“From Nbody1 to Nbody6 : The Growth of an Industry” is the title of a 1999 invited review by Sverre Aarseth, for Publications of the Astronomical Society of the Pacific (PASP). I took this as an inspiration for the title of this paper; it describes how Sverre’s Nbody Industry has further grown since 90s of the previous century, and how it is further flourishing and hopefully developing, in his spirit, even after the sad news of his passing away reached us. My contact and friendship with Sverre started a few decades ago being sent to Cambridge to learn Nbody5, counting input parameters, and learning about the fact that even a sophisticated code (which had already at that time quite a history) requires permanent maintenance and bug fixes. Managed by Sverre, who relentlessly ran his code and responded to the widely spread crowd of “customer” colleagues. There has been a phase of massive and fast development and improvements due to vectorization, parallelization, GRAPE and GPU acceleration, and Sverre has been always on top of it if not ahead, but also fully adopting ideas of collaborators, once they tested well. Nbody6++GPU and Nbody7 entered the scene, and also recent new competitors, such as PeTar or BiFrost. We all have learnt a lot from Sverre, and strive to continue in his open-minded spirit, for open source and exchange. A striking evidence for the further growth of the “industry” is the number of papers here (and two of them follow in this session, but also in other sessions) using and further developing the aforementioned codes, as well as the occurrence of new and competing codes, which keep the field alive.

The solar convective envelope generates, by the dynamo effect, a surface magnetic field whose strength evolves on an 11-year cycle, with a change in polarity at the end of each cycle. A similar magnetic variability exists in other solar-type stars, influencing their dynamics. Furthermore, solar-like oscillators exhibit acoustic modes whose properties, such as their frequency, amplitude, and energy, vary over time in relation to the activity cycles. The turbulent motions in the convective envelope of these stars have so far been considered the only mechanism for exciting the modes. In this work, we investigate the variation of mode excitation during Cycles 23, 24, and the beginning of Cycle 25 for the Sun. To do so, we analyze data obtained since 1996 by the VIRGO SunPhotometers on the SoHO satellite, using a method that provides a better time resolution than classical methods such as spectrum fitting. By combining the small-time-scale variations in energy for several low-degree modes, we found a statistical discrepancy in the observed excitation rate compared to the one expected under the hypothesis that modes are only stochastically excited by convection. Our results indeed show that several modes can be excited simultaneously. During this presentation, we explored the possible sources of high energy in the modes, i.e., instrumental problems or other exciting mechanisms, which may be linked to the magnetic cycle of the star, such as flares or Coronal Mass Ejections, and compared the results with data from the GOLF spectrometer, also on board SoHO.

Astronomical image deconvolution has evolved from early linear regularization to modern deep learning approaches. This article provides a concise overview of the historical milestones, methodological breakthroughs, and current challenges in the field, with emphasis on sparsity, compressed sensing, and the role of deep learning in large astronomical surveys.

Globular clusters are a prototype of old, dense stellar systems that host diverse compact objects and binaries, including black holes and neutron stars. There are about 200 globular clusters in the Milky Way Galaxy, but many more beyond the Milky Way. They are divided into two groups according to their location: galactic globular clusters and intracluster globular clusters. On the small scale the globular clusters are an excellent laboratory to study 1) the nature of compact objects and binaries, 2) the dynamics of dense stellar systems, and 3) stellar evolution. On the larger scale, they are an efficient tool to investigate formation and evolution of galaxies and galaxy clusters. I present an overview of the current understanding of globular clusters in the universe: from the globular clusters in the Milky Way Galaxy to the young massive star clusters in high-redshift galaxies, and from galactic globular clusters to intracluster globular clusters.

Weak gravitational lensing of background galaxies traces the growth of dark matter and is key to constraining the cosmological model. This proceedings paper provides an overview of the analysis presented in Voigt (2024), applying convolutional neural networks (CNNs) to the challenge of precision shear estimation. Using a shallow CNN, we investigate the sensitivity of shear recovery to point spread function (PSF) misestimation, with reference to the accuracy requirements of Euclid. To mitigate noise bias we employ a non-standard loss function, and assess performance assuming a linear relation between predicted and true shears.

Helio- and asteroseismic observations show that stellar cores rotate much more slowly than predicted by hydrodynamical standard rotating models, revealing the need for additional angular momentum transport processes. Internal gravity waves (IGWs) are a strong candidate for this transport. We extend their classical treatment by including the effect of the Coriolis acceleration in the damping of waves, entering the gravito-inertial regime. The new formalism, implemented in the STAREVOL code, is tested on a $1M_{\odot}$ pre-main and main-sequence model. Rotation modifies the spatial distribution of wave dissipation, moving the damping region closer to the excitation zone. This represents a first step toward a consistent treatment of gravito-inertial waves (GIWs) and an understanding of their role in the rotational evolution of low-mass stars.

Photometric redshift estimation (photo-z) offers a scalable alternative to spectroscopic measurements, at the expense of reduced accuracy and increased uncertainty. In this work, we investigate machine learning approaches that explicitly incorporate uncertainty quantification into photo-z estimation using Sloan Digital Sky Survey (SDSS) data. Starting from a baseline neural network, we extend to Gaussian regression, quantile regression, and conformalized quantile regression (CQR). Our experiments demonstrate that, while the baseline model achieves competitive accuracy, uncertainty-aware methods provide more reliable and interpretable predictions.

Finding low-mass planets around solar-type stars requires understanding the physical variability of the host star, which greatly exceeds the planet-induced radial-velocity modulation. This project aims to analyze – observationally and theoretically – the character and physical origins of fluctuations in solar photospheric absorption lines.

Observationally, photospheric equivalent-width variations were measured in 1000 selected spectra from three years of HARPS-N data of the Sun-as-a-star, showing changes that largely shadow the chromospheric Ca II H&K activity-cycle signal, but with much smaller amplitudes on sub-percent levels. Among iron lines, the greatest are for Fe II in the blue, while the trends change sign among lines in the green Mg I triplet and between Balmer lines. No variation was seen in the semi-forbidden Mg I 457.1 nm.

Theoretically, hydrodynamic 3D modeling of solar surface convection produces time sequences of synthetic high-resolution spectral atlases. Radial velocities averaged over small simulation areas jitter by some $\pm$150 m s$^{-1}$, scaling to $\sim$$\pm$2 m s$^{-1}$ for the full solar disk on timescales of granular convection. Among different lines, jittering is in phase, but the amplitudes differ by about one tenth of their values: greater for stronger and ionized lines, decreasing at longer wavelengths

The extended Main Sequence Turn-off (eMSTO) in the open cluster NGC 2355 is investigated using precise astrometry and photometry from Gaia DR3 and spectroscopic data from the Gaia-ESO Survey. We find a clear positive correlation between the rotational velocity ($v \sin i$) and color of eMSTO stars, supporting the role of stellar rotation and gravity darkening in causing the observed color spread. Contrary to predictions from binary tidal-locking scenarios, spatial distribution analysis reveals that fast-rotating stars are preferentially concentrated in the cluster’s central regions. This finding provides the observational evidence favoring star-disk interactions during the pre-main-sequence phase, rather than tidal locking in binaries, as the primary mechanism responsible for the origin of rotational spread in eMSTO stars.

Machine Learning applications have gained significant interest in the solar physics community for solving previously intractable problems and accelerating computationally demanding tasks that could otherwise require years to complete. This work presents a novel approach for spectropolarimetric inversions using machine learning models, specifically neural networks, trained with classical supervised methodology enhanced by a physical constraint in the loss function based on the WFA approximation. The results show that a weight of w=0.01 for the physical constraint term yields superior performance in some cases compared to unconstrained models when evaluating metrics such as RRMSE and correlation across different atmospheric depths on synthetic test data. When applied to observational data, although the model struggles with thermodynamic and velocity parameters, the magnetic field generation in the line-of-sight—which relates to the WFA constraint—shows improved generalization when the constraint is applied. These results suggest a promising new approach for developing supervised machine learning models that incorporate physical constraints, guiding them toward a deeper “understanding” of the underlying physics in the analyzed phenomena.

The intracluster medium (ICM) records the history of galaxy clusters through its complex dynamical properties. To effectively interpret these properties, robust methods are needed to compare observational data with theoretical models. We present a novel machine learning framework for comparing ICM line-of-sight velocity maps derived from X-ray observations. Our approach uses convolutional and Siamese neural networks to identify similarities between different kinematic fields. We outline the architecture of this framework and perform a series of sanity checks to validate its performance. These checks demonstrate the model’s ability to correctly identify and quantify kinematic features, establishing a powerful new tool for future comparative studies of the ICM.

We present a study of eight Galactic open clusters, some with an age comparable to those of globular clusters yet harboring thousands of members. The sample challenges the conventional view that such systems need to survive constant disk perturbation and are expected to dissociate by 1 Gyr. Using Gaia DR3, we determined a reliable member list for each cluster, with which the age is derived by fitting to stellar evolution models. These clusters exhibit advanced positional and kinematic mass segregation. Moreover, the mass function for each cluster, after a “false-positive” sample is considered, shows a deficit of low-mass members, signifying stellar evaporation. A possible explanation for the survival of these clusters is their large galactocentric distance and orbital isolation from disk perturbations.

