discuss-gnustep
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: GNUstep for Engineers


From: Stefan Bidigaray
Subject: Re: GNUstep for Engineers
Date: Wed, 14 Mar 2007 12:13:53 -0400

OK, I've come up with a working name: gnustep-cae!  Since I'd like this to be extremely extensible I have decided to separate it into a bunch of smaller libraries, the first one I'll work on will be the FE library.  It will pretty much be in charge to holding node and element info and create a stiffness matrix as well as a force array.  I will also need a meshing library to create and handle meshes, including all the files types, an issue brought up earlier.  To create a decent software a CAD/Modeling library will also be required.

I've done some poking around and I've come up with a preliminary set of classes/types.  It still needs some major work, and what I'm going to have to do is come up with some simple FE problems and solve them by hand to check what I'm missing.  I can say right away that I still need to create a class like NSApplication which will hold all FENodes, FEElements, etc.  There can only be 1 instance of this class for a particular problem (that's why I said it's kind of like a NSApplication).  I will also need a solver, which will probably just be a function, but it needs to invert and multiply matrices, which I'm hoping MathArray can do since a simple FE problem creates a 1000x1000+ matrix which needs to be solved.  I'm not sure how efficient the program will be at first seeing as it will use a load of memory just to store all the info in these matrices.

I have one question, what do you guys think would be better, to create libraries or bundles?  From what I've read bundles would be a better idea for this case and they can be dynamically loaded, so a program can use one, two or all.

I attached what I'd like to do, of course it's missing a lot of things, but this will give you an idea of how everything will work.  However I think I will need to work on optimizing the code.  Also keep in mind it's still very mechanically oriented, seeing as FE can be used to solve thermo and flow as well, and not just deformation.

Let me get some feedback!  I know some of you guys might not know a word of what I'm talking about, but you have development experience, which I don't.

Stefan

reply via email to

[Prev in Thread] Current Thread [Next in Thread]