Global goal and main bottlenecks
The primary scientific aim of this work is to provide a scalable and efficient implementation of innovative analysis and visualization tools, entirely suitable for use with exascale simulations of molecular systems and to provide the means to make them cooperate.
These tools are targeted at life and materials science and will be designed such that analysis carried out on PC clusters or supercomputers can be accessed from multiple platforms, including desktop, VR CAVE, the web and mobile devices.
The scientific blockages to be overcome concern the implementation of new approaches such as visual analytics and in situ visualization within existing workflows. New concepts incorporating post-processing, analysis and user interaction might be needed to efficiently integrate visualization into the exploration phase of molecular data. ExaViz aims to pioneer such work. The technical blockages are found in the aspects of efficiently parallelizing analysis and visualization components, optimizing memory handling and disk I/O and achieving low latency for real time data exploration. Another important technical blockage concerns the component composition to build an efficient application according to the available resources.
The originality of the project lies in the development of tools associated to a programming environment, which plunge the researcher into a virtual environment and facilitate the analysis of models, making the interaction with molecular simulation data natural and intuitive. The ambitious intention is to make the system scale up to future simulations at the exascale.
To summarise, we propose to develop a chain of tools to enable experts in nanoscale life and materials science to incorporate immersive visual analysis in their everyday work :
1) Interactive real-time visual analysis of molecular dynamics trajectories
2) Intuitive and interactive simulation of solid-state NMR spectra based on ab initio calculation results by simple clicks and selections in the molecular structure.
3) Set the bases of an evolutive platform for the validation, the refinement, or the determination of molecular or periodic solid-state structures based on comparisons between experimental and calculated (by ab initio methods) NMR parameters.
4) In situ visualization for molecular dynamics, ab initio structure refinements, or NMR-based structure-determination protocols.
The tools developed will be tested in real case studies relevant to human health (influenza, tobacco addiction, antivomiting drugs) and the design of new materials (glasses, catalytic & biomaterials). We expect significant scientific advances, notably in the modelling of membrane proteins, which are of major interest from a public health standpoint. These advances concern the private sector (e.g. Servier, Pierre Fabre, Novartis, etc.) as well as the public sector (INSERM, Institut Pasteur, etc.). In solid-state NMR of materials, the complexity, heterogeneity and absence of bridges between simulation programs have restricted their use to the academic community (with the exception of DMfit, whose GUI makes it accessible to a wider range of users among materials scientists). New tools offering interoperability through a graphical interface for the visualization of molecular structures, properties, and NMR spectra have the potential to attract a considerably larger range of scientists and industrial partners focusing on materials applications in fields as crucial as catalysis, optics, electronics, nanotechnologies, or health.
