Search

Citizen science is gaining popularity as a valuable tool for labelling large collections of astronomical images by the general public. This is often achieved at the cost of poorer quality classifications made by amateur participants, which are usually verified by employing smaller data sets labelled by professional astronomers. Despite its success, citizen science alone will not be able to handle the classification of current and upcoming surveys. To alleviate this issue, citizen science projects have been coupled with machine learning techniques in pursuit of a more robust automated classification. However, existing approaches have neglected the fact that, apart from the data labelled by amateurs, (limited) expert knowledge of the problem is also available along with vast amounts of unlabelled data that have not yet been exploited within a unified learning framework. This paper presents an innovative learning methodology for citizen science capable of taking advantage of expert- and amateur-labelled data, featuring a transfer of labels between experts and amateurs. The proposed approach first learns from unlabelled data with a convolutional autoencoder and then exploits amateur and expert labels via the pre-training and fine-tuning of a convolutional neural network, respectively. We focus on the classification of galaxy images from the Galaxy Zoo project, from which we test binary, multi-class, and imbalanced classification scenarios. The results demonstrate that our solution is able to improve classification performance compared to a set of baseline approaches, deploying a promising methodology for learning from different confidence levels in data labelling.

WEAVE, the new wide-field, massively multiplexed spectroscopic survey facility for the William Herschel Telescope, will see first light in late 2022. WEAVE comprises a new 2-degree field-of-view prime-focus corrector system, a nearly 1000-multiplex fibre positioner, 20 individually deployable 'mini' integral field units (IFUs), and a single large IFU. These fibre systems feed a dual-beam spectrograph covering the wavelength range 366$-$959\,nm at $R\sim5000$, or two shorter ranges at $R\sim20\,000$. After summarising the design and implementation of WEAVE and its data systems, we present the organisation, science drivers and design of a five- to seven-year programme of eight individual surveys to: (i) study our Galaxy's origins by completing Gaia's phase-space information, providing metallicities to its limiting magnitude for $\sim$3 million stars and detailed abundances for $\sim1.5$ million brighter field and open-cluster stars; (ii) survey $\sim0.4$ million Galactic-plane OBA stars, young stellar objects and nearby gas to understand the evolution of young stars and their environments; (iii) perform an extensive spectral survey of white dwarfs; (iv) survey $\sim400$ neutral-hydrogen-selected galaxies with the IFUs; (v) study properties and kinematics of stellar populations and ionised gas in $z<0.5$ cluster galaxies; (vi) survey stellar populations and kinematics in $\sim25\,000$ field galaxies at $0.3\lesssim z \lesssim 0.7$; (vii) study the cosmic evolution of accretion and star formation using $>1$ million spectra of LOFAR-selected radio sources; (viii) trace structures using intergalactic/circumgalactic gas at $z>2$. Finally, we describe the WEAVE Operational Rehearsals using the WEAVE Simulator., Comment: 41 pages, 27 figures, accepted for publication by MNRAS; updated version including information on individual grants in a revised Acknowledgements section, corrections to the affiliation list, and an updated references list

The accuracy in determining the spatial-kinematical parameters of open clusters makes them ideal tracers of the Galactic structure. Young open clusters (YOCs) are the main representative of the clustered star formation mode, which identifies how most of the stars in the Galaxy form. We apply the Kriging technique to a sample of Gaia YOCs within a 3.5 kpc radius around the Sun and log(age) $\leq$ 7.5, age in years, to obtain the $Z(X,Y)$ and $V_Z(X,Y)$ maps. The previous work by Alfaro et al. (1991) showed that Kriging can provide reliable results even with small data samples ($N \sim 100$). We approach the 3D spatial and vertical velocity field structure of the Galactic disk defined by YOCs and analyze the hierarchy of the stellar cluster formation, which shows a rich hierarchical structure, displaying complexes embedded within each other. We discuss the fundamental characteristics of the methodology used to perform the mapping and point out the main results obtained in phenomenological terms. Both the 3D spatial distribution and the vertical velocity field reveal a complex disk structure with a high degree of substructures. Their analysis provides clues about the main physical mechanisms that shape the phase space of the clustered star formation in this Galactic area. Warp, corrugations, and high local deviations in $Z$ and $V_Z$, appear intimately connected in a single but intricate scenario.

(Abridged) We have used Gaia EDR3 data, together with spectra of a large sample of luminous stars in the field surrounding Westerlund 1, to explore the extent of the cluster. We carry out a non-parametric analysis of proper motions and membership determination. We investigate the reddening and proper motions of several dozen OB stars and red supergiants < 1 deg away from Wd 1. We identify a population of kinematic members of Wd 1 that largely includes the known spectroscopic members. From their EDR3 parallaxes, we derive a distance to the cluster of $4.23^{+0.23}_{-0.21}$kpc. Extinction in this direction increases by a large amount around 2.8 kpc, due to dark clouds associated to the Scutum-Crux arm. As a consequence, we hardly see any stars at distances comparable to that of the cluster. The proper motions of Wd 1, however, are very similar to those of stars in the field surrounding it, but distinct. We find a second, astrometrically well-defined population in the foreground ($d\approx2\:$kpc), likely connected to the possible open cluster BH197. Wd 1 is very elongated, an effect not driven by the very heavy extinction to the East and South. We find a low-density halo extending up to 10' from the cluster centre, mainly in the NW quadrant. A few OB stars at larger distances from the cluster, most notably the LBV MN48, share its proper motions, suggesting that Wd 1 has little or no peculiar motion with respect to the field population of the Norma arm. However, we are unable to find any red supergiant that could belong to an extended population related to the cluster, although we observe several dozen such objects in the foreground. We find a substantial population of luminous OB members obscured by several more magnitudes of extinction than most known members. These objects, mostly located in the central region of the cluster, increase the population of OB supergiants by about 25%., Comment: Accepted for publication in Astronomy & Astrophysics. Figures have been converted to PNG and are of lower quality than in the version accepted. Three tables in the appendices will be published electronically; only table C1 is given here. 25 pages (including appendices),23 figures. V2 incorporates style and language edits and some comments

