Colloquiums and Seminars

Cosmological hydrodynamical simulations are essential for understanding the interplay between dark matter and baryons, yet they remain computationally expensive and struggle to fully reproduce observational constraints. At the same time, tensions in key cosmological parameters, such as σ₈, raise the question of whether new fundamental physics or baryonic feedback effects are responsible. Recent advances in machine learning and differentiable modeling offer new approaches to improving these simulations and connecting them to observations, from GPU acceleration to field-level inference techniques that bypass traditional summary statistics. In this talk, I will explore how integrating ML and related optimization techniques with hydrodynamical simulations can enhance our ability to extract cosmological information, discuss the challenges and limitations of these methods, and outline the path toward a more efficient and robust framework for large-scale structure inference.

Over the past decade, advanced wide-field and multi-messenger transient surveys have revolutionized our ability to watch stellar explosions in real time. Coupled with rapid follow-up observations and extensive numerical models, I will highlight recent breakthroughs — such as the first robust electron-capture supernova at the boundary between white dwarf formation and iron core collapse, mysterious yet ubiquitous mass-loss activity in the final stages of massive stars, and kilonova emission from a binary neutron star merger detected in gravitational waves — that reshape our understanding of stellar evolution and explosion physics. I will also discuss their wide-ranging implications, including nucleosynthesis, galactic chemical evolution, and compact object populations. Finally, I will conclude with the promises in the coming golden decade of time-domain astronomy with the Vera C. Rubin Observatory’s LSST, Nancy Grace Roman Space Telescope, LIGO-Virgo-KAGRA’s fifth observing run, and the Bustling Universe Radio Survey Telescope in Taiwan.

XRISM (X-Ray Imaging and Spectroscopy Mission) is an X-ray observatory led by JAXA and NASA. XRISM carries two X-ray observing systems, called Resolve and Xtend. Resolve consists of an X-ray mirror and an X-ray microcalorimeter, while Xtend combines an X-ray mirror with an X-ray CCD. Resolve has high energy resolution of 5 eV at 6 keV X-ray photons, and it is effective even for spatially extended objects since it is a non-dispersive spectrometer. Xtend has a wide field of view of more than 900 arcmin2 (larger than the full moon!). XRISM was successfully launched in September 2023. However, due to a gate valve in front of the microcalorimeter failing to open, Resolve is unfortunately unable to observe X-ray photons below 2 keV. Nevertheless, XRISM observations are yielding exciting results. In this talk, I will present some of the recent results from XRISM.

The early growth of black holes (BHs) in atomic cooling halos is likely influenced by feedback on the surrounding gas. While the effects of radiative feedback are well-documented, mechanical feedback, particularly from AGN jets, has been comparatively less explored. Building on our previous work that examined the growth of a 100-solar-mass black hole in a constant density environment regulated by AGN jets, we have expanded the black hole mass range from 1 to 10,000 solar masses and adopted a more realistic density profile for atomic cooling halos. We provide an analytic models for jet cocoon propagation and feedback regulation. We also identify several critical radii—namely, the terminal radius of jet cocoon propagation, the isotropization radius of the jet cocoon, and the core radius of the atomic cooling halo—that are crucial in determining black hole growth, given specific gas properties and jet feedback models. In a significant portion of the parameter space, our findings show that jet feedback substantially disrupts the halo-core gas density during the initial feedback episode, halting black hole growth beyond 10,000 solar masses. Conversely, conditions characterized by low jet velocities and high gas densities enable sustained black hole growth over extended periods. We have identified the parameter space that allows a stellar-mass black hole to grow into a supermassive black hole at high redshift by accreting gas from an atomic cooling halo.

