Zuse-Institut Berlin
Full-time
Temporary
61,193.82 - 86,189.54 € per year
Within the research group the Interactive Optimization and Learning research group (IOL), we are offering a position, starting January 1, 2027, on a full-time basis (39,4 hours per week), limited until December 31, 2028 for a Researcher (f/m/d). If the applicant meets the relevant wage requirements and personal qualifications, the salary will be based on remuneration group 13 TV-L of the pay scale for the German public sector.
Mixed-integer programming (MIP) is one of the most widely used tools for decision-making in industry and science, and branch-and-bound is the algorithm at the core of every modern MIP solver. Infeasibilities arise throughout the search: a branching decision can lead to a subproblem with no feasible solution, and the node is pruned. Yet most of the information contained in those prunings is discarded.
Conflict analysis is the main mechanism for reusing this information. Inherited from state-of-the-art algorithms used to solve the Boolean Satisfiability (SAT) problem, it learns constraints that propagation uses to avoid similar infeasibilities later in the search. A conflict can be read as a shorter alternative dive that would also have ended in infeasibility, indicating which decisions caused it. Branch-and-bound, however, keeps a fixed parent–child structure: a decision taken high in the tree stays above everything beneath it, even when conflict analysis shows it played no role in the infeasibilities below. This project investigates whether conflict analysis can restructure the tree, deciding where a decision is placed rather than only which variable to branch on next.
You will investigate and develop innovative algorithms for mixed-integer programming, focusing on conflict analysis and branch-and-bound methods. You will explore new approaches to dynamically restructuring branch-and-bound trees, implement and evaluate the proposed algorithms, and contribute to scientific publications and presentations at international conferences.
Preferred
We offer a friendly work environment with flexible work and meeting times, excellent equipment and a challenging professional environment as well as:
Please apply via our online application form by November 25, 2026, submitting your complete application including curriculum vitae in tabular form and the standard supporting documents.
For further information about the position, please refer to our website or contact Mr. Gioni Mexi ().
The Zuse Institute Berlin (ZIB) is an interdisciplinary research institute of the State of Berlin, combining artificial intelligence and scientific computing with applied mathematics and high-performance computing. Together with partners from academia, technology and industry, we develop computational methods to address challenges of societal relevance and provide state-of-the-art infrastructure for high-performance computing, artificial intelligence and data management.