セミナー

Optimal transport: Thermodynamics, biological information processing and generative models

2027年1月18日(月) 13:30 - 15:00

伊藤 創祐 (東京大学 大学院理学系研究科 准教授)

This is a joint seminar co-hosted by Institute for Physics of Intelligence (iπ), UTokyo, and the Venue is #1042,10F of Science Bldg.1 Optimal transport theory is a useful theoretical framework for geometrically characterizing the distance between probability distributions. It is not only used in recent research on generative models but has also been applied to detect information-processing capabilities in biological systems from a statistical perspective, and to discuss thermodynamic efficiency. In this seminar, I will begin with an introduction to optimal transport theory, and also discuss our research on its various applications. More specifically, I will discuss the following topics: 1) The problem of minimizing entropy production within a finite time and its experimental testing. 2) The application of optimal transport to the training of generative models, specifically diffusion models, and their optimal protocols. 3) Quantifying information processing performance using optimal transport in biological signal transduction.

会場: 東京大学 本郷キャンパス 理学部１号館 (メイン会場) / via Zoom

イベント公式言語: 英語

Hamiltonian simulation: Standard algorithms and recent developments

2026年12月17日(木) 10:30 - 17:00

水田 郁 (大阪大学 量子情報・量子生命研究センター 准教授)

This is a lecture-style seminar of quantum computation. The tentative schedule consists of three slots: 10:30-12:00, 13:30-15:00, and 15:30-17:00. Due to equipment limitations, the blackboard might not be visible on Zoom, so we recommend attending in person.

会場: セミナー室 (359号室) (メイン会場) / via Zoom

イベント公式言語: 英語

Quantum Simulation of Non-Equilibrium Dynamics in SU(2) and SU(3) Lattice Gauge Theories

2026年11月24日(火) 15:00 - 16:30

Xiaojun Yao (Research Assistant Professor, University of Washington, USA)

会場: セミナー室 (359号室) (メイン会場) / via Zoom

イベント公式言語: 英語

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

2026年11月16日(月) 14:00 - 15:00

ドロシー・エリス (理化学研究所 生命医科学研究センター (IMS) 統合ゲノミクス研究チーム 特別研究員)

会場: via Zoom

イベント公式言語: 英語

Scaling regimes of the 1D Kuramoto-Sivashinsky equation

2026年11月5日(木) 17:00 - 18:30

Liubov Gosteva (Ph.D. Student, LPMMC, Université Grenoble Alpes, France)

The Kuramoto-Sivashinsky (KS) equation describes various phenomena exhibiting chaos and instabilities, including flames, reaction-diffusion processes, and thin liquid film flows. It resembles the Kardar-Parisi-Zhang (KPZ) equation, with the key difference being the negative "viscosity" in the KS equation. The KS equation is known to belong to the KPZ universality class: at large scales, its statistical properties are governed by the KPZ fixed point, characterized by the dynamical exponent $z=3/2$ and a known PDF of fluctuations. In this talk, I will analyze the stochastic KS equation using the functional renormalization group (FRG). In this framework, attractive fixed points of the FRG equation determine large-scale universal behavior, while unstable fixed points govern small-scale universal behavior. I will show that the two-point correlation function exhibits, besides the KPZ scaling at large scale, another universal scaling regime at small scale, with $z=1$. This scaling is very robust as it is intrinsic to the KS dynamics. It arises from the vanishing of the effective viscosity when evolving from its microscopic negative KS value, to its macroscopic effective positive KPZ value. I will also present results from direct numerical simulations, which confirm these findings.

