MMMarcus Mohr

Marcus Mohr

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Marcus Mohr has a total of 74 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    TERRA-NG v1.0: Extreme-Scale, GPU-accelerated Mantle Convection

    by Fabian Böhm, Nils Kohl, Ponsuganth Ilangovan, Gabriel Robl, Fatemeh Rezaei, Marcus Mohr, Bernhard Schuberth, Harald Köstler, Hans-Peter Bunge, Ulrich Rüde

    We present TERRA-NG, a portable, GPU-accelerated, matrix-free mantle-convection code. A single Kokkos C++ implementation runs at scale on NVIDIA, AMD, and Intel GPU supercomputers. TERRA-NG has a deliberately narrow design: built on a radially extruded mesh of spherical wedges, tailored to the spherical shell geometry, which enables domain-specific optimizations like single quadrature-point integral-evaluations, radial coordinate storage compression and radial shared-memory tiling. The corresponding low-order W1-iso-W2/W1 wedge-based Stokes--energy discretisation is verified against the Zhong et al.(2008) spherical-shell convection benchmark suite. We showcase TERRA-NG through strong- and weak-scaling on the JUWELS Booster (NVIDIA A100), MareNostrum 5 (NVIDIA H100), LUMI-G (AMD MI250X), Hunter (AMD MI300A APU), and SuperMUC-NG Phase 2 (Intel PVC) supercomputers. Coupled mantle convection simulations at ∼11 km and ∼5.6 km radial spacing (∼2.8 B and ∼22 B DoFs) can be run routinely on standard node partitions of all considered systems. Global ∼1 km-per-gridpoint mantle convection (∼1.4 T DoFs) is feasible on an extreme-scale allocation, and a sub-km hero-run at ∼0.7 km grid spacing scaling up to ∼11,000 GPUs of LUMI-G (∼11 T DoFs) shows the potential of the code on future, larger machines.

  • Open Access

    Assessing the effect of weak tectonic plate boundaries in 3D global mantle circulation models

    by Fatemeh Rezaei, Hans-Peter Bunge, Ponsuganth Ilangovan Ponkumar Ilango, Berta Vilacís, Gabriel Robl, Nils Kohl, Marcus Mohr

    A key characteristic of plate tectonics is strain localization along narrow, weak boundaries between otherwise rigid tectonic plates. This localization enables efficient deformation, subduction, and plate motion, and plays a central role in the dynamic evolution of Earth. However, in mantle circulation models, plate velocities are often assimilated as surface boundary conditions without accounting for the rheological weakness of plate boundaries, relative to the surrounding lithosphere.Weak plate boundaries can be reproduced via sophisticated strain weakening rheologies. While effective, this strategy makes the Stokes system nonlinear and incurs substantial computational cost.Here, we exploit the fact that data assimilation implies that the locations of plate boundaries are known a priori and introduce specifically prescribed weak zones along plate boundaries in the models. These low-viscosity zones allow us to mimic the natural strain localization of Earth’s lithosphere, allowing deformation to focus at plate margins. We show that this approach can provide a computationally efficient and robust framework for bridging the gap between simplified convection models and the complex tectonic behavior of the real Earth.

  • Open Access

    Geodynamic Validation and Scalability of TerraNeo: Matrix-Free Mantle Convection Framework

    by Ponsuganth Ilangovan, Gabriel Robl, Fatemeh Rezaei, Berta Vilacis, Andreas Burkhart, Nils Kohl, Marcus Mohr, Hans-Peter Bunge

    Mantle convection models are of utmost importance in understanding the physics governing major geological processes of our planet such as earthquakes, mountain building, etc. The TerraNeo framework is focussed on creating extreme-scale high-resolution geodynamic models which it achieveswith the massively parallel matrix-free finite element package HyTeG. To handle the Stokes system which arises from the conservation of mass andmomentum equations, a multigrid preconditioned Krylov subspace solver is used, whereas to handle the advection term in the conservation of energyequation, an operator splitting approach based on the modified method of characteristics (particles) is used.We first present standard numerical benchmark experiments for geodynamic validation of the framework against other community codes. In addition, we verify order of convergence of error in velocity and pressure against highly accurate solutions for the Stokes system computed with the propagator matrix method for radially varying viscosity and density cases. Next, a mantle circulation model with spatially varying physical parameters (viscosity and density) and assimilated plate velocities is simulated from a past physical state to present day and assessed for geodynamic correctness. Finally, we present scalability studies performed on the supercomputer SuperMUC-NG Phase 1 at LRZ (91st in TOP500, Nov’ 25). In these experiments, we were able to scale the framework to a global model resolution of ≃ 7.5 km on > 300, 000 MPI processes. These results combined with the numerical benchmarking of the framework clearly show that TerraNeo is well suited for creating large-scale geodynamic models.

