How to Acquire Demo Version of Fortran

[Nhelzki]

Hello all!!! How can we acquire a demo version or evaluation version of Fortran? Is it available in the internet? I want to study how to program the stiffness matrix method for evaluating forces and moments on the structure....Thanks in advance...Cheers!!

 

[IDS]

Did you try searching the Internet for Fortran download?

Anyway, as well as the several open source versions of Fortran I can recommend Silverfrost Fortran, which has a free version for non-commercial use.

But if you are wanting it for self education about structural analysis, why not use a spreadsheet?

e.g.

http://newtonexcelbach.wordpress.com/2009/01/31/frame-analysis-with-excel-1-single-beam/

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[xwb]

There are quiet a few free ones: gfortran, g95 and silverfrost (or salford). Silverfrost is quite slow if it is a big number crunching app but it makes up for that by having a reasonable IDE.

The evaluation ones are Intel (you will need at least a P4 or a dual core otherwise it won't even install), Lahey.

 

[johnhors]

"Silverfrost is quite slow" - ummm, I've found silverfrost to be siginficantly faster than g95 and gfortran on pure number crunching tasks! (I don't believe the figures on timnings published by polyhedron software are correct)

http://www.Roshaz.com

 

[Nhelzki]

Thanks guys for that valuable input...Another question fellows...Is silverfrost capable of doing routines for numerical computation using matrix such as gaussian elimination, inversing matrix, transforming matrix of the form Ff=T^T*Qf?..Thanks in advance for your time...Cheers...I am studying to program analysis of structures using matrix stiffness method linearly at the moment...

 

[xwb]

Re: Silverfrost - I'm writing about the version from 4 years back. Back then, it ran about 3-4 times slower than G95 (both running on W2K on a P120 with 32Mb). Maybe the new version is a lot faster.

Nhelzki - Fortran doesn't have built in libraries for the stuff you want. You need to go to a site like

http://www.physics.unlv.edu/~pang/cp_f77.html.

If you're doing lots of simultaneous equations, use Choleski (or LU decomposition). It is a lot faster than any of the Gaussian techniques, especially for big matrices.

You could always use the Matlab type tools like freemat if want to get the algorithm right first. Unlike Fortran, FreeMat has most of the stuff you require built in. Only problem is it is quite slow. It leaves you time to concentrate on the problem rather than the programming. Once you've fixed the problem, and you want a faster solution, translate to Fortran for speed.