Context. Computational techniques are essential for mining large databases produced in modern surveys with value-added products. Aims. This paper presents a machine learning procedure to carry out simultaneously galaxy morphological classification and photometric redshift estimates. Currently, only spectral energy distribution (SED) fitting has been used to obtain these results all at once. Methods. We used the ancillary data gathered in the OTELO catalog and designed a non-sequential neural network that accepts optical and near-infrared photometry as input. The network transfers the results of the morphological classification task to the redshift fitting process to ensure consistency between both procedures. Results. The results successfully recover the morphological classification and the redshifts of the test sample, reducing catastrophic redshift outliers produced by SED fitting and avoiding possible discrepancies between independent classification and redshift estimates. Our technique may be adapted to include galaxy images to improve the classification., Comment: Astronomy and Astrophysics (A&A) accepted

We present the analysis of a sample of \ha\,, \hb\ and \oii\ emission line galaxies from the \otelo\ survey, with masses typically below $log(M_*/M_\sun) \sim 9.4$ and redshifts between $z \sim 0.4$ and 1.43. We study the star formation rate, star formation rate density, and number density and their evolution with redshift. We obtain a robust estimate of the specific star formation rate -- stellar mass relation based on the lowest mass sample published so far. We also determine a flat trend of the star formation rate density and number density with redshift. Our results suggest a scenario of no evolution of the number density of galaxies, regardless of their masses, up to redshift $z\sim1.4$. This implies a gradual change of the relative importance of the star forming processes, from high-mass galaxies to low-mass galaxies, with decreasing redshift. We also find little or no variation of the star formation rate density in the redshift range of $0.4

We report deep imaging observations with DOLoRes@TNG of an ultra-faint dwarf satellite candidate of the Triangulum galaxy (M33) found by visual inspection of the public imaging data release of the DESI Legacy Imaging Surveys. Pisces VII/Triangulum (Tri) III is found at a projected distance of 72 kpc from M33, and using the tip of the red giant branch method we estimate a distance of D=1.0 +0.3,-0.2 Mpc, meaning the galaxy could either be an isolated ultra-faint or the second known satellite of M33. We estimate an absolute magnitude of M_V=-6.1+/-0.2 if Pisces VII/Tri II is at the distance of M33, or as bright as M_V=-6.8+/-0.2 if the galaxy is isolated. At the isolated distance, it has a physical half-light radius of r_h=131+/-61 pc consistent with similarly faint galaxies around the Milky Way. As the tip of the red giant branch is sparsely populated, constraining a precision distance is not possible, but if Pisces VII/Tri III can be confirmed as a true satellite of M33 it is a significant finding. With only one potential satellite detected around M33 previously (Andromeda XXII/Tri I), it lacks a significant satellite population in stark contrast to the similarly massive Large Magellanic Cloud. The detection of more satellites in the outskirts of M33 could help to better illuminate if this discrepancy between expectation and observations is due to a poor understanding of the galaxy formation process, or if it is due to the low luminosity and surface brightness of the M33 satellite population which has thus far fallen below the detection limits of previous surveys. If it is truly isolated, it would be the faintest known field dwarf detected to date., Comment: 10 pages, 9 figures, 2 tables, accepted for publication in MNRAS. This final version includes a moderate revision after the referee's comments and a correct title

In this paper, we aim to study the main properties and luminosity function (LF) of the [OII] emitters detected in the OTELO survey in order to characterise the star formation processes in low-mass galaxies at $z\sim1.43$ and to constrain the faint-end of the LF. Here, we describe the selection method and analysis of the emitters obtained from narrow-band scanning techniques. In addition, we present several relevant properties of the emitters and discuss the selection biases and uncertainties in the determination of the LF and the star formation rate density (SFRD). We confirmed a total of 60 sources from a preliminary list of 332 candidates as [OII] emitters. Approximately 93% of the emitters have masses in the range of $10^{8}

The INT Photometric H$\alpha$ Survey (IPHAS) has provided us with a number of new-emission line sources, among which planetary nebulae (PNe) constitute an important fraction. Here we present a detailed analysis of the IPHAS nebula Sab\,19 (IPHASX\,J055242.8+262116) based on radio, infrared, and optical images and intermediate- and high-dispersion longslit spectra. Sab\,19 consists of a roundish 0.10 pc in radius double-shell nebula surrounded by a much larger 2.8 pc in radius external shell with a prominent H-shaped filament. We confirm the nature of the main nebula as a PN whose sub-solar N/O ratio abundances, low ionized mass, peculiar radial velocity, and low-mass central star allow us to catalog it as a type III PN. Apparently, the progenitor star of Sab\,19 became a PN when crossing the Perseus Arm during a brief visit of a few Myr. The higher N/O ratio and velocity shift $\simeq$40 \kms\ of the external shell with respect to the main nebula and its large ionized mass suggest that it is not truly associated with Sab\,19, but it is rather dominated by a Str\"omgren zone in the interstellar medium ionized by the PN central star., Comment: 12 pages, 8 figures, accepted for publication in MNRAS

We present DECam imaging combined with Gaia DR2 data to study the Canis Major overdensity. The presence of the so-called Blue Plume stars in a low-pollution area of the color-magnitude diagram allows us to derive the distance and proper motions of this stellar feature along the line of sight of its hypothetical core. The stellar overdensity extends on a large area of the sky at low Galactic latitudes, below the plane, and between 230$^\circ < \ell < 255^\circ$. According to the orbit derived for Canis Major, it presents an on-plane rotation around the Milky Way. Moreover, additional overdensities of Blue Plume stars are found around the plane and across the Galaxy, proving that these objects are not only associated with that structure. The spatial distribution of these stars, derived using Gaia astrometric data, confirms that the detection of the Canis Major overdensity results more from the warped structure of the Milky Way disk than from the accretion of a dwarf galaxy., Comment: 11 pages, 9 figures, accepted for publication in MNRAS

Context. The accurate classification of hundreds of thousands of galaxies observed in modern deep surveys is imperative if we want to understand the universe and its evolution. Aims. Here, we report the use of machine learning techniques to classify early- and late-type galaxies in the OTELO and COSMOS databases using optical and infrared photometry and available shape parameters: either the Sersic index or the concentration index. Methods. We used three classification methods for the OTELO database: 1) u-r color separation , 2) linear discriminant analysis using u-r and a shape parameter classification, and 3) a deep neural network using the r magnitude, several colors, and a shape parameter. We analyzed the performance of each method by sample bootstrapping and tested the performance of our neural network architecture using COSMOS data. Results. The accuracy achieved by the deep neural network is greater than that of the other classification methods, and it can also operate with missing data. Our neural network architecture is able to classify both OTELO and COSMOS datasets regardless of small differences in the photometric bands used in each catalog. Conclusions. In this study we show that the use of deep neural networks is a robust method to mine the cataloged data, Comment: 20 pages, 10 tables, 14 figures, Astronomy and Astrophysics (in press)

