 10$^{21}$ cm$^{-2}$. About half of the candidate prestellar cores in Cepheus may have formed due to the widespread fragmentation expected to occur within filaments of "transcritical" line mass. The lognormal robust prestellar core mass function (CMF) drawn from all five Cepheus clouds peaks at 0.56 M$_{\odot}$ and has a width of $\sim$0.5 dex, similar to that of Aquila's CMF. Indeed, the width of Cepheus' aggregate CMF is similar to the stellar system Initial Mass Function (IMF). The similarity of CMF widths in different clouds and the system IMF suggests a common, possibly turbulent origin for seeding the fluctuations that evolve into prestellar cores and stars., Comment: 36 pages, 21 figures, 2 online text tables (not yet included here), accepted for publication in The Astrophysical Journal

The role played by magnetic field during star formation is an important topic in astrophysics. We investigate the correlation between the orientation of star-forming cores (as defined by the core major axes) and ambient magnetic field directions in 1) a 3D MHD simulation, 2) synthetic observations generated from the simulation at different viewing angles, and 3) observations of nearby molecular clouds. We find that the results on relative alignment between cores and background magnetic field in synthetic observations slightly disagree with those measured in fully 3D simulation data, which is partly because cores identified in projected 2D maps tend to coexist within filamentary structures, while 3D cores are generally more rounded. In addition, we examine the progression of magnetic field from pc- to core-scale in the simulation, which is consistent with the anisotropic core formation model that gas preferably flow along the magnetic field toward dense cores. When comparing the observed cores identified from the GBT Ammonia Survey (GAS) and Planck polarization-inferred magnetic field orientations, we find that the relative core-field alignment has a regional dependence among different clouds. More specifically, we find that dense cores in the Taurus molecular cloud tend to align perpendicular to the background magnetic field, while those in Perseus and Ophiuchus tend to have random (Perseus) or slightly parallel (Ophiuchus) orientations with respect to the field. We argue that this feature of relative core-field orientation could be used to probe the relative significance of the magnetic field within the cloud., Comment: 18 pages, 11 figures, accepted for publication in MNRAS

Recent observations of global velocity gradients across and along molecular filaments have been interpreted as signs of gas accreting onto and along these filaments, potentially feeding star-forming cores and proto-clusters. The behavior of velocity gradients in filaments, however, has not been studied in detail, particularly on small scales (< 0.1 pc). In this paper, we present MUFASA, an efficient, robust, and automatic method to fit ammonia lines with multiple velocity components, generalizable to other molecular species. We also present CRISPy, a Python package to identify filament spines in 3D images (e.g., position-position-velocity cubes), along with a complementary technique to sort fitted velocity components into velocity-coherent filaments. In NGC 1333, we find a wealth of velocity gradient structures on a beam-resolved scale of ~0.05 pc. Interestingly, these local velocity gradients are not randomly oriented with respect to filament spines and their perpendicular, i.e., radial, component decreases in magnitude towards the spine for many filaments. Together with remarkably constant velocity gradients on larger scales along many filaments, these results suggest a scenario in which gas falling onto filaments is progressively damped and redirected to flow along these filaments., Comment: Accepted to ApJ

When multiple star-forming gas structures overlap along the line-of-sight and emit optically thin emission at significantly different radial velocities, the emission can become non-Gaussian and often exhibits two distinct peaks. Traditional line-fitting techniques can fail to account adequately for these double-peaked profiles, providing inaccurate cloud kinematics measurements. We present a new method called Convnet Line-fitting Of Velocities in Emission-line Regions (CLOVER) for distinguishing between one-component, two-component, and noise-only emission lines using 1D convolutional neural networks trained with synthetic spectral cubes. CLOVER utilizes spatial information in spectral cubes by predicting on $3\times3$ pixel sub-cubes, using both the central pixel's spectrum and the average spectrum over the $3\times3$ grid as input. On an unseen set of 10,000 synthetic spectral cubes in each predicted class, CLOVER has classification accuracies of $\sim99\%$ for the one-component class and $\sim97\%$ for the two-component class. For the noise-only class, which is analogous to a signal-to-noise cutoff of four for traditional line-fitting methods, CLOVER has classification accuracy of $100\%$. CLOVER also has exceptional performance on real observations, correctly distinguishing between the three classes across a variety of star-forming regions. In addition, CLOVER quickly and accurately extracts kinematics directly from spectra identified as two-component class members. Moreover, we show that CLOVER is easily scalable to emission lines with hyperfine splitting, making it an attractive tool in the new era of large-scale NH$_3$ and N$_2$H$^+$ mapping surveys., Comment: Accepted for publication in ApJ

