Multibody dynamics is the key capability required for our macroscale DBP on neuroprosthetic dynamics, and is also used in our coarse-grain molecular DBP and mesoscale DBP. As described in our Progress Report, we have developed and successfully deployed within SimTK the Simbody multibody dynamics code, which forms the core of the OpenSim application for biomechanics. Hundreds of biomechanics labs are now using OpenSim as a major research tool. We have also built the Molmodel API, which integrates Simbody and OpenMM to allow macromolecule simulations using large articulated bodies (e.g. rigid helices) for representing motion, while using fast OpenMM particle computations for calculating forces.
Multibody dynamics refers to sets of rigid and articulated bodies forming the primary mass-carrying framework onto which force-generating elements are positioned. The resulting forces then act on the multibody framework and create motion in accordance with Newton’s laws. For neuromuscular models, the skeleton forms a natural multibody system, to which muscles, tendons, and contact surfaces are attached to generate forces acting on the skeleton. For macromolecular complexes, a multibody system may represent articulated subdomains and rigid substructures that move as rigid bodies. For mesoscale applications, a multibody system may be the set of connected scaffold elements that create, modify, and connect the cell wall and membranes. The open source Simbody code has been used in macro and microscale contexts in the previous grant period. We propose to extend Simbody to provide an efficient multibody framework for a much wider class of mesoscale and multiscale systems, where performance challenges remain in three areas addressed below: (1) integrator step size, (2) cost per step, and (3) specification of efficient models in a hardware-independent manner. We will encapsulate our results in a multibody DSL.
Share with your friends: |