The OTELO survey is a very deep, blind exploration of a selected region of the Extended Groth Strip and is designed for finding emission-line sources (ELSs). The survey design, observations, data reduction, astrometry, and photometry, as well as the correlation with ancillary data used to obtain a final catalogue, including photo-z estimates and a preliminary selection of ELS, were described in a previous contribution. Here, we aim to determine the main properties and luminosity function (LF) of the [O III] ELS sample of OTELO as a scientific demonstration of its capabilities, advantages, and complementarity with respect to other surveys. The selection and analysis procedures of ELS candidates obtained using tunable filter (TF) pseudo-spectra are described. We performed simulations in the parameter space of the survey to obtain emission-line detection probabilities. Relevant characteristics of [O III] emitters and the LF([O III]), including the main selection biases and uncertainties, are presented. A total of 184 sources were confirmed as [O III] emitters at a mean redshift z=0.83. The minimum detectable line flux and equivalent width (EW) in this ELS sample are $\sim$5 $\times$ 10$^{-19}$ erg s$^{-1}$ cm$^{2}$ and $\sim$6 \AA, respectively. We are able to constrain the faint-end slope ($\alpha = -1.03\pm0.08$) of the observed LF([O III]) at z=0.83. This LF reaches values that are approximately ten times lower than those from other surveys. The vast majority (84\%) of the morphologically classified [O III] ELSs are disc-like sources, and 87\% of this sample is comprised of galaxies with stellar masses of M$_\star$ $<$ 10$^{10}$ M$_{\odot}$., Comment: v1: 16 pages, 6 figures. Accepted in Astronomy \& Astrophysics. v2: Author added in metadata

We report results of proper motions of 15 known Large Magellanic Cloud (LMC) old globular clusters (GCs) derived from the Gaia DR2 data sets. When these mean proper motions are gathered with existent radial velocity measurements to compose the GCs' velocity vectors, we found that the projection of the velocity vectors onto the LMC plane and those perpendicular to it tell us about two distinct kinematic GC populations. Such a distinction becomes clear if the GCs are split at a perpendicular velocity of 10 km/s (absolute value). The two different kinematic groups also exhibit different spatial distributions. Those with smaller vertical velocities are part of the LMC disc, while those with larger values are closely distributed like a spherical component. Since GCs in both kinematic-structural components share similar ages and metallicities, we speculate with the possibility that their origins could have occurred through a fast collapse that formed halo and disc concurrently., Comment: 5 pages, 2 figures. Accepted for publication in Monthly Notices of the Royal Astronomical Society Letters

One of the fundamental goals of modern astrophysics is to estimate the physical parameters of galaxies. We present a hierarchical Bayesian model to compute age maps from images in the \Ha\ line (taken with Taurus Tunable Filter, TTF), ultraviolet band (GALEX far UV, FUV), and infrared bands (Spitzer 24, 70, and 160 $\mu$m). We present the burst ages for young stellar populations in a sample of nearby and nearly face-on galaxies. The \Ha\ to FUV flux ratio is a good relative indicator of the very recent star formation history (SFH). As a nascent star-forming region evolves, the \Ha\ line emission declines earlier than the UV continuum, leading to a decrease in the \Ha/FUV ratio. Using star-forming galaxy models, sampled with a probabilistic formalism, and allowing for a variable fraction of ionizing photons in the clusters, we obtain the corresponding theoretical ratio \Ha/FUV to compare with our observed flux ratios, and thus to estimate the ages of the observed regions. We take into account the mean uncertainties and the interrelationships between parameters when computing \Ha/FUV. We propose a Bayesian hierarchical model where a joint probability distribution is defined to determine the parameters (age, metallicity, IMF) from the observed data (the observed flux ratios \Ha/FUV). The joint distribution of the parameters is described through independent and identically distributed (i.i.d.) random variables generated through MCMC (Markov Chain Monte Carlo) techniques., Comment: 31 pages; 25 figures

We propose a method for calculating the radius of an open cluster in an objective way from an astrometric catalogue containing, at least, positions and proper motions. It uses the minimum spanning tree (hereinafter MST) in the proper motion space to discriminate cluster stars from field stars and it quantifies the strength of the cluster-field separation by means of a statistical parameter defined for the first time in this paper. This is done for a range of different sampling radii from where the cluster radius is obtained as the size at which the best cluster-field separation is achieved. The novelty of this strategy is that the cluster radius is obtained independently of how its stars are spatially distributed. We test the reliability and robustness of the method with both simulated and real data from a well-studied open cluster (NGC 188), and apply it to UCAC4 data for five other open clusters with different catalogued radius values. NGC 188, NGC 1647, NGC 6603 and Ruprecht 155 yielded unambiguous radius values of 15.2+/-1.8, 29.4+/-3.4, 4.2+/-1.7 and 7.0+/-0.3 arcmin, respectively. ASCC 19 and Collinder 471 showed more than one possible solution but it is not possible to know whether this is due to the involved uncertainties or to the presence of complex patterns in their proper motion distributions, something that could be inherent to the physical object or due to the way in which the catalogue was sampled., Comment: 12 pages including 14 figures and 1 table. Accepted for publication in MNRAS

We present ELDAR, a new method that exploits the potential of medium- and narrow-band filter surveys to securely identify active galactic nuclei (AGN) and determine their redshifts. Our methodology improves on traditional approaches by looking for AGN emission lines expected to be identified against the continuum, thanks to the width of the filters. To assess its performance, we apply ELDAR to the data of the ALHAMBRA survey, which covered an effective area of $2.38\,{\rm deg}^2$ with 20 contiguous medium-band optical filters down to F814W$\simeq 24.5$. Using two different configurations of ELDAR in which we require the detection of at least 2 and 3 emission lines, respectively, we extract two catalogues of type-I AGN. The first is composed of 585 sources ($79\,\%$ of them spectroscopically-unknown) down to F814W$=22.5$ at $z_{\rm phot}>1$, which corresponds to a surface density of $209\,{\rm deg}^{-2}$. In the second, the 494 selected sources ($83\,\%$ of them spectroscopically-unknown) reach F814W$=23$ at $z_{\rm phot}>1.5$, for a corresponding number density of $176\,{\rm deg}^{-2}$. Then, using samples of spectroscopically-known AGN in the ALHAMBRA fields, for the two catalogues we estimate a completeness of $73\,\%$ and $67\,\%$, and a redshift precision of $1.01\,\%$ and $0.86\,\%$ (with outliers fractions of $8.1\,\%$ and $5.8\,\%$). At $z>2$, where our selection performs best, we reach $85\,\%$ and $77\,\%$ completeness and we find no contamination from galaxies., Comment: 23 pages, 18 figures, submitted to MNRAS

