7

I want to create a finite element code in fortran. There are many libraries out there on internet. Can someone suggest me a good finite element library in fortran?

I am looking for

  • Ease of use (using fortran language specifically)
  • Well documented
  • free
Christian Clason
  • 12,301
  • 3
  • 48
  • 68
343_458
  • 171
  • 2

2 Answers2

5

I would recommend writing your own code using PETSc. The most complex part of a parallel FE code is parallel assembly and solve and PETSc takes care of both. The latest version of PETSc also has routines for managing unstructured meshes meaning you can write a performant, scalable FE code in a just a few hundred lines.

stali
  • 1,759
  • 1
  • 10
  • 13
2

There is of course a good question why exactly you'd like to create a new finite element library in 2015 when there are already excellent ones out there. There's also the question why you would want to use Fortran in 2015 when that's simply not any more what the community believes is the language to use for new projects.

But if you believe that you have good reasons for both of these questions, then it may be worthwhile looking into PLTMG -- the grandfather of all modern finite element libraries, and written in Fortran. As others have pointed out, PETSc also provides interfaces for Fortran and has finite element support built in.

Wolfgang Bangerth
  • 55,373
  • 59
  • 119
  • 2
    thanks for your suggestions. There's also the question why you would want to use Fortran in 2015 when that's simply not any more what the community believes is the language to use for new projects. Really? There shouldn't be any doubt that optimized fortran programs will be computationally cheaper as compared to c++; papers show benchmarking. And fortran is still language of modern scientific programming. Many commercial finite element codes are still in fortran. – 343_458 Oct 02 '15 at 23:22
  • No, every evidence over the last ten years shows that neither of the two languages is more efficient than the other. And it isn't, in all practice, the language of modern scientific computing: all widely used software libraries that were started over the past 15 years are in fact written in C or C++. This is true for basically all finite element libraries, as well as packages such as PETSc and Trilinos. Codes that are still written in Fortran are overwhelmingly legacy codes that have been around since at least the 1980s. – Wolfgang Bangerth Oct 04 '15 at 17:45