We present initial results from the K-band focal plane array Examinations of Young STellar Object Natal Environments (KEYSTONE) survey, a large project on the 100-m Green Bank Telescope mapping ammonia emission across eleven giant molecular clouds at distances of $0.9-3.0$ kpc (Cygnus X North, Cygnus X South, M16, M17, MonR1, MonR2, NGC2264, NGC7538, Rosette, W3, and W48). This data release includes the NH$_3$ (1,1) and (2,2) maps for each cloud, which are modeled to produce maps of kinetic temperature, centroid velocity, velocity dispersion, and ammonia column density. Median cloud kinetic temperatures range from $11.4\pm2.2$ K in the coldest cloud (MonR1) to $23.0\pm6.5$ K in the warmest cloud (M17). Using dendrograms on the NH$_3$ (1,1) integrated intensity maps, we identify 856 dense gas clumps across the eleven clouds. Depending on the cloud observed, $40-100\%$ of the clumps are aligned spatially with filaments identified in H$_2$ column density maps derived from SED-fitting of dust continuum emission. A virial analysis reveals that 523 of the 835 clumps ($\sim63\%$) with mass estimates are bound by gravity alone. We find no significant difference between the virial parameter distributions for clumps aligned with the dust-continuum filaments and those unaligned with filaments. In some clouds, however, hubs or ridges of dense gas with unusually high mass and low virial parameters are located within a single filament or at the intersection of multiple filaments. These hubs and ridges tend to host water maser emission, multiple 70$\mu$m-detected protostars, and have masses and radii above an empirical threshold for forming massive stars., Comment: Accepted for publication in ApJ

We perform a virial analysis of starless dense cores in three nearby star-forming regions : L1688 in Ophiuchus, NGC 1333 in Perseus, and B18 in Taurus. Our analysis takes advantage of comprehensive kinematic information for the dense gas in all of these regions made publicly available through the Green Bank Ammonia Survey Data Release 1, which used to estimate internal support against collapse. We combine this information with ancillary data used to estimate other important properties of the cores, including continuum data from the James Clerk Maxwell Telescope Gould Belt Survey for core identification, core masses, and core sizes. Additionally, we used \textit{Planck} and \textit{Herschel}-based column density maps for external cloud weight pressure, and Five College Radio Astronomy Observatory $^{13}$CO observations for external turbulent pressure. Our self-consistent analysis suggests that many dense cores in all three star-forming regions are not bound by gravity alone, but rather require additional pressure confinement to remain bound. Unlike a recent, similar study in Orion~A, we find that turbulent pressure represents a significant portion of the external pressure budget. Our broad conclusion emphasizing the importance of pressure confinement in dense core evolution, however, agrees with earlier work., Comment: 35 pages, 8 tables, and 14 figures consisting of 16 .pdf files. Accepted for publication in the Astrophysical Journal

Continuum observations of molecular clouds have revealed a surprising amount of substructure in the form of filaments of a few pc length and cores of ~0.1 pc diameter. Understanding the evolution of these substructures towards star formation requires the kinematic and dynamical insights provided uniquely by sensitive line observations at high angular and spectral resolution. In this short paper, we describe how an ngVLA can probe effectively the dynamics of filaments and cores in nearby star-forming molecular clouds using the NH3 rotation-inversion transitions at 24 GHz. Such emission has been proven to trace well the high column density environments of star-forming cores and filaments but higher-resolution observations are needed to reveal important details of how dense gas is flowing within and onto these substructures. In particular, we describe how 150 x 18-m antennas with a maximum baseline of 1 km can be used to map sensitively NH3 emission across high column density locations in clouds in roughly an order of magnitude less time than with the current Jansky VLA., Comment: 9 pages, 3 figures, to be published in the ASP Monograph Series, "Science with a Next-Generation VLA," ed. E. J. Murphy (ASP, San Francisco, CA)

