Demon's variational principle for informational active matter
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Condensed Matter. The day in condensed matter: materials, phases and the measurements under them.
Kai: We're looking at the paper: "Demon's variational principle for informational active matter".
Mira: Here's the gist of the paper. Demon’s variational principle for informational active matter develops a unified variational framework to quantify path entropy under specified memory states and enables the derivation of cumulant generating functions for arbitrary observables in systems with measurement and feedback The proposed Informational Onsager-Machlup principle (IOMP) provides a systematic method for analyzing information-driven processes in nonequilibrium systems.
Kai: Theoretical Framework The IOMP is formulated as a “demon’s variational principle” extending the Generalized Onsager-Machlup principle (GOMP) It introduces the conditioned Onsager-Machlup integral (OMI), which represents the path entropy conditioned on the memory state The framework systematically incorporates four quantities: free energy F, active power W¤, dissipation Φ, and information M.
Mira: The CGF of an observable A is computed by maximizing the modified OMI ΩA with respect to system variables x(t), v(t), and a(t) under a given memory state y.
Kai: Thermodynamic Consistency The IOMP is confirmed to be consistent with stochastic thermodynamics and information thermodynamics The system's energy dissipation Ψ(y) into both the thermal bath and the microscopic degree of freedom is given by a relation involving conditioned free energy change ΔF(y) and conditioned active work W(y).
Paper summary: Lev: So, looking at this paper, "Demon's variational principle for informational active matter," the core contribution is really that they’ve developed this unified framework—the IOMP—to quantify path entropy under specific memory states and to derive cumulant generating functions for any observable in systems with measurement and feedback.
Kai: They use a variational principle, extending the Generalized Onsager-Machlup principle, to create this structure that neatly ties together the energetic, dissipative, and informational contributions into one formalism. It’s a very systematic way to handle non-equilibrium thermodynamics.
Mira: And they confirm that this whole thing is consistent with both stochastic thermodynamics and information thermodynamics by relating the entropy production in a specific way involving conditioned free energy change and conditioned active work. This gives it thermodynamic credibility.
Lev: For me, what this means for hardware development is that if you're designing a real system with adaptive drag, you have this tool to calculate the expected swimming velocity using these recurrence relations derived from the Gaussian closure.
Kai: But we need to be careful because they point out their limitation: the Gaussian closure only works under certain conditions and it breaks down when there's extreme drag asymmetry. Plus, at small time scales, they neglect higher order cumulants of the velocity distribution, which affects accuracy compared to simulations.
Mira: So while this paper gives us a powerful systematic method to analyze information-driven processes—the IOMP—it’s not a universal solution for every single non-equilibrium system. It tells us what the physics looks like under specific simplifying assumptions.
Lev: I think the most important thing is that it shows how explicitly the information content affects the dynamics of these active matter systems, moving beyond just treating them as purely dissipative processes and showing how measurement really changes things.
Kai: Yeah, so this paper gives us a solid theoretical tool to explore those complex feedback-regulated systems where information isn't just a label; it actively shapes the physical motion.
Conclusion: Kai: So, this paper, "Demon's variational principle for informational active matter," is basically about creating this new way to look at how things move when they have memory and feedback from measurements.
Mira: Yeah, it uses a variational principle that extends an older idea called the Generalized Onsager-Machlup principle. It's trying to unify how we count path entropy, especially when there's some information involved.
Lev: From my side, what’s interesting is that they get this framework to handle active matter systems where the drag coefficient changes based on how fast the particle is moving. That’s actually something we try to model in real hardware.
Kai: Right, so it takes free energy, power input, dissipation, and information itself and puts them all into one structure to find the most likely path. It's about finding the cumulant generating function for any observable you pick.
Mira: And they prove this new way of doing things actually lines up with established thermodynamics—stochastic thermodynamics and information thermodynamics—by showing how entropy production adds up correctly, considering both the thermal bath and the internal microscopic degrees of freedom.
Lev: But there’s a catch, because they use a Gaussian closure for repeated measurements to get steady-state results. They have to be careful because that method breaks down if the drag asymmetry gets too extreme.
Kai: So, what does this actually mean when we look at these informational swimmers? It shows us how the act of measuring something doesn't just give you data; it actively shapes the path the system takes through its memory states.
Mira: It gives us a systematic tool to analyze information-driven processes in non-equilibrium systems, which is pretty powerful for understanding how active matter behaves when it’s constantly reacting to its own measurements.
Lev: And that steady-state velocity they calculate, where it stays finite even as the time scale goes to zero, suggests there are predictable limits on how fast these things can swim under certain conditions.
Kai: It opens up a new way for us to predict the behavior of systems where information isn't just an input; it’s a fundamental part of how the matter moves. Next, we look at what happens when we try to actually build this kind of system in a lab.
Laboratory of Physics, Nihon University · Department of Mathematics, Kyoto University · Zhejiang Key Laboratory of Soft Matter Biomedical Materials, University of Chinese Academy of Sciences
cond-mat.soft, cond-mat.stat-mech
Submitted: 2025-10-15
Updated: 2026-11-11
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 94/100
The gist: Demon’s variational principle for informational active matter develops a unified variational framework to quantify path entropy under specified memory states and enables the derivation of cumulant…
Key concepts
- Demon's Variational Principle
- This is a core idea that extends the Generalized Onsager-Machlup principle. It uses a variational approach to quantify path entropy under specific memory states within systems exhibiting active matter, allowing for the derivation of key thermodynamic properties.
