Zum Hauptinhalt springen

PhD Position R3-1

The RTG π3 is a collaborative project between mathematicians of the Center for Industrial Mathematics (ZeTeM); mathematicians in analysis, topology, and statistics; and applied scientists of the University of Bremen. We invite applications for a Ph.D. position (75% of a full position) in the area of data analysis in the framework of project R3-1: Regularisation properties of invertible residual networks.

Recently several architectures for invertible neural networks have been proposed. In this project, we will analyze in particular the concepts of invertible coupling layers and invertible residual networks in the context of non-linear inverse problems. Such problems are mathematically formulated as operator equations and arise in a multitude of applications such as medical imaging or non-destructive testing. The aim of this project is to bridge the gap between the functional analytic background of inverse problems and the more experimental basis of neural networks for solving them. Hence this project combines theoretical challenges for e.g. obtaining convergence results for specific network architectures with a substantial part in the high-level application

We are searching for an enthusiastic and committed researcher with an interest in inverse problems and learning-based methods as well as in developing and applying new mathematical models and algorithms. Within the research training group, the Ph.D. student will be part of the Industrial Mathematics group at the Center for Industrial Mathematics, working under the supervision of Prof. Peter Maaß.


  • M.Sc. or equivalent degree with excellent grades in mathematical sciences or related fields.
  • Skills in scientific computer programming, preferably experience with Python.
  • Experience in the fields of inverse problems and data analysis is advantageous.
  • Industry or research internships are advantageous.
  • Fluency in English.
  • Desire to work in an international and interdisciplinary team.

The position is for a fixed term of 3 years. The earliest starting date for each position in the research training group is 1 October 2022. The salary is according to the German federal employee scale TV-L E13, 75% of a full position (i.e., approximately € 1.900 – 2.100 monthly net income). This call is open until all positions are filled.

Applicants are invited to submit their letter of motivation including a reference to Ph.D. project R3-1 an extended CV including copies of certificates, a publication list (as far as applicable), one recommendation letter from a math professor, and contact information of two more scientists as possible referees.

The recommendation letter should be sent by the math professor directly to us (pi3-applicationprotect me ?!mathprotect me ?!.uni- bremen.de), while the application file should only include her/his name and affiliation.

All relevant documents, quoting the official reference number A 106/22, should be submitted by April 25, 2022,

– preferably electronically as a single PDF file to pi3-applicationprotect me ?!math.uni-bremenprotect me ?!.de – to the π3-coordination: Dr. Daniel Otero Baguer, Zentrum für Technomathematik, Universität Bremen, Bibliothekstr. 5, 28359 Bremen.

The University of Bremen has received several awards for its gender and diversity policies and is particularly aiming to increase the number of female researchers. Gender equality will be given special emphasis within this research training group. Applications from female candidates, international applications, and applications of academics with a migrant background are explicitly welcome.

Disabled persons with the same professional and personal qualifications will be given preference.


Further inquiries for this position may be addressed to:

Prof. Dr. Peter Maaß

Center for Industrial Mathematics

pmaassprotect me ?!uni-bremenprotect me ?!.de


Applications may be addressed to: 

Dr. Daniel Otero Baguer

Zentrum für Technomathematik

Bibliothekstr. 5

28359 Bremen

pi3-applicationprotect me ?!math.uni-bremenprotect me ?!.de