ABSTRACT
Classification of galaxies is traditionally associated with their morphologies through visual inspection of images. The amount of data to come render this task, inhuman and Machine Learning ...(mainly Deep Learning) has been called to the rescue for more than a decade. However, the results look mitigate and there seems to be a shift away from the paradigm of the traditional morphological classification of galaxies. In this paper, I want to show that the algorithms indeed are very sensitive to the features present in images, features that do not necessarily correspond to the Hubble or de Vaucouleurs vision of a galaxy. However, this does not preclude to get the correct insights into the physics of galaxies. I have applied a state-of-the-art ‘traditional’ Machine Learning clustering tool, called Fisher-EM, a latent discriminant subspace Gaussian mixture model algorithm to 4458 galaxies carefully classified into 18 types by the EFIGI project. The optimum number of clusters given by the integrated complete likelihood criterion is 47. The correspondence with the EFIGI classification is correct, but it appears that the Fisher-EM algorithm gives a great importance to the distribution of light which translates to characteristics such as the bulge to disc ratio, the inclination or the presence of foreground stars. The discrimination of some physical parameters (bulge-to-total luminosity ratio, (B–V)T, intrinsic diameter, presence of flocculence or dust, and arm strength) is very comparable in the two classifications.
Context.
Defining templates of galaxy spectra is useful to quickly characterise new observations and organise databases from surveys. These templates are usually built from a pre-defined ...classification based on other criteria.
Aims.
We present an unsupervised classification of 702 248 spectra of galaxies and quasars with redshifts smaller than 0.25 that were retrieved from the Sloan Digital Sky Survey (SDSS) database, release 7.
Methods.
The spectra were first corrected for redshift, then wavelet-filtered to reduce the noise, and finally binned to obtain about 1437 wavelengths per spectrum. The unsupervised clustering algorithm Fisher-EM, relying on a discriminative latent mixture model, was applied on these corrected spectra. The full set and several subsets of 100 000 and 300 000 spectra were analysed.
Results.
The optimum number of classes given by a penalised likelihood criterion is 86 classes, of which the 37 most populated gather 99% of the sample. These classes are established from a subset of 302 214 spectra. Using several cross-validation techniques we find that this classification agrees with the results obtained on the other subsets with an average misclassification error of about 15%. The large number of very small classes tends to increase this error rate. In this paper, we do an initial quick comparison of our classes with literature templates.
Conclusions.
This is the first time that an automatic, objective and robust unsupervised classification is established on such a large number of galaxy spectra. The mean spectra of the classes can be used as templates for a large majority of galaxies in our Universe.
Aims.
Our study aims at providing deeper insight into the power and limitation of an unsupervised classification algorithm (called Fisher-EM) on spectra of galaxies. This algorithm uses a Gaussian ...mixture in a discriminative latent subspace. To this end, we investigate the capacity of this algorithm to segregate the physical parameters used to generate mock spectra and the influence of the noise on the classification.
Methods.
With the code CIGALE and different values for nine input parameters characterising the stellar population, we simulated a sample of 11 475 optical spectra of galaxies containing 496 monochromatic fluxes. The statistical model and the optimum number of clusters are given in Fisher-EM by the integrated completed likelihood (ICL) criterion. We repeated the analyses several times to assess the robustness of the results.
Results.
Two distinct classifications can be distinguished in the case of the noiseless spectra. The classification with more than 13 clusters disappears when noise is added, while the classification with 12 clusters is very robust against noise down to a signal-to-noise ratio (S/N) of 3. At
S
/
N
= 1, the optimum is 5 clusters, but the classification is still compatible with the previous classification. The distribution of the parameters used for the simulation shows an excellent discrimination between classes. A higher dispersion both in the spectra within each class and in the parameter distribution leads us to conclude that despite a much higher ICL, the classification with more than 13 clusters in the noiseless case is not physically relevant.
Conclusions.
This study yields two conclusions that are valid at least for the Fisher-EM algorithm. Firstly, the unsupervised classification of spectra of galaxies is both reliable and robust to noise. Secondly, such analyses are able to extract the useful physical information contained in the spectra and to build highly meaningful classifications. In an epoch of data-driven astrophysics, it is important to trust unsupervised machine-learning approaches that do not require training samples that are unavoidably biased.
