302PD Surrogate model devpt for thematic Earth Observation Digital Twins

  • Full Time
  • Toulouse

Job title: 302PD Surrogate model devpt for thematic Earth Observation Digital Twins

Company: Centre National d’Etudes Spatiales

Job description: 24-302PD Surrogate model devpt for thematic Earth Observation Digital Twins


24-302PD Surrogate model devpt for thematic Earth Observation Digital Twins

  • Post-doctorat, 24 mois
  • Temps plein
  • Moins de 2 ans d’expĂ©rience
  • Doctorat, Bac+8
  • Propulsion



Your application must include a recommendation letter from your Ph.D. supervisor, a detailed CV including university education and work experience, a list of publications, a 2-page description of the work undertaken during the course of your PhD.

For more Information, contact : Directeur de Recherche

Submit the complete application online (Apply) before March 15th, 2024 Midnight Paris time

Digital Twin (DT) technology holds significant promise and potential for many soaring sectors in Earth Observation (environmental monitoring, climate change adaptation, smart cities, ecological mobility, agriculture, etc.). It can address critical challenges due to its potential to enhance our understanding of complex and intricate systems and its ability to simulate several predictive scenarios.

Many DT definitions coexist. Among them, the three-fold NASA formulation is particularly synthetic and interesting. First, a DT provides a digital replica of the past and current states of the system of interest. It answers to the What-now? Question. Second, it allows for computing forecasts of future states based on the current replica, it answers to the What-next question. Finally, it offers the possibility to explore and investigate the impacts of predictive scenarios simulated.

The ability of a DT to efficiently predict future states depends on two main factors: accuracy (to be sure that results are correct), and performance (wall time needed to get the result). However, the best numerical simulation models require complex high-performance computing platforms and are computationally intensive leading to non-real-time insights. That’s where surrogate models can provide a solution. Often based on machine learning and data-driven techniques, they offer the right compromise between sufficient accuracy and acceptable performance by reducing computational overhead while maintaining the minimum fidelity required by the simulation.

In this context, CNES has identified several areas of research that could lead to innovative work with a strong scientific impact, as they are in line with the current state of the art :

– Machine Learning and Data-driven Surrogate Models. It’s certainly one of the most active research fields. It consists of developing and training AI models that can efficiently replace complex and high-dimensional predictive functions.

– Uncertainty quantification. As the surrogate model is supposed to be less accurate than the nominal physical one (based on numerical simulations), it is very important to quantify the loss of accuracy to assess whether it is still acceptable.

– Multi-fidelity surrogate models. One of the solutions to balance accuracy and computational costs is to integrate results from simulations of different levels of fidelity.

– Adaptive Surrogate models. In the same sense, the objective is to develop surrogate models that can automatically decide where to refine and improve their accuracy based on some parameters.

– Hybrid surrogate models. It combines the best of both worlds, for instance, physical informed AI, the idea is to use what we know about the physics of our world to guide the AI model to an optimal solution.

– Non-linearity and rare events. One of the main challenges of AI-trained surrogate models is to deal with discontinuities and rare events in simulations.

One of the goals of the post-doctorate will be to find and develop general methods and algorithms with the potential to be applied to multiple use cases even if the surrogate model itself is often strongly related to a particular physics.


Applied Mathematical Modelling, Artificial Intelligence, Numerical simulation

Expected salary:

Location: Toulouse

Job date: Thu, 01 Feb 2024 23:35:54 GMT

Apply for the job now!

Submit your Resume!

Do you like the ai jobs 24 ?

Powered By Wischi | CW from Jobs in Germany.net