Complexity analysis of simulations with analytic bond-order potentials

Teijeiro, Carlos; Hammerschmidt, Thomas; Seiser, B.; Drautz, Ralf; Sutmann, Godehard

doi:10.1088/0965-0393/24/2/025008

Cited by 4 publications

(16 citation statements)

References 16 publications

(34 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For typical choices of the number of moments, the complexity of the calculations increases with the number of moments to the power of approximately 4.5. The implementation of analytic BOPs in BOPfox reaches this theoretical scaling limit 23 . The increase in the computational effort with the number of atoms is linear ( Fig.…”

Section: A Scalabilitymentioning

confidence: 92%

“…The theoretical scalability of the computational effort with respect to the number of atoms and the number of moments is discussed in a detailed complexity analysis and systematic benchmarks in Ref. 23 . For typical choices of the number of moments, the complexity of the calculations increases with the number of moments to the power of approximately 4.5.…”

Section: A Scalabilitymentioning

confidence: 99%

“…The implementation of this approach in BOPfox reaches the theoretical scaling limits of the required execution time and is discussed in detail in Ref. 23 .…”

Section: Expansion Coefficients and Momentsmentioning

confidence: 99%

“…Here we describe our implementation of analytic BOPs in the software package BOPfox 10 . BOPfox has already been used in several publications 8,[11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26] and is being continuously extended and optimised. We point out similarities of TB/BOP calculations and computations carried out using other electronic structure methods, and discuss the peculiarities of analytic BOPs in detail.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

BOPfox program for tight-binding and analytic bond-order potential calculations

Hammerschmidt

Seiser

Ford

et al. 2019

Computer Physics Communications

View full text Add to dashboard Cite

Bond-order potentials (BOPs) provide a local and physically transparent description of the interatomic interaction. Here we describe the efficient implementation of analytic BOPs in the BOPfox program and library. We discuss the integration of the underlying non-magnetic, collinear-magnetic and noncollinear-magnetic tight-binding models that are evaluated by the analytic BOPs. We summarize the flow of an analytic BOP calculation including the determination of self-returning paths for computing the moments, the self-consistency cycle, the estimation of the band-width from the recursion coefficients, and the termination of the BOP expansion. We discuss the implementation of the calculations of forces, stresses and magnetic torques with analytic BOPs. We show the scaling of analytic BOP calculations with the number of atoms and moments, present options for speeding up the calculations and outline different concepts of parallelisation. In the appendix we compile the implemented equations of the analytic BOP methodology and comments on the implementation. This description should be relevant for other implementations and further developments of analytic bond-order potentials.arXiv:1803.07491v2 [physics.comp-ph]

show abstract

Section: A Scalabilitymentioning

confidence: 92%

Section: A Scalabilitymentioning

confidence: 99%

“…The implementation of this approach in BOPfox reaches the theoretical scaling limits of the required execution time and is discussed in detail in Ref. 23 .…”

Section: Expansion Coefficients and Momentsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

BOPfox program for tight-binding and analytic bond-order potential calculations

Hammerschmidt

Seiser

Ford

et al. 2019

Computer Physics Communications

View full text Add to dashboard Cite

show abstract

“…. , j n , which is a task that is guided using a to-do list (Teijeiro et al, 2016b) and therefore can be safely executed independently of other iterations. Some partial force contributions associated with the half-length paths are already obtained at this point.…”

Section: Openmp Algorithmsmentioning

confidence: 99%

Optimized parallel simulations of analytic bond-order potentials on hybrid shared/distributed memory with MPI and OpenMP

Teijeiro

Hammerschmidt

Drautz

et al. 2017

The International Journal of High Performance Computing Applica

Self Cite

View full text Add to dashboard Cite

Analytic bond-order potentials (BOPs) allow to obtain a highly accurate description of interatomic interactions at a reasonable computational cost. However, for simulations with very large systems, the high memory demands require the use of a parallel implementation, which at the same time also optimizes the use of computational resources. The calculations of analytic BOPs are performed for a restricted volume around every atom and therefore have shown to be well suited for a message passing interface (MPI)-based parallelization based on a domain decomposition scheme, in which one process manages one big domain using the entire memory of a compute node. On the basis of this approach, the present work focuses on the analysis and enhancement of its performance on shared memory by using OpenMP threads on each MPI process, in order to use many cores per node to speed up computations and minimize memory bottlenecks. Different algorithms are described and their corresponding performance results are presented, showing significant performance gains for highly parallel systems with hybrid MPI/OpenMP simulations up to several thousands of threads.

show abstract