(This is a joint seminar with the Information Theory Study Group.)

Collective motion is a fundamental phenomenon observed in various biological systems, characterized by the coordinated movement of individual entities. Such dynamics are especially crucial in understanding cellular behaviors, which can now be observed at an individual level in complex tissue formations involving multiple types of cells, thanks to recent advancements in imaging technology. To harness this rich data and uncover the hidden mechanisms of such dynamics, we developed a deep learning framework that estimates equations of motion from observed trajectories. By integrating graph neural networks with neural differential equations, our framework effectively predicts the two-body interactions as a function of the states of the interacting entities.
In this seminar, I will first introduce the structure and hyperparameters of our framework. Subsequently, I will detail two numerical experiments. The first is a simple toy model that was employed to generate data for testing our framework to refine the hyperparameters. The second explores a more complex scenario mimicking the collective motion of cellular slime molds, highlighting our model's ability to adapt to mixed-species interactions.

Galaxy simulations have found the interdependence of multiscale gas physics, such as star formation, stellar feedback, inflow/outflow, and so on, by improving the physical models and resolution. The mass resolution remains capped at around 1,000 solar masses (e.g., Applebaum et al. 2021). To overcome the limitations, we are developing a new N-body/SPH code, ASURA-FDPS, to leverage exascale computing (e.g., Fugaku), handle approximately one billion particles, and simulate individual stars and stellar feedback within the galaxy. However, the emergence of communication costs hinders scalability beyond one thousand CPU cores. One of the causes is short timescale events localized in tiny regions, such as supernova explosions. In response, we have developed a surrogate model using machine learning to duplicate supernova feedback quickly (Hirashima et al., 2023a,b). In the presentation, I report the fidelity and progress of the simulations with our new machine-learning technique.

(The speaker is also the team leader of Laboratory for Medical Science Mathematics, RIKEN Center for Integrative Medical Sciences. This is a joint seminar with the iTHEMS Biology Group.)

In medical science, the recent explosive development of omics technologies has enabled the measurement not only of bulk data from entire tissues, but also data for individual cells and their spatial location information, and even allowed collection of such information in real-time. Meaningful interpretation of these rich data requires an ability to understand high-order and complex relationships that underpin biological phenomena such as drug response, simulating their dynamics, and selecting the optimal treatment for each patient based on these results. While these data are large-scale and of ultra-high dimensionality, they are also often sparse, with many missing values in the measurements and frequent higher-order interactions among variables, making them hard to handle with conventional statistics. To make further progress, machine learning – especially deep learning – is emerging as one of the promising ways forward. We have developed a method to transform omics data into an image-like representation for analysis with deep learning (DeepInsight) and have successfully used it to predict drug response and to identify original cell types from single-cell RNA-seq data. However, anticipation of the vast amount of medical data being accumulated gives particular urgency to addressing the problems of the time it actually takes to train deep learning models and the complexity of the necessary computational solutions. One possible way to resolve many of these problems is “quantum transcendence”, which is made possible by quantum superposition computation. Among all the different ways to apply quantum computation to medical science, we are particularly interested in quantum deep learning based on optimization and search problems, quantum modeling of single nucleotide detection by observational systems and statistical techniques such as regression analysis by inverse matrix computation and eigenvalue computation. In this seminar, I will first present an overview of how quantum machine learning and quantum deep learning can be used to formulate treatment strategies in medicine. We will discuss how to implement the quantum DeepInsight method, the challenges of noise in quantum computation when training QCNNs, feature mapping issues, problems of pretraining in quantum deep learning, and concerns relating to handling sensitive data such as genomic sequences. I hope this seminar will enhance our understanding of how to effectively facilitate medical research with quantum computing.

We will learn about nuclear fusion and related subjects, such as turbulence in astronomy and astrophysics, from experts and discuss possible interdisciplinary collaborations in the near future. Some researchers will visit RIKEN iTHEMS from the National Institute for Fusion Science (NIFS) and other universities and research institutes for the workshop. We will have the workshop in a hybrid style so that many researchers in Japan can hear the presentations even remotely. This workshop is supported by Moonshot Goal 10 (Program Director Yoshida Zensho (NIFS)).

Program
Session１
9:00-9:35 (25+10: 25 mins for Presentation, 10 mins for Q&A)：Shinya Maeyama
9:35-10:10 (25+10): Naoki Sato
10:10-10:45 (25+10): Yohei Kawazura
10:45-11:15 Coffee Break
Session２
11:15-11:50 (25+10): Takanobu Amano
11:50-12:25 (25+10): Yosuke Matsumoto
12:25-13:00 (25+10): Akira Mizuta
13:00-14:00 Lunch Break
Session３
14:00-14:35 (25+10): Chiho Nonaka
14:35-15:10 (25+10): Takeo Hoshi
15:10-15:45 (25+10): Motoki Nakata
15:45-16:15 Coffee Break
Session ４
16:15-16:50 (25+10): Kumiko Hori
16:50-17:25 (25+10): Yutaka Ohira
17:25-18:00 (25+10): Camilia Demidem (TBC)
18:30-20:30: Dinner in the Main Research Building.

iTHEMS Cosmology Forum Workshop is a series of short workshops, each focused on an emerging topics in cosmology. The targeted audience is cosmologists, high-energy physicists and astronomers interested in learning about the subject, not just those who have already worked on the topic. The goal of the workshop is to provide working knowledge of the topic and leave dedicated time for discussions to encourage mutual interactions among participants.