In this work we analyse the structure of a subspace of the phase space of the star-forming region NGC~ 2264 using the Spectrum of Kinematic Groupings (SKG). We show that the SKG can be used to process a collection of star data to find substructure at different scales. We have found structure associated with the NGC~ 2264 region and also with the background area. In the NGC~ 2264 region, a hierarchical analysis shows substructure compatible with that found in previous specific studies of the area but with an objective, compact methodology that allows us to homogeneously compare the structure of different clusters and star-forming regions. Moreover, this structure is compatible with the different ages of the main NGC~ 2264 star-forming populations. The structure found in the field can be roughly associated with giant stars far in the background, dynamically decoupled from NGC~ 2264, which could be related either with the Outer Arm or Monoceros Ring. The results in this paper confirm the relationship between structure in the RV phase-space subspace and different kinds of populations, defined by other variables not necessarily analysed with the SKG, such as age or distance, showing the importance of detecting phase-space substructure in order to trace stellar populations in the broadest sense of the word., Comment: Accepted in MNRAS

NGC 2548 is a 400-500 Myr old open cluster with evidence of spatial substructures likely caused by its interaction with the Galactic disk. In this work we use precise astrometric data from the Carte du Ciel - San Fernando (CdC-SF) catalogue to study the clumpy structure in this cluster. We confirm the fragmented structure of NGC 2548 but, additionally, the relatively high precision of our kinematic data lead us to the first detection of substructures in the proper motion space of a stellar cluster. There are three spatially separated cores each of which has its own counterpart in the proper motion distribution. The two main cores lie nearly parallel to the Galactic plane whereas the third one is significantly fainter than the others and it moves toward the Galactic plane separating from the rest of the cluster. We derive core positions and proper motions, as well as the stars belonging to each core., Comment: 9 pages, 8 figures, 4 tables. Accepted for publication in MNRAS

On the basis of the Galactic O-Star Spectroscopic Survey (GOSSS), a detailed systematic investigation of the O Vz stars is presented. The currently used spectral classification criteria are rediscussed, and the Vz phenomenon is recalibrated through the addition of a quantitative criterion based on the equivalent widths of the He I 4471, He II 4542, and He II 4686 spectral lines. The GOSSS O Vz and O V populations resulting from the newly adopted spectral classification criteria are comparatively analyzed. The locations of the O Vz stars are probed, showing a concentration of the most extreme cases toward the youngest star forming regions. The occurrence of the Vz spectral peculiarity in a solar-metallicity environment, as predicted by the fastwind code, is also investigated, confirming the importance of taking into account several processes for the correct interpretation of the phenomenon., Comment: Accepted for publication in The Astronomical Journal

We present a new geometrical method aimed at determining the members of open clusters. The methodology estimates, in an N-dimensional space, the membership probabilities by means of the distances between every star and the cluster central overdensity. It can handle different sets of variables, which have to satisfy the simple condition of being more densely distributed for the cluster members than for the field stars (as positions, proper motions, radial velocities and/or parallaxes are). Unlike other existing techniques, this fact makes the method more flexible and so can be easily applied to different datasets. To quantify how the method identifies the clus- ter members, we design series of realistic simulations recreating sky regions in both position and proper motion subspaces populated by clusters and field stars. The re- sults, using different simulated datasets (N = 1, 2 and 4 variables), show that the method properly recovers a very high fraction of simulated cluster members, with a low number of misclassified stars. To compare the goodness of our methodology, we also run other existing algorithms on the same simulated data. The results show that our method has a similar or even better performance than the other techniques. We study the robustness of the new methodology from different subsamplings of the ini- tial sample, showing a progressive deterioration of the capability of our method as the fraction of missing objects increases. Finally, we apply all the methodologies to the real cluster NGC 2682, indicating that our methodology is again in good agreement with preceding studies., Comment: 15 pages, 9 figures, 6 tables, accepted for publication in MNRAS

We present a method for analysing the phase space of star-forming regions. In particular we are searching for clumpy structures in the 3D subspace formed by two position coordinates and radial velocity. The aim of the method is the detection of kinematic segregated radial velocity groups, that is, radial velocity intervals whose associated stars are spatially concentrated. To this end we define a kinematic segregation index, $\tilde{\Lambda}$(RV), based on the Minimum Spanning Tree (MST) graph algorithm, which is estimated for a set of radial velocity intervals in the region. When $\tilde{\Lambda}$(RV) is significantly greater than 1 we consider that this bin represents a grouping in the phase space. We split a star-forming region into radial velocity bins and calculate the kinematic segregation index for each bin, and then we obtain the spectrum of kinematic groupings, which enables a quick visualization of the kinematic behaviour of the region under study. We carried out numerical models of different configurations in the subspace of the phase space formed by the coordinates and the radial velocity that various case studies illustrate. The analysis of the test cases demonstrates the potential of the new methodology for detecting different kind of groupings in phase space., Comment: 8 pages, 4 figures, accepted for publication in MNRAS

One of the fundamental goals of modern Astronomy is to estimate the physical parameters of galaxies from images in different spectral bands. We present a hierarchical Bayesian model for obtaining age maps from images in the \Ha\ line (taken with Taurus Tunable Filter (TTF)), ultraviolet band (far UV or FUV, from GALEX) and infrared bands (24, 70 and 160 microns ($\mu$m), from Spitzer). As shown in S\'anchez-Gil et al. (2011), we present the burst ages for young stellar populations in the nearby and nearly face on galaxy M74. As it is shown in the previous work, the \Ha\ to FUV flux ratio gives a good relative indicator of very recent star formation history (SFH). As a nascent star-forming region evolves, the \Ha\ line emission declines earlier than the UV continuum, leading to a decrease in the \Ha\/FUV ratio. Through a specific star-forming galaxy model (Starburst 99, SB99), we can obtain the corresponding theoretical ratio \Ha\ / FUV to compare with our observed flux ratios, and thus to estimate the ages of the observed regions. Due to the nature of the problem, it is necessary to propose a model of high complexity to take into account the mean uncertainties, and the interrelationship between parameters when the \Ha\ / FUV flux ratio mentioned above is obtained. To address the complexity of the model, we propose a Bayesian hierarchical model, where a joint probability distribution is defined to determine the parameters (age, metallicity, IMF), from the observed data, in this case the observed flux ratios \Ha\ / FUV. The joint distribution of the parameters is described through an i.i.d. (independent and identically distributed random variables), generated through MCMC (Markov Chain Monte Carlo) techniques., Comment: 12 pages, 6 figures, conference

