[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Help-gsl] example of linear fitting does not release dinamical memo
Re: [Help-gsl] example of linear fitting does not release dinamical memory!
Tue, 20 Jul 2004 12:23:07 -0800
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7) Gecko/20040616
Thank you, Brian. That was written that PDE algorithms are to vary and
huge. It is correct. Probably, I will better start a different library
based upon GSL for that.
A few words are about alloc and free. I think it is better style ever
have a coupled alloc-free functions. Never leave it for compilers. In my
own program (separate function) it might not work, for example. You are
right that it does not matter in that example, but I would add this anyway.
Brian Gough wrote:
Anton Kulchitsky writes:
> There are 3 calls of gsl_vector_alloc and gsl_matrix_alloc. However,
> there are no calls of gsl_vector_free and gsl_matrix_free. Why there are
> no needs for this?
As you probably know, memory is automatically freed when a program
exits so it is not essential to use a free at the end of main().
> PS. I have another question. Do authors plan to include some PDE work
> into the library? I have my own rather comprehensive and good tested
> code for hyperbolic equations (different one-step finite volume methods
> from upwind to FCT and TVD schemes, 2D+time and 1D+time). That could be
> adopted for GSL. And some simple code for parabolic equations (1D+time
> and 2D+time). I also could write and test some code for elliptic
> equations (something lile the code presented in Numerical Recepies
> library and maybe for complex numbers). I could work on it (during this
> year, this is a necessary part of my current project) and submit the
> code and docs (all I need for my documentation is verify my terrible
> English). Do you need it? Do you have any docs for programmers (like GSL
> programming guide or so)?
See the "GSL Design Document" at http://sources.redhat.com/gsl/
which discusses PDEs.