High-Performance and Scalable Agent-Based Simulation with BioDynaMo

Abstract

Agent-based modeling plays an essential role in gaining insights into biology, sociology, economics, and other fields. However, many existing agent-based simulation platforms are not suitable for large-scale studies due to the low performance of the underlying simulation engines. To overcome this limitation, we present a novel high-performance simulation engine.We identify three key challenges for which we present the following solutions. First, to maximize parallelization, we present an optimized grid to search for neighbors and parallelize the merging of thread-local results. Second, we reduce the memory access latency with a NUMA-aware agent iterator, agent sorting with a space-filling curve, and a custom heap memory allocator. Third, we present a mechanism to omit the collision force calculation under certain conditions.Our evaluation shows an order of magnitude improvement over Biocellion, three orders of magnitude speedup over Cortex3D and NetLogo, and the ability to simulate 1.72 billion agents on a single server.Supplementary Materials, including instructions to reproduce the results, are available at: https://doi.org/10.5281/ zenodo.6463816 CCS Concepts: • Computing methodologies → Massively parallel and high-performance simulations; Agent / discrete models; Parallel algorithms; • Software and its engineering → Software performance.

Country-Wide Agent-Based Epidemiological Modeling Using 17 Million Individual-Level Microdata

“…Our decision to focus on computational efficiency stems from the pivotal role the BioDynaMo framework played in enabling our research. The advanced computational capabilities of BioDynaMo Breitwieser et al (2023 ); Hesam et al (2021 ) significantly contributed to the feasibility of our study, allowing us to undertake complex simulations that would otherwise have been unattainable. The details on the hardware that was used for the performance benchmarks results can be found in Table 1.…”

“…The platform BioDynaMo Breitwieser et al (2021 ) focuses on supporting high-performance and modular agent-based simulations. BioDynaMo demonstrated unprecedented performance in biomedical applications, efficiently executing simulations of up to 1 billion agents on a single server Breitwieser et al (2023 ).…”

Agent-based modeling in cancer biomedicine: applications and tools for calibration and validation

“…An aspect that requires consideration for the usage of mechanistic modeling and ML-assisted mechanistic modeling is ‘ scalability .’ Given that mechanistic modeling should be, for the sake of biological plausibility, based on local information exchange only, its simulation can naturally make use of parallelized and distributed computing. 101 , 102 For numerous ML methods, their adaptation for large-scale applications requires often custom efforts, since every algorithm has a distinct communication pattern, as demonstrated in the work of Verbraeken et al 103 Along those lines, synchronization requirements among nodes can vary across ML methods, as well as suitability for specific hardware (e.g., CPUs versus GPUs). Overall, the smooth and efficient interfacing with mechanistic modeling constitutes nevertheless a challenging task that remains to be addressed in the future.…”

Mechanistic model of radiotherapy-induced lung fibrosis using coupled 3D Agent-Based and Monte Carlo simulations

“…Besides, the average runtime required to perform 1200-days long simulations was 3 hours. The reduced simulation times (enabled mainly by the optimizations made in the BioDynaMo framework 51 and the longer simulation steps) allowed us to perform longer simulations (with respect to our previous model, up to 1200 days) and improve the statistics by performing more experiments.…”