We address the study of the \Ha\ vertical velocity field in a sample of four nearly face-on galaxies using long slit spectroscopy taken with the ISIS spectrograph attached to the WHT at the Roque de los Muchachos Observatory (Spain). The spatial structure of the velocity vertical component shows a radial corrugated pattern with spatial scales higher or within the order of { one} kiloparsec. The gas is mainly ionized by high-energy photons: only in some locations of NGC~278 and NGC~1058 is there some evidence of ionization by low-velocity shocks, which, in the case of NGC~278, could be due to minor mergers. The behaviour of the gas in the neighbourhood of the spiral arms fits, in the majority of the observed cases, with that predicted by the so-called hydraulic bore mechanism, where a thick magnetized disk encounters a spiral density perturbation. The results obtained show that it is { difficult to explain the \Ha\ large scale velocity field without the presence of a magnetized, thick galactic disk}. Larger samples and spatial covering of the galaxy disks are needed to provide further insight into this problem., Comment: 19 pages, 9 figures, 3 tables

We discuss the properties of H$\alpha$ emission stars across the sample of 22035 spectra from the Gaia-ESO Survey internal data release, observed with the GIRAFFE instrument and largely belonging to stars in young open clusters. Automated fits using two independent Gaussian profiles and a third component that accounts for the nebular emission allow us to discern distinct morphological types of H$\alpha$ line profiles with the introduction of a simplified classification scheme. All in all we find 3765 stars with intrinsic emission and sort their spectra into eight distinct morphological categories: single--component emission, emission blend, sharp emission peaks, double emission, P-Cygni, inverted P-Cygni, self--absorption, and emission in absorption. We have more than one observation for 1430 stars in our sample, thus allowing a quantitative discussion of the degree of variability of H$\alpha$ emission profiles, which is expected for young, active objects. We present a catalogue of stars with properties of their H$\alpha$ emission line profiles, morphological classification, analysis of variability with time and the supplementary information from the SIMBAD, VizieR, and ADS databases. The records in SIMBAD indicate the presence of H$\alpha$ emission for roughly 25% of all stars in our catalogue, while at least 305 of them have already been more thoroughly investigated according to the references in ADS. The most frequently identified morphological categories in our sample of spectra are emission blend (23%), emission in absorption (22%), and self--absorption (16%). Objects with repeated observations demonstrate that our classification into discrete categories is generally stable through time, but categories P-Cygni and self--absorption seem less stable, which is the consequence of discrete classification rules, as well as of the fundamental change in profile shape.

The mass distribution of prestellar cores is obtained for clouds with arbitrary internal mass distributions using a selection criterion based on the thermal and turbulent Jeans mass and applied hierarchically from small to large scales. We have checked this methodology comparing our results for a lognormal density PDF with the theoretical CMF derived by Hennebelle and Chabrier, namely a power-law at large scales and a log-normal cutoff at low scales, but our method can be applied to any mass distributions representing a star-forming cloud. This methodology enables us to connect the parental cloud structure with the mass distribution of the cores and their spatial distribution, providing an efficient tool for investigating the physical properties of the molecular clouds that give rise to the prestellar core distributions observed. Simulated fBm clouds with the Hurst exponent close to the value H=1/3 give the best agreement with the theoretical CMF derived by Hennebelle and Chabrier and Chabrier's system IMF. Likewise, the spatial distribution of the cores derived from our methodology show a surface density of companions compatible with those observed in Trapezium and Ophiucus star-forming regions. This method also allows us to analyze the properties of the mass distribution of cores for different realizations. We found that the variations in the number of cores formed in different realizations of fBm clouds (with the same Hurst exponent) are much larger than the expected root N statistical fluctuations, increasing with H., Comment: 27 pages including 6 figures and 2 tables. Accepted for publication in The Astrophysical Journal

With new data from the Galactic O-Star Spectroscopic Survey, we confirm and expand the ONn category of late-O, nitrogen-enriched (N), rapidly rotating (n) giants. In particular, we have discovered two "clones" (HD 102415 and HD 117490) of one of the most rapidly rotating O stars previously known (HD 191423, "Howarth's Star"). We compare the locations of these objects in the theoretical HR Diagram to those of slowly rotating ON dwarfs and supergiants. All ON giants known to date are rapid rotators, whereas no ON dwarf or supergiant is; but all ON stars are small fractions of their respective spectral-type/luminosity-class/rotational subcategories. The ONn giants, displaying both substantial processed material and high rotation at an intermediate evolutionary stage, may provide significant information about the development of those properties. They may have preserved high initial rotational velocities or been spun up by TAMS core contraction; but alternatively and perhaps more likely, they may be products of binary mass transfer. At least some of them are also runaway stars., Comment: 18 pages, 2 figures, 2 tables; accepted for publication in the November 2011 issue of The Astronomical Journal

We present the burst ages for young stellar populations in a sample of six nearby (< 10 Mpc) spiral galaxies using a differential pixel-based analysis of the ionized gas emission. We explore this as an alternative approach for connecting large-scale dynamical mechanisms with star formation processes in disk galaxies, based on burst ages derived from the Ha to far UV (FUV) flux ratio. Images of each galaxy in Ha were taken with Taurus Tunable Filter (TTF) and matched to FUV imaging from GALEX. The resulting flux ratio provides a robust measure of relative age across the disk which we discuss in terms of the large-scale dynamical motions. Systematic effects, such as a variable initial mass function (IMF), non-solar metallicities, variable star-formation history (SFHs), and dust attenuation, have been used to derive estimates of the systematic uncertainty. The resulting age maps show a wide range of patterns outside of those galaxies with the strongest spiral structure, confirming the idea that star formation is driven one by several processes, largely determined by the individual circumstances of the galaxy. Generally, grand design spirals such as M74, M100, and M51 exhibit age gradients across the main spiral arms, with the youngest star formation regions along the central and inner edges. In M63 and M74 galaxy-wide trends emerge, suggesting that although most star formation is located along spiral arms, spiral density waves are not the only driver in these cases. We argue that despite appearances, galaxy morphology is not an absolute discriminator of the star formation history of an individual galaxy, nor of the processes triggering it. We conclude that Ha-to-FUV flux ratios are a relatively direct way to probe burst ages across galaxies and infer something of their dynamical histories, provided that sources of systematics are properly taken into account., Comment: The paper contains 22 pages, 13 figures and 5 tables