The first workshop is devoted to cosmic birefringence, a newly establishing cosmological probe of the nature of our universe. Cosmic birefringence is the rotation of the linear polarization plane of the cosmic microwave background (CMB) radiation and, thanks to its origin, inherently measures the degree of parity violation in the cosmic history. This one-day workshop gathers both the observational and theoretical aspects of this growing topic.

The workshop will be in English. The venue is on RIKEN Wako Campus, and the exact room is yet to be determined, depending on the number of registered participants.

The workshops are organised by the iTHEMS Cosmology Forum working group, which is the successor of the Dark Matter Working Group at RIKEN iTHEMS.

Important dates:
30th April - Registration deadline
14th May - Workshop Day

Invited Speakers:
Toshiya Namikawa (Kavli IPMU)
Maresuke Shiraishi (Suwa University of Science)
Fuminobu Takahashi (Tohoku University)

Organisers:
Kohei Hayashi, Nagisa Hiroshima, Derek Inman, Amaury Micheli, Ryo Namba

Drawing from a thought experiment that we conduct, we propose that a virtual graviton gives rise to a black hole geometry when its momentum surpasses a certain threshold value on the Planck scale. This hypothesis implies that the propagator of a virtual graviton, that possesses momentum surpassing this threshold, vanishes. Consequently, a Feynman diagram containing this type of graviton propagator does not add to the overall amplitude. This mechanism suggests the feasibility of formulating an ultraviolet-finite four-dimensional quantum gravitational theory. The elementary particles including the gravitons are treated as point particles in this formulation.

I will give a quick introduction to operator algebras. Operator algebras in this talk consist of linear operators over some Hilbert space. Their study was initiated by Murray and von Neumann, motivated partially by the mathematical foundation of quantum mechanics. Starting from the definitions of a few basic notions, I will explain that commutative operator algebras can be interpreted as spaces. On the other hand, simple operator algebras (i.e., those without non-trivial ideals) form a class of operator algebras opposite to commutative ones and have attracted many operator algebraists. I will try to introduce several examples of simple operator algebras, some of which appear in mathematical physics. If time permits, I will also give recent results on ideals in C*-algebras. People with any scientific background are welcome.

Particle cosmology is a discipline seeking a fundamental understanding of the Universe based on particle physics. Five mysteries drive our research today: cosmic inflation, baryon asymmetry, neutrino properties, dark matter, and dark energy.
Resolving any of the five mysteries will revolutionize our picture of the Universe. Numerous interesting theoretical hypotheses have been proposed to this end. Many require new scalar quantum fields, such as inflatons, axions, supersymmetric particles, etc. They are, in a sense, an attempt to expand the role of the vacuum. Since we have not found such spin-zero fields yet, we shall invent new eyes to make an experimental or observational breakthrough.
The International Center for Quantum-field Measurement Systems for Studies of the Universe and Particles (QUP) was established in December 2021 at KEK under the WPI program of MEXT and JSPS. With its tagline of "bring new eyes to humanity," one of the primary missions of QUP is inventing and developing such new eyes for particle cosmology. In this seminar, after briefly introducing QUP, I focus on research topics I have contributed, including the LiteBIRD satellite to study inflatons and light scalar quantum field searches with novel methods using quantum sensing techniques.

Thanks to intensive research efforts, topology has been established as a fundamental concept in physics. For closed quantum systems, the classification of gapped topological phases has matured. Moreover, the importance of topology is not limited to isolated quantum systems. Recently, the topology of non-Hermitian Hamiltonians, which effectively describe systems with dissipation, has attracted much attention worldwide. This fascination is exemplified by topological phases and topological phenomena unique to non-Hermitian systems.

Against this background, the primary purpose of this workshop is to bring together researchers working on topological phases and to discuss (i) open questions in topological phases of closed quantum systems and (ii) the role of topology in open quantum systems and measurements.

Disease is caused by genetic factors and environmental factors. Genome-wide association study (GWAS) is a powerful method to find genetic factors associated with disease and human complex traits. One conceptual finding GWAS revealed is that many common diseases are caused by a combination of multiple genetic factors (polygenic), rather than a single causal mutation (monogenic). I have been working on finding genetic factors causing polygenic diseases by developing software that accurately finds sequence insertions and deletions from human population-scale sequencing datasets. In this talk, first, I will introduce some examples of disease-causing variants we recently discovered. Next I will also introduce my current research theme aiming to untangle how multiple genetic factors coordinately change cellular homeostasis, which I would like to have a collaboration with mathematical scientists.