Seminar

A non-negative matrix factorization-based approach to integrating multimodal omics data with automatically tuned hyperparameters

November 16 (Mon) 14:00 - 15:00, 2026

Dorothy Ellis (Postdoctoral Researcher, Laboratory for Integrative Genomics, RIKEN Center for Integrative Medical Sciences (IMS))

Venue: via Zoom

Event Official Language: English

AI and Scientific Discovery

October 19 (Mon) 14:00 - 15:30, 2026

Joseph Ledsam (Google Health Lead, Japan, Google Japan)

Artificial intelligence is having a transformative impact on health and scientific discovery. This presentation will trace the evolution from foundational breakthroughs to the sophisticated capabilities of today's large-scale AI models. It will explore how these advanced systems are creating new possibilities across the healthcare landscape, from accelerating therapeutic development to enhancing diagnostic processes and interpreting complex medical data. The session will also take a deeper look at the future possibilities for AI in health and explore the emerging role of agentic AI in scientific discovery. The core theme is the responsible development of AI to create tools that assist scientists, support healthcare professionals, and empower users. Bio: Dr Joseph Ledsam leads Google Health in Japan, where he works across AI research, digital health and health in Google products. He has led research in medical AI, genomics and drug discovery published in journals including Nature, Nature Medicine and Nature Methods. Before moving to Japan he worked as a medical doctor in the UK, and founded the Health Research and Genomics teams in Google DeepMind. He obtained his medical degree from The University of Leeds, UK, and was a research fellow at University College London during his clinical residency.

Venue: #435-437, Main Research Building (Main Venue) / via Zoom

Event Official Language: English

From data to cohomologies of digraphs, to representations of categories

October 13 (Tue) 10:30 - 12:00, 2026

Luigi Caputi (Research Fellow, Department of Mathematics, University of Bologna, Italy)

Motivated by applications to topological data analysis, in the last years we have been witnessing a growing interest in new (topological, homological, categorical) features associated to graphs. The aim of the talk is to describe a recently developed (co)homology theory of directed graphs, called multipath cohomology, whose definition is inspired by Khovanov homology of knots. We will show its relations with more established invariants, such as chromatic homology of graphs, Hochschild homology of algebras, and matching complexes. If time allows it, we will show that the ranks of multipath cohomology, when restricted to graphs with bounded genus, grow at most polynomially in the number of edges, and that the order of its torsion is bounded. This will depend upon structural properties of (the representation category of) graphs, such as its Noetherianity properties.

Venue: #359, 3F, Main Research Building (Main Venue) / via Zoom

Event Official Language: English

Primtive form of singularity and mirror symmetry

September 11 (Fri) 15:30 - 17:30, 2026

Zhe Wang (Research Scientist, Division of Fundamental Mathematical Science, RIKEN Center for Interdisciplinary Theoretical and Mathematical Sciences (iTHEMS))

In 1980s, K. Saito introduced the primitive forms in his study of periods as to generalize the elliptic period integral theory to higher dimensions. An important consequence was that there exists a flat structure on the space of universal unfolding of a singularity. Later after Witten proposed his famous conjecture relating the intersection theory on moduli space to the integrable hierarchy, this flat structure was re-discovered by Dubrovin, and nowadays it is referred as a Frobenius mafniold structure. Frobenius mafniolds provide a convenient tool for formulating the mirror symmetry, which was a special type of duality among string theories orginally studied by physicits. In this talk, I will explain the related constructions, and show by examples how the study of primitive forms gives prediction of geometric quantities such as Gromov-Witten invariants.

Venue: #359, 3F, Seminar Room #359

Event Official Language: English

Making robust reionisation 21 cm signal predictions with C21LRT

September 11 (Fri) 14:00 - 15:15, 2026

Qin Han (JSPS International Research Fellow, Shibaura Institute of Technology)

The redshifted 21 cm line signal is a unique probe for the Epoch of Reionisation by tracking the evolution of the averaged intergalactic medium properties as well as their fluctuations, enabling tomographic studies. Accurate modelling of the 21 cm global and power spectrum signals is crucial for interpreting measurements from current and forthcoming 21 cm experiments. Theoretical predictions usually post-process reionisation simulations with optical depth approximation, which treats local line broadening and peculiar velocities approximately and often leads to divergences due to its velocity gradient term. I will present our cosmological 21 cm line radiative transfer (C21LRT) formulation which resolve these issues naturally. I will compare the 21-cm signals calculated with C21LRT and the conventional optical depth method, and highlight the need to fully quantify the uncertainties of using the optical depth method for exploring the 21-cm signals associated from various possible reionisation scenarios.