In this poster we show our preliminary analysis of DIBs (Diffuse Interstellar Bands) and other interstellar absorption lines with the purpose of understanding their origin and their relationship with extinction. We use the biggest Galactic O-star blue-violet spectroscopic sample ever (GOSSS, see contribution by Ma\'iz Apell\'aniz at this meeting). This sample allows a new insight on this topic because of the adequacy of O-star spectra, the sample number (700 and increasing, 400 used here), and their distribution in the MW disk. We confirm the high correlation coefficients between different DIBs and E(B - V), though the detailed behavior of each case shows small differences. We also detect a moderately low correlation coefficient between Ca II {\lambda}3934 (Ca K) and E(B - V) with a peculiar spatial distribution that we ascribe to the relationship between line saturation and velocity profiles for Ca II {\lambda}3934., Comment: 6 pages, 3 figures, 1 table. Poster at the IX Scientific Meeting of the Spanish Astronomical Society held on September 13-17, 2010, in Madrid, Spain

The Galactic O-Star Spectroscopic Survey (GOSSS) is a project that is observing all known Galactic O stars with B < 13 (~2000 objects) in the blue-violet part of the spectrum with R~2500. It also includes two companion surveys (a spectroscopic one at R~1500 and a high resolution imaging one). It is based on v2.0 of the Galactic O star catalog (v1, Ma\'iz Apell\'aniz et al. 2004; v2, Sota et al. 2008). We have completed the first part of the main project. Here we present results on the first 400 objects of the sample., Comment: 5 pages, 5 figues. Poster contribution to the proceedings of the LIAC2010 conference "The multi-wavelength view of hot, massive stars"

The Interstellar Medium has a fractal structure, in the sense that gas and dust distribute in a hierarchical and self-similar manner. Stars in new-born cluster probably follow the same fractal patterns of their parent molecular clouds. Moreover, it seems that older clusters tend to distribute their stars with radial density profiles. Thus, it is expected that clusters form with an initial fractal distribution of stars that eventually evolves toward centrally concentrated distributions. Is this really the case? This simple picture on to the origin and early evolution of star clusters and associations is very far from being clearly understood. There can be both young clusters exhibiting radial patterns and evolved clusters showing fractal structure. Additionally, the fractal structure of some open clusters is very different from that of the Interstellar Medium in the Milky Way. Here we summarize and discuss observational and numerical evidences concerning this subject., Comment: 11 pages (two columns). To appear in Lecture Notes and Essays in Astrophysics IV, A. Ulla and M. Manteiga (editors)

We study the distribution of stars, HII regions, molecular gas, and individual giant molecular clouds in M33 over a wide range of spatial scales. The clustering strength of these components is systematically estimated through the fractal dimension. We find scale-free behavior at small spatial scales and a transition to a larger correlation dimension (consistent with a nearly uniform distribution) at larger scales. The transition region lies in the range 500-1000 pc. This transition defines a characteristic size that separates the regime of small-scale turbulent motion from that of large-scale galactic dynamics. At small spatial scales, bright young stars and molecular gas are distributed with nearly the same three-dimensional fractal dimension (Df <= 1.9), whereas fainter stars and HII regions exhibit higher values (Df = 2.2-2.5). Our results indicate that the interstellar medium in M33 is on average more fragmented and irregular than in the Milky Way., Comment: 18 pages including 4 figures. Accepted for publication in ApJ

On the basis of an extensive new spectroscopic survey of Galactic O stars, we introduce the Ofc category, which consists of normal spectra with C III \lambda\lambda4647-4650-4652 emission lines of comparable intensity to those of the Of defining lines N III \lambda\lambda4634-4640-4642. The former feature is strongly peaked to spectral type O5, at all luminosity classes, but preferentially in some associations or clusters and not others. The relationships of this phenomenon to the selective C III \lambda5696 emission throughout the normal Of domain, and to the peculiar, variable Of?p category, for which strong C III \lambda\lambda4647-4650-4652 emission is a defining characteristic, are discussed. Magnetic fields have recently been detected on two members of the latter category. We also present two new extreme Of?p stars, NGC 1624-2 and CPD -28^{\circ}2561, bringing the number known in the Galaxy to five. Modeling of the behavior of these spectral features can be expected to better define the physical parameters of both normal and peculiar objects, as well as the atomic physics involved., Comment: 12 pages, 1 table, 2 figures; accepted for ApJL, vol. 711

We present the analysis of high-resolution optical spectroscopic observations of the zero-age main-sequence O star Herschel 36 spanning six years. This star is definitely a multiple system, with at least three components detected in its spectrum. Based on our radial-velocity (RV) study, we propose a picture of a close massive binary and a more distant companion, most probably in wide orbit about each other. The orbital solution for the binary, whose components we identify as O9 V and B0.5 V, is characterized by a period of 1.5415 +/- 0.0006 days. With a spectral type O7.5 V, the third body is the most luminous component of the system and also presents RV variations with a period close to 498 days. Some possible hypotheses to explain the variability are briefly addressed and further observations are suggested., Comment: 6 pages, 2 figures

We analyze the dependence of the membership probabilities obtained from kinematical variables on the radius of the field of view around open clusters (the sampling radius, Rs). From simulated data, we show that the best discrimination between cluster members and non-members is obtained when the sampling radius is very close to the cluster radius. At higher Rs values more field stars tend to be erroneously assigned as cluster members. From real data of two open clusters (NGC 2323 and NGC 2311) we obtain that the number of identified cluster members always increases with increasing Rs. However, there is a threshold Rs value above which the identified cluster members are severely contaminated by field stars and the effectiveness of membership determination is relatively small. This optimal sampling radius is \sim 14 arcmin for NGC 2323 and \sim 13 arcmin for NGC 2311. We discuss the reasons for such behavior and the relationship between cluster radius and optimal sampling radius. We suggest that, independently of the method used to estimate membership probabilities, several tests using different sampling radius should be performed in order to evaluate the existence of possible biases., Comment: 8 pages (two columns) including 9 figures. Accepted for publication in A&A

