Aktuell

• von Dr. Alexandra Nießen
• 09.07.2026

What are neural networks, and how do they work? Prof. Dr. Sorin-Mihai Grad, the new Professor of Discrete and Stochastic Optimization at the University of Duisburg-Essen, aims to improve the understanding of mathematical mechanisms underlying these algorithms, making them more efficient and precise.

[Dr. Alexandra Nießen:] Could you briefly summarise your research?

[Prof. Dr. Sorin-Mihai Grad:] My work focuses on optimisation. This means I operate at the interface between theoretical and applied mathematics.

Numerous applications – for example, in business, finance, logistics, image processing and AI – can be modelled as minimisation or maximisation problems. When it comes to solving these, my primary aim is to characterise and find optimal solutions to such problems.

Are there any issues that are particularly close to your heart?

I am particularly interested in vector optimisation. There, one has to optimise several objectives simultaneously. This is also known as multi-criteria optimisation. The aim is to find a solution that represents the best compromise for all the criteria at hand, rather than the absolute optimum for any single one.

The classic example comes from the world of finance, from the economist Harry Markowitz: the aim is to find an investment portfolio that minimises risk and maximises expected return. Such a portfolio does not exist in reality, which is why various solution approaches have been developed for such optimisation problems.

I have already achieved various theoretical results in vector optimisation and published some of them in the monograph ‚Duality in Vector Optimisation‘. I am currently working on methods that allow one to approach solutions to such problems via iterative procedures. The existing algorithms for this purpose involve too much heuristics, and I am very confident that multi-criteria optimisation problems can indeed be solved by means of mathematically accurate and fully implementable iterative methods.

What would you like to develop further in your department at the UDE?

I would like to make some of the mathematical mechanisms underlying neural networks easier to understand. As a byproduct, some of the algorithms currently used in AI are expected to become more efficient and precise.

As there are several professorships in the Department of Mathematics at the UDE specialising in optimisation, each with their own areas of focus, I also see many opportunities for joint research projects and seminars.

Many pupils find maths difficult. How could you attract students to your subject?

It is important to me to explain mathematical concepts clearly and in a way that is easy to understand. I see teaching as a dialogue rather than a one-way street. Questions are always welcome during and after my lectures. When it comes to final-year projects, I also try to work with students to find a topic that genuinely interests them and gives them scope to develop their own ideas.

About the person:
From Cluj-Napoca via Chemnitz, Vienna and Paris to Duisburg-Essen: Sorin-Mihai Grad studied mathematics in Romania, completed his PhD and habilitation at Chemnitz University of Technology, and subsequently worked in research and teaching in Leipzig, Vienna and Paris. Most recently, he served as Professor of Optimisation at ENSTA Paris and as part-time Associate Professor at the École Polytechnique. Since 2026, he has been strengthening the Department of Mathematics at the University of Duisburg-Essen (UDE) as Professor of Discrete and Stochastic Optimisation.

More information:
Prof. Dr. Sorin-Mihai Grad, Faculty of Mathematics, Optimisation Research Group, phone. +49 (0)201/18-36880,sorinmihai.grad@uni-due.de

Editor: Dr. Alexandra Niessen, phone +49 (0)203/37-91487, alexandra.niessen@uni-due.de