Venue: Hybrid Format (2F #220 and Zoom), Main Research Building

Event Official Language: English

Magic state distillation and cultivation #2

September 10 (Thu) 13:30 - 17:00, 2026

Yutaka Hirano (Research Scientist, Nanofiber Quantum Technologies)

Venue: Seminar Room #359 (Main Venue) / via Zoom

Event Official Language: Japanese

Magic state distillation and cultivation #1

September 10 (Thu) 10:30 - 12:00, 2026

Yutaka Hirano (Research Scientist, Nanofiber Quantum Technologies)

Venue: Seminar Room #359 (Main Venue) / via Zoom

Event Official Language: Japanese

How fungi generate one of the highest pressures in biology: the cell wall as a living nanoporous material

September 7 (Mon) 14:00 - 15:00, 2026

Naoyoshi Kumakura (Senior Research Scientist, RIKEN Center for Sustainable Resource Science (CSRS))

Many devastating crop diseases, including anthracnose and rice blast, begin when a fungal pathogen builds a single infection cell called an appressorium. This single cell generates up to ~8 megapascals of turgor pressure (roughly 40 times a car tire’s air pressure), among the highest documented in any living organism, to physically rupture the plant's tough surface. This pressure arises because the cell wall acts as a semipermeable membrane, retaining internal osmolytes such as glycerol while allowing water to pass. What confers this selective permeability had remained unknown for decades: melanin was long assumed responsible, but turgor still forms in melanin-deficient mutants. Using genome editing and direct biochemical analysis of these fungi, we identified two enzymes, PKS2 and PBG13, that synthesize a novel polymer from dihydroxyhexanoic acid (DHHA), which narrows cell-wall pore size at the nanometer scale and is required for turgor generation (Kumakura* et al., Science, 2026). I will frame this system as a living adaptive porous material and discuss questions open to mathematical and physical modeling, including how pore geometry sets the limits of selective permeability.

Venue: Hybrid Format (3F #359 and Zoom), Seminar Room #359

Event Official Language: English

Computational demands shape seizure susceptibility in recurrent neural networks

September 3 (Thu) 13:00 - 14:00, 2026

Sebastian Eydam (Research Scientist, Neural Circuits and Computations Unit, RIKEN Center for Brain Science (CBS))

Why do some brain areas slip into seizures more readily than others? Anatomy and physiology are part of the story, but in this talk we argue that the computation a network performs is itself a determinant of its vulnerability. Working in the language of recurrent neural networks and attractor dynamics, we contrast two computational regimes: networks that represent continuous variables (e.g. tracking a continuous position in space) establish barrier-free state manifolds, whereas networks that store discrete memories have to separate different states by establishing deep wells. A simple energy-landscape picture suggests that continuous networks can amplify perturbations into runaway activity more easily while discrete networks contain them. We test this idea across three systems: handcrafted spiking attractor networks, recurrent networks trained on continuous or discrete computational tasks, and in vivo Neuropixels recordings comparing medial entorhinal cortex (continuous, grid-cell dynamics) with hippocampal CA3 (discrete memory). Under a shared disinhibitory "seizure perturbation," the continuous systems destabilize sooner and drive stronger epileptiform activity, and causal silencing shows this depends on intact entorhinal output. Together, these results establish a direct link between the computation a network is built to perform and its susceptibility to seizures, showing that the very features that enable a network to process information also shape its vulnerability to pathological transitions.

Venue: via Zoom / #359, 3F, Seminar Room #359

Event Official Language: English

Finite cut-off holography and Matrix models

September 3 (Thu) 10:30 - 11:45, 2026

Sounak Pal (Student, Indian Institute of Technology Gandhinagar, India)

In this talk, I will discuss finite-cutoff models of two-dimensional topological gravity and their relation to random matrix theories. Random matrix models have emerged as a useful tool for understanding and refining semiclassical holographic dualities, particularly in low-dimensional quantum gravity. I will begin with a brief introduction to large-(N) matrix models and then describe their connection to low-dimensional topological quantum field theories. Along the way, I will highlight how introducing a finite cutoff can lead to significant modifications in the growth of the black-hole interior compared with the matrix-model description of pure JT gravity. Finally, I will discuss the relation between (TT-bar)-deformed Schwarzian theories and their gravitational bulk duals, emphasizing how these deformations provide an avenue for exploring finite-cutoff holography and possible departures from the standard JT-gravity picture.

Venue: #359, 3F, Main Research Building / via Zoom

Event Official Language: English

Coulomb branches in supersymmetric gauge theories

August 28 (Fri) 15:30 - 17:00, 2026

Hiraku Nakajima (Professor, Kavli Institute for the Physics and Mathematics of the Universe (Kavli IPMU), The University of Tokyo)

Supersymmetric gauge theories in 3 and 4 dimensions have no mathematically rigorous foundation. Coulomb branches of gauge theories originally defined via low energy effective theories by non-rigorous physical argument in 90s. I will explain how I interpret this original argument and make a mathematically rigorous definition of Coulomb branches in my joint work with Braverman and Finkelberg.

Venue: Okochi Hall (Main Venue) / via Zoom

Event Official Language: English

Multistability in Biochemical Reaction Systems: Structural Approach

August 27 (Thu) 16:00 - 17:00, 2026

Takashi Okada (Associate Professor, Program of Mathematical and Life Sciences, Graduate School of Integrated Sciences for Life, Hiroshima University)

In living cells, biochemical reactions form intricate networks that can exhibit multistability, allowing a single system to maintain multiple stable states under identical conditions. However, determining theoretically whether a given reaction system can exhibit multistability is challenging because detailed kinetic parameters are often unknown. In this presentation, we introduce a structural framework that decomposes biochemical networks into subnetworks based solely on network topology. For each subnetwork, we define a scalar quantity, and an index-theoretic argument shows that the sign of this quantity determines whether the subnetwork can contribute to multistability. Our method identifies indicator species whose concentrations distinguish all steady states without detailed kinetic information, thereby enabling efficient measurement design in large networks.

Venue: via Zoom

Event Official Language: English