DGSOL is a software package for solving large distance geometry problems in macromolecular modeling. DGSOL has been developed for both sequential and parallel architectures.
Distance geometry problems are interesting mathematical problems with important applications in computational biology, the interpretation of NMR data, and the determination of protein structure. Distance geometry problems are difficult to solve. From a complexity viewpoint, obtaining an approximate solution to a distance geometry is NP-hard. From an optimization viewpoint, distance geometry problems have many local minimizers.
DGSOL solves distance geometry problems with a global continuation algorithm, with Gaussian smoothing of a merit function that only depends on the sparse distance data. Several minimizers are traced in the continuation process, and the best minimizer is selected as the global minimizer. In the current release of DGSOL, a variable-metric limited-memory code is used to trace the minimizers. More robust algorithms will be added in future releases.
For additional information on the solution of distance geometry problems with global continuation and Gaussian smoothing, see
DGSOL is written in C and Fortran. The parallel version uses the MPICH message-passing library for communication.
The current release of DGSOL can be obtained by downloading the compressed file dgsol-1.3.tar.gz. The README file contains instructions on how to install the software.
Argonne National Laboratory / email@example.com