Periodic Reporting for period 2 - InfAct (Active matter information machines)
Reporting period: 2022-07-01 to 2023-12-31
Thermodynamic principles lie at the heart of the design of every engine and refrigerator around us. As technology progresses toward miniaturization, it becomes essential to develop thermodynamic principles that apply to engines operating in microscopic environments. This requirement triggered the development of stochastic thermodynamics, treating small systems in which the role of fluctuations is significant.
The provocative thought experiments by Maxwell and Szilard have pioneered the study of the relationship between information and thermodynamics. Ever since, information engines have served as a main tool to investigate this relation. Yet, this relationship remains unresolved. Information engines use feedback loops based on measured information to lower the entropy or extract work from a single heat bath. To date, a description of information engines based on the Jarzynski fluctuation theorem has been used to derive a generalized form of the second law of thermodynamics under the framework of stochastic thermodynamics.
A crucial question that goes beyond the currently explored regime in the field is how information machines perform when working on an active system whose dynamics are not governed by contact with a thermal bath. This is the general scenario of relevance to biological systems and stands at the focus of this project.
In this project, we realize experimentally active matter information engines and study their dynamics and operation in unprecedented detail. It is imperative that we understand the information in this context since it is one observable that remains well-defined out of equilibrium, unlike, for example, pressure. We maintain that experimental realizations of information engines working in complex out-of-equilibrium environments are essential for this task, as they will provide a solid footing for future theoretical development.
In particular, we study three experimental systems of increasing complexity: (1) An optically manipulated information engine operating on a self-propelled colloidal suspension, (2) A mechanically manipulated information engine operating on active macroscopic bristle robots, and (3) An information engine designed to convert information to work in the form of directed motion.
This project has already provided new insights into the thermodynamics of systems and processes far from thermal equilibrium, both in microscopic thermalized systems and in macroscopic athermal systems. The third part of the project is inspired conceptually by food transport in ant colonies. This is clearly our most risky objective; however, initial results provide a new design concept for decentralized robotics. We envision miniature active matter machines consuming chemical fuel and operating in complex environments such as the human body. Such robots will be self-sufficient and independent once they are introduced into the environment.
The provocative thought experiments by Maxwell and Szilard have pioneered the study of the relationship between information and thermodynamics. Ever since, information engines have served as a main tool to investigate this relation. Yet, this relationship remains unresolved. Information engines use feedback loops based on measured information to lower the entropy or extract work from a single heat bath. To date, a description of information engines based on the Jarzynski fluctuation theorem has been used to derive a generalized form of the second law of thermodynamics under the framework of stochastic thermodynamics.
A crucial question that goes beyond the currently explored regime in the field is how information machines perform when working on an active system whose dynamics are not governed by contact with a thermal bath. This is the general scenario of relevance to biological systems and stands at the focus of this project.
In this project, we realize experimentally active matter information engines and study their dynamics and operation in unprecedented detail. It is imperative that we understand the information in this context since it is one observable that remains well-defined out of equilibrium, unlike, for example, pressure. We maintain that experimental realizations of information engines working in complex out-of-equilibrium environments are essential for this task, as they will provide a solid footing for future theoretical development.
In particular, we study three experimental systems of increasing complexity: (1) An optically manipulated information engine operating on a self-propelled colloidal suspension, (2) A mechanically manipulated information engine operating on active macroscopic bristle robots, and (3) An information engine designed to convert information to work in the form of directed motion.
This project has already provided new insights into the thermodynamics of systems and processes far from thermal equilibrium, both in microscopic thermalized systems and in macroscopic athermal systems. The third part of the project is inspired conceptually by food transport in ant colonies. This is clearly our most risky objective; however, initial results provide a new design concept for decentralized robotics. We envision miniature active matter machines consuming chemical fuel and operating in complex environments such as the human body. Such robots will be self-sufficient and independent once they are introduced into the environment.
In this reporting period, the progress of the research proceeded according to plan, with the addition of two related topics.
Aim 1: Stage 1 – passive many-body microscopic information engine – manuscript is in preparation. We describe the experimental and numerical implementation. We discover and prove analytically, experimentally, and numerically a new universal form for the extracted work per measurement when recast as a function of the probability of being able to move the wall.
Stage 2 – active, many-body microscopic information engine – we are currently performing the experiments, numeric studies show that active colloids do not follow the universal behavior found in stage 1.
Aim 2: Part 1 – many-body active macroscopic Szilard engine (see figure) – manuscript under review and deposited in the arxiv. We realize experimentally a macroscopic many-particle Szilard engine that consists of active particles and uses it to lift a mass against gravity. We show that the extractable work per cycle increases when the raised weight is changed more gradually during the process. Interestingly, we find that the ideal extractable work grows with the number of particles due to giant number fluctuations. This is in contrast to the calculated behavior of a similar engine operating on thermal particles.
