Periodic Reporting for period 3 - C0PEP0D (Life and death of a virtual copepod in turbulence)
Período documentado: 2022-09-01 hasta 2024-02-29
The objective of the project is to decipher how planktonic copepods exploit hydrodynamic and chemical sensing to detect and track targets in turbulent flows.
Copepods are millimetric crustaceans that play a crucial role in marine ecosystems. They live in all seas and oceans and are thought to be the most abundant multi-cellular organism on the planet. Yet, copepods are blind. To detect preys, predators, and mates, copepods use hydrodynamic and chemical sensing (Kiørboe, 2010). How are they able to distinguish a meaningful signal in oceanic turbulence? Copepods being one of the greatest success stories of marine evolution, they likely evolved smart algorithms to process this sensing information. But today, these algorithms are poorly understood.
C0PEP0D aims at deciphering these algorithms by addressing three questions:
Q1: Mating. How do male copepods follow the pheromone trail left by females?
Q2: Finding. How do copepods use hydrodynamic signals to ‘see’?
Q3: Feeding. What are the best feeding strategies in turbulent flow?
Copepods are millimetric crustaceans that play a crucial role in marine ecosystems. They live in all seas and oceans and are thought to be the most abundant multi-cellular organism on the planet. Yet, copepods are blind. To detect preys, predators, and mates, copepods use hydrodynamic and chemical sensing (Kiørboe, 2010). How are they able to distinguish a meaningful signal in oceanic turbulence? Copepods being one of the greatest success stories of marine evolution, they likely evolved smart algorithms to process this sensing information. But today, these algorithms are poorly understood.
C0PEP0D aims at deciphering these algorithms by addressing three questions:
Q1: Mating. How do male copepods follow the pheromone trail left by females?
Q2: Finding. How do copepods use hydrodynamic signals to ‘see’?
Q3: Feeding. What are the best feeding strategies in turbulent flow?
The C0PEP0D project has started in Sept. 2019. Since then, three PhD students and two postdocs have been recruited.
We have currently submitted four publications that address the first results achieved on navigation in a turbulent environment and hydrodynamic signature at a moderate Reynolds number. These publications are available on arXiv.org (see also the webpage of the project c0pep0d.github.io).
1. M. Geiger, C. Eloy, M. Wyart (2021) How memory architecture affects learning in a simple POMDP: the two-hypothesis testing problem, arXiv:2106.08849
2. R. Monthiller, A. Loisy, M.A.R. Koehl, B. Favier, C. Eloy (2021) Surfing on turbulence, arXiv:2110.10409
3. T. Redaelli, F. Candelier, R. Mehaddi, B. Mehlig (2021) Unsteady and inertial dynamics of an active particle in a fluid, arXiv:2105.01408
4. A. Loisy, C. Eloy (2021) Searching for a source without gradients: how good is infotaxis and how to beat it, arXiv:2105.01408
We have currently submitted four publications that address the first results achieved on navigation in a turbulent environment and hydrodynamic signature at a moderate Reynolds number. These publications are available on arXiv.org (see also the webpage of the project c0pep0d.github.io).
1. M. Geiger, C. Eloy, M. Wyart (2021) How memory architecture affects learning in a simple POMDP: the two-hypothesis testing problem, arXiv:2106.08849
2. R. Monthiller, A. Loisy, M.A.R. Koehl, B. Favier, C. Eloy (2021) Surfing on turbulence, arXiv:2110.10409
3. T. Redaelli, F. Candelier, R. Mehaddi, B. Mehlig (2021) Unsteady and inertial dynamics of an active particle in a fluid, arXiv:2105.01408
4. A. Loisy, C. Eloy (2021) Searching for a source without gradients: how good is infotaxis and how to beat it, arXiv:2105.01408
Publication 1 resulted from discussions with the group of Matthieu Wyart in EPFL on deep reinforcement learning. We realized that the navigation problems faced by planktonic organisms are very stochastic (because of the turbulent flow). We wondered whether a limited memory would help to learn in this context. We discovered that even in the simplest task (the canonical two-armed bandit problem for which we derive optimal solutions analytically), learning can be very difficult even with small memories because of the presence of multiple local minima.
Publication 2 describes a new strategy, called "surfing", that would allow planktonic organisms to exploit the turbulent flow to move faster in the desired direction. We showed that the gain in effective speed can be as large as 100%, a result quite unexpected and with applicability to a wide range of organisms in the marine plankton world.
Publication 3 shows how to compute unsteady effects and inertial effects on a small active particle at a small but finite Reynolds number. This work can be viewed as an extension of previous work on passive particles or particular time-dependent but spatially homogeneous flows. It is needed to understand how the hydrodynamic signals emitted by copepods or their prey are affected by unsteadiness and inertia.
Publication 4 provides a follow-up development to a path-breaking paper, which introduced infotaxis in 2007. Infotaxis is a strategy for olfactory navigation in turbulence with numerous applications to robotics and ethology. In our paper, we assess its performance. We connect infotaxis, a physics-based navigation problem, to partially observable Markov decision processes, a framework used in artificial intelligence and operations research. By leveraging techniques of these fields, we showed three methods to improve infotaxis.
The expected results until the end of the project concern the use of reinforcement learning to address navigation problems faced by planktonic organisms. We also plan to experiment with real copepods in model flows and turbulent flow to better understand their behaviors.
Publication 2 describes a new strategy, called "surfing", that would allow planktonic organisms to exploit the turbulent flow to move faster in the desired direction. We showed that the gain in effective speed can be as large as 100%, a result quite unexpected and with applicability to a wide range of organisms in the marine plankton world.
Publication 3 shows how to compute unsteady effects and inertial effects on a small active particle at a small but finite Reynolds number. This work can be viewed as an extension of previous work on passive particles or particular time-dependent but spatially homogeneous flows. It is needed to understand how the hydrodynamic signals emitted by copepods or their prey are affected by unsteadiness and inertia.
Publication 4 provides a follow-up development to a path-breaking paper, which introduced infotaxis in 2007. Infotaxis is a strategy for olfactory navigation in turbulence with numerous applications to robotics and ethology. In our paper, we assess its performance. We connect infotaxis, a physics-based navigation problem, to partially observable Markov decision processes, a framework used in artificial intelligence and operations research. By leveraging techniques of these fields, we showed three methods to improve infotaxis.
The expected results until the end of the project concern the use of reinforcement learning to address navigation problems faced by planktonic organisms. We also plan to experiment with real copepods in model flows and turbulent flow to better understand their behaviors.