会場: via Zoom

イベント公式言語: 英語

Quantum Creation of the Universe: Reassessing the No-Boundary and Tunneling Proposals

2026年10月30日(金) 14:00 - 15:00

松井 宏樹 (大阪公立大学 数学研究所 特別研究員)

How was our Universe created? This is one of the most fundamental questions in cosmology. Quantum cosmology attempts to answer it by using a "wave function of the Universe," and for four decades the discussion has been dominated by two proposals: the no-boundary proposal of Hartle and Hawking, and the tunneling proposal of Vilenkin. The two predict very different initial states for our Universe, yet for a long time there was no common framework in which they could be exactly analyzed and compared. In this talk I will begin with a pedagogical introduction to the wave function of the Universe and to the Lorentzian path integral formulation based on Picard-Lefschetz theory, which makes the gravitational path integral well defined without the ambiguities of the traditional Euclidean approach and allows both proposals to be treated on the same footing. I will then present our recent results. Using Lefschetz-thimble analysis together with resurgence theory, we resolve the Stokes-line ambiguity that had obscured earlier analyses and show that the path integral uniquely yields the tunneling wave function rather than the no-boundary one. The no-boundary saddles nevertheless leave a trace: their contributions exactly cancel the Borel ambiguity of the perturbative quantum-gravity corrections around the tunneling saddle, demonstrating that resurgence is at work in quantum cosmology. I will also discuss what changes beyond general relativity, using Hořava-Lifshitz gravity as a concrete UV-complete example.

会場: 研究本館 3階 359号室とZoomのハイブリッド開催 / via Zoom

イベント公式言語: 英語

Maximal Enhancements in Eight-Dimensional Non-Supersymmetric Heterotic Strings

2026年10月30日(金) 10:30 - 12:00

Yuefeng Liu (九州大学 高等研究院 学術研究員)

We classify maximally enhanced, rank-preserving non-supersymmetric heterotic strings in eight dimensions by studying orbifolds of supersymmetric heterotic strings on T^2.We first review the construction in nine dimensions, where it reproduces the 95 maximally semisimple enhancement points obtained from previous extended-Dynkin analyses. We then start from the maximally enhanced supersymmetric Narain points in eight dimensions and use Kac's theorem to enumerate candidate order-two inner actions, retaining only those that lift to consistent Narain-lattice shifts. This gives 1210 maximal enhancement points, of which 71 are tachyon-free, and 25 have neither tachyons nor shifted-sector massless scalars. For each entry we determine the gauge algebra and the low-lying scalar and fermion spectra, and for the tachyon-free cases we evaluate the one-loop cosmological constant. For the 25 subcases without shifted-sector massless scalars, we further compute the Hessian of the one-loop potential and test the refined de Sitter swampland conjecture.

会場: セミナー室 (359号室) (メイン会場) / via Zoom

イベント公式言語: 英語

Born-Oppenheimer EFT: an unified description of ordinary and exotic quarkonia

2026年10月27日(火) 13:30 - 15:00

Nora Brambilla (Professor, TUM School of Natural Sciences, Technical University of Munich, Germany)

We show how the Born--Oppenheimer effective field theory (BOEFT) provides a unified description of ordinary and exotic quarkonia grounded on the non-relativistic expansions of QCD and supplemented with lattice QCD inputs. We show applications of the BOEFT to tetraquarks, pentaquarks, quarkonium hybrids and on threshold effects in the quarkonium spectrum. We present the insights that BOEFT delivers us on the mature of these XYZ exotics states.

会場: via Zoom

イベント公式言語: 英語

Data assimilation with the full covariance on a fault-tolerant quantum computer

2026年10月21日(水) 15:00 - 16:30

三好 建正 (理化学研究所 計算科学研究センター (R-CCS) データ同化研究チーム チームプリンシパル)

