Are you interested in combining the fields of Machine Learning and Computational Catalysis, Chemical Physics or Materials Science? In the respective groups of Assoc. Prof. Mie Andersen and Prof. Bjørk Hammer at Aarhus University, we have a number of open post-doctoral positions within these research topics. Applications range from the prediction of atomic-scale structures of materials to the catalytic activity and selectivity of those materials.

Starting Date and Period
Starting date is June 1st 2021 or sometime thereafter.
The duration of the employment may be from one to three years.

Machine Learning for Catalysis
In this project, the successful candidate will contribute to the development of a computational framework based on active learning to guide the search for materials that catalyze the conversion of CO2 to methanol at low temperatures. In recent work, Mie Andersen’s group has developed machine learning models for the prediction of catalytically relevant parameters such as adsorption energies for a wide range of molecules and active site motifs at metal and oxide catalysts. The predictive models can, once trained, replace expensive first-principles calculations and provide direct input to thermodynamic or microkinetic models of catalytic activity and selectivity. The postdoc will be introduced to these methods and is expected to further develop them into an active learning framework that, among others, will include the use of uncertainty estimates to carry out global sensitivity analysis and uncertainty quantification of microkinetic models.

You can learn more about Mie Andersen’s research activities here.

Machine Learning for Chemical Physics
In this project, the successful candidate(s) will identify and develop machine learning techniques that may speed up first principles calculations of equilibrium structures and reaction pathways in chemical physics. In recent years, Bjørk Hammer’s group has developed a number of techniques for global optimization (see: https://gofee.au.dk). One method, GOFEE, does its structural search in a model potential using Gaussian Process Regression and is guided by Bayesian statistics to perform occasional sanity checks with full density functional theory (DFT) calculations. Another method, ASLA (see: https://asla.au.dk), does self-training of image recognition for neural network agents that interact with a DFT program. The postdoc(s) will be introduced to these methods and be expected to develop their own improvements and to apply these in the search for the reactive state of matter ranging from interstellar dust clouds to industrial heterogeneous catalysts.

You can learn more about Bjørk Hammer’s group here.

Your Profile
Applicants must hold a PhD degree in physics, chemistry, nanoscience, computer science or equivalent. Previous experience with machine learning methods and/or first principles energy calculations in physical chemistry is required. Experience with programming in python is highly desired.

Place of Work and Area of Employment
The place of work is Ny Munkegade 120, 8000 Aarhus C and the area of employment is Aarhus University with related departments.

Contact Information
Further information may be obtained by e-mailing Mie Andersen ([email protected]) or Bjørk Hammer ([email protected]). Please use as e-mail subject: "postdoctoral position 21".

Application
The application must be submitted via Aarhus University’s web page before April 20th 2021. All interested candidates are encouraged to apply, regardless of their personal background.