We present the observation and analysis of newly discovered coherent structures in the L1688 region of Ophiuchus and the B18 region of Taurus. Using data from the Green Bank Ammonia Survey (GAS), we identify regions of high density and near-constant, almost-thermal, velocity dispersion. Eighteen coherent structures are revealed, twelve in L1688 and six in B18, each of which shows a sharp "transition to coherence" in velocity dispersion around its periphery. The identification of these structures provides a chance to study the coherent structures in molecular clouds statistically. The identified coherent structures have a typical radius of 0.04 pc and a typical mass of 0.4 Msun, generally smaller than previously known coherent cores identified by Goodman et al. (1998), Caselli et al. (2002), and Pineda et al. (2010). We call these structures "droplets." We find that unlike previously known coherent cores, these structures are not virially bound by self-gravity and are instead predominantly confined by ambient pressure. The droplets have density profiles shallower than a critical Bonnor-Ebert sphere, and they have a velocity (VLSR) distribution consistent with the dense gas motions traced by NH3 emission. These results point to a potential formation mechanism through pressure compression and turbulent processes in the dense gas. We present a comparison with a magnetohydrodynamic simulation of a star-forming region, and we speculate on the relationship of droplets with larger, gravitationally bound coherent cores, as well as on the role that droplets and other coherent structures play in the star formation process., Comment: Accepted by ApJ in April, 2019

In the absence of the two emission lines H$\alpha$ and [NII] (6584\AA) in a BPT diagram, we show that other spectral information is sufficiently informative to distinguish AGN galaxies from star-forming galaxies. We use pattern recognition methods and a sample of galaxy spectra from the Sloan Digital Sky Survey (SDSS) to show that, in this survey, the flux and equivalent width of [OIII] (5007\AA) and H$\beta$, along with the 4000\AA-break, can be used to classify galaxies in a BPT diagram. This method provides a higher accuracy of predictions than those which use stellar mass and [OIII]/H$\beta$. First, we use BPT diagrams and various physical parameters to re-classify the galaxies. Next, using confusion matrices, we determine the `correctly' predicted classes as well as confused cases. In this way, we investigate the effect of each parameter in the confusion matrices and rank the physical parameters used in the discrimination of the different classes. We show that in this survey, for example, $\rm{g - r}$ colour can provide the same accuracy as galaxy stellar mass to predict whether or not a galaxy hosts an AGN. Finally, with the same information, we also rank the parameters involved in the discrimination of Seyfert and LINER galaxies., Comment: Accepted for publication in MNRAS. 12 pages, 14 figures

Understanding the early stages of star formation is a research field of ongoing development, both theoretically and observationally. In this context, molecular data have been continuously providing observational constraints on the gas dynamics at different excitation conditions and depths in the sources. We have investigated the Barnard 59 core, the only active site of star formation in the Pipe Nebula, to achieve a comprehensive view of the kinematic properties of the source. These information were derived by simultaneously fitting ammonia inversion transition lines (1,1) and (2,2). Our analysis unveils the imprint of protostellar feedback, such as increasing line widths, temperature and turbulent motions in our molecular data. Combined with complementary observations of dust thermal emission, we estimate that the core is gravitationally bound following a virial analysis. If the core is not contracting, another source of internal pressure, most likely the magnetic field, is supporting it against gravitational collapse and limits its star formation efficiency., Comment: 18 pages, 18 figures

We use Green Bank Ammonia Survey observations of NH$_3$ (1,1) and (2,2) emission with 32'' FWHM resolution from a ~ 10 pc$^{2}$ portion of the Cepheus-L1251 molecular cloud to identify hierarchical dense gas structures. Our dendrogram analysis of the NH$_3$ data results in 22 top-level structures, which reside within 13 lower-level, parent structures. The structures are compact (0.01 pc $\lesssim R_{eff} \lesssim$ 0.1 pc) and are spatially correlated with the highest H$_2$ column density portions of the cloud. We also compare the ammonia data to a catalog of dense cores identified by higher-resolution (18.2'' FWHM) Herschel Space Observatory observations of dust continuum emission from Cepheus-L1251. Maps of kinetic gas temperature, velocity dispersion, and NH$_3$ column density, derived from detailed modeling of the NH$_3$ data, are used to investigate the stability and chemistry of the ammonia-identified and Herschel-identified structures. We show that the dust and dense gas in the structures have similar temperatures, with median $T_{dust}$ and $T_K$ measurements of 11.7 $\pm$ 1.1 K and 10.3 $\pm$ 2.0 K, respectively. Based on a virial analysis, we find that the ammonia-identified structures are gravitationally dominated, yet may be in or near a state of virial equilibrium. Meanwhile, the majority of the Herschel-identified dense cores appear to be not bound by their own gravity and instead confined by external pressure. CCS $(2_0-1_0)$ and HC$_5$N $(9-8)$ emission from the region reveal broader line widths and centroid velocity offsets when compared to the NH$_3$ (1,1) emission in some cases, likely due to these carbon-based molecules tracing the turbulent outer layers of the dense cores., Comment: Accepted for publication in ApJ