Part 2 - We performed a macroscopic experiment using bristle robots and compared these results to the microscopic engine of aim 1, verifying the importance of how energy is pumped into the system.
Extension 1: Related to Aims 1 and 2, ongoing work, combining experiments, simulation, and numerical calculations. Algorithm to design the optimal work extraction from a piston-like engine given position probability distribution taken from experiments.
Extension 2: Related to Aims 1 and 2, ongoing work, combining experiments, simulation, and numerical calculations. Study the effect of clustering on active matter information engines using Kilbots – programable robots.
Aim 3: We have followed exactly the proposed research and demonstrated that injected information can result in the rectification and of a random multi-robot system and result in cargo transfer. We showed that there is a compromise between accuracy and speed of the cargo transfer. The manuscript is in preparation.
Extension 3: Related to aims 2 and 3. A better understanding of temperature in active systems. The first manuscript was published in PRX in which we demonstrate a non-linear fluctuation-dissipation relation that tests for Markovianity. The second manuscript is in preparation.
Extension 4: Related to aim 3. How swarms of simple robots can mimic ants to find food and other targets faster. A manuscript describing environment-assisted search is under review and published in the arxiv.
Aim 1: Stage 1 – passive many-body microscopic information engine – manuscript is in preparation. We describe the experimental and numerical implementation. We discover and prove analytically, experimentally, and numerically a new universal form for the extracted work per measurement when recast as a function of the probability of being able to move the wall.
Stage 2 – active, many-body microscopic information engine – we are currently performing the experiments, numeric studies show that active colloids do not follow the universal behavior found in stage 1.
Aim 2: Part 1 – many-body active macroscopic Szilard engine (see figure) – manuscript under review and deposited in the arxiv. We realize experimentally a macroscopic many-particle Szilard engine that consists of active particles and uses it to lift a mass against gravity. We show that the extractable work per cycle increases when the raised weight is changed more gradually during the process. Interestingly, we find that the ideal extractable work grows with the number of particles due to giant number fluctuations. This is in contrast to the calculated behavior of a similar engine operating on thermal particles.
Part 2 - We performed a macroscopic experiment using bristle robots and compared these results to the microscopic engine of aim 1, verifying the importance of how energy is pumped into the system.
Extension 1: Related to Aims 1 and 2, ongoing work, combining experiments, simulation, and numerical calculations. Algorithm to design the optimal work extraction from a piston-like engine given position probability distribution taken from experiments.
Extension 2: Related to Aims 1 and 2, ongoing work, combining experiments, simulation, and numerical calculations. Study the effect of clustering on active matter information engines using Kilbots – programable robots.
Aim 3: We have followed exactly the proposed research and demonstrated that injected information can result in the rectification and of a random multi-robot system and result in cargo transfer. We showed that there is a compromise between accuracy and speed of the cargo transfer. The manuscript is in preparation.
Extension 3: Related to aims 2 and 3. A better understanding of temperature in active systems. The first manuscript was published in PRX in which we demonstrate a non-linear fluctuation-dissipation relation that tests for Markovianity. The second manuscript is in preparation.
Extension 4: Related to aim 3. How swarms of simple robots can mimic ants to find food and other targets faster. A manuscript describing environment-assisted search is under review and published in the arxiv.
Our main progress beyond the state of the art so far can be summarized thus:
1) We have realized experimentally and analyzed the first many-body information engines, in two very different configurations.
2) We have realized experimentally and analyzed the first active matter information engines, in two very different configurations.
3) We show that increasing the number of elements in these engines is beneficial in the case of active constituents and detrimental in the equilibrium case for a Szilard engine. The situation is different for a piston-like engine.
4) We discovered optimal work extraction criteria for these engines at the different configurations.
5) We introduced a novel idea for improving search in general and for active searchers in particular which we term “environment-mediated search”.
6) We derived and proved a new and much more practical test for Markovianity.
We expect the following to be achieved by the end of the project:
1) Experimental realization of microscopic active many-body information engines, and principles for their optimal operation and design.
2) Additional insights into the stochastic thermodynamic description of active matter.
3) Design principles for efficient use of information in swarm robotics.
1) We have realized experimentally and analyzed the first many-body information engines, in two very different configurations.
2) We have realized experimentally and analyzed the first active matter information engines, in two very different configurations.
3) We show that increasing the number of elements in these engines is beneficial in the case of active constituents and detrimental in the equilibrium case for a Szilard engine. The situation is different for a piston-like engine.
4) We discovered optimal work extraction criteria for these engines at the different configurations.
5) We introduced a novel idea for improving search in general and for active searchers in particular which we term “environment-mediated search”.
6) We derived and proved a new and much more practical test for Markovianity.
We expect the following to be achieved by the end of the project:
1) Experimental realization of microscopic active many-body information engines, and principles for their optimal operation and design.
2) Additional insights into the stochastic thermodynamic description of active matter.
3) Design principles for efficient use of information in swarm robotics.