Eda Oktay studied mathematics at Middle East Technical University in Turkey, graduating in 2017. She received her master’s degree in scientific computing from the same university at the Institute of Applied Mathematics, where she focused on parallel spectral graph partitioning algorithms.
From 2020 to 2024, she was a PhD student under the supervision of Erin Carson in the Department of Computational Mathematics at Charles University in Prague. She studied mixed-precision computations in numerical linear algebra, with a particular focus on iterative methods. She developed algorithms using multiprecision arithmetic to enhance the performance and efficiency of iterative and orthogonalization methods for solving various linear and least-squares problems. In addition, she conducted detailed performance and finite-precision error analyses. Her doctoral research was supported by the ERC Starting Grant inEXASCALE and the Exascale Computing Project.
During the second half of her PhD, she was also affiliated with the Chair of Scientific Computing at Chemnitz University of Technology, serving as a research and teaching assistant. There, she worked on mixed-precision training of Gaussian processes, applying mixed-precision techniques to data science problems.
Since August 2024, Eda has been a postdoctoral researcher at Otto-von-Guericke University Magdeburg and a guest researcher at the Max Planck Institute for Dynamics of Complex Technical Systems in Magdeburg, within the Computational Methods in Systems and Control Theory group. Her current research focuses on mixed-precision computations for low-rank tensor approximations, combining mixed precision with other high-performance computing techniques, such as parallel computing, to develop efficient high-performance methods for tensor operations.