arXiv subjects

Benedikt Dorschner

Publications and source records attributed to Benedikt Dorschner.

12 recordsLinked to original sources

arXiv2025

Simulating many-engine spacecraft: Exceeding 1 quadrillion degrees of freedom via information geometric regularization

We present an optimized implementation of the recently proposed information geometric regularization (IGR) for unprecedented scale simulation of compressible fluid flows applied to multi-engine spacecraft boosters. We improve upon state-of-the-art computational fluid dynamics (CFD) techniques along computational cost, memory footprint, and energy-to-solution metrics. Unified memory on coupled CPU--GPU or APU platforms increases problem size with negligible overhead. Mixed half/single-precision storage and computation on well-conditioned numerics is used. We simulate flow at 200 trillion grid points and 1 quadrillion degrees of freedom, exceeding the current record by a factor of 20. A factor of 4 wall-time speedup is achieved over optimized baselines. Ideal weak scaling is seen on OLCF Frontier, LLNL El Capitan, and CSCS Alps using the full systems. Strong scaling is near ideal at extreme conditions, including 80% efficiency on CSCS Alps with an 8-node baseline and stretching to the full system.

physics.comp-ph↗

arXiv2025

Benchmarking the ability of a controller to execute quantum error corrected non-Clifford circuits

Reaching fault-tolerant quantum computation relies on the successful implementation of non-Clifford circuits with quantum error correction (QEC). In QEC, quantum gates and measurements encode quantum information into an error-protected Hilbert space, while classical processing decodes the measurements into logical errors. QEC non-Clifford gates pose the greatest computation challenge from the classical controller perspective, as they require mid-circuit decoding-dependent feed-forward, modifying the physical gate sequence based on the decoding outcome of previous measurements within the same circuit. In this work, we introduce the first benchmarks to holistically evaluate the capability of a combined controller-decoder system to run non-Clifford QEC circuits. We show that executing an error-corrected non-Clifford circuit, comprised of numerous non-Clifford gates, strictly hinges upon the classical controller-decoder system. Particularly, its ability to perform decoding-based feed-forward with low-latency, defined as the time between the last measurement required for decoding and the dependent mid-circuit quantum operation. We analyze how the system latency dictates the circuit operational regime: latency divergence, classical-controller-limited runtime, or quantum-operation-limited runtime. Based on this understanding, we introduce latency-based benchmarks to set a standard for developing QEC control systems as an essential components of fault-tolerant quantum computation.

quant-ph↗

arXiv2024

Transition time of a bouncing drop

Contact time of bouncing drops is one of the most essential parameters to quantify the water-repellency of surfaces. Generally, the contact time on superhydrophobic surfaces is known to be Weber number-independent. Here, we probe an additional characteristic time, \emph{transition time} inherent in water drop impacting on superhydrophobic surfaces, marking a switch from a predominantly lateral to an axial motion. Systematic experiments and numerical simulations show that the transition time is also Weber number-independent and accounts for half the contact time. Additionally we identify a Weber-independent partition of volume at the maximum spreading state between the rim and lamella and show that the latter contains 1/4 of the total volume of the drop.

physics.flu-dyn↗

arXiv2024

Toward Capturing Genetic Epistasis From Multivariate Genome-Wide Association Studies Using Mixed-Precision Kernel Ridge Regression

We exploit the widening margin in tensor-core performance between [FP64/FP32/FP16/INT8,FP64/FP32/FP16/FP8/INT8] on NVIDIA [Ampere,Hopper] GPUs to boost the performance of output accuracy-preserving mixed-precision computation of Genome-Wide Association Studies (GWAS) of 305K patients from the UK BioBank, the largest-ever GWAS cohort studied for genetic epistasis using a multivariate approach. Tile-centric adaptive-precision linear algebraic techniques motivated by reducing data motion gain enhanced significance with low-precision GPU arithmetic. At the core of Kernel Ridge Regression (KRR) techniques for GWAS lie compute-bound cubic-complexity matrix operations that inhibit scaling to aspirational dimensions of the population, genotypes, and phenotypes. We accelerate KRR matrix generation by redesigning the computation for Euclidean distances to engage INT8 tensor cores while exploiting symmetry.We accelerate solution of the regularized KRR systems by deploying a new four-precision Cholesky-based solver, which, at 1.805 mixed-precision ExaOp/s on a nearly full Alps system, outperforms the state-of-the-art CPU-only REGENIE GWAS software by five orders of magnitude.

q-bio.GN↗

arXiv2022

Exploring shock-capturing schemes for Particles on Demand simulation of compressible flows

In this exploratory study, we apply shock-capturing schemes within the framework of the Particles on Demand kinetic model to simulate compressible flows with mild and strong shock waves and discontinuities. The model is based on the semi-Lagrangian method where the information propagates along the characteristics while a set of shock-capturing concepts such as the total variation diminishing and weighted essentially non-oscillatory schemes are employed to capture the discontinuities and the shock-waves. The results show that the reconstruction schemes are able to remove the oscillations at the location of the shock waves and together with the Galilean invariance nature of the Particles on Demand model, stable simulations of mild to extreme compressible benchmarks can be carried out. Moreover, the essential numerical properties of the reconstruction schemes such as their spectral analysis and order of accuracy are discussed.

