MaMiCo: Parallel Noise Reduction for Multi-instance Molecular-Continuum Flow Simulation

Jarmatz, Piet; Neumann, Philipp

doi:10.1007/978-3-030-22747-0_34

Cited by 4 publications

(2 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An overview of the implementation for the coupling with MaMiCo is given below. For further information on MaMiCo and its functionalities, we refer to [21][22][23].…”

Section: Molecular-continuum Couplingmentioning

confidence: 99%

“…with density ρ = 0.81, macroscopic cells of size dx = 5.0 and 50 MD time steps per coupling time step, our multi-instance approach yielded approximately 100 million/25 million molecule samples in the default simulation/simulation 1, per cell. This justifies (a) the use of high-performance and supercomputing for general molecular-continuum setups and (b) the need for more enhanced algorithms to reduce Brownian noise, such as algorithms incorporating noise filters; see also, amongst others, [23,28].…”

Section: Lesson 1: Averaging Matters!mentioning

confidence: 99%

See 1 more Smart Citation

Transient Two-Way Molecular-Continuum Coupling with OpenFOAM and MaMiCo: A Sensitivity Study

Wittenberg

Neumann

2021

Computation

Self Cite

View full text Add to dashboard Cite

Molecular-continuum methods, as considered in this work, decompose the computational domain into continuum and molecular dynamics (MD) sub-domains. Compared to plain MD simulations, they greatly reduce computational effort. However, the quality of a fully two-way coupled simulation result strongly depends on a variety of system-specific parameters, and the corresponding sensitivity is only rarely addressed in the literature. Using a state-flux molecular-continuum coupling algorithm, we investigated the influences of various parameters, such as the size of the overlapping region, the coupling time step and the quality of ensemble-based sampling of flow velocities, in a Couette flow scenario. In particular, we considered a big setup in terms of domain size and number of time steps, which allowed us to investigate the long-term behavior of the coupling algorithm close to the incompressible regime. While mostly good agreement was reached on short time scales, it was the long-term behavior which differed even with slightly differently parametrized simulations. We demonstrated our findings by measuring the error in velocity, and we summarize our main observations with a few lessons learned.

show abstract

“…An overview of the implementation for the coupling with MaMiCo is given below. For further information on MaMiCo and its functionalities, we refer to [21][22][23].…”

Section: Molecular-continuum Couplingmentioning

confidence: 99%

Section: Lesson 1: Averaging Matters!mentioning

confidence: 99%