We have performed multivariate statistical analyses of photometric and chemical abundance parameters of three large samples of stars in the globular cluster ω Centauri. The statistical analysis of a ...sample of 735 stars based on seven chemical abundances with the method of Maximum Parsimony (cladistics) yields the most promising results: seven groups are found, distributed along three branches with distinct chemical, spatial and kinematical properties. A progressive chemical evolution can be traced from one group to the next, but also within groups, suggestive of an inhomogeneous chemical enrichment of the initial interstellar matter. The adjustment of stellar evolution models shows that the groups with metallicities Fe/H > −1.5 are Helium enriched, thus presumably of second generation. The spatial concentration of the groups increases with chemical evolution, except for two groups, which stand out in their other properties as well. The amplitude of rotation decreases with chemical evolution, except for two of the three metal-rich groups, which rotate fastest, as predicted by recent hydrodynamical simulations. The properties of the groups are interpreted in terms of star formation in gas clouds of different origins. In conclusion, our multivariate analysis has shown that metallicity alone cannot segregate the different populations of ω Centauri.
We present the first unsupervised classification of spaxels in hyperspectral images of individual galaxies. Classes identify regions by spectral similarity and thus take all the information into ...account that is contained in the data cubes (spatial and spectral). We used Gaussian mixture models in a latent discriminant subspace to find clusters of spaxels. The spectra were corrected for small-scale motions within the galaxy based on emission lines with an automatic algorithm. Our data consist of two MUSE/VLT data cubes of JKB 18 and NGC 1068 and one NIRSpec/JWST data cube of NGC 4151. Our classes identify many regions that are most often easily interpreted. Most of the 11 classes that we find for JKB 18 are identified as photoionised by stars. Some of them are known regions, but we mapped them as extended, with gradients of ionisation intensities. One compact structure has not been reported before, and according to diagnostic diagrams, it might be a planetary nebula or a denser region. For NGC 1068, our 16 classes are of active galactic nucleus-type (AGN) or star-forming regions. Their spatial distribution corresponds perfectly to well-known structures such as spiral arms and a ring with giant molecular clouds. A subclassification in the nuclear region reveals several structures and gradients in the AGN spectra. Our unsupervised classification of the MUSE data of NGC 1068 helps visualise the complex interaction of the AGN and the jet with the interstellar medium in a single map. The centre of NGC 4151 is very complex, but our classes can easily be related to ionisation cones, the jet, or H$_2$ emission. We find a new elongated structure that is ionised by the AGN along the N-S axis perpendicular to the jet direction. It is rotated counterclockwise with respect to the axis of the H$_2$ emission. Our work shows that the unsupervised classification of spaxels takes full advantage of the richness of the information in the data cubes by presenting the spectral and spatial information in a combined and synthetic way.
Abstract
Early-type galaxies are characterized by many scaling relations. One of them, the so-called Fundamental Plane, is a relatively tight correlation between three variables and has resisted a ...clear physical understanding, despite many years of intensive research. Here, we show that the correlation between the three variables of the Fundamental Plane can be the artefact of the effect of another parameter influencing all, so that the Fundamental Plane may be understood as a confounding correlation. Indeed, the complexity of the physics of galaxies and of their evolution suggests that the main confounding parameter must be related to the level of diversification reached by the galaxies. Consequently, many scaling relations for galaxies are probably evolutionary correlations.
Active galactic nuclei (AGNs) display many energetic phenomena-broad emission lines, X-rays, relativistic jets, radio lobes-originating from matter falling onto a supermassive black hole. It is ...widely accepted that orientation effects play a major role in explaining the observational appearance of AGNs. Seen from certain directions, circum-nuclear dust clouds would block our view of the central powerhouse. Indirect evidence suggests that the dust clouds form a parsec-sized torus-shaped distribution. This explanation, however, remains unproved, as even the largest telescopes have not been able to resolve the dust structures. Here we report interferometric mid-infrared observations that spatially resolve these structures in the galaxy NGC 1068. The observations reveal warm (320 K) dust in a structure 2.1 parsec thick and 3.4 parsec in diameter, surrounding a smaller hot structure. As such a configuration of dust clouds would collapse in a time much shorter than the active phase of the AGN, this observation requires a continual input of kinetic energy to the cloud system from a source coexistent with the AGN.
