Skip to content

Expandable And Scalable Infrastructure for Finite Element Methods, EASIFEM, is [Modern Fortran](https://fortran-lang.org) framework for solving partial differential equations (PDEs) using finite element methods. EASIFEM "eases" the efforts to develop scientific programs in Fortran.

License

easifem/base

Repository files navigation

π‘—• easifemBase

easifemBase (or, Base) library is the low level component in easifem. It contains routines and derived types which are helpful for implementing numerical methods for solving differential equation.|

In Base library, we do not use object-oriented programming concepts and mainly use multiple dispatch approach. This approach improves the flexibility and speed of easifemBase. All user-defined datatypes are declared in the BaseType module, and all methods are exposed through BaseMethods modules. In the Base library string_class is the only exception, wherein Object-oriented paradigm has been used. Currently, easifemBase has interface with BLAS95, Lapack95, Sparsekit, Metis, PlPlot, SuperLU, ARPACK, etc.

Usage

Use association

USE easifemBase

or

USE BaseType
USE BaseMethods

System requirements

EASIFEM requires following software packages to be installed on the system.

Component Version Latest tested version Comment
Gfortran >=9.0 12.0 GNU Fortran compiler
GCC >=9.0 12.0 GNU-compiler collection
OpenMP >= 4.5 Multithread shared memory parallelisation
Curl >=7.87 7.87 A command-line utility for transferring data from or to a remote server
Git >=2.34 2.34.1 A version control system and command-line utility for downloading packages from GitHub
Cmake >=3.19 3.22.4 Cross-platform family of tools designed to build, test and package software
Ninja-build >=1.10 1.11.0 Build system
Python3 >=3.7 3.11.0 Scripting language
Pip >=20 23.1.0 Command line tool for downloading python packages
LAPACK >=3.11.0 3.11.0 Linear algebra package
OpenBlas >= 0.3.20 0.3.30 Optimize BLAS library
HDF5 >=1.10 1.10.7 High-performance data software-library and file-format
PlPlot >=5.15.0 5.15.0 Cross-platform, scientific graphics plotting library
Boost
Gnuplot >=5.0 5.4 Portable command-line driven graphing utility
Doxygen >=1.9.1 1.9.1 documentation generation
GTK-4 n

External packages

EASIFEM depends upon the following external packages (extpkgs) that are not shipped with the source-code.

extpkg description command
OpenBlas Highly optimized BLAS easifem install openblas
SuperLU Direct solution of large, sparse, nonsymmetric systems of linear equations easifem install superlu
LIS Linear interative solver easifem install lis
METIS Mesh partitioning library easifem install metis
SCOTCH Mesh partitioning library easifem install scotch
ARPACK Eigensolver for sparse matrices easifem install arpack
FFTW Fast Fourier Transform easifem install fftw
GTK-Fortran Fortran bindings for GTK-4 library easifem install gtk-fortran
LAPACK95 Fortran 95 interface for Lapack library easifem install lapack95
Sparsekit Fortran library for sparse matrices easifem install sparsekit
Gmsh Finite element mesh generator easifem install gmsh

Installation

You can use following instructions to install easifemBase depending upon your system.

Structure

The Base library consists two components:

  1. BaseType BaseType.F90, which contains the user-defined data-type. You can see the list of user-defined data type here
  2. BaseMethods BaseMethods.F90, contains the modules (each module defines the routines for data-types defined in BaseType.F90.) The list of modules defined in BaseMethods can be found here

The source directory is shown in figure given below. The source directory has two directories

  1. πŸ“ modules
  2. πŸ“ submodules

The modules directory mainly contains header and interface of methods. The implementation is given in submodules directory.

:::info Both BaseType.F90 and BaseMethods.F90 are included in modules directory. :::

Let us understand the structure of the Base library by an example of CSRSparsity_ data type.

  1. First, we define CSRSparsity_ in BaseType.F90 as
TYPE :: CSRSparsity_
  INTEGER(I4B) :: nnz = 0
  INTEGER(I4B) :: ncol = 0
  INTEGER(I4B) :: nrow = 0
  LOGICAL(LGT) :: isSorted = .FALSE.
  LOGICAL(LGT) :: isInitiated = .FALSE.
  LOGICAL(LGT) :: isSparsityLock = .FALSE.
  LOGICAL(LGT) :: isDiagStored = .FALSE.
  INTEGER(I4B), ALLOCATABLE :: IA(:)
  INTEGER(I4B), ALLOCATABLE :: JA(:)
  INTEGER(I4B), ALLOCATABLE :: idiag(:)
  TYPE(IntVector_), ALLOCATABLE :: row(:)
  TYPE(DOF_) :: idof
  !! DOF for row
  TYPE(DOF_) :: jdof
  !! DOF for columns
END TYPE CSRSparsity_
  1. Then we create a directory called CSRSparsity in both modules and submodules directory.
  2. In modules/CSRSparsity we create CSRSparsity_Method.F90 file.
  3. In modules/CSRSparsity/CSRSparsity_Method.F90 we define a module CSRSparsity_Method (same name as file).
  4. In CSRSparsity_Method module, we only define interface of methods. In this way, this file can be considered as header file. See, the example given below:
  5. In submodules/CSRSparsity, we create CSRSparsity_Method@ConstructorMethods.F90, which contains the contruction related routines.
  6. Also, we create CSRSparsity_Method@IOMethods.F90, which include methods related to input and output.
MODULE CSRSparsity_Method
USE GlobalData
USE BaseType
IMPLICIT NONE
PRIVATE

INTERFACE Initiate
  MODULE SUBROUTINE csr_initiate1(obj, ncol, nrow, idof, jdof)
    TYPE(CSRSparsity_), INTENT(INOUT) :: obj
    INTEGER(I4B), INTENT(IN) :: ncol, nrow
    TYPE(DOF_), OPTIONAL, INTENT(IN) :: idof
    !! DOF for row
    TYPE(DOF_), OPTIONAL, INTENT(IN) :: jdof
    !! DOF for column
  END SUBROUTINE csr_initiate1
END INTERFACE Initiate

INTERFACE Display
  MODULE SUBROUTINE csr_Display(obj, Msg, UnitNo)
    TYPE(CSRSparsity_), INTENT(IN) :: obj
    CHARACTER(*), INTENT(IN) :: Msg
    INTEGER(I4B), OPTIONAL, INTENT(IN) :: UnitNo
  END SUBROUTINE csr_Display
END INTERFACE Display

END MODULE CSRSparsity_Method

CSRSparsity_Method@ConstructorMethods.F90

SUBMODULE(CSRSparsity_Method) ConstructorMethods
USE BaseMethod
IMPLICIT NONE
CONTAINS

MODULE PROCEDURE csr_initiate1
obj%nnz = 0
obj%ncol = ncol
obj%nrow = nrow
END PROCEDURE csr_initiate1

END SUBMODULE ConstructorMethods

CSRSparsity_Method@IOMethods.F90

SUBMODULE(CSRSparsity_Method) IOMethods
USE BaseMethod
IMPLICIT NONE
CONTAINS

MODULE PROCEDURE csr_Display
CALL Display(Msg, unitNo=unitNo)
CALL Display(obj%nnz, "# NNZ : ", unitNo=unitNo)
END PROCEDURE csr_Display

END SUBMODULE IOMethods

Run on Cloud

Coming soon.

Contributing

Credits

License

License

About

Expandable And Scalable Infrastructure for Finite Element Methods, EASIFEM, is [Modern Fortran](https://fortran-lang.org) framework for solving partial differential equations (PDEs) using finite element methods. EASIFEM "eases" the efforts to develop scientific programs in Fortran.

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published