In the last decades, simulations have been established in several fields of science and industry to study various phenomena by solving, inter alia, partial differential equations. For an efficient use of current and future high performance computing systems, with many thousands of computation ranks, high node-level performance, scalable communication, and the omission of unnecessary calculations are of high priority in the development of new solvers. The challenge of contemporary simulation applications is to bridge the gap between the scales of the various physical processes. We introduce the NAStJA framework, a block-based MPI parallel solver for arbitrary algorithms, based on stencil code or other regular grid methods. NAStJA decomposes the domain of spatially complex structures into small cuboid blocks. A special feature of NAStJA is the dynamic block adaption which modifies the calculation domain around the region where the computation currently takes place, and hence avoids unnecessary calculations. This often occurs, inter alia, in phase-field simulations. Block creation and deletion is managed autonomously within local neighborhoods. A basic load balancing mechanism allows a re-distribution of newly created blocks to the involved computing ranks. The use of a multi-hop network, to distribute information to the entire domain, avoids collective all-gather communications. Thus, we can demonstrate excellent scaling. The present scaling tests substantiate the enormous advantage of this adaptive method. For certain simulation scenarios, we can show that the calculation effort and memory consumption can be reduced to only 3.5 percent, compared to the classical full-domain reference simulation. The overhead of 70-100 percent for the dynamic adapting block creation is significantly lower than the gain. The approach is not restricted to phase-field simulations, and can be employed in other domains of computational science to exploit sparsity of computing regions.

Development of composite workflow applications in science and engineering is troublesome and costly due to high heterogeneity of data representations and data access interfaces of the underlying individual components. As an effective solution we present a generic toolkit enabling domain experts to develop data models and automatically generate a self-contained data access service. We defined a custom metamodel based on Ecore which can be readily used to create domain-specific data models. Using the generated data access service, instances of the modeled data residing on heterogeneous and distributed resources, such as databases and cloud data stores, are accessible from the individual application components via a language-independent Web service interface. We discuss the framework architecture, the toolkit implementation, the deployment process, as well as the performance of the data access service. Workflow designers as target users would benefit from the toolkit by using it for rapid and cost-efficient application integration.

BACKGROUND: Molecular dynamics (MD) simulations provide valuable insight into biomolecular systems at the atomic level. Notwithstanding the ever-increasing power of high performance computers current MD simulations face several challenges: the fastest atomic movements require time steps of a few femtoseconds which are small compared to biomolecular relevant timescales of milliseconds or even seconds for large conformational motions. At the same time, scalability to a large number of cores is limited mostly due to long-range interactions. An appealing alternative to atomic-level simulations is coarse-graining the resolution of the system or reducing the complexity of the Hamiltonian to improve sampling while decreasing computational costs. Native structure-based models, also called Gō-type models, are based on energy landscape theory and the principle of minimal frustration. They have been tremendously successful in explaining fundamental questions of, e.g., protein folding, RNA folding or protein function. At the same time, they are computationally sufficiently inexpensive to run complex simulations on smaller computing systems or even commodity hardware. Still, their setup and evaluation is quite complex even though sophisticated software packages support their realization. RESULTS: Here, we establish an efficient infrastructure for native structure-based models to support the community and enable high-throughput simulations on remote computing resources via GridBeans and UNICORE middleware. This infrastructure organizes the setup of such simulations resulting in increased comparability of simulation results. At the same time, complete workflows for advanced simulation protocols can be established and managed on remote resources by a graphical interface which increases reusability of protocols and additionally lowers the entry barrier into such simulations for, e.g., experimental scientists who want to compare their results against simulations. We demonstrate the power of this approach by illustrating it for protein folding simulations for a range of proteins. CONCLUSIONS: We present software enhancing the entire workflow for native structure-based simulations including exception-handling and evaluations. Extending the capability and improving the accessibility of existing simulation packages the software goes beyond the state of the art in the domain of biomolecular simulations. Thus we expect that it will stimulate more individuals from the community to employ more confidently modeling in their research.