Context. The diversification of galaxies is caused by transforming events such as accretion, interaction, or mergers. These explain the formation and evolution of galaxies, which can now be described ...by many observables. Multivariate analyses are the obvious tools to tackle the available datasets and understand the differences between different kinds of objects. However, depending on the method used, redundancies, incompatibilities, or subjective choices of the parameters can diminish the usefulness of these analyses. The behaviour of the available parameters should be analysed before any objective reduction in the dimensionality and any subsequent clustering analyses can be undertaken, especially in an evolutionary context. Aims. We study a sample of 424 early-type galaxies described by 25 parameters, 10 of which are Lick indices, to identify the most discriminant parameters and construct an evolutionary classification of these objects. Methods. Four independent statistical methods are used to investigate the discriminant properties of the observables and the partitioning of the 424 galaxies: principal component analysis, K-means cluster analysis, minimum contradiction analysis, and Cladistics. Results. The methods agree in terms of six parameters: central velocity dispersion, disc-to-bulge ratio, effective surface brightness, metallicity, and the line indices NaD and OIII. The partitioning found using these six parameters, when projected onto the fundamental plane, looks very similar to the partitioning obtained previously for a totally different sample and based only on the parameters of the fundamental plane. Two additional groups are identified here, and we are able to provide some more constraints on the assembly history of galaxies within each group thanks to the larger number of parameters. We also identify another “fundamental plane” with the absolute K magnitude, the linear diameter, and the Lick index Hβ. We confirm that the Mg b vs. velocity dispersion correlation is very probably an evolutionary correlation, in addition to several other scaling relations. Finally, combining the results of our two papers, we obtain a classification of galaxies that is based on the transforming processes that are at the origin of the different groups. Conclusions. By taking into account that galaxies are evolving complex objects and using appropriate tools, we are able to derive an explanatory classification of galaxies, based on the physical causes of the diverse properties of galaxies, as opposed to the descriptive classifications that are quite common in astrophysics.
We perform an evolutionary multivariate analysis of a sample of 54 Galactic globular clusters with high-quality colour–magnitude diagrams and well-determined ages. The four parameters adopted for the ...analysis are: metallicity, age, maximum temperature on the horizontal branch and absolute V magnitude. Our cladistic analysis breaks the sample into three novel groups. An a posteriori kinematical analysis puts groups 1 and 2 in the halo, and group 3 in the thick disc. The halo and disc clusters separately follow a luminosity–metallicity relation of much weaker slope than galaxies. This property is used to propose a new criterion for distinguishing halo and disc clusters. A comparison of the distinct properties of the two halo groups with those of Galactic halo field stars indicates that the clusters of group 1 originated in the inner halo, while those of group 2 formed in the outer halo of the Galaxy. The inner halo clusters were presumably initially the most massive one, which allowed the formation of more strongly helium-enriched second generation stars, thus explaining the presence of Cepheids and of very hot horizontal-branch stars exclusively in this group. We thus conclude that the ‘second parameter’ is linked to the environment in which globular clusters form, the inner halo favouring the formation of the most massive clusters which subsequently become more strongly self-enriched than their counterparts of the galactic outer halo and disc.
The fundamental plane of early-type galaxies is a rather tight three-parameter correlation discovered more than 20 yr ago. It has resisted both a global and precise physical interpretation despite a ...consequent number of works, observational, theoretical or using numerical simulations. It appears that its precise properties depend on the population of galaxies in study. Instead of selecting a priori these populations, we propose to objectively construct homologous populations from multivariate analyses. We have undertaken multivariate cluster and cladistic analyses of a sample of 56 low-redshift galaxy clusters containing 699 early-type galaxies, using four parameters: effective radius, velocity dispersion, surface brightness averaged over effective radius and Mg2 index. All our analyses are consistent with seven groups that define separate regions on the global fundamental plane, not across its thickness. In fact, each group shows its own fundamental plane, which is more loosely defined for less diversified groups. We conclude that the global fundamental plane is not a bent surface, but made of a collection of several groups characterizing several fundamental planes with different thicknesses and orientations in the parameter space. Our diversification scenario probably indicates that the level of diversity is linked to the number and the nature of transforming events and that the fundamental plane is the result of several transforming events. We also show that our classification, not the fundamental planes, is universal within our redshift range (0.007–0.053). We find that the three groups with the thinnest fundamental planes presumably formed through dissipative (wet) mergers. In one of them, this(ese) merger(s) must have been quite ancient because of the relatively low metallicity of its galaxies, Two of these groups have subsequently undergone dry mergers to increase their masses. In the k-space, the third one clearly occupies the region where bulges (of lenticular or spiral galaxies) lie and might also have formed through minor mergers and accretions. The two least diversified groups probably did not form by major mergers and must have been strongly affected by interactions, some of the gas in the objects of one of these groups having possibly been swept out. The interpretation, based on specific assembly histories of galaxies of our seven groups, shows that they are truly homologous. They were obtained directly from several observables, thus independently of any a priori classification. The diversification scenario relating these groups does not depend on models or numerical simulations, but is objectively provided by the cladistic analysis. Consequently, our classification is more easily compared to models and numerical simulations, and our work can be readily repeated with additional observables.
PDF