  • Open Access

    Highly scalable geodynamic simulations with HyTeG

    by Ponsuganth Ilangovan, Nils Kohl, Marcus Mohr

    High-resolution geodynamic simulations of mantle convection are essential to quantitatively assess the complex physical mechanisms driving the large-scale tectonic processes that shape Earth's surface. Accurately capturing small-scale features such as unstable thermal boundary layers requires global resolution on the order of 1 km. This renders traditional sparse matrix methods impractical due to their prohibitively high memory demands and low arithmetic intensity. Matrix-free methods offer a scalable alternative, enabling the solution of large-scale linear systems efficiently. In this work, we leverage the matrix-free Finite Element framework HyTeG to conduct large-scale geodynamic simulations that incorporate realistic physical models. We validate the framework through a combination of convergence studies of the Finite Element approximations against analytical solutions and through geophysical community benchmarks. The latter include test cases with temperature-dependent and nonlinear rheologies. Our scalability studies demonstrate excellent performance, scaling up to problems with about 100 billion (1011) unknowns in the Stokes system.


  • Open Access

    Reply on RC1

    by Ponsuganth Ilangovan, Nils Kohl, Marcus Mohr

    High-resolution geodynamic simulations of mantle convection are essential to quantitatively assess the complex physical processes driving the large-scale tectonic phenomena that shape Earth’s surface. Accurately capturing small-scale features such as unstable thermal boundary layers requires global resolution on the order of 1 km, which renders traditional sparse matrix methods impractical due to prohibitive memory demands and low arithmetic intensity. Matrix-free methods offer a scalable alternative, enabling the solution of large-scale linear systems efficiently. In this work, we leverage the matrix-free Finite Element framework HyTeG to conduct large-scale geodynamic simulations that incorporate realistic physical models. We validate the framework through a combination of convergence studies and geophysical benchmarks. These include verifying the convergence rates of Finite Element solutions against analytical solutions and through community benchmarks, including test cases with temperature-dependent and nonlinear rheologies. Our scalability studies demonstrate excellent performance, scaling up to problems with about 1011 unknowns in the Stokes system.

  • Open Access

    Reply on RC2

    by Ponsuganth Ilangovan, Nils Kohl, Marcus Mohr

    High-resolution geodynamic simulations of mantle convection are essential to quantitatively assess the complex physical processes driving the large-scale tectonic phenomena that shape Earth’s surface. Accurately capturing small-scale features such as unstable thermal boundary layers requires global resolution on the order of 1 km, which renders traditional sparse matrix methods impractical due to prohibitive memory demands and low arithmetic intensity. Matrix-free methods offer a scalable alternative, enabling the solution of large-scale linear systems efficiently. In this work, we leverage the matrix-free Finite Element framework HyTeG to conduct large-scale geodynamic simulations that incorporate realistic physical models. We validate the framework through a combination of convergence studies and geophysical benchmarks. These include verifying the convergence rates of Finite Element solutions against analytical solutions and through community benchmarks, including test cases with temperature-dependent and nonlinear rheologies. Our scalability studies demonstrate excellent performance, scaling up to problems with about 1011 unknowns in the Stokes system.

