Loading Events


Lunch Pitches with Eric Capo and Kevin Kamm

To encourage cross pollination of ideas between researchers from different disciplines, IceLab hosts interdisciplinary research lunches with the vision of allowing ideas to meet and mate. During the Lunch Pitch Season, the creative lunches take place at KBC on a Wednesday.
Place: KBC Glasburen
Time: Wednesday 10 April at 12:00.

Pitch 1: Eric Capo: Water deoxygenation changing microbial services in past, present and future coastal systems

Assistant professor, Department of Ecology and Environmental Sciences

Water deoxygenation as a consequence of global warming is currently one of the major problems faced by oceans, besides warming and acidification. In coastal waters, oxygen decline is enhanced by inputs from agriculture, industries and urbanization and cause the presence of low-oxygen ‘hypoxic’ or oxygen depleted ‘anoxic’ zones with dramatic consequences on food webs and ecosystem services. The future impact of coastal deoxygenation on their microbial communities is yet to be unravelled as this may modify their contribution to greenhouse gas emissions with further effects of Earth’s climate.

In this talk, I will discuss combining molecular ecology, paleogenomics, bioinformatics and modelling approaches to study the past, present and future impact of oxygen decline on the microbial ecosystem services from coastal seas. This project relies on the production of new data from the water column and sedimentary archives of coastal systems (Baltic Sea, Black Sea and Norwegian fjords) to characterize their resident microorganisms and how they have been modified by past and ongoing deoxygenation events.

In this project, microbial genomes will be obtained from water columns to study the taxonomic and functional diversity of coastal ecosystems. Then, RNA molecules will be sequenced to study microbial processes characteristic of deoxygenated water systems i.e., denitrification, methanogenesis, sulfate reduction, anoxygenic anaerobic photosynthesis and mercury methylation. The long-term history of these microbial functions will be investigated using DNA preserved in the sedimentary archives during the Holocene i.e., the past 11 thousand years. This data will be used to develop models aiming to project the impact of future coastal deoxygenation on the production and emission of greenhouse gases.

Overall, this project will pave the way for using environmental (paleo)genomics data to develop models to project the impact of environmental change on marine ecosystem services and thus on the future of human societies.


What I am looking for: Eric is seeking collaborators with expertise in climate modelling that could help to develop the last step of this project (using ancient environmental DNA data to project the future impact of water deoxygenation on global biogeochemical processes.


Pitch 2: Kevin Kamm: Deep Learning for High-Dimensional Joint Optimal Stopping and Stochastic Control Problems with Applications in Finance

Postdoctoral researcher, Department of Mathematics and Mathematical Statistics

The project is inspired by optimal harvesting decisions in aquaculture and combines insights from biology, economics, and mathematics. As shown in Figure 1, fish farms pose significant modeling challenges due to their intricate nature. They involve living organisms susceptible to parasites, and their underlying price dynamics are stochastic processes.

The timing of the harvest is crucial for maximizing the farm owner’s revenue. Furthermore, it is necessary to develop strategies for feeding and treatment to increase the farm’s biomass. However, the costs and effectiveness of these methods fluctuate stochastically. This reasoning can also be transferred to forestry and agricultural harvests. For instance, in forestry, decisions are made about the optimal time to fell trees (optimal stopping) and implementing measures to protect against diseases such as fungi or improve growth rates (optimal control).

Mathematically this problem can be solved by a joint optimal stopping and stochastic optimal control problem and the ultimate goal is to develop a numerical method applicable to all of these use-cases.

In this talk, we will discuss the current state of the project, major challenges, and our idea how to solve this problem with Deep-Reinforcement-Learning.

Interested in: We would like to find collaborators with expertise in aquatic, forest or agricultural ecology, to develop realistic models to determine optimal feeding/fertilizing and pest reduction strategies.





Share This Post