Victor Boussange

Victor Boussange

PhD candidate

ETH Zürich

Biography

I’m Victor, a $4^{th}$ year Ph.D candidate in the Landscape Ecology Group at ETH Zürich and at the Swiss Federal Institute for Forest, Snow & Landscape (WSL), Switzerland.

My Ph.D aims at better understanding evolutionary processes that affect the dynamics of ecosystems and economic systems. I conduct my investigations with mathematical models capturing eco-evolutionary dynamics. In parallel, I develop machine learning methods to confront these models with empirical data and infer scientific knowledge. I believe that the combination of mechanistic models and machine learning provides a powerful approach to better understand our world. This is crucial, in the face of potentially important ecosystem changes and accelerating threats.

Besides work, I am passionate about alpine adventures and spend my freetime climbing and going down mountains 🏔 , be it with chalk, ice-axes, skis, or a mountainbike. You can check out my alpine CV here.

Interests
  • Ecology and evolution
  • Mathematical modeling
  • Scientific machine learning
  • Complex systems
Education

  • PhD in Environmental Sciences, expected 2022

    ETH Zürich, Switzerland

  • MSc in Energy and Environmental Sciences, 2018

    INSA Lyon, France

Research projects 🔬

Click on each project to learn more.

*
All Scientific Machine Learning Ecology and evolution Economics
Ecology informed machine learning
In order to anticipate the responses of ecosystems to anthropogenic pressure and climate change, models that can extrapolate ecological dynamics beyond observations are required. While we know very well the laws of physics, we are very far from a predictive theory of life. Machine learning can help filling this huge gap! 🧠 🤖 🌱
Ecology informed machine learning
Econobiology: What can economics learn from biology?
Economic actors develop competitive and mutualistic interactions and evolve through time, in similar ways as groups of species. I rely on scientific machine learning to understand how evolutionary processes arising in economic systems connect to those observed in biological systems.
Econobiology: What can economics learn from biology?
Machine learning to solve highly dimensional non-local nonlinear PDEs
Partial Differential Equations (PDEs) are equations that arise in a variety of models in physics, engineering, finance and biology. I develop numerical schemes based on machine learning techniques to solve for a special class of PDEs (cf below) in high dimension.
Machine learning to solve highly dimensional non-local nonlinear PDEs
Gaining a mechanistic understanding of spatial biodiversity patterns
Biodiversity results from differentiation mechanisms developing within biological populations. Such mechanisms are influenced by the properties of the landscape over which individuals interact, disperse and evolve. The documentation of high levels of species diversity in complex mountain region or riverine systems suggest that some peculiar landscape properties foster diversity. I use mathematical models to investigate how landscape connectivity and habitat heterogeneity foster diversity.
Gaining a mechanistic understanding of spatial biodiversity patterns

Open source softwares 🧑🏽‍💻

Open source as a philosphy.

EvoId.jl

Evolutionary Individual based modelling, mathematically grounded. A user friendly package aimed at simulating the evolutionary dynamics of a population structured over a complex spatio-evolutionary structures.

Star

HighDimPDE.jl

Solver for highly dimensional, non-local, nonlinear PDEs. It is integrated within the SciML ecosystem (see below). Try it out! 😃 If you want to learn more about the algorithms implemented, check out my research interests.

Star

MiniBatchInference.jl

Suite for parameter inference and model selection with dynamical models characterised by complex dynamics.

Star

SciML

I am a member of the SciML organisation, an open source ecosystem for Scientific Machine Learning in the Julia programming language. On top of being the main author of HighDimPDE.jl, I actively participate in the development of other packages such as DiffEqFlux.jl, a library to train differential equations with data.

Star


I am also a reviewer at the Journal of Open Source Software Science (JOSS).

