rank() needs Numerical optimization!

#1 Опубликовано: 13.02.2012 16:52:16

325 сообщений из 2052 понравились пользователям.

Группа: Moderator

Hello,

It seems that function rank() will fail with Symbolic optimization.

[MATH=eng]S←mat(2,0,2,0,0,0,1,-1,2,5,-1,1,-2,-3,0,-2,-2,0,0,-6,0,0,0,6,6,0,0,0,-4,-4,6,5)[/MATH]
[MATH=eng]N.r←rank(S)[/MATH]
Symbolic:
[MATH=eng]N.r=0[/MATH]
Numeric:
[MATH=eng]N.r=3[/MATH]
None:
[MATH=eng]N.r=3[/MATH]

Regards,
Radovan

When Sisyphus climbed to the top