Re: [tlug] [OT] Japan and ?? credit cards

[Edward Middleton <emiddleton@example.com>]

Stephen J. Turnbull wrote:
> Edward Middleton writes:
> 
>  > Standard FORTRAN77 doesn't support dynamic memory allocation, i.e. you
>  > have to know the size of all matrix when you write the program
> 
> This is usually not a problem for the kinds of analysis I'm talking
> about.  The data sets are infinitely large, but the whole point of the
> analysis is to summarize them statistically and that reduces to a
> matrix of known dimensions, independent of the number of observations.

Fare enought, it was a deal breaker for me because every calculation
required a different size matrix.  After using both I am glad I went
with fortran90 on a sparc because it was about 100x faster then the
equivalent in g77 and produced more accurate results.

>  > FORTRANS major advantage is speed, but depending on what your doing
>  > it is debatable whether it will be any faster then Matlab,
> 
> That depends on whether Matlab already has the optimized libraries.
> This is unlikely for state-of-the-art crisis-causing derivative
> securities and/or theoretical advances.

I don't know, most numerical calculation break down to a few basic
matrix calculations which have optimized libraries for the specific
hardware.  It is obviously possible to go further then this in optimize
specific algorithms for specific hardware, and there has been some
interesting haskell research into automatically doing this for video
card graphics processors and ocaml for producing hardware optimized FFT
but it is probably more cost effective to just throw more hardware at
the problem.

Edward