Detailed studies of our Local Group benchmark our understanding of galaxy formation. Stellar metallicities are key tracers of the baryonic astrophysics shaping galactic properties, but they are challenging to measure in distant and faint galaxies that are pushing push our understanding of dwarf galaxy formation to new regimes in luminosity, star formation history, and environment. For my thesis, I present hundreds of new stellar metallicities in faint, Local Group dwarf galaxies, measured through a novel use of HST narrowband Ca H&K imaging. Our imaging includes: 1) 463 stars in 13 ultra-faint dwarf galaxies (UFDs) around the Milky Way, which effectively doubles the number of stellar metallicities in all known UFDs, b) 374 stellar metallicities in the quenched field dwarf galaxy Tucana (Mv = -8.8, D = 1 Mpc), a factor of ~7 increase over literature spectroscopy, and c) 286 stellar metallicities in two M31 dwarfs And XVI and And XXVIII. I will present highlights from the wide range of science cases enabled by our data, which include: 1) chemical evolution modeling to put novel constraints on the baryon cycle in UFDs, 2) new metallicity benchmarks for cosmological simulations of the faintest galaxies, 3) high-fidelity metallicity gradients that constrain stellar feedback and DM core formation models in dwarf galaxies. I conclude with a discussion on the immense scientific potential for narrowband Ca H&K imaging to transform stellar metallicity studies at the edge of the LG and beyond.

Recent studies have proposed that the nuclear millimeter continuum emission observed in nearby active galactic nuclei (AGN) could be created by the same population of electrons that gives rise to the X-ray emission that is ubiquitously observed in accreting black holes. In my talk I will present the results of several dedicated high spatial resolution (~60-100 milliarcsecond) 100 GHz ALMA campaigns focussed on nearby radio-quiet AGN. We find an extremely high detection rate (~95%), which shows that nuclear emission at mm-wavelengths is nearly ubiquitous in accreting SMBHs. Our high-resolution observations show a tight correlation between the nuclear (1-23 pc) 100GHz and the intrinsic X-ray emission. This shows the potential of ALMA continuum observations to detect heavily obscured AGN (up to an optical depth of one at 100GHz, i.e. ~1e27 cm^-2), and to identify binary SMBHs with separations <100 pc, which cannot be probed by current X-ray facilities.

The initial mass function (IMF) is crucial for our understanding of star and galaxy formation and evolution, while our knowledge of the IMF in the early universe remains limited. Recent advancements in numerical simulations have revealed that stellar masses in primordial environments are significantly larger, reaching several hundred solar masses, compared to observations in present-day environments. This suggests the existence of an IMF transition throughout cosmic evolution. Through three-dimensional hydrodynamics simulations, we have discovered that both metallicity and redshift play crucial roles in determining the IMF. Metallicity influences the cooling capability, while the redshift affects the temperature floors by altering the cosmic microwave background temperature. Our simulations have revealed that the IMF becomes top-heavy in environments with Z/Zsun < 0.01 or z > 10. These results may provide an explanation for recent observations by the James Webb Space Telescope (JWST), which have revealed an unexpected abundance of high-redshift luminous galaxies, showing an increase in UV luminosity at fixed stellar mass.

The formation of supermassive black holes (SMBHs) is one of the biggest challenges in astrophysics. The Direct Collapse (DC) model provides seed BHs with a mass of 10^5 Msun and explains the origin of high-redshift SMBHs. It has been assumed that these seed BHs can only form in primordial gas. However, this condition is very restrictive, resulting in a low number density of seed BHs. Consequently, the DC model cannot account for the entire population of SMBHs. In this seminar, we investigate whether massive seed BHs can form in a metal-enriched universe, which is more prevalent. By conducting hydrodynamical simulations across various metallicity enrichments, we found that stars with masses greater than 10^4 - 10^5 Msun can form when Z/Z_sun <= 10^-3. In finite metallicity cases, dust cooling promotes low-mass star formation. However, the gas preferentially feeds the massive stars, similar to the primordial case, resulting in the formation of supermassive stars. This model offers a larger number of seed BHs compared to the previous DC model, providing a universal explanation for the origin of SMBHs in terms of their abundance.