- Informational Onsager-Machlup Principle (IOMP)
- The IOMP is a systematic method for analyzing information-driven processes in non-equilibrium systems. It is formulated as an extension of the GOMP and incorporates four main quantities: free energy, active power, dissipation, and information.
- Conditioned Onsager-Machlup Integral (OMI)
- The OMI represents the path entropy specifically conditioned on a given memory state. This integral is central to the IOMP framework. By maximizing the modified OMI with respect to system variables, researchers can compute cumulant generating functions for observables.
- Cumulant Generating Function (CGF)
- The CGF is a tool used to characterize the statistical properties of an observable in a system. In this paper, it is computed by maximizing the modified OMI. It helps determine how different observables behave under measurement and feedback conditions.
Terminology
Summary
Demon’s variational principle for informational active matter develops a unified variational framework to quantify path entropy under specified memory states and enables the derivation of cumulant generating functions for arbitrary observables in systems with measurement and feedback The proposed Informational Onsager-Machlup principle (IOMP) provides a systematic method for analyzing information-driven processes in nonequilibrium systems. p. 0
Theoretical Framework
The IOMP is formulated as a “demon’s variational principle” extending the Generalized Onsager-Machlup principle (GOMP) It introduces the conditioned Onsager-Machlup integral (OMI), which represents the path entropy conditioned on the memory state The framework systematically incorporates four quantities: free energy F, active power W¤, dissipation Φ, and information M. The CGF of an observable A is computed by maximizing the modified OMI ΩA with respect to system variables x(t), v(t), and a(t) under a given memory state y. p. 1
Thermodynamic Consistency
The IOMP is confirmed to be consistent with stochastic thermodynamics and information thermodynamics The system's energy dissipation Ψ(y) into both the thermal bath and the microscopic degree of freedom is given by a relation involving conditioned free energy change ΔF(y) and conditioned active work W(y). The entropy production in Eq. p. 2 (12) is expressed as σ = ΔSbath + ΔSmic + ΔSmes, confirming consistency with stochastic thermodynamics. p. 2
Single Measurement Results
For a single measurement at t=0 for V0, the CGF for the observable V1 can be calculated analytically as Eq. (23) The first cumulant is given by Eq. p. 4 (24), which shows that feedback through drag coefficients leads to a finite swimming velocity ⟨V1⟩c ≠ 0 when the drag coefficients differ. p. 4
Multiple Measurements and Steady State
For repeated measurements, a Gaussian closure is employed, leading to recurrence relations for the first and second cumulants The steady-state behavior of the information swimmer is obtained by taking the limit N → ∞ yielding an expression for ⟨V∞⟩c/VT. This steady-state velocity remains finite when τ → 0 with an expression ⟨V∞⟩c/VT ≈ √2/(π(γ− + γ+))/(γ+ + γ-). p. 5
Model Application and Limitations
The IOMP is applied to the information swimmer model, where the particle’s drag coefficient adaptively modulates based on measurements of its instantaneous velocity The Gaussian closure sets the range of validity for multi-measurement and steady-state predictions, breaking down only for extreme drag asymmetry. The deviation between the analytical result and Langevin simulations at small τ is attributed to the neglect of higher-order cumulants of the distribution p(Vn). p. 6
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Variational framework for non-equilibrium systems: The informational Onsager-Machlup principle (IOMP) provides a unified approach to compute cumulant generating functions for stochastic processes involving measurement and feedback. This allows AI systems to derive analytical expressions for mean velocity and higher-order cumulants in information-driven swimmer models.
-
Modeling active matter dynamics: The IOMP can be applied to minimal models of informational active matter, such as the information swimmer, enabling the derivation of analytical expressions for mean velocity and higher-order cumulants under single and multiple measurement cases.
-
Prediction of steady-state behavior: The paper provides approximate analytical expressions for repeated measurements and steady states, showing that
feedback through memory-dependent drag can generate a finite average velocity when the drag coefficients differ, 1/γ+ ≠ 1/γ-.
This allows AI to predict persistent motion in fluctuating environments where measurement and feedback are present. -
Distribution of observable statistics: The derivation provides the cumulant generating function for arbitrary observables in measurement-feedback dynamics, allowing the calculation of statistical properties like variance and higher-order cumulants for measured velocities.
Sources
- Active particles in moving traps: minimum work protocols and information efficiency of work extraction
- Neural optimization of the most probable paths of 3D active Brownian particles
- Covariant Onsager and Onsager-Machlup principles for active and inertial dynamics
Related papers
- Coarse-graining dense, deformable active particles
- Three-body corrections to the Brownian viscosity of hard spheres without hydrodynamic interactions
- Thin active nematohydrodynamic layers: asymptotic theories and instabilities
- Origins of suppressed self-diffusion of nanoscale constituents of a complex liquid
- Soft material mechanical property degradation at high strain rates
- Avalanche criticality emerges by thermal fluctuation in a quiescent glass