Data assimilation (DA) is the core of numerical weather prediction, estimating an accurate state of the weather model by combining forecast and observations. DA provides a general approach and is being applied to various problems beyond weather. The state estimation is a linear solve where the background error covariance B is essential. Since B is an N × N matrix with N ~ 10^8 to 10^9 for the weather model, the full B has never been used. For decades the community has approximated B by parameterized covariance models or by low-rank ensembles with localization and thereby discards genuine structures that we know exist from large-ensemble experiments. Fault-tolerant quantum computing (FTQC) could remove this constraint, since quantum linear system solvers scale polylogarithmically in N. However, the DA equations cannot be applied naively: the observation-space form squares the block-encoding normalization factor for observations with wide footprints, and the state-space form contains B^{-1}, whose condition number ~10^8. This talk presents a combined state-observation formulation that avoids both difficulties: it contains no B^{-1}, and its condition number is governed by the observation error covariance R rather than by B. Numerical experiments with the Lorenz-96 model reproduce the exact solution with the full B and outperform localized ensemble Kalman filters. The formulation assumes that B and the observation operator are provided as block-encoded oracles, and how to construct such oracles without ever storing the matrices is the open problem. A first concrete target is an end-to-end demonstration with Lorenz-96 on a quantum simulator with only seven data qubits, which I would like to discuss with the iTHEMS community.

会場: セミナー室 (359号室) (メイン会場) / via Zoom

イベント公式言語: 英語

AI and Scientific Discovery

2026年10月19日(月) 14:00 - 15:30

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.

会場: 研究本館 435-437号室 (メイン会場) / via Zoom

イベント公式言語: 英語

Finding your rhythm in a complex acoustic environment

2026年10月14日(水) 13:00 - 14:00

Kim Hoke (Professor, Department of Biology, Colorado State University, USA)

My lab integrates genetic, transcriptomic, morphological, and physiological mechanisms to understand the striking diversity of behavior. This talk will focus on recent work on frog acoustic communication. Although the basic acoustic features of frog calls differ reliably among species, individual males of the same species differ in some acoustic properties and female frogs use those male-male differences in mate choice. And for many species, this communication is occurring in very dense choruses which can reach hundreds of males! Our current work assesses which features of calls are characteristic of individuals and how males adjust their calling based on their social environments. We hope to trace how overall chorus dynamics emerge from individual call strategies playing out in time and space, and link those individual strategies to neurobiological mechanisms. We also address the complementary question of female preferences: how do females integrate over a series of signals in this cacophony to make mate choice decisions? Together these lines of research adopt a systems approach to understand how the robust genetic and developmental processes that produce species-specific behaviors have evolved to generate the striking diversity of communication strategies frogs employ.

会場: セミナー室 (359号室) / via Zoom

イベント公式言語: 英語

From data to cohomologies of digraphs, to representations of categories

2026年10月13日(火) 10:30 - 12:00

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.

会場: 研究本館 3階 359号室 (メイン会場) / via Zoom

イベント公式言語: 英語

Adaptive Regulation through Model-Observation Mismatch: From Process Control to Cognitive Neurorobotics

2026年10月8日(木) 13:00 - 14:00

Henrique Oyama (沖縄科学技術大学院大学 (OIST) 認知脳ロボティクス研究ユニット ポスドク研究員)

Adaptive behavior requires biological and artificial systems to respond when their internal models no longer match what they observe. How can such discrepancies be used not only to correct predictions, but also to regulate how a system interacts with its environment? In this talk, I will explore this question across two research directions spanning process control and cognitive neurorobotics. First, I will introduce a model predictive control framework in which a controller actively guides a nonlinear process toward informative conditions to distinguish among competing models while maintaining safe operation. I will then move to cognitive neurorobotics, where brain-inspired computational principles are used to investigate perception, prediction, and action in embodied artificial agents. In physical human-robot interaction, I will show how a predictive-coding-inspired recurrent neural network enables a humanoid robot to regulate the balance between its learned expectations and sensory information arising from human-guided movements. In particular, recent mismatches between sensory observations and the robot's internal model are used to regulate how strongly it relies on its learned dynamics or accommodates ongoing interaction. Together, these studies illustrate how model-observation mismatch can provide a general signal for adaptive regulation, connecting ideas from control engineering, artificial intelligence, neuroscience, and embodied cognition.

会場: 研究本館 3階 359号室とZoomのハイブリッド開催 / via Zoom

イベント公式言語: 英語