Foundations of distributed multiscale computing: Formalization, specification, and analysis

Abstract

a b s t r a c tInherently complex problems from many scientific disciplines require a multiscale modeling approach. Yet its practical contents remain unclear and inconsistent. Moreover, multiscale models can be very computationally expensive, and may have potential to be executed on distributed infrastructure. In this paper we propose firm foundations for multiscale modeling and distributed multiscale computing. Useful interaction patterns of multiscale models are made predictable with a submodel execution loop (SEL), four coupling templates, and coupling topology properties. We enhance a high-level and well-defined Multiscale Modeling Language (MML) that describes and specifies multiscale models and their computational architecture in a modular way. The architecture is analyzed using directed acyclic task graphs, facilitating validity checking, scheduling distributed computing resources, estimating computational costs, and predicting deadlocks. Distributed execution using the multiscale coupling library and environment (MUSCLE) is outlined. The methodology is applied to two selected applications in nanotechnology and biophysics, showing its capabilities.

Multiscale modeling: recent progress and open questions

“…By adhering to a single framework, not tied to a specific discipline, groups of researchers ensure that their respective contribution may seamlessly integrate with others. MMSF is described in detail in Borgdorf et al (2013b) and references therein. Here, we review the main ideas of the formalism.…”

“…We define multiscale computing patterns as high-level call sequences that exploit the functional decomposition of multiscale models in terms of single scale models. We have identified three MCPs MCPs can be expressed at the level of the task graph, which is the directed acyclic graph used to determine the execution order of submodels, to schedule submodel dependencies, and to estimate runtime and communication cost (Borgdorf et al 2013b). We formulate generic task graphs for each MCP and use them to obtain an optimized mapping of the multiscale simulation on HPC resources.…”

“…And that is the notion of a dynamic number of instantiations of single scale models. Borgdorf et al (2013b) already identified coupling templates where the number of instantiations of a single scale model can be dynamic and unknown at compile time. Important classes of mutliscale models that fall under this category are heterogeneous multiscale.…”

“…A step towards the solution to the above identified requirements can be found in the so-called multiscale modeling and simulation framework (MMSF) that some of us have been developing over the last few years Borgdorf et al 2013b). This framework encourage multiscale modelers to express problems composed of a wide range of scales as a set of separate "single scale" models, and to dis-tinguish these single scale models from the "scale bridging" methods that need to be developed to couple them.…”

On the Common Conceptual and Computational Frameworks for Multiscale Modeling

“…It should be noted that both the GST-and the CxA-based frameworks have already been taken as a basis for developing tools to support multiscale modelling, as will be discussed in Section 4. Besides, the CxAbased framework has recently been extended to integrate not only cellular automata but also other types of models 24 .…”