Intentions The TerraNeon project (PI : Jean-Daniel Kant) proposes to build an innovative decision-making tool aimed at reducing the impact of human activities on climate and biodiversity by adopting a highly systemic and multi-agent approach. TerraNeon is supported by CNRS Innovation, and Sorbonne University.
TerraNeon includes TerraSim, an agent-based simulator (composed of interacting software agents) to model and simulate: human activities (individual and social), economic activities (individuals and companies), and their impacts on climate and biodiversity; the impact of policies (measures, rules, laws) on the system as a whole. Multi-Agent Simulation is the result of recent work in collective artificial intelligence.
The strong point of Multi-Agent Simulations is that they allow collective emergences that cannot be easily predicted from individual behaviours [1,2]. At LIP6, we have significant experience in multi-agent simulation of human activities. Thus, with WorkSim , we were the only ones to propose a quantitative evaluation of the “El Khomri” labour law, an evaluation cited and recognised by the IMF  and by the European Commission . Furthermore, we have developed an efficient method for validating multi-agent models, based on the realism of the modelling assumptions, the automatic calibration of the model parameters, and robustness and sensitivity analyses .
Description You will join the TerraNeon team (6 members now), interact and benefit from the current advances in designing TerraSim.
In TerraSim, 4 main Human activities - called “blocks” - are primarily modeled: Housing / urban planning, Energy, Agriculture, and Transport (grouped under the acronym HEAT in the following).
We follow an incremental approach, which is essential to master the complexity inherent in the chosen systemic approach. Starting from a macroscopic level model (provided), it will be progressively enriched by modelling HEAT activities with a finer granularity, chosen to better represent and capture the societal and environmental impacts of the policies studied. An Agile methodology will be used to ensure both the consistency of the model with its implementation, and the adequacy of the level of granularity chosen for each HEAT component.
This post-doc focuses on the deepening of some of the blocks of the HEAT model. After a phase of analysis of the model and its implementation coupled with a bibliographic study, each work cycle will be composed of the following stages:
- Selection of the block (among HEAT) to be developed; design of the agents and their interactions in TerraSim
- Climate : Data Collection for each of the previous points, also integrating GHG emissions and Life Cycle Analyses for the products considered, etc.
- Biodiversity: Data Collection on Land Occupation, Critical Level Ressources, etc.
- Modelling the population: initialisation, demography, …
- Definition of output indicators: medium and long term sustainability, etc.
- Implementation of the simulation
- Analysis of results
Validation Particular emphasis will be placed on the validation processes of the product model, as TerraSim aims to produce a model that is as realistic as possible. We adopt the MOSIMA method, which includes :
- The collection of data and empirical elements for the construction of the model;
- The collection of data for the calibration of parameters and the implementation of this automatic calibration using optimization algorithms ;
- Generalization tests to reproduce known stylized facts.
Technical Environment The implementations will be carried out in Scala and developed in continuous integration, following an agile method and in interaction with other developments carried out around the TerraSim tool. Significant vigilance will be paid to the quality of the code and the rigor of testing procedures, and software engineering in general.
Conditions The post-doc will take place in Paris (France), within the MAS team of LIP6 (Sorbonne University, Jussieu Campus) for a period of 18 months. The gross monthly remuneration will be 3400 €.
Requirements The ideal candidate must have a PhD degree and a strong background in computer science, agent-based modelling and simulation.
The successful candidate should have:
- Solid Experience in agent-based modelling and simulation (applied to social science, economics or environmental sciences)
- Experience in Data Collection
- Knowledge / experience in Climate Science and/or Biodiversity Sciences and/or Economics is a plus
- Good knowledge of experimental design and simulation analysis
- Excellent publication record
- Strong skills in software development object-oriented programming: Scala (ideally) or Java or similar
- Willing to work in a multi-disciplinary environment
- Excellent communication (oral and written, English) and interpersonal skills
This is a companion discussion topic for the original entry at https://www.comses.net/jobs/603/