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.

One contribution of 13 to a Theme Issue 'Multi-scale systems in fluids and soft matter: approaches, numerics and applications' . We review a methodology to design, implement and execute multi-scale and multi-science numerical simulations. We identify important ingredients of multi-scale modelling and give a precise definition of them. Our framework assumes that a multi-scale model can be formulated in terms of a collection of coupled single-scale submodels. With concepts such as the scale separation map, the generic submodel execution loop (SEL) and the coupling templates, one can define a multi-scale modelling language which is a bridge between the application design and the computer implementation. Our approach has been successfully applied to an increasing number of applications from different fields of science and technology.

One contribution of 25 to a Theme Issue 'The virtual physiological human: integrative approaches to computational biomedicine'. Multiscale simulations are essential in the biomedical domain to accurately model human physiology. We present a modular approach for designing, constructing and executing multiscale simulations on a wide range of resources, from laptops to petascale supercomputers, including combinations of these. Our work features two multiscale applications, in-stent restenosis and cerebrovascular bloodflow, which combine multiple existing single-scale applications to create a multiscale simulation. These applications can be efficiently coupled, deployed and executed on computers up to the largest ( peta) scale, incurring a coupling overhead of 1-10% of the total execution time.

We present the Multiscale Coupling Library and Environment: MUSCLE 2. This multiscale component-based execution environment has a simple to use Java, C++, C, Python and Fortran API, compatible with MPI, OpenMP and threading codes. We demonstrate its local and distributed computing capabilities and compare its performance to MUSCLE 1, file copy, MPI, MPWide, and GridFTP. The local throughput of MPI is about two times higher, so very tightly coupled code should use MPI as a single submodel of MUSCLE 2; the distributed performance of GridFTP is lower, especially for small messages. We test the performance of a canal system model with MUSCLE 2, where it introduces an overhead as small as 5% compared to MP

Multiscale simulations model phenomena across natural scales using monolithic or component-based code, running on local or distributed resources. In this work, we investigate the performance of distributed multiscale computing of component-based models, guided by six multiscale applications with different characteristics and from several disciplines. Three modes of distributed multiscale computing are identified: supplementing local dependencies with large-scale resources, load distribution over multiple resources, and load balancing of small- and large-scale resources. We find that the first mode has the apparent benefit of increasing simulation speed, and the second mode can increase simulation speed if local resources are limited. Depending on resource reservation and model coupling topology, the third mode may result in a reduction of resource consumption.