Cold molecular gas is a key component in galaxy evolution, as it forms stars, bears feedbacks, and feeds supermassive blackholes. Interferometric observaions of the Atacama Large Millimeter/submillimeter Array (ALMA) have remarkably advanced this field in the past decade. For spiral galaxies, a larger sample with higher physical resolution than before is systematically surveyed. I will discuss the molecular gas morphology and kinematics in three megamaser (Seyfert-II) galaxies at resolutions of around 100 pc. We found prevelant irregularities, potentially related to active galactic nucleus feedback and supermassive blackhole feeding. For early-type galaxies, it is now feasible to spatially resolve giant molecular clouds (GMCs). I will talk about GMCs at 15-pc resolution in the lenticular galaxy, NGC1387. Their dynamical states (Larson relations, virial parameters, etc.) are surprisingly similar to those in spiral galaxies. For our own Milky Way centre, a new ALMA large programme has mapped the central molecular zone (CMZ) at unprecedented spatial (sub-pc) and spectral (0.2 km/s) resolutions. Here, we found evidence of galactic shear effects and magnetic fields driving gas structure morphologies. I will conclude by summarising the physical drivers of molecular gas properties at different scales and in different environments.

Nearby low-metallicity dwarf galaxies are excellent laboratories to study astrophysical processes in the first galaxies at high redshift. I will present the initial results of our multiwavelength study of the dwarf galaxy KUG 1138 + 327 at 24.5 Mpc and its group environment. This galaxy shows an unusual tadpole morphology as a result of a very young starburst with extremely low (6\% solar) metallicity. We obtained Chandra X-ray observations as well as a JVLA continuum and HI survey of the region in several array configurations. The Chandra data reveal a dominant point-like source with an average 0.3-10 keV luminosity of $10^{40.3}$ erg/s and a variability by a factor of $\sim 2$ over months. This extremely luminous X-ray source is apparently associated with the young central cluster of the starburst and has an elongated nonthermal radio continuum counterpart of the order of $\sim 200$~pc. The radio, optical, and X-ray results suggest that the X-ray source could well be an intermediate-mass black hole undergoing sub-Eddington accretion. This scenario also explains the prominent emission lines [He\texttt{II}]$\lambda$4658 and [Ar\texttt{IV}]$\lambda$4711 in the starburst spectrum. These findings provide insights into questions such as: What is the radiation environment produced by the first generation of stars? What are the feedback processes in high-z star-forming regions at very low metallicity? We also find that KUG 1138+327 is associated with two additional far-UV bright dwarf galaxies, a rare phenomenon in the local Universe. The HI morphology of these galaxies strongly suggests that they have undergone tidal interactions, which may explain the recent low-metallicity gas feeding and starburst in KUG 1138+327.

The interstellar medium (ISM) is highly structured and multiphase, and our knowledge of it stems from multi-wavelength radiation dictated by ISM chemistry. Hydrodynamical simulations coupled with chemistry networks have been our primary tools for predicting these observables. However, interstellar dust, the catalyst of ISM chemistry, is commonly treated as a non-evolving species. In reality, dust is constantly created, destroyed, and reformed, with significant spatial variations in its abundance. In this talk, I will discuss recent progress in resolved galaxy-scale simulations of low-metallicity dwarf galaxies coupled with a dust-evolving chemistry network where dust evolution is followed explicitly rather than treated in a sub-grid fashion. I will demonstrate how dust evolution helps explain the observed CO luminosity at low metallicity, which has important implications for galaxies in the early universe observed by JWST. I will also discuss the survival of dust in galactic outflows and its role in the observed molecular outflows, whose origin remains a puzzle.

Galaxy survey analyses often infer lower values of two cosmological parameters, omega_matter and sigma_8, compared to the constraints from the cosmic microwave background. Is that a hint of new physics? Before any such conclusion we need to robustly probe this tension using novel and independent techniques. Moreover, ongoing and upcoming surveys of galaxies will contain an immense wealth of information in the smallest scales, which is under-utilised in most cosmological analyses. Vast amounts of information is often discarded because the smallest scales in survey data are most affected by systematics such as baryonic feedback, and because of the inherent difficulty involved in modelling the fully nonlinear regime. Because of these two key motivations, we developed Basilisk, a new Bayesian hierarchical forward-modelling technique to optimally extract maximum information from the smallest and fully non-linear scales of redshift survey data. It uses that complementary information to constrain galaxy-halo connection with extreme precision, and to constrain cosmological parameters in a completely novel approach that is free of halo assembly bias systematics and incorporates baryonic effects. I will discuss the methodology behind the new probe, the key results with SDSS data, and what can be achieved with upcoming galaxy surveys.

