Conference rusure::math

    iam not sure exactly what you mean by numerical math? but 
    this is a good book on numerical methods and software , it  comes
    with a floppy disk of lots of FORTRAN routines from Linpack, Quadpack
    Mindpack and SLATEC (I only heard of Linpack, dont know what the 
    other 3 are). the book is called "numerical methods and software"
    by David Kahaner and Cleve moler and stephen Nash.

    some of its contents: - computer arithmetic and computational errors,
    interpolation, numerical quadrature, linear least squares data fitting,
    solution of nonlinear equations, ODE, optimization and nonlinear
    least squares, simulation and random numbers, FFT.


    there is also offcourse the yellow book. "numerical recipes in X"
    where X is C, FORTRAN or Pascal. my advice is to get the FORTRAN one
    if you want to get this. the C version they shift all their arrays
    off by one to make the index start at 1 not zero (which is what C uses)
    and it is kind'a confusing. so I'd advice to get the FORTRAN or Pascal 
    if you must.  (any one wants to trade their FORTRAN numerical recipes
    with a C one? I'd swap with you ;-)

    \nasser


    

    I think one of the "Numerical Recipes" series (there are versions
    in fortran, c, pascal) would be what you are looking for.  New
    editions have just come out; they're published by Cambridge
    University Press (Flannery, Press, et al...)

    The algorithms are not the very best (public domain implentations
    of the state of the art algorithms for common things are available
    from netlib), but the book is nonetheless a good introduction
    and the routines are not bad either.

    Stoer and Bulirsch is a very good book for theory and covers a
    lot of interesting material, but you also need some more practical
    books to get a balanced treatment.

    I have an old copy of the Schaums book you mention, the English
    version doesn't seem to be too bad errorwise.  It lacks a lot
    of modern material, but does have some neat stuff, like a good
    discussion of finite differences.

    The book by Moler et al mentioned previously doesn't cover as
    much material as the Recipes books, which are really self contained
    in that they have full listing of the programs.  Moler treats the
    routines as magic black boxes, something I am not in agreement with.

    There are books on every specialty in numerical analysis
    (linear equations, eigenvalue problems, sparse and iterative methods,
    polynomial and other equation root finding, signal processing,
    numerical integration, solving of ordinary differential equations,
    statistics, etc.)  I'm familiar with this literature and can give some
    recommendations if you are interested in a specific topic, but
    the bibiographies in the sections of the Numerical Recipes books
    are a good pointer to the literature.

    - Jim

    i forgot to mention also the "problem solver" book on numerical
    analysis. i know this is not a Text book , but going over problems
    in problems solver always been a great learning for me, i have many
    of these books, and they are really great, they solve many problems
    for you in details, so you learn this way. i always
    find that i learn much better by working out problems that just
    reading the theory or the method. i like to see how it is
    used in practical examples...you know what i mean. 

    so, if you want to learn numerical analysis, or many other techcnial
    subjects for that matter, check out the problem solver books, 
    and you'll thank me for it too.

    \bye
    \nasser

    New extended and corrected versions of the Num. Rec. books have come
    out recently.  I have not yet seen any of them, but do intend on
    purchasing one.

				    Topher

    when i talked to num. recipes. people here in mass. about a year
    or so ago, they said they might get one out in ADA. 
    
    \nasser