PLTMG_DOUBLE
Piecewise Linear Multigrid
Finite Element Program
(Double precision)

PLTMG_DOUBLE is a library of FORTRAN77 routines, using double precision arithmetic, which can be used to define and solve a finite element model of a partial differential equation, using piecewise linear basis functions and the multigrid method.

PLTMG_DOUBLE can be compiled with the MPI library, in which case it can be run in parallel.

PLTMG_DOUBLE can be compiled with the X11 library, in which case it will use an interactive X Window based interface. However, in this case it relies on the Motif interface as well, which is likely to be available on Sun machines, but not on many others.

PLTMG_DOUBLE can be compiled with an OpenGL interface, which can be used to display the graphics created by the program.

Related Data and Programs:

The FEM data directory contains a description of the data files that can be used to describe a finite element model.

FEM_50 is a MATLAB finite element program in just 50 lines of code.

FEM_50_HEAT is modified version of FEM_50 suitable for solving the heat equation.

FEM_IO is a set of MATLAB routines for reading or writing the node, element and data files that define a finite element model.

FEM_TO_TEC is a MATLAB program that can convert an FEM model into a TEC graphics file.

FEM2D_POISSON is a FORTRAN90 program for solving Poisson's equation on a square, using the finite element method.

FEMPACK is a FORTRAN90 library of routines for various basis finite element calculations.

FREE_FEM_POISSON is a FORTRAN90 program for solving Poisson's equation on a triangulated region, using the finite element method.

FREE_FEM_STOKES is a FORTRAN90 program for solving the Stokes equations on a triangulated region, using the finite element method.

HOT_PIPE is a sample problem that can be run with the MATLAB script FEM_50_HEAT.

HOT_POINT is a sample problem that can be run with the MATLAB script FEM_50_HEAT.

The MPI library may be included with PLTMG_DOUBLE, in which case the problem will run in parallel.

The OpenGL library may be included with PLTMG_DOUBLE, to allow various kinds of graphics displays.

PLTMG_SINGLE is a single precision version of the PLTMG_DOUBLE library.

The TRIANGULATION_ORDER3 directory contains a description and examples of how an order 3 (linear) triangulation is defined by a node and element file.

The TRIANGULATION_ORDER6 directory contains a description and examples of how an order 6 (quadratic) triangulation is defined by a node and element file.

TRIANGULATION_ORDER6_CONTOUR is a MATLAB script which can make contours of a scalar quantity defined on the nodes of an order 6 triangulation.

The X_WINDOW library is needed to implement some of the related graphics display programs.

The XDR library may be needed to run PLTMG_DOUBLE in parallel.

Author:

Randy Bank

Reference:

1. Randolph Bank,
   PLTMG: A Software Package for Solving Elliptic Partial Differential Equations,
   User's Guide, Version 8.0
   SIAM, 1994.
2. Randolph Bank,
   PLTMG: A Software Package for Solving Elliptic Partial Differential Equations,
   User's Guide, Version 9.0
   Available online.
3. The PLTMG Quick Reference Web Pages

Source Code:

• pltmg_double.f, the vector code;
• mgxdr.c, an interface to the XDR library (external data representation), which may also require the RPCSVC library;
• pltmg_double.csh, commands to compile the source code;

Examples and Tests:

ATEST is a main program that is required when running any of the following test programs. ATEST can operate in one of three modes, determined by the value of the MODE parameter:

• MODE=-1, run as an interactive program, with a simple terminal interface;
• MODE=0, run as an interactive program, with an X Window interface which allows graphical output;
• MODE=+1, run as a batch program, with input read from a "journal" file;

ATEST can be run with a true or dummy MPI interface, depending on whether MPI is available. ATEST can be run with a true SG OpenGl graphics display interface, or a dummy interface, depending on whether the OpenGL graphics library is available. ATEST can be run with a true X Window graphics display interface, or a dummy interface, depending on whether the X Window library is available.

• atest.f, a calling program;
• mgmpi.f, the MPI interface; if this is used, the user must also make the MPI libraries available;
• mgmpi_stubs.f, the dummy MPI interface, which keeps the compiler and loader happy, but means that the MPI library is not needed and cannot be used;
• mgvio.f, the OpenGL interface; if this is used, the user must also make the OpenGL graphics libraries available;
• mgvio_stubs.f, the dummy OpenGL interface, which keeps the compiler and loader happy, but means that the OpenGL graphics library is not needed and cannot be used;
• xgui.c, the X Window interface; if this is used, the user must also make the X Window libraries available;
• xgui_stubs.c, the dummy X Window interface, which keeps the compiler and loader happy, but means that the X Window library is not needed and cannot be used;

BATTERY defines the thermal battery problem.

• battery.f, a calling program;
• battery.csh, commands to compile, link, load and run the calling program;
• battery.out, the output.

BURGER defines the Burgers equation problem.

• burger.f, a calling program;
• burger.csh, commands to compile, link, load and run the calling program;
• burger.out, the output.

CIRCLE defines the circle problem.

• circle.f, a calling program;
• circle.csh, commands to compile, link, load and run the calling program;
• circle.out, the output.

CONTROL defines the control problem.

• control.f, a calling program;
• control.csh, commands to compile, link, load and run the calling program;
• control.out, the output.

DOMAINS defines the domains problem.

• domains.f, a calling program;
• domains.csh, commands to compile, link, load and run the calling program;
• domains.out, the output.

IDENT defines the IDENT problem.

• ident.f, a calling program;
• ident.csh, commands to compile, link, load and run the calling program;
• ident.out, the output.

JCN defines the JCN problem.

• jcn.f, a calling program;
• jcn.csh, commands to compile, link, load and run the calling program;
• jcn.out, the output.

MESSAGE defines the message problem.

• message.f, a calling program;
• message.csh, commands to compile, link, load and run the calling program;
• message.out, the output.

MNSURF defines the MNSURF problem.

• mnsurf.f, a calling program;
• mnsurf.csh, commands to compile, link, load and run the calling program;
• mnsurf.out, the output.

NACA defines the NACA problem.

• naca.f, a calling program;
• naca.csh, commands to compile, link, load and run the calling program;
• naca.out, the output.

OB defines the OB problem.

• ob.f, a calling program;
• ob.csh, commands to compile, link, load and run the calling program;
• ob.out, the output.

SQUARE defines the square problem.

• square.f, a calling program;
• square.csh, commands to compile, link, load and run the calling program;
• square.out, the output.

You can go up one level to the FORTRAN77 source codes.