The World of SPLectrum
Home > Positioning > Persons > Ito
Sosuke Ito
Ito works on what thermodynamics becomes when the systems are small, fluctuating and coupled to one another — and specifically on how information flowing between such systems enters the thermodynamic accounting. His results generalise the second law to networks of interacting systems, and his more recent programme recasts stochastic thermodynamics in geometric terms, drawing on information geometry and the mathematics of optimal transport.
Life
Associate Professor at the Universal Biology Institute, Graduate School of Science, University of Tokyo. His doctoral and early postdoctoral work was with Takahiro Sagawa, whose work with Masahito Ueda on measurement and feedback in thermodynamics is one of the foundations of the field Ito works in. He has hosted visiting researchers at the Universal Biology Institute, including Artemy Kolchinsky in 2022–2023.
Information thermodynamics on causal networks
His landmark result, with Sagawa, appeared in Physical Review Letters in 2013. The second law and the fluctuation theorem had been extended to systems subject to measurement and feedback — the Maxwell’s demon setting, one system observing another. Ito and Sagawa generalised this to arbitrary networks of interacting systems described by causal, Bayesian-network structure: the entropy production of any one system in such a network is bounded not by zero but by the information flow between it and the others.
The significance is that it turns a two-body result into a structural one. Where the earlier work treated a demon and its target, the causal-network formulation handles biochemical signalling networks, sensory systems and any collection of components exchanging information while dissipating — settings where which system is measuring which is not fixed in advance. The framework was applied to a real case in 2015, again with Sagawa: the adaptation network of E. coli chemotaxis, where the second law with information flow included gives a tighter and more meaningful bound on the signalling component than the conventional second law.
Thermodynamic uncertainty relations
A second strand concerns the trade-off between precision and dissipation. The thermodynamic uncertainty relation — that the relative fluctuation of any current in a non-equilibrium steady state is bounded by the entropy production — establishes that precision has a thermodynamic price, and Ito has contributed to its extensions and to the mapping of where it holds and fails. The relation matters because it is a constraint not derivable from the second law, and because its violation in quantum systems has become a proposed signature of genuinely quantum thermodynamic behaviour.
Geometric thermodynamics
The programme he is currently best known for within the field. Stochastic thermodynamics can be reformulated geometrically: the space of probability distributions carries a natural metric structure, studied as information geometry, and thermodynamic quantities acquire geometric readings on it — entropy production as a length or a divergence, optimal protocols as geodesics. Ito’s work connects this to the theory of optimal transport, where the Wasserstein distance between distributions turns out to have direct thermodynamic meaning, giving bounds on dissipation in terms of how far a distribution must be moved and how quickly.
Where Ito stops
The results are constraints and structural relations: given a network of interacting systems, what the second law becomes; given a required precision, what must be dissipated; given a transformation, what it costs at minimum. What the framework does not supply is why particular biological or engineered systems adopt the architectures they do. The causal-network formulation will tell you the thermodynamic consequences of a signalling structure once the structure is specified; it does not say why evolution arrived at that structure rather than another, and the papers are careful not to claim otherwise.
Key works
- Ito, S. & Sagawa, T., “Information Thermodynamics on Causal Networks,” Physical Review Letters 111:180603 (2013) — the second law and fluctuation theorem on Bayesian networks.
- Ito, S. & Sagawa, T., “Maxwell’s demon in biochemical signal transduction with feedback loop,” Nature Communications 6:7498 (2015) — the causal-network second law applied to E. coli chemotaxis.
- Ito, S., “Stochastic thermodynamic interpretation of information geometry,” Physical Review Letters (2018) — the geometric programme.
- Ito, S. & Dechant, A., “Stochastic time evolution, information geometry, and the Cramér–Rao bound,” Physical Review X (2020).
See also: Thermodynamics · Kolchinsky · Wolpert · Landauer