Postdoctoral Researcher – Sparse Linear Systems (Algorithms) (HLRS_08_2026)

As one of Europe's leading facilities for high-performance computing, HLRS is a diverse community of scientists, engineers, and other professionals focused on discovering and developing new applications of powerful digital technologies and computer science methods.

Postdoctoral Researcher – Sparse Linear Systems (Algorithms)

High-Performance Computing Center Stuttgart (HLRS)
Future Computing Group

About HLRS
The High-Performance Computing Center Stuttgart (HLRS) is Germany's first federal high-performance computing center. It operates one of the world's fastest supercomputers and provides universities, research institutions, and industrial partners with access to high-performance computing resources. Furthermore, HLRS holds a leading global position in the research areas of parallel computing, cloud computing, as well as big data and artificial intelligence, and plays a key role in international and national research projects. Within HLRS, the Future Computing Group focuses on next-generation technologies for HPC, including research on photonic processors that promise to revolutionize computational performance for specific classes of problems.

Postdoctoral Researcher – Sparse Linear Systems (Algorithms) (m/f/x, 100%, TV‑L 13) HLRS_08_2026

The position is a 2.5-year fixed-term contract.

Background and Motivation

Sparse linear systems arise ubiquitously in scientific simulation: finite element and finite volume discretizations of PDEs, graph problems, network flows, and many machine learning inference tasks all reduce to solving Ax = b where A is sparse. Despite decades of progress, solving very large sparse systems remains a dominant computational cost in HPC. At HLRS, we are investigating how photonic processors can be leveraged to accelerate sparse linear algebra operations that are fundamental to scientific computing. Krylov subspace methods are key to solving large, sparse linear systems in HPC. Their computational core, the sparse matrix-vector product and the application of a preconditioner, is notoriously memory bandwidth‑bound on classical hardware. Photonic accelerators promise to accelerate these numerical implementations by natively supporting the computational kernels. This research develops library backends targeting dominant frameworks for parallel sparse linear algebra in HPC with wide adoption. Position A focuses on developing the algorithmic foundations and working closely with Position B (Implementation) to ensure translation into production‑quality software.

Research Objectives

• Identify which components of PETSc‑based sparse solvers can benefit from photonic offloading
• Develop corresponding algorithmic strategies, including compression, reformulation, and accuracy control
• Investigate mixed‑precision strategies for photonic acceleration (e.g., low‑precision smoothers in multigrid, inner iterations in flexible Krylov methods)
• Validate algorithmic approaches on benchmark problems from CFD and structural mechanics
• Contribute to hardware‑software co‑design through performance profiling and identification of optimization opportunities

Key Tasks

• Development of photonic‑focused reformulations of preconditioners and solvers
• Algorithmic design of library components for sparse linear systems
• Numerical validation on benchmark problems from CFD and structural mechanics
• Close collaboration with Position B for joint library
• Benchmarking and evaluation of photonic hardware suitability for HPC tasks
• Development support for photonic hardware to enable fast execution of numerical implementations on photonic processors
• Publication of methods and results in peer‑reviewed journals and conferences

Requirements

Essential Qualifications

• PhD in mathematics, computer science, or engineering with a strong focus on iterative methods for linear systems
• Strong analytical and problem‑solving skills
• Ability to work independently and as part of an interdisciplinary team
• Excellent communication skills in English (written and spoken)
• Willingness to publish and communicate results to academia

Beneficial Experience

• Expertise in Krylov solvers, preconditioning, and multigrid methods
• Hands‑on experience with PETSc, Ginkgo or comparable large‑scale solver libraries
• Experience with mixed‑precision numerical methods
• Familiarity with high‑performance computing environmentsBackground in numerical linear algebra or computational science

What We Offer

• Opportunity to conduct cutting‑edge research at the intersection of numerical methods and emerging photonic computing technologies
• Access to world‑class supercomputing infrastructure including the Vulcan, Hunter, and HammerHAI systems
• Collaborative research environment with connections to academic and industry
• Professional development through participation in international conferences and publication opportunities
• Contribution to pioneering work in photonic HPC acceleration
• Competitive salary according to TV‑L 13 (100%)

Please submit your application including: cover letter explaining your motivation and research interests, detailed CV, copies of academic transcripts and degree certificates, names and contact details of at least two references (optional), links to GitHub/GitLab repositories or other code samples (if available), relevant publications or technical reports (optional). Please send your application via email (as a single PDF file) by May 15, 2026 , with the subject line “HLRS_08_2026 ” to bewerbungen(at)hlrs.de. The University of Stuttgart aims to increase the proportion of women in academic and academic support roles and is therefore particularly interested in applications from women. Full‑time positions may generally be split into part‑time positions. Individuals with severe disabilities will be given priority in hiring if equally qualified. The hiring of academic and non‑academic staff is handled by the Central Administration of the University of Stuttgart (Office of the President). For information on the handling of applicant data in accordance with Article 13 of the GDPR, please visit:

HLRS is a central unit of the University of Stuttgart. A member of the Gauss Centre for Supercomputing, HLRS is one of three German national centers for high‑performance computing.