There is a lack of state-of-the-art information on very young open clusters, with implications for determining the structure of the Galaxy. Our main objective is to study the timing and location of the star formation processes which yielded the generation of the giant HII region Sh 2-284. The analysis is based on UBVRcIc CCD measurements and JHKs photometry in the central part of the HII region, where the cluster Dolidze 25 is located.The determination of cluster distance, reddening and age is carried out through comparison with ZAMS, post-MS and PMS isochrones. Reference lines for metallicity Z=0.004 are used, in agreement with spectroscopic metallicity determination published for several cluster members. The results are: E(B-V)=0.78+-0.02, M=12.8+-0.2, LogAge(yr)=6.51+-0.07. A PMS member sequence is proposed, coeval within the errors with the post-MS cluster age (LogAge(yr)=6.7+-0.2). The mass function for this PMS population in the mass range above 1.3-3.5 Msun is well fitted by a Salpeter mass function.The presence of a different star generation in the cluster with a distinctly older age, around 40 Myr, is suggested. The NIR results indicate a large number of sources with H-Ks excess, practically distinct from the optical PMS candidate members. The distance determined for the cluster is distinctly lower than previously published values. This result originates in the consistent use of low metallicity models for ZAMS fitting, applying published metallicity values for the cluster., Comment: gzipped tar file: 1 aa.cls, 1 tex file, 1 bbl file, 13 figures. Accepted in Astronomy & Astrophysics (9.11.2009)

In this work we study the internal spatial structure of 16 open clusters in the Milky Way spanning a wide range of ages. For this, we use the minimum spanning tree method (the Q parameter, which enables one to classify the star distribution as either radially or fractally clustered), King profile fitting, and the correlation dimension (Dc) for those clusters with fractal patterns. On average, clusters with fractal-like structure are younger than those exhibiting radial star density profiles. There is a significant correlation between Q and the cluster age measured in crossing time units. For fractal clusters there is a significant correlation between the fractal dimension and age. These results support the idea that stars in new-born clusters likely follow the fractal patterns of their parent molecular clouds, and eventually evolve toward more centrally concentrated structures. However, there can exist stellar clusters as old as 100 Myr that have not totally destroyed their fractal structure. Finally, we have found the intriguing result that the lowest fractal dimensions obtained for the open clusters seem to be considerably smaller than the average value measured in galactic molecular cloud complexes., Comment: To appear in the proceedings of IAU Symposium 266: "Star Clusters: Basic Galactic Building Blocks Throughout Time And Space", eds. R. de Grijs and J. Lepine

The galaxy NGC 6946 contains a gas-star complex of 700 pc in diameter which appears populated by tens of young stellar clusters, and a Super Star Cluster (SSC) as massive as 10$^6$ M$\odot$. The ionized gas, as drawn by the H$\alpha$ emission, delineates an almost circular shape which we show here to be in expansion. Previous studies have analyzed the stellar component of the complex, as well as the structure of the atomic and ionized gas; these analyses were restricted to the blueshifted component along the whole extent of the bubble or to a smaller inner region where both sides of an expanding bubble were seen. In this work we present a complete spectroscopic study of this object for two position angles crossing each other close to the young massive SSC. We have obtained new data with a spectral resolution six times better than previous spectroscopic studies, taken under atmospheric conditions better than those previously reported, allowing us to detect the approaching and receding walls of one the largest bubbles in external galaxies ever studied in detail. The kinematical analysis shows a large expanding bubble, whose walls appear to be highly structured with superposed smaller shells, likely originated as the result of star forming events occurring at the edges of the larger scale shell, a la Huygens. We also study some diagnostic diagrams of the ionized gas and conclude that most of the observed ionization is originated by photons from hot stars, but with clear evidence that some of the gas is shock ionized. This peculiar complex is an excellent laboratory for the analysis of the interaction and feedback between the gas where the stars were formed and the young and massive generation of new born stars., Comment: 41 pages, 12 figures

In this work we quantify the degree to which star-forming events are clumped. We apply a precise and accurate technique to calculate the correlation dimension Dc of the distribution of HII regions in a sample of disk galaxies. Our reliable results are distributed in the range 1.5

It is known that the gas has a fractal structure in a wide range of spatial scales with a fractal dimension that seems to be a constant around Df = 2.7. It is expected that stars forming from this fractal medium exhibit similar fractal patterns. Here we address this issue by quantifying the degree to which star-forming events are clumped. We develop, test, and apply a precise and accurate technique to calculate the correlation dimension Dc of the distribution of HII regions in a sample of disk galaxies. We find that the determination of Dc is limited by the number of HII regions, since if there are < 100 regions available then a bias tending to underestimate the dimension is produced. The reliable results are distributed in the range 1.5 < Dc < 2.0 with an average value Dc = 1.81. This corresponds to a three-dimensional dimension of Df = 2.73, very similar to the value measured in the interstellar clouds. However, we get significant variations in the fractal dimension among galaxies, contrary to a universal picture sometimes claimed in literature. The fractal dimension exhibits a weak but significant correlation with the absolute magnitude and, to a lesser extent, with the galactic radius. The faintest galaxies tend to distribute their HII regions in more clustered (less uniform) patterns. The fractal dimension for the brightest HII regions within the same galaxy seems to be smaller than for the faintest ones suggesting some kind of evolutionary efffect, but the obtained correlation remains unchanged if only the brightest regions are taken into account., Comment: 41 pages including 6 figures and 5 tables. Accepted for publication in ApJS

The Interstellar Medium seems to have an underlying fractal structure, which can be characterized through its fractal dimension (Df). However, several factors may affect the determination of Df, such as distortions due to projection, low image resolution, opacity of the cloud, and low signal-to-noise ratio. Here we use both simulated clouds and real molecular cloud maps to study these effects in order to estimate Df in a reliable way. Our results indicate in a self-consistent way that the fractal dimension of the Interstellar Medium is in the range 2.6 < Df < 2.8, which is significantly higher than the value Df = 2.3 usually assumed in the literature., Comment: 2 pages including 2 figures. To appear in RMxAC: proceedings of the XII Latin American Regional IAU Meeting, Venezuela, 22-26 October 2007

