Hi!
I'm not a programmer, I'm actually working with Matlab, and I'm dealing with symbolic matrices: I need to work with 5000x5000 matrices and invert them, I have to solve some linear systems.
But, since I have problems with it -it's very time-consuming, already 10x10 symbolic matrices crash everything, it's not like using matrices filled with numbers- I was thinking whether I can change my programming language: probably c++ or something else could be better suited for solving simple but large systems?
Thank you very much for your suggestions! :w00t:
Jacopo
Suggestions about large Matrices???
Started by jakopo, Feb 26 2010 08:29 AM
10 replies to this topic
#1
Posted 26 February 2010 - 08:29 AM
|
|
|
#2
Posted 26 February 2010 - 05:08 PM
What exactly do you want to do in the other programming language? Only to invert matrices?
#3
Posted 27 February 2010 - 01:46 AM
I'd suggest C/C++, but how are you going to input the matrices? A file? By hand?
sudo rm -rf /
#4
Posted 27 February 2010 - 06:32 AM
Hi!
Yes, I just need to invert a matrix and then to change some of its entries when I need to set them equal to zero. I know it's easy, my problem is just its size.
Well, in Matlab I've written a little file that creates the matrix and then invert it. It works and I can transfer it to c++. I will try!
Thank you very much for the suggestion!
Jacopo
Yes, I just need to invert a matrix and then to change some of its entries when I need to set them equal to zero. I know it's easy, my problem is just its size.
Well, in Matlab I've written a little file that creates the matrix and then invert it. It works and I can transfer it to c++. I will try!
Thank you very much for the suggestion!
Jacopo
#5
Posted 27 February 2010 - 12:25 PM
#6
Posted 28 February 2010 - 07:20 AM
No, it's actually my program that creates the entries of the matrix from a statistical distribution, set them to the proper place and then invert the matrix. I have no files from which I read it... :confused:
#7
Posted 28 February 2010 - 02:18 PM
Oh, ok. So as long as you can code the statistical distribution into C/C++ then you should be fine. I strongly suggest that you compile with maximum speed optimizations. I can tell you right now, though, that coding the RREF bit is going to be a PAIN.
sudo rm -rf /
#8
Posted 01 March 2010 - 12:16 AM
Thank you very much Dargueta, now I've a path to follow! Your indications are precious, and I know what to expect.
So, let's delve into C++ now a face the pain! ;-)
So, let's delve into C++ now a face the pain! ;-)
#9
Posted 01 March 2010 - 12:27 AM
Ooh, pain. I love pain. :D Luckily I found an RREF algorithm already implemented in C (written over six years ago, mind you) so we should be able to tweak it a bit to work with what you need. Most of the work is already done for you.
Attached Files
sudo rm -rf /
#10
Posted 01 March 2010 - 01:18 AM
Uh, many thanks! That's big stuff! :thumbup: I've a lot to learn from this code, as far as I can see.
With this, I can try to begin to translate everything from Matlab.
This is huge Help!!:lol:
Many many thanks!!
Jacopo
With this, I can try to begin to translate everything from Matlab.
This is huge Help!!:lol:
Many many thanks!!
Jacopo
#11
Posted 01 March 2010 - 01:23 AM


Sign In
Create Account

Back to top










