談話会
Colloquium

Kazuyuki Sugimura (Tohoku U.)
Simulating the first galaxy formation
The first galaxies in the Universe are considered to form as low-mass galaxies at redshift z ~ 10 (~ 500 Myr after the Big Bang). While they are hard to observe with the current technology, next-generation observations, such as JWST and TMT, are expected to detect the first galaxies for the first time. Here, I will present the current status of our simulations of the first galaxy formation with a modified version of the public adaptive-mesh-refinement (AMR) cosmological radiation hydrodynamics code, RAMSES-RT. In the code, we treat the first-star formation, first-star-remnant black hole (BH) formation and growth, and low-metallicity stellar cluster formation using sub-grid models based on the latest understanding of these processes. In each run, we follow the cosmological formation of a first galaxy from z > 100 to z = 9. We find that bursty star formation occurs at the early stage of the first galaxy formation and that BHs do not significantly grow until the end of the simulation. Finally, I will discuss the physical mechanism that leads to burst star formation.

Yutaka Hirai (Tohoku U.)
Toward next-generation simulations of galaxy formation
Numerical simulation of galaxies is a powerful tool to study galaxy formation. Thanks to the advancement of computational facilities, the resolution of the simulations is greatly improved in the last two decades. Although galaxy formation simulations rely on sub-grid physics, wide ranges of observed properties have been reproduced. These simulations usually adopt the simple stellar population approximation for star particles and individual timestep schemes. Recently, the supercomputer Fugaku with over 7 million cores commenced operations. The mass resolution of galaxy formation simulations is expected to reach less than 10 solar mass by using such systems. However, models and codes usually used in previous galaxy formation simulations cannot be applied to simulations with this resolution. In this talk, I will present the design of our new model and code for extremely high-resolution galaxy formation simulations. In our simulations, we stochastically assign the stellar mass to each star particle. I will show that the expected IMF is correctly sampled if we set the maximum search radius appropriately. For efficient computation, we have developed ASURA-FDPS code. In this code, the region around supernovae, which usually has short timesteps, is computed separately to reduce global communication among computational nodes. The supernova region is automatically selected based on deep learning prediction. We have tested this code using the full node (7 million cores) of the supercomputer Fugaku using one trillion particles. The performance was 31 Pflops, suggesting that the cosmological hydrodynamics simulations with 100 billion particles can be performed within three months by using 10% of the Fugaku system. Such high-resolution galaxy formation simulations will open up a new window to comprehensively understand galaxy formation from the scale of ultrafaint dwarf galaxies to the Milky Way.

Satoshi Ohashi (RIKEN)
Grain growth and ring formation in early stage of the disk formation around protostars
Recent ALMA observations have revealed that a variety of disk structures have already formed in young protostellar disks. These observations allow us to consider the planet formation much earlier than classical theories. It is important to investigate the formation mechanism of such structures in the protostellar disks and the relation to the planet formation. In this talk, I will introduce one of the possible ring formation scenarios, which is coagulation of dust grains in protostellar disks. The dust grains are supposed to grow by coagulation and lead to planetesimals, which occurs in an inside-out manner because the growth timescale is roughly proportional to the orbital period. The boundary of the dust evolution can be regarded as a growth front, where the growth time is comparable to the disk age. With radiative transfer calculations based on the dust coagulation model, we found that the growth front can be observed as a ring structure. I will show compare the current observations and model, and discuss this ring formation scenario.