Featured Publications

Victor Boussange, Pau Vilimelis Aceituno, Loïc Pellissier
July, 2022 bioRxiv
Mini-batching ecological data to improve ecosystem models with machine learning
Mini-batching ecological data to improve ecosystem models with machine learning

Ecosystems are involved in global biogeochemical cycles that regulate climate and provide essential services to human societies. Mechanistic models are required to describe ecosystem dynamics and anticipate their response to anthropogenic pressure, but their adoption has been limited in practice because of issues with parameter identification and because of model inaccuracies. We propose a machine learning (ML) framework relying on a mini-batch method combined with automatic differentiation and state-of-the-art optimizers. The proposed ML framework can efficiently learn from data and elucidate mechanistic pathways to improve our understanding and predictions of ecosystem dynamics.
PDF Cite Code Project DOI
Victor Boussange, Loïc Pellissier
July, 2022 Commun. Biol.
Eco-evolutionary model on spatial graphs reveals how habitat structure affects phenotypic differentiation
Eco-evolutionary model on spatial graphs reveals how habitat structure affects phenotypic differentiation

Differentiation mechanisms are influenced by the properties of the landscape over which individuals interact, disperse and evolve. Here, we investigate how habitat connectivity and habitat heterogeneity affect phenotypic differentiation by formulating a stochastic eco-evolutionary model where individuals are structured over a spatial graph. By formalising the eco-evolutionary and spatial dynamics of biological populations on graphs, our study establishes fundamental links between landscape features and phenotypic differentiation.
PDF Cite Code Project DOI
See all publications

Recent Posts

Create and deploy your Hugo website
This website is based on the wowchemy template that relies on Hugo, a fast framework for building static websites. Although Hugo has many advantages over its brother Jekyll, a Hugo website is not as easy to deploy on GitHub Pages, the free service offered by GitHub to host a personal website.
Mar 27, 2022 3 min read
Parameter Inference in dynamical systems
One of the challenges modellers face in biological sciences is to calibrate models in order to match as closely as possible observations and gain predictive power. Scientific machine learning addresses this problem by applying optimisation techniques originally developed within the field of machine learning to mechanistic models, allowing to infer parameters directly from observation data.
Victor Boussange
Jan 9, 2021 6 min read Demo
Parameter Inference in dynamical systems

Publications & Talks

Publications

  • Boussange, V. & Pellissier, L., Eco-evolutionary model on spatial graphs reveals how habitat structure affects phenotypic differentiation. Commun Biol 5, 668 (2022). [bioRxiv]

Preprints

  • Boussange, V., Vilimelis-Aceituno, P., Pellissier, L., Mini-batching ecological data to improve ecosystem models with machine learning [bioRxiv] (2022), 46 pages. In review.
  • Boussange, V., Becker, S., Jentzen, A., Kuckuck, B., Pellissier, L., Deep learning approximations for non-local nonlinear PDEs with Neumann boundary conditions. [arXiv] (2022), 59 pages. Revision requested from Partial Differential Equations and Applications.

Proceedings

  • Poulet, T., Alevizos, S., Veveakis, M., Boussange, V., Regenauer-Lieb, K., Episodic mineralising fluid injection through chemical shear zones, ASEG Extended Abstracts (2018), 5 pages.

Works in progress

  • Boussange, V., Sornette, D., Lischke, H., Pellissier, L., Quantifiying eco-evolutionary dynamics in economic systems.

Talks

  • HighDimPDE.jl: A Julia package for solving high-dimensional PDEs, JuliaCon2022, online video 📺
  • Interpretable machine learning for forecasting dynamical processes in ecosystems, World Biodiversity Forum, Davos, Switzerland (June 2022). [slides]
  • Investigating empirical patterns of biodiversity with mechanistic eco-evolutionary models, Seminar at the Theoretical Ecology and Evolution group, invited speaker, Universität Bern (June 2022).
  • Deep learning approximations for non-local nonlinear PDEs, StAMBio seminar, invited speaker, St Andrews, UK (November 2021).
  • Graph topology and habitat assortativity drive phenotypic differentiation in an eco-evolutionary model, Conference on Complex Systems, Lyon, France (October 2021). [slides]
  • Using graph-based metrics to assess the effect of landscape topography on diversification, ECBC, Amsterdam, Netherlands (October 2021). [slides]
  • Solving non-local nonlinear Partial Differential Equations in high dimensions with HighDimPDE.jl, International Conference on Computational Methods in Systems Biology, Bordeaux, France (October 2021). [poster]
  • Responses of neutral and adaptive diversity to complex geographic population structure, Mathematical Population Dynamics, Ecology and Evolution, CIRM Marseille, France (April 2021). [poster]

Popular Topics

Applied Mathematics Biodiversity Deep Learning Ecology and evolution Economics Ecosystem modelling Evolutionary ecology Food-webs Inverse modelling Julia Mathematical modelling Model selection Network topology PDE PDEs Population dynamics Scientific Machine Learnig Scientific Machine Learning