  • Open Access

    Reply on CC1

    by Ponsuganth Ilangovan, Nils Kohl, Marcus Mohr

    High-resolution geodynamic simulations of mantle convection are essential to quantitatively assess the complex physical processes driving the large-scale tectonic phenomena that shape Earth’s surface. Accurately capturing small-scale features such as unstable thermal boundary layers requires global resolution on the order of 1 km, which renders traditional sparse matrix methods impractical due to prohibitive memory demands and low arithmetic intensity. Matrix-free methods offer a scalable alternative, enabling the solution of large-scale linear systems efficiently. In this work, we leverage the matrix-free Finite Element framework HyTeG to conduct large-scale geodynamic simulations that incorporate realistic physical models. We validate the framework through a combination of convergence studies and geophysical benchmarks. These include verifying the convergence rates of Finite Element solutions against analytical solutions and through community benchmarks, including test cases with temperature-dependent and nonlinear rheologies. Our scalability studies demonstrate excellent performance, scaling up to problems with about 1011 unknowns in the Stokes system.

  • Open Access

    Comment on egusphere-2025-2552

    by Ponsuganth Ilangovan, Nils Kohl, Marcus Mohr

    High-resolution geodynamic simulations of mantle convection are essential to quantitatively assess the complex physical processes driving the large-scale tectonic phenomena that shape Earth’s surface. Accurately capturing small-scale features such as unstable thermal boundary layers requires global resolution on the order of 1 km, which renders traditional sparse matrix methods impractical due to prohibitive memory demands and low arithmetic intensity. Matrix-free methods offer a scalable alternative, enabling the solution of large-scale linear systems efficiently. In this work, we leverage the matrix-free Finite Element framework HyTeG to conduct large-scale geodynamic simulations that incorporate realistic physical models. We validate the framework through a combination of convergence studies and geophysical benchmarks. These include verifying the convergence rates of Finite Element solutions against analytical solutions and through community benchmarks, including test cases with temperature-dependent and nonlinear rheologies. Our scalability studies demonstrate excellent performance, scaling up to problems with about 1011 unknowns in the Stokes system.

  • Open Access

    Highly Scalable Geodynamic Simulations with HyTeG

    by Ponsuganth Ilangovan, Nils Kohl, Marcus Mohr

    High-resolution geodynamic simulations of mantle convection are essential to quantitatively assess the complex physical processes driving the large-scale tectonic phenomena that shape Earth’s surface. Accurately capturing small-scale features such as unstable thermal boundary layers requires global resolution on the order of 1 km, which renders traditional sparse matrix methods impractical due to prohibitive memory demands and low arithmetic intensity. Matrix-free methods offer a scalable alternative, enabling the solution of large-scale linear systems efficiently. In this work, we leverage the matrix-free Finite Element framework HyTeG to conduct large-scale geodynamic simulations that incorporate realistic physical models. We validate the framework through a combination of convergence studies and geophysical benchmarks. These include verifying the convergence rates of Finite Element solutions against analytical solutions and through community benchmarks, including test cases with temperature-dependent and nonlinear rheologies. Our scalability studies demonstrate excellent performance, scaling up to problems with about 1011 unknowns in the Stokes system.


  • Open Access

    Extreme-Scale Geodynamic Modelling with TerraNeo

    by Ponsuganth Ilangovan, Nils Kohl, Marcus Mohr, Hamish Brown, Eugenio D'AScoli, Isabel Papanagnou, Berta Vilacis, Hans-Peter Bunge

    Finely resolved Geodynamic mantle convection models are crucial to understandthe detailed physics governing major geological process and to inferthe mineralogical state of the Earth. Given the scale needed to globallyresolve features of the mantle such as unstable boundary layersand asthenospheric flows where viscosity can change by around four orders ofmagnitude within 50 km, traditional sparse matrix methods become unsuitabledue to their immense memory requirements. TerraNeo builds upon themassively parallel matrix-free finite element framework HyTeG which usesgeometric multigrid solvers on block-structured hybrid tetrahedral grids.The extreme scalability of HyTeG has been demonstrated previously, solvingStokes problems with trillions (1e12) of unknowns, corresponding to ≃ 1km global resolution of Earth’s mantle.We discuss the features and scalability of TerraNeo with respect to thenumerical treatment of the truncated anelastic liquid approximation as amodel for mantle convection. The compute kernels are evaluated and verifiedthrough geophysical applications, convergence studies, and communitybenchmarks covering sharp viscosity variations, nonlinear rheologies andmixed boundary conditions.