Ali Karakus

Ali Karakus serves as a Research Fellow at Argonne National Laboratory, actively contributing to the Computational Engineering Education and Design (CEED) project within the Exascale Computing Project framework. His work bridges advanced computational mathematics with high-performance engineering applications, specializing in GPU-accelerated numerical methods for fluid dynamics simulations.

His research program centers on high-order discontinuous Galerkin methods, with particular expertise in multiphase flow modeling using adaptive level set techniques, polymer processing simulations, and exascale solver development. Key methodological innovations include local time-stepping algorithms, subcell stabilization approaches, and memory-optimized multigrid preconditioners that maintain design-order accuracy while maximizing GPU utilization. His publications consistently demonstrate rigorous validation against benchmark problems including dam-break scenarios, Rayleigh-Taylor instabilities, and thermal convection systems.

Analysis of Karakus's 15 most recent publications (2020-2024) reveals three dominant research trajectories: (1) performance optimization of high-order methods for GPU architectures, (2) extension of discontinuous Galerkin frameworks to complex physical phenomena like weakly compressible flows, and (3) pioneering integration of physics-informed neural networks with traditional CFD solvers. This evolution reflects the broader computational science community's shift toward hybrid AI-numerical approaches while maintaining rigorous mathematical foundations.

Within Argonne's computational mathematics division, Karakus collaborates with multidisciplinary teams on open-source software development including NekRS (spectral element solver) and libParanumal (high-order finite element library). His work directly supports Department of Energy initiatives for next-generation exascale applications, with documented performance improvements on Summit and Sierra supercomputers through the CEED project's milestone deliverables.