1601 Computational Contact Mechanics: Advances and Frontiers in Modeling Contact
Danny Kaufman, Adobe Research
Florence Bertails-Descoubes, NRIA Rhône-Alpes / LJK
Jeff Trinkle, RPI
We propose a mini-symposium to communicate advances and address ongoing challenges in the design and analysis of predictive and efficient computational methods for contact simulation.
Our objective is to bring active researchers in contact mechanics together from a wide variety of disciplines, along with practitioners in aligned areas. Advancing computational contact mechanics is a fundamentally multidisciplinary effort. Nevertheless, disparate research groups in contact modeling have evolved independently in many disciplines with largely complementary scientific skill sets. Thus, as a follow-up to our previous BIRS workshop http://www.birs.ca/events/2014/5-day-workshops/14w5147, we expect this mini-symposium to continue to both forge new interdisciplinary links between mathematicians, computational scientists, and mechanicians, and to enrich ongoing collaborative efforts.
In particular, this workshop will include presentations on a trio of identified high-need areas:
1. Well-posedness, validation, and verification of contact models.
Standard notions of well-posedness are challenged by contact problems where non-uniqueness is often the norm and existence remains an open question for many models regularly employed. Nevertheless, physical systems must be modeled and, in turn, these models will necessarily be discretized and simulated. It is thus important to consider questions of how these models can be best validated and verified. Presentations will cover theoretical considerations of model formulation, analysis, and existence questions as well as the engineering and validation of contact simulation pipelines employed in compute-intensive environments for computational experiments and industrial applications.
2. Numerical integration of nonsmooth systems.
Collisions and impacts introduce sharp, often discontinuous, jumps in state. Under discretization these contact forces are often generated by corresponding nonsmooth interaction potentials with unbounded second derivatives. Such models thus challenge many of the standard smooth assumptions that guarantee good behavior in favored numerical integration methods. Bringing together foremost experts in the area, presentations will review recent developments in nonsmooth numerical integration and consider next steps towards ensuring stability, structure preservation, and accuracy when simulating contact.
3. Scalable numerical methods for solving inequality constrained dynamics.
Discretized contacting systems generally require the solution of constrained optimization problems whose form and scale exercise the limits of existing technologies. Indeed, robust and accurate numerical solutions to many of the large scale optimization problems we generally encounter in high-dimensional contact simulations are often lacking. Nevertheless, many unique structural properties exist and can be leveraged in such problems. Presentations will cover recent aligned developments in optimization methods and theory, discuss the limitations and suitability of current methods, and the specialized properties and requirements of contact derived optimization problems for suitable numerical methods.
Towards these objectives we plan to invite active researchers evenly distributed from the many diverse communities currently contributing to the area of contact modeling including computer graphics, robotics, computational science, mathematics, and numerical optimization.
By bridging communities and bringing these researchers together in the focused environment of a mini-symposium, we hope to foster the rapid advance of our field well beyond the expected progress of independent research threads cross-fertilized only by publication and sporadic conference exchanges. We anticipate that these advances will lead to the accurate and reliable computational tools currently demanded by today’s contact intensive scientific and industrial applications.
Our objective is to bring active researchers in contact mechanics together from a wide variety of disciplines, along with practitioners in aligned areas. Advancing computational contact mechanics is a fundamentally multidisciplinary effort. Nevertheless, disparate research groups in contact modeling have evolved independently in many disciplines with largely complementary scientific skill sets. Thus, as a follow-up to our previous BIRS workshop http://www.birs.ca/events/2014/5-day-workshops/14w5147, we expect this mini-symposium to continue to both forge new interdisciplinary links between mathematicians, computational scientists, and mechanicians, and to enrich ongoing collaborative efforts.
In particular, this workshop will include presentations on a trio of identified high-need areas:
1. Well-posedness, validation, and verification of contact models.
Standard notions of well-posedness are challenged by contact problems where non-uniqueness is often the norm and existence remains an open question for many models regularly employed. Nevertheless, physical systems must be modeled and, in turn, these models will necessarily be discretized and simulated. It is thus important to consider questions of how these models can be best validated and verified. Presentations will cover theoretical considerations of model formulation, analysis, and existence questions as well as the engineering and validation of contact simulation pipelines employed in compute-intensive environments for computational experiments and industrial applications.
2. Numerical integration of nonsmooth systems.
Collisions and impacts introduce sharp, often discontinuous, jumps in state. Under discretization these contact forces are often generated by corresponding nonsmooth interaction potentials with unbounded second derivatives. Such models thus challenge many of the standard smooth assumptions that guarantee good behavior in favored numerical integration methods. Bringing together foremost experts in the area, presentations will review recent developments in nonsmooth numerical integration and consider next steps towards ensuring stability, structure preservation, and accuracy when simulating contact.
3. Scalable numerical methods for solving inequality constrained dynamics.
Discretized contacting systems generally require the solution of constrained optimization problems whose form and scale exercise the limits of existing technologies. Indeed, robust and accurate numerical solutions to many of the large scale optimization problems we generally encounter in high-dimensional contact simulations are often lacking. Nevertheless, many unique structural properties exist and can be leveraged in such problems. Presentations will cover recent aligned developments in optimization methods and theory, discuss the limitations and suitability of current methods, and the specialized properties and requirements of contact derived optimization problems for suitable numerical methods.
Towards these objectives we plan to invite active researchers evenly distributed from the many diverse communities currently contributing to the area of contact modeling including computer graphics, robotics, computational science, mathematics, and numerical optimization.
By bridging communities and bringing these researchers together in the focused environment of a mini-symposium, we hope to foster the rapid advance of our field well beyond the expected progress of independent research threads cross-fertilized only by publication and sporadic conference exchanges. We anticipate that these advances will lead to the accurate and reliable computational tools currently demanded by today’s contact intensive scientific and industrial applications.