We use gas temperature and velocity dispersion data from the Green Bank Ammonia Survey and core masses and sizes from the James Clerk Maxwell Telescope Gould Belt Survey to estimate the virial states of dense cores within the Orion A molecular cloud. Surprisingly, we find that almost none of the dense cores are sufficiently massive to be bound when considering only the balance between self-gravity and the thermal and non-thermal motions present in the dense gas. Including the additional pressure binding imposed by the weight of the ambient molecular cloud material and additional smaller pressure terms, however, suggests that most of the dense cores are pressure confined., Comment: Accepted for publication in ApJ Aug 10, 2017

We present an overview of the first data release (DR1) and first-look science from the Green Bank Ammonia Survey (GAS). GAS is a Large Program at the Green Bank Telescope to map all Gould Belt star-forming regions with $A_V \gtrsim 7$ mag visible from the northern hemisphere in emission from NH$_3$ and other key molecular tracers. This first release includes the data for four regions in Gould Belt clouds: B18 in Taurus, NGC 1333 in Perseus, L1688 in Ophiuchus, and Orion A North in Orion. We compare the NH$_3$ emission to dust continuum emission from Herschel, and find that the two tracers correspond closely. NH$_3$ is present in over 60\% of lines-of-sight with $A_V \gtrsim 7$ mag in three of the four DR1 regions, in agreement with expectations from previous observations. The sole exception is B18, where NH$_3$ is detected toward ~ 40\% of lines-of-sight with $A_V \gtrsim 7$ mag. Moreover, we find that the NH$_3$ emission is generally extended beyond the typical 0.1 pc length scales of dense cores. We produce maps of the gas kinematics, temperature, and NH$_3$ column densities through forward modeling of the hyperfine structure of the NH$_3$ (1,1) and (2,2) lines. We show that the NH$_3$ velocity dispersion, ${\sigma}_v$, and gas kinetic temperature, $T_K$, vary systematically between the regions included in this release, with an increase in both the mean value and spread of ${\sigma}_v$ and $T_K$ with increasing star formation activity. The data presented in this paper are publicly available., Comment: 33 pages, 27 figures, accepted to ApJS. Datasets are publicly available: https://dataverse.harvard.edu/dataverse/GAS_DR1

We present observations of the Cepheus Flare obtained as part of the James Clerk Maxwell Telescope (JCMT) Gould Belt Legacy Survey (GBLS) with the SCUBA-2 instrument. We produce a catalogue of sources found by SCUBA-2, and separate these into starless cores and protostars. We determine masses and densities for each of our sources, using source temperatures determined by the Herschel Gould Belt Survey. We compare the properties of starless cores in four different molecular clouds: L1147/58, L1172/74, L1251 and L1228. We find that the core mass functions for each region typically show shallower-than-Salpeter behaviour. We find that L1147/58 and L1228 have a high ratio of starless cores to Class II protostars, while L1251 and L1174 have a low ratio, consistent with the latter regions being more active sites of current star formation, while the former are forming stars less actively. We determine that, if modelled as thermally-supported Bonnor-Ebert spheres, most of our cores have stable configurations accessible to them. We estimate the external pressures on our cores using archival $^{13}$CO velocity dispersion measurements and find that our cores are typically pressure-confined, rather than gravitationally bound. We perform a virial analysis on our cores, and find that they typically cannot be supported against collapse by internal thermal energy alone, due primarily to the measured external pressures. This suggests that the dominant mode of internal support in starless cores in the Cepheus Flare is either non-thermal motions or internal magnetic fields., Comment: Accepted for publication by MNRAS, 29 pages, 16 figures, 8 tables