Upcoming spectroscopic surveys like PFS and DESI are beginning to release their first data, revealing an unprecedented three-dimensional view of the cosmic large-scale structure through the spatial distribution of (billions of) galaxies. Such surveys hold the potential to unveil key insights into the nature of dark matter, dark energy, and gravity—but fully realizing this potential requires extracting the maximal (reliable) information from galaxy maps. Traditional analyses compress such maps into summary statistics, while uncertainties in galaxy formation and evolution further complicate interpretation.
I will introduce the field-level inference program, which directly models the full galaxy maps on scales where astrophysical complexities can be robustly marginalized over. I will also discuss ongoing efforts to apply FLI to DESI and PFS data, including developments to account for observational systematics. If time allows, I will highlight how FLI can contribute to studies of galaxy formation and evolution through cross-correlation analyses.

Turbulence in protoplanetary disks affects dust evolution and planetesimal formation. The vertical shear instability (VSI) is one of the candidate turbulence-driving mechanisms in the outer part of the disks. Since the VSI requires rapid gas cooling, dust particles in disks can influence and potentially control VSI-driven turbulence. However, VSI-driven turbulence has strong vertical motion, causing vertical diffusion of dust particles. We perform global two-dimensional hydrodynamical simulations of an axisymmetric protoplanetary disk to investigate how the VSI drives turbulence and maintains a balance between dust settling and diffusion. These simulations account for the dynamic interplay between dust distribution, cooling rates, and VSI-driven turbulence. We find that VSI mixing, dust settling, and the local dust cooling reach an equilibrium, forming a thick dust layer with a dimensionless vertical mixing coefficient of approximately $10^{-3}$. The ability of VSI to sustain this equilibrium depends on dust size and dust-to-gas mass ratio. Larger grains or lower mass ratios weaken the turbulence, leading to dust settling. Our results suggest that in VSI-dominated disks, dust grows under turbulence with intensity varying by dust size. Finally, I will talk about the application of the early disk stage and further planet formation.

The presence of supermassive black holes (SMBHs) in galaxies is well known. But why most of them remain silent in today's Universe is poorly understood. Sgr A* at the center of our Milky Way Galaxy is an asymptotic example of such a low-luminosity SMBH. The proximity of Sgr A* provides a unique opportunity to observe and understand the dynamics of black hole activity and its interplay with its Galactic nuclear environment. Based primarily on deep X-ray observations and computer simulations, I will discuss what Sgr A* has been doing recently and how this interplay may have determined the life cycle of black hole activity and other galactic nuclear processes that profoundly affect the structure and evolution of our Galaxy. Such studies, complemented by observations of other nearby SMBHs and their environments, provide insights into the functioning of galactic ecosystems and astrophysical processes under extreme conditions.

1. Reference HOPS Calibration Pipeline for millimetre-VLBI Arrays
We present a new version of the EHT-HOPS calibration pipeline (Blackburn et al. 2019), designed to process data from the evolving EHT array, for handling both current and future configurations. This updated pipeline supports the calibration of multiple years of EHT data and similar VLBI arrays, accommodating changes in instrumentation, frequency bands, and array configurations (with and without ALMA). The refactored pipeline improves performance, code integration, and usability, features updated diagnostics, and supports hybrid polarization bases, tested using 2017 M87 data. Planned future developments include complex bandpass and full polarization calibration, parallel processing, and integration with new data formats to generate standardized calibrated data products for all EHT campaigns.

