Tramonto Software

Welcome to the Tramonto home page: Software for Nanostructured Fluids in materials and biology

Project goals

This project is based at Sandia National Laboratories, and is focused on developing molecular theory based computational tools for predicting the structure and properties of fluids at the nanoscale near surfaces and macromolecules. At this length scale fluids are inhomogeneous and common approximations for bulk fluids such as incompressibility do not apply. The specific capabilities of Tramonto and the related FasTram software packages are detailed in the Capability links to the left. In both cases, the molecular theories treated by the codes are fluid density functional theories (F-DFTs). These theories compute fluid structure near surfaces or as a result of self-assembly in contrast to quantum density functional theories (Q-DFTs) which are widely used to compute electronic structure of materials.

Applications of Fluids-DFTs

Fluids Density Functional Theory approaches have been used to study a wide range of physical systems. Some examples are: fluids at interfaces, surface forces, colloidal fluids, wetting, porous media, capillary condensation, interfacial phase transitions, nucleation phenomena, freezing, self-assembly, lipid bilayers, ion channel proteins, solvation of surfaces and molecules. The characteristic particle size in F-DFT models ranges from atoms (e.g Argon) to colloidal particles, proteins, or cells. Thus these F-DFT approaches provide a multiscale framework for studying the physcis of many complex fluid systems. Some of these applications are represented in the publication list on the left, others may be found in a very diverse literature. The Tramonto code does not capture all of the F-DFT approaches that have been developed to date, but can be extended to new theories and models.

Motivation - a Scientific Computing perspective.

Until recently, application of F-DFTs to problems in inhomogeneous fluids was limited primarily to systems with two dimensions of symmetry allowing for 1-dimensional computations. In that domain, fast calculations can be performed on single processer computers using algorithms of limited sophistication (e.g. Picard iterations).

Two and three dimensional calculations for F-DFTs are much more costly due to the integral nature of the systems of equations. To understand the computational cost, consider the differences between partial differential equations (PDEs) and the integral equations associated with F-DFTs. The nodes in PDEs generally interact only with nearest neighbors or next nearest neighbors often resulting in diagonally dominant sparse matrices. As the mesh is refined the number of interactions remains constant although there are more nodes to process. In the case of DFTs the range of the integration stencils is significantly longer based on the underlying physics included in a given calculation. Furthermore as the mesh is refined, the integration stencils become larger and more costly because the range of the stencils is fixed by the physics of the problem so with mesh refinement more nodes fall within the reach of a given integration stencil. Finally, as the complexity of the physics and the system dimension increases, simple iterative schemes such as the Picard scheme become less satisfactory for solving the nonlinear systems.

We have addressed these challenges with a combination of parallel computing and specialized linear solver algorithms that are coupled with a Newton's method approach where quadratic convergence is obtained. By utilizing the tools in the Trilinos framework, the molecular theory codes can be coupled to a variety of engineering analysis tools such as arc-length continuation and optimization. These algorithms expand the utility of the software for engineering purposes. One example is that phase diagrams for interfacial fluids can often be generated quickly once two phase coexistence at one state point is established.

Status of the codes

Tramonto v2.1 is available as of March 2007, and has been released with the Lesser Gnu Public Licence. If bugs are found, please send a note to tramonto-help@software.sandia.gov. The details of the license can be obtained here.

Interacting with the SNL Development Team

Contact us if you are interested in contributing to the development of Tramonto. We request that published results obtained from the Tramonto code cite the URL for this web site, and when appropriate the original publications. We invite all users, collaborators, and developers to send a URL or e-mail address, and a brief description of their work for inclusion on this site.

Acknowledgements

The development of the Tramonto code has been funded by:

The Applied Mathematics Program of the Advanced Scientific Computing Research (ASCR) Office at the Department of Energy.
The Laboratory Directed Research and Development (LDRD) program at Sandia National Laboratories.

The Tramonto development team would also like to acknowledge Sandia management personnel who have provided consistent support for the development of this code and the related applications research, including (but not limited to): William Camp, Terry Michalske, Grant Heffelfinger, David Womble, Scott Collis, Bruce Hendrickson, Mark Rintoul, and Kathe Andrews-Kremer.