We report the properties of a new young double stellar cluster in the region towards IRAS 07141-0920 contained in the HII region Sh2-294. High-resolution optical UBVRI and Halpha images, near-infrared JHKs and H2 filter images were used to make photometric and morphological studies of the point sources and the nebula seen towards Sh2-294. The optical images reveal an emission nebula with very rich morphological details, composed mainly of UV scattered light and of Halpha emission. Contrasting with the bright parts of the nebula, opaque, elongated patches are seen. Our optical photometry confirms that the illuminator of the nebula is likely to be a B0.5V star located at a distance of about 3.2 kpc. Our near-IR images reveal an embedded cluster, extending for about 2 pc and exhibiting sub-clustering: a denser, more condensed, sub-cluster surrounding the optical high-mass B0.5V illuminator star; and a more embedded, optically invisible, sub-cluster located towards the eastern, dark part of the nebula and including the luminous MSX source G224.1880+01.2407, a massive protostellar candidate that could be the origin of jets and extended features seen at 2.12 micron. The double cluster appears to be clearing the remaining molecular material of the parent cloud, creating patches of lower extinction and allowing some of the least reddened members to be detected in the optical images. We find 12 MS and 143 PMS members using 3 different methods: comparison with isochrones in optical colour-magnitude diagrams, detection of near-IR excess, and presence of Halpha emission. The most massive star fits a 4 Myr post-MS isochrone. The age of the optically selected PMS population is estimated to be 7-8 Myr. The IR-excess population shows sub-clustering on scales as small as 0.23 pc and is probably much younger., Comment: 15 pages, 16 figures

There exists observational evidence that the interstellar medium has a fractal structure in a wide range of spatial scales. The measurement of the fractal dimension (Df) of interstellar clouds is a simple way to characterize this fractal structure, but several factors, both intrinsic to the clouds and to the observations, may contribute to affect the values obtained. In this work we study the effects that opacity and noise have on the determination of Df. We focus on two different fractal dimension estimators: the perimeter-area based dimension (Dper) and the mass-size dimension (Dm). We first use simulated fractal clouds to show that opacity does not affect the estimation of Dper. However, Dm tends to increase as opacity increases and this estimator fails when applied to optically thick regions. In addition, very noisy maps can seriously affect the estimation of both Dper and Dm, decreasing the final estimation of Df. We apply these methods to emission maps of Ophiuchus, Perseus and Orion molecular clouds in different molecular lines and we obtain that the fractal dimension is always in the range 2.6 < Df < 2.8 for these regions. These results support the idea of a relatively high (> 2.3) average fractal dimension for the interstellar medium, as traced by different chemical species., Comment: 17 pages including 6 figures and 1 table. Accepted for publication in ApJ

The recently discovered Virgo stellar over-density, which expands over \~1000deg^2 perpendicularly to the Galactic disk plane (7< Z <15 kpc, R~7 kpc), is the largest clump of tidal debris ever detected in the outer halo and is likely related with the accretion of a nearby dwarf galaxy by the Milky Way. We carry out N-body simulations of the Sagittarius stream to show that this giant stellar over-density is a confirmation of theoretical model predictions for the leading tail of the Sagittarius stream to cross the Milky Way plane in the Solar neighborhood. Radial velocity measurements are needed to confirm this association and to further constrain the shape of the Milky Way dark matter halo through a new generation of theoretical models. If the identification of Virgo over-density and the Sagittarius leading arm is correct, we predict highly negative radial velocities for the stars of Virgo over-density. The detection of this new portion of the Sagittarius tidal stream would represent an excellent target for the on-going and future kinematic surveys and for dark matter direct detection experiments in the proximity of the Sun., Comment: Accepted for publication in The Astrophysical Journal. Supplementary movies to this paper are available at: http://www.astro.princeton.edu/~mjuric/sgr-virgo/

We study the physical properties derived from interstellar cloud complexes having a fractal structure. We first generate fractal clouds with a given fractal dimension and associate each clump with a maximum in the resulting density field. Then, we discuss the effect that different criteria for clump selection has on the derived global properties. We calculate the masses, sizes and average densities of the clumps as a function of the fractal dimension (D_f) and the fraction of the total mass in the form of clumps (epsilon). In general, clump mass does not fulfill a simple power law with size of the type M_cl ~ (R_cl)**(gamma), instead the power changes, from gamma ~ 3 at small sizes to gamma<3 at larger sizes. The number of clumps per logarithmic mass interval can be fitted to a power law N_cl ~ (M_cl)**(-alpha_M) in the range of relatively large masses, and the corresponding size distribution is N_cl ~ (R_cl)**(-alpha_R) at large sizes. When all the mass is forming clumps (epsilon=1) we obtain that as D_f increases from 2 to 3 alpha_M increases from ~0.3 to ~0.6 and alpha_R increases from ~1.0 to ~2.1. Comparison with observations suggests that D_f ~ 2.6 is roughly consistent with the average properties of the ISM. On the other hand, as the fraction of mass in clumps decreases (epsilon<1) alpha_M increases and alpha_R decreases. When only ~10% of the complex mass is in the form of dense clumps we obtain alpha_M ~ 1.2 for D_f=2.6 (not very different from the Salpeter value 1.35), suggesting this a likely link between the stellar initial mass function and the internal structure of molecular cloud complexes., Comment: 32 pages, 13 figures, 1 table. Accepted for publication in ApJ

The interstellar medium seems to have an underlying fractal structure which can be characterized through its fractal dimension. However, interstellar clouds are observed as projected two-dimensional images, and the projection of a tri-dimensional fractal distorts its measured properties. Here we use simulated fractal clouds to study the relationship between the tri-dimensional fractal dimension (D_f) of modeled clouds and the dimension resulting from their projected images. We analyze different fractal dimension estimators: the correlation and mass dimensions of the clouds, and the perimeter-based dimension of their boundaries (D_per). We find the functional forms relating D_f with the projected fractal dimensions, as well as the dependence on the image resolution, which allow to estimatethe "real" D_f value of a cloud from its projection. The application of these results to Orion A indicates in a self-consistent way that 2.5 < D_f < 2.7 for this molecular cloud, a value higher than the result D_per+1 = 2.3 some times assumed in literature for interstellar clouds., Comment: 27 pages, 13 figures, 1 table. Accepted for publication in ApJ. Minor changes