matrix algebra functions

Looking into Feedback I’ve seen this entry:

[quote]Marley Watkins Yesterday at 23:00
matrix algebra is quite useful in many scientific fields. Having built-in functions to perform matrix addition, multiplication, inverse, etc. would be a useful addition to Xojo. Relying on the plug-ins created by a retired math professor is not safe over the long-term.
[/quote]
Well, I completely agree with this, but I don’t think Xojo is going to implement it ever. I remember when I asked to preserve Vector3D class (I use vectors a lot in geometry), but after some time deprecated it was removed.
However, I have something that I like to know: Bob Delaney’s software is a plug-in. I suppose this is faster because this plug-in is written in C. Is it true?
However, there is a handicap since I can not use it in 64 bits. When we are talking about matrix functions and speed we must think in 64 bits, mainly in very big or huge matrices. If it was written in plain Xojo, would it be this fast?
If it were, I would be interested in participating in an open source class for matrices in Xojo. But I fear that plain Xojo code wouldn’t be as faster. Although with the new 64 bits … who knows.

Any comment?

I’m pretty far along with my own matrix plugin for a project here at work. I’ve based it on a free Objective-C class called YCMatrix, which uses vDSP and BLAS to accelerate certain operations quite a bit. It’s 32/64 bit universal. It is currently Mac only, however. If anyone is interested, I might be persuaded to post it, and or the XCode source for anyone to use.

//However, I have something that I like to know: Bob Delaney’s software is a plug-in. I suppose this is faster because this plug-in is written in C. Is it true?
However, there is a handicap since I can not use it in 64 bits. When we are talking about matrix functions and speed we must think in 64 bits, mainly in very big or huge matrices. If it was written in plain Xojo, would it be this fast?
//

My fp plugin 10.0 is written in C++. It supports 64-bit applications. It handles integer, rational, complex, and floating point matrices with an extensive set of functions.

Bob

The integer and rational matrices in fp plugin use my BigInteger type which has unlimited length. Complex and floating point matrices use my BigFloat type with precision (unlimited) set by the user.

Since fp plugin supports 64-bit applications, it should be useful for years to come.

I am a retired physics professor.
Bob

Does that mean you teach retired physics? :slight_smile:

I didn’t know physics had wheels.

The standard for highly optimized math functions, e.g. for linear algebra
nowadays seems to be LaPack; it takes over from LinPack.

See the link for further info :[url=http://www.netlib.org/lapack/]

It contains c-callable functions for almost every need.

I didn’t mean to imply that there was anything lacking in your plugin, Robert. In fact, I’m certain it’s far more mature than this thing I banged together in two days. My only impulse here was a willingness to contribute.