physics.flu-dyn↗

arXiv2021

Lattice Boltzmann method for fluid-structure interaction in compressible flow

We present a two-way coupled fluid-structure interaction scheme for rigid bodies using a two-population lattice Boltzmann formulation for compressible flows. Arbitrary Lagrangian-Eulerian formulation of the discrete Boltzmann equation on body-fitted meshes is used in a combination with polynomial blending functions. The blending function approach localizes mesh deformation and allows treating multiple moving bodies with a minimal computational overhead. We validate the model with several test cases of vortex induced vibrations of single and tandem cylinders and show that it can accurately describe dynamic behavior of these systems. Finally, in the fully compressible regime, we demonstrate that the proposed model accurately captures complex phenomena such as transonic flutter over an airfoil.

physics.flu-dyn↗

arXiv2021

Kinetic Simulations of Compressible Non-Ideal Fluids: From Supercritical Flows to Phase-Change and Exotic Behavior

We investigate a kinetic model for compressible non-ideal fluids [DOI:10.1103/PhysRevE.102.020103]. The model imposes the local thermodynamic pressure through a rescaling of the particles velocities, which accounts for both long- and short-range effects and hence full thermodynamic consistency. The model is fully Galilean invariant and treats mass, momentum and energy as local conservation laws. The analysis and derivation of the hydrodynamic limit is followed by the assessment of accuracy and robustness through benchmark simulations ranging from Joule-Thompson effect to a phase-change and high-speed flows. In particular, we show the direct simulation of the inversion line of a van der Waals gas followed by simulations of phase-change such as the one-dimensional evaporation of a saturated liquid, nucleate and film boiling and eventually,we investigate the stability of a perturbed strong shock front in two different fluid mediums. In all of the cases, we find excellent agreement with the corresponding theoretical analysis and experimental correlations. We show that our model can operate in the entire phase diagram, including super- as well as sub-critical regimes and inherently captures phase-change phenomena.

physics.flu-dyn↗

arXiv2020

Multi-resolution lattice Green's function method for incompressible flows

We propose a multi-resolution strategy that is compatible with the lattice Green's function (LGF) technique for solving viscous, incompressible flows on unbounded domains. The LGF method exploits the regularity of a finite-volume scheme on a formally unbounded Cartesian mesh to yield robust and computationally efficient solutions. The original method is spatially adaptive, but challenging to integrate with embedded mesh refinement as the underlying LGF is only defined for a fixed resolution. We present an ansatz for adaptive mesh refinement, where the solutions to the pressure Poisson equation are approximated using the LGF technique on a composite mesh constructed from a series of infinite lattices of differing resolution. To solve the incompressible Navier-Stokes equations, this is further combined with an integrating factor for the viscous terms and an appropriate Runge-Kutta scheme for the resulting differential-algebraic equations. The parallelized algorithm is verified through with numerical simulations of vortex rings, and the collision of vortex rings at high Reynolds number is simulated to demonstrate the reduction in computational cells achievable with both spatial and refinement adaptivity.

physics.comp-ph↗

arXiv2020

On the formation and recurrent shedding of ligaments in droplet aerobreakup

The breakup of water droplets when exposed to high-speed gas flows is investigated using both high-magnification shadowgraphy experiments as well as fully three-dimensional numerical simulations, which account for viscous as well as capillary effects. After thorough validation of the simulations with respect to the experiments, we elucidate the ligament formation process and the effect of surface tension. By Fourier decomposition of the flow field, we observe the development of specific azimuthal modes, which destabilize the liquid sheet surrounding the droplet. Eventually, the liquid sheet is ruptured, which leads to the formation of ligaments. We further observe the ligament formation and shedding to be a recurrent process. While the first ligament shedding weakly depends on the Weber number, subsequent shedding processes seem to be driven primarily by inertia and the vortex shedding in the wake of the deformed droplet.

physics.flu-dyn↗

arXiv2020

Thermokinetic model of compressible multiphase flows

We present a novel approach to kinetic theory modeling enabling the simulation of a generic, real gas presented by its corresponding equation of state. The model is based on mass, momentum and energy conservation, and unlike the lattice Boltzmann models enables simulations of fluids exhibiting a liquid-vapor phase transition in both super- and subcritical states. In this new kinetic theory,an arbitrary equation of state can be introduced through rescaling the discrete particle velocities.Different benchmarks associated with real-gas thermodynamics illustrate that the proposed scheme can handle a wide range of compressible multiphase flows.

physics.flu-dyn↗

arXiv2019

A fast multi-resolution lattice Green's function method for elliptic difference equations

We propose a mesh refinement technique for solving elliptic difference equations on unbounded domains based on the fast lattice Green's function (FLGF) method. The FLGF method exploits the regularity of the Cartesian mesh and uses the fast multipole method in conjunction with fast Fourier transforms to yield linear complexity and decrease time-to-solution. We extend this method to a multi-resolution scheme and allow for locally refined Cartesian blocks embedded in the computational domain. Appropriately chosen interpolation and regularization operators retain consistency between the discrete Laplace operator and its inverse on the unbounded domain. Second-order accuracy and linear complexity are maintained, while significantly reducing the number of degrees of freedom and hence the computational cost.

physics.comp-ph↗

arXiv2017

Fluid-Structure Interaction with the Entropic Lattice Boltzmann Method

We propose a novel fluid-structure interaction (FSI) scheme using the entropic multi-relaxation time lattice Boltzmann (KBC) model for the fluid domain in combination with a nonlinear finite element solver for the structural part. We show validity of the proposed scheme for various challenging set-ups by comparison to literature data. Beyond validation, we extend the KBC model to multiphase flows and couple it with FEM solver. Robustness and viability of the entropic multi-relaxation time model for complex FSI applications is shown by simulations of droplet impact on elastic superhydrophobic surfaces.

physics.flu-dyn↗