Solar Spectral Irradiance (SSI) provides information about various properties of the solar atmosphere, including physical, chemical, thermal, and dynamic aspects. However, its importance extends beyond this, as solar emission in EUV wavelengths is the primary source of energy input to planetary atmospheres, significantly impacting their structure, composition, and dynamics, particularly in the Ionosphere-Thermosphere-Mesosphere (ITM) system. To study the impact of SSI on planetary atmospheres, it is important to provide long and continuous SSI time series. In the case of Earth, this can be used in ITM models to understand their impact. Since the photospheric magnetic flux drives the solar irradiance variability, it is possible to use the magnetic field to model the Solar Spectral Irradiance (SSI), e.g., the COronal DEnsity and Temperature (CODET) model. It is a physics-based model that uses the relationship between the magnetic field, density, temperature, and EUV emission. This work shows the most recent results and applications, e.g., new SSI time series in Earth’s exosphere and at Mars’ distances. New perspectives and improvements are also discussed.

One of the significant problems in cosmology is the morphological classification of galaxies in order to gain insights about the origin and evolution of our Cosmos. Here, we investigate the classification accuracy of several modern convolutional neural network architectures (some of them pretrained on non-astronomical data and others trained from scratch to galaxy data), and of transformers with primary goal to understand their performance on conditions of fast training (i.e., in a few rounds). We experimented with Transfer Learning, and Vision Transformers (ViT). We contrasted their performance with that of traditional Convolutional Neural Networks (CNN) methods, i.e., AlexNet, and VGG. Our results show a 75% classification accuracy (by the ViT and VGG), beating the rest of the competitors. Moreover, one pretrained VGG model showed also excellent performance.

This research presents unsupervised machine learning and statistical techniques to detect and analyze tidal tails in five open clusters, BH 164, Alessi 2, NGC 2281, NGC 2354, and M67, with ages ranging from 60 Myr to 4 Gyr, using Gaia DR3 data. Color-magnitude diagram (CMD) matching, principal component analysis (PCA), and density-based clustering were used to detect the tails. Tidal tails were found in all clusters, extending 40 to 100 pc and containing 100 to 200 stars. Other statistical methods were used to study the stellar kinematics, photometry, and spatial distribution of the detected features. The tails had luminosity functions similar to their parent clusters, generally lacked massive stars, and showed higher binary fractions. Significant rotation was detected in M67 and NGC 2281 for the first time.

Recent advances in high-resolution solar observations have led us into a transformative era for understanding the complex interplay of magnetic fields and energy release processes in the solar atmosphere. Building upon the contributions of missions such as SDO, IRIS, and Hinode, we combine ground-based spectro-polarimetric measurements from SST and GREGOR with space-borne data. This coordinated approach enables us to track the evolution of magnetic fields from the photosphere to the corona and to investigate small- to intermediate-scale events—driven by processes such as magnetic reconnection and flux cancelation—that contribute to atmospheric heating and possibly to solar wind acceleration.

In this context, we explore events observed by IRIS in the ultraviolet that caused an increase in Si IV and O IV line intensity (by factors of 100 to 1000) and significant line broadening (up to $\pm 150 \,\mathrm{km \,s}^{-1}$), indicating bi-directional plasma flows from magnetic reconnection. SDO/AIA imaging reveals bright counterparts across all channels, underscoring the upper atmospheric response. High-resolution spectro-polarimetry in the photosphere and chromosphere details the magnetic configuration during flux emergence and decay. Our findings demonstrate the vital role of multi-instrument campaigns in addressing key challenges in solar and stellar physics and pave the way for future studies with advanced facilities like IBIS2.0, EST, and the Solar-C and MUSE satellites.

Efforts are underway to measure the global 21 cm signal from neutral hydrogen, which is a powerful probe of the early universe, using NASA radio telescopes on the far side of the Moon. Physics-based models of the signal are computationally expensive to perform Bayesian multi-parameter inferences, for which we have developed novel, publicly-available neural network emulators utilizing a Long Short-Term Memory (LSTM) network and a Kolmogorov-Arnold Network (KAN). <sc>21cmLSTM</sc> is currently the most accurate emulator in the community by leveraging the signal’s temporally-correlated structure, and <sc>21cmKAN</sc> maintains similar accuracy while training 75 times faster, by learning expressive functional transformations. Each emulator can fit realistic mock signals and obtain unbiased physical parameter constraints, with <sc>21cmKAN</sc> able to complete end-to-end training and inference in under 30 minutes. The implementation of machine learning tools like these in data analysis pipelines is important to fully exploit upcoming measurements of the cosmological 21 cm signal.