GetDP

GetDP is an open source and totally free command-line software project that provides a general environment for the treatment of discrete problems using mixed elements to discretize de Rham-type complexes in 1, 2, and 3 dimensions. It offers closeness between the input data, which defines symbolic mathematical expressions and the discrete problems of the respective problems.

Offers powerful command-line options

Being a command-line program, GetDP comes with various command-line options that allow the user to treat discrete problems better. Among these, we can mention the pre-processing, processing and post-processing options, support for reading meshes in the MSH format from a given file, support for reading Gmsh data, and support for resuming processing from the last position.

Additionally, you will be able to use command-line options to save processing results in separate files, to load processing results from a file or multiple files, to use strings as generic filename, to specify parameter file, to create binary output files, to read adaptation constraints from a certain file, to restrict the maximum interpolation order, as well as to cache network computations to disk.

It is also possible to use SLEPc instead of Arpack as eigensolver, to create mesh-based Gmsh output files, to interactively check the structure for problems, to communicate with OneLab using a server address or file, to set the constant number and string, and to set the progress indicator update and verbosity level.

Mandatory example

The following example will give you an idea on how to solve the Poisson equation div(a grad(v)) = f on a domain D, using an input file (".pro" file):

FunctionSpace {
 { Name H1; Type Form0;
  BasisFunction {
   { Name sn; NameOfCoef vn; Function BF_Node; Support D; Entity NodesOf[All]; }
  }
 }
}
Formulation{
 { Name Poisson; Type FemEquation;
  Quantity {
   { Name v; Type Local; NameOfSpace H1; }
  }
 Equation {
  Galerkin { [ a[] * Dof{d v}, {d v} ] ; In D; Jacobian V; Integration I; }
  Galerkin { [ f[], {v} ] ; In D; Jacobian V; Integration I; }
  }
 }
}

What is new in this release:

• added option to embed Octave and Python interpreters;
• extended "Field" functions with gradient; extended string and list handling functions;
• new resolution and postprocessing functions (RenameFile, While, ...);
• extended EigenSolve with eigenvalue filter and high order polynomial EV problems;
• small bug fixes.

What is new in version 2.4.4:

• Better stability
• Updated onelab API version and inline parameter definitions
• Fixed UpdateConstraint in harmonic case
• Improve performance of multi-harmonic assembly
• Fixed memory leak in parallel MPI version
• Improved EigenSolve (quadratic EVP with SLEPC, EVP on real matrices)
• New CosineTransform, MPI_Printf, SendMergeFileRequest parser commands
• Small improvements and bug fixes

What is new in version 2.4.3:

• New mandatory 'Name' attribute to define onelab variables in DefineConstant[] & co; minor bug fixes.

What is new in version 2.4.2:

• Fixed function arguments in nested expressions; minor improvements.

What is new in version 2.4.0:

• This version introduces new two-step Init constraints, faster network computations, an improved Update operation, Unicode support, and various small improvements and bugfixes.

What is new in version 2.3.1:

• This version updates ONELAB and fixes various bugs.

What is new in version 2.3.0:

• This version introduces a new build system based on cmake, a new family of Field functions to use data imported from Gmsh, and improved list handling capabilities.

What is new in version 2.2.1:

• This version contains small improvements and bugfixes.

What is new in version 2.1.0:

• This version adds support for parallel resolution using PETSc solvers, Gmsh2 output format, and experimental SLEPc-based eigensolvers.

What is new in version 2.0.0:

• general code cleanup (separated interface from legacy
• code; removed various undocumented, unstable and otherwise
• experimental features; moved to C++); updated input file formats;
• default solvers are now based on PETSc; small bug fixes (binary .res
• read, Newmark -restart).