Search NASA⌕ Search

Engineering topics

Chevance, Mélanie

Publications and source records attributed to Chevance, Mélanie.

Star cluster classification using deep transfer learning with PHANGS- HST

Currently available star cluster catalogues from the Hubble Space Telescope (HST) imaging of nearby galaxies heavily rely on visual inspection and classification of candidate clusters. Here, the time-consuming nature of this process has limited the production of reliable catalogues and thus also post-observation analysis. To address this problem, deep transfer learning has recently been used to create neural network models that accurately classify star cluster morphologies at production scale for nearby spiral galaxies (D ≲ 20 Mpc). Here, we use HST ultraviolet (UV)–optical imaging of over 20 000 sources in 23 galaxies from the Physics at High Angular resolution in Nearby GalaxieS (PHANGS) survey to train and evaluate two new sets of models: (i) distance-dependent models, based on cluster candidates binned by galaxy distance (9–12, 14–18, and 18–24 Mpc), and (ii) distance-independent models, based on the combined sample of candidates from all galaxies. We find that the overall accuracy of both sets of models is comparable to previous automated star cluster classification studies (~60–80 per cent) and shows improvement by a factor of 2 in classifying asymmetric and multipeaked clusters from PHANGS-HST. Somewhat surprisingly, while we observe a weak negative correlation between model accuracy and galactic distance, we find that training separate models for the three distance bins does not significantly improve classification accuracy. We also evaluate model accuracy as a function of cluster properties such as brightness, colour, and spectral energy distribution (SED)-fit age. Based on the success of these experiments, our models will provide classifications for the full set of PHANGS-HST candidate clusters (N ~ 200 000) for public release.

79 ASTRONOMY AND ASTROPHYSICS↗

Applying the Tremaine–Weinberg Method to Nearby Galaxies: Stellar-mass-based Pattern Speeds and Comparisons with ISM Kinematics

We apply the Tremaine–Weinberg method to 19 nearby galaxies using stellar mass surface densities and velocities derived from the PHANGS-MUSE survey, to calculate (primarily bar) pattern speeds (Ω{sub P}). After quality checks, we find that around half (10) of these stellar-mass-based measurements are reliable. For those galaxies, we find good agreement between our results and previously published pattern speeds, and we use rotation curves to calculate major resonance locations (corotation radii and Lindblad resonances). We also compare these stellar-mass-derived pattern speeds with Hα (from MUSE) and CO(J = 2 − 1) emission from the PHANGS-ALMA survey. We find that in the case of these clumpy interstellar medium (ISM) tracers, this method erroneously gives a signal that is simply the angular frequency at a representative radius set by the distribution of these clumps (Ω{sub clump}), and that this Ω{sub clump} is significantly different from Ω{sub P} (∼20% in the case of Hα, and ∼50% in the case of CO). Thus, we conclude that it is inadvisable to use “pattern speeds” derived from ISM kinematics. Finally, we compare our derived pattern speeds and corotation radii, along with bar properties, to the global parameters of these galaxies. Consistent with previous studies, we find that galaxies with a later Hubble type have a larger ratio of corotation radius to bar length, more molecular-gas-rich galaxies have higher Ω{sub P}, and more bulge-dominated galaxies have lower Ω{sub P}. Unlike earlier works, however, there are no clear trends between the bar strength and Ω{sub P}, nor between the total stellar mass surface density and the pattern speed.

74 ATOMIC AND MOLECULAR PHYSICS↗

Deep transfer learning for star cluster classification: I. application to the PHANGS– HST survey

ABSTRACT We present the results of a proof-of-concept experiment that demonstrates that deep learning can successfully be used for production-scale classification of compact star clusters detected in Hubble Space Telescope(HST) ultraviolet-optical imaging of nearby spiral galaxies ($D\lesssim 20\, \textrm{Mpc}$) in the Physics at High Angular Resolution in Nearby GalaxieS (PHANGS)–HST survey. Given the relatively small nature of existing, human-labelled star cluster samples, we transfer the knowledge of state-of-the-art neural network models for real-object recognition to classify star clusters candidates into four morphological classes. We perform a series of experiments to determine the dependence of classification performance on neural network architecture (ResNet18 and VGG19-BN), training data sets curated by either a single expert or three astronomers, and the size of the images used for training. We find that the overall classification accuracies are not significantly affected by these choices. The networks are used to classify star cluster candidates in the PHANGS–HST galaxy NGC 1559, which was not included in the training samples. The resulting prediction accuracies are 70 per cent, 40 per cent, 40–50 per cent, and 50–70 per cent for class 1, 2, 3 star clusters, and class 4 non-clusters, respectively. This performance is competitive with consistency achieved in previously published human and automated quantitative classification of star cluster candidate samples (70–80 per cent, 40–50 per cent, 40–50 per cent, and 60–70 per cent). The methods introduced herein lay the foundations to automate classification for star clusters at scale, and exhibit the need to prepare a standardized data set of human-labelled star cluster classifications, agreed upon by a full range of experts in the field, to further improve the performance of the networks introduced in this study.

Wei, Wei↗

Cloud-scale ISM Structure and Star Formation in M51

We compare the structure of molecular gas at 40 pc resolution to the ability of gas to form stars across the disk of the spiral galaxy M51. We break the PAWS survey into 370 pc and 1.1 kpc resolution elements, and within each we estimate the molecular gas depletion time (τ{sub Dep}{sup mol}), the star-formation efficiency per free-fall time (ϵ{sub ff}), and the mass-weighted cloud-scale (40 pc) properties of the molecular gas: surface density, Σ, line width, σ, and b≡Σ/σ{sup 2}∝α{sub vir}{sup −1}, a parameter that traces the boundedness of the gas. We show that the cloud-scale surface density appears to be a reasonable proxy for mean volume density. Applying this, we find a typical star-formation efficiency per free-fall time, ϵ{sub ff}(〈Σ{sub 40pc}〉)∼0.3%--0.36%, lower than adopted in many models and found for local clouds. Furthermore, the efficiency per free-fall time anti-correlates with both Σ and σ, in some tension with turbulent star-formation models. The best predictor of the rate of star formation per unit gas mass in our analysis is b≡Σ/σ{sup 2}, tracing the strength of self-gravity, with τ{sub Dep}{sup mol}∝b{sup −0.9}. The sense of the correlation is that gas with stronger self-gravity (higher b) forms stars at a higher rate (low τ{sub Dep}{sup mol}). The different regions of the galaxy mostly overlap in τ{sub Dep}{sup mol} as a function of b, so that low b explains the surprisingly high τ{sub Dep}{sup mol} found toward the inner spiral arms found by Meidt et al. (2013).

79 ASTRONOMY AND ASTROPHYSICS↗