saving a matrix

Hi,

I am trying to compare my two codes , which should give same results. Both code have the same variable names such as vx vz. I am trying to calculate velecotiy profiles. For that I open to Matlabs and run both codes . Than I write

vz1=vz; save ( ' FT','vza')

and I load FT in the other matlab window, but it directly takes vz values in the the program. And I get 0 as difference, where I should be getting something around 1e-15 ( I know that because I copy paste the results from one to other).

Is there a way to solve my problem witouth changing the variable names?

[Paulo Silva]

Your question isn't very clear and you didn't provided the needed code.

[Kamuran]

Let me try to explain, because I can not provide my all code (it is too long) . I have two different codes , one is evolved from the other, therefore, they have the same variable names, only difference is the way I present my equations.

I am calculating a velocity profile in a cavity such as vz. And I want to compare two results, for that I open two matlabs at the same time and run the codes. Then, in one Command window I write

FT=vz; % vz is (200*200) matrix

save FT

After that I load FT in the second command window;

Load FT

max(max(abs(vz-FT))) % this should be around 1e-15 but matlab gives me 0 . As Matt point out Matlab is overwritting all the variables in the second command window when I load it.

Answer 1

[Matt Fig]

When you load, use the functional form of LOAD in order to avoid overwriting your variables.

X = load('FT');
% Then use X.vz, etc.

.

.

.

.

.

EDIT

As I state below in the comment, copy paste is not reliable with floating points. I want to say it again here so the formatting is better.

format long
pi  % copy this and paste below
pi- 3.14159265358979
ans =
     3.10862446895044e-015

[Kamuran]

It is not working.

I wrote

vza=vz; % just to change name hoping that it will work.

FT=vza;

save FT

and in the other window

X=load ('FT')

max(max(abs(vz-X.vza)))

result is 0

[Matt Fig]

If you did what I showed above and got zero difference, I have to question the method you used to find a real difference. Copy and paste with floating point numbers is often unreliable. Try this:

format long

pi % Now copy this using ctrl+c

pi- 3.14159265358979

ans =

3.10862446895044e-015

Answer 2

[Robert Cumming]

The error could be coming from the copy/paste between the two matlabs.

Are you 100% sure that you have different values?

When I do the same as your doing I get a max of 0 (as you do), and if I copy the data from the editor from 1 matlab to another I get an difference of ~1e-15.

But I know my numbers are exactly the same, so the difference is definately coming from the copy and paste step.

Answer 3

[Kamuran]

Yes you are right, the difference is coming from copy paste .

Thank you