Keywords: level set method, parallel fast marching, domain decomposition, signed distance, contact mechanics.

This paper presents a parallel library supporting large scale engineering problems which deal with free boundaries and distance computation using efficient solvers for level set methods and Eikonal equations. Some applications that will benefit from such a framework are design and material optimization, contact problems, modelling of flood lakes, air pollution simulation, combustion modeling, computer vision or image registration. All the mentioned problems share a common drawback: limited problem size and computational speed. Both these factors are taken into account in the design of efficient parallel algorithms. Another important aspect is the ease of use, which is also taken care of in our code design.

The first part of this paper deals with design of a new parallel algorithm for fast marching methods based on the domain decomposition approach. The newly designed algorithm is created in a way that avoids most of the communication. Our parallel fast marching algorithm is based on the simple idea of setting appropriate boundary condition for each subdomain, so that the fast marching has sufficient data to run the computation independently.

The second part of this paper presents our parallel level set library. This library is based on the parallel level set method and our parallel fast marching method. An important part of the library consist of the programming interface used to provide easy connection to our third party libraries.

The last part of this paper presents an existing application of the parallel level set library in contact mechanics. Our library is already used by UG solver library developed at ISC in Lugano.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £40 +P&P)