2. Interpol: A Polarized Imaging and Calibration Code for Heterogeneous VLBI Arrays
Very Long Baseline Interferometry (VLBI) has paved the way for groundbreaking astrophysical discoveries, such as black hole jets, planet formation, and imaging of the shadow of a black hole. However, transforming the raw VLBI data into an interpretable image is a formidable challenge. VLBI imaging requires modeling an entire image from a sparse sampling of Fourier components and the complicated instrument response to the incident electric field. While the instrument response and image are traditionally analyzed separately, distinguishing instrumental effects from source structure is complex and non-linear for heterogeneous VLBI arrays such as the EHT or GMVA. In this talk, I will present Interpol, a novel Bayesian simultaneous imaging and calibration algorithm implemented in the VLBI software Comrade. Interpol can reconstruct the full Stokes polarized image and instrumental response for circular, linear, and mixed feed arrays. We demonstrate Interpol on various VLBI arrays, including the VLBA, Event Horizon Telescope, and VLBA sources, demonstrating improved image resolution with fewer image artifacts and discuss its place relative to Polconvert, a common tool used to convert ALMA data from linear to a circular feed basis.

I will discuss a perspective on counting and locating the majority of our Universe's baryons from the scant signatures imprinted on the light and signals of distant sources (quasars and fast radio bursts). These data offer what I consider the most robust measurement of a cosmological parameter (the baryonic mass density) and I'll describe previous and ongoing efforts to establish where the majority of this matter resides. If time permits, I'll briefly detail my new adventure to measure – again in silhouette – the constituents of our Earth's oceans (e.g. phytoplankton).

Transneptunian Automated Occultation Survey (TAOS II) is a blind occultation survey with the primary goal to measure the size distribution of small (~1 km diameter) Trans-Neptunian Objects (TNOs). Such events are extremely rare (< 0.001 per star per year) and very short in duration (~200 ms). To enable serendipitous discovery many stars must be monitored simultaneously at high cadence. TAOS II will monitor 2,000 to 10,000 stars simultaneously at a cadence of 20 Hz using three 1.3 meter F/4 telescopes operating at San Pedro Mártir Observatory (SPM) in Baja California, Mexico. Over five years of operations, the TAOS II dataset will collect ~3 PB of multi-telescope 20 Hz lightcurve data for hundreds of thousands of objects. This unique dataset will provide ample opportunity for science outside of the primary goal to measure the TNO size distribution. TAOS II will achieve first light in Spring 2025. The current status and the near future plan will be presented in this talk.

Stable isotope measurements of natural samples over the past 80 years have been applied to numerous systems across space and time. Oxygen isotopes for example are used in polar ice and oceanic foraminifera to determine global temperatures from yearly to millions of year time scales. Atmospheric greenhouse gases stable isotope measurements are used to track sources and sinks of molecules and to define their chemistry. Meteorite isotope measurements identify pre solar system species such as grains of stardust to define their stellar sources and also to resolve the formation of the solar system chemical processes These applications require high level physical chemical understanding of the processes. Of note are those involved with photodissociation. Most recently we have partnered with Raphy Levines theory group in Jerusalem and made synchrotron measurements to develop the most complete isotopic theory to date, which will be presented.

Observational constraints on gas surface density in protoplanetary disks are crucial for understanding planet formation. However, directly measuring it is challenging due to the lack of a robust tracer, leading to uncertainties of orders of magnitude even in total disk mass estimates. To address this issue, we propose using the pressure broadening of CO emission lines. Since pressure-broadened CO lines are sensitive to hydrogen volume density, they enable a direct determination of the total gas surface density. We applied this method to three transition disks and successfully constrained their gas surface density profiles for the first time. By comparing these profiles with dust surface density distributions, we directly demonstrate that dust grains are trapped at gas pressure maxima in these disks. Additionally, our estimates of total gas mass exceed the minimum-mass solar nebula, suggesting that further planet formation remains quantitatively feasible.

We derive an effective equation and action for the propagation of gravitational waves (GW), encoding the effects of interaction and self-interaction in a time, frequency and polarization dependent effective speed.This effective approach generalizes the effective theory of dark energy and allows to make model independent predictions of observable quantities. We discuss how the frequency and polarization dependence of the GW-EM distance ratio and time delay provide a new test of general relativity and its modifications, and more in general of the interaction of GWs with other fields. As an application, consistent with the effective theory of dark energy, we show that for a luminal modified gravity theory, the gravitational-wave propagation and luminosity distance are the same as in general relativity, but depending on the matter gravity coupling, the electromagnetic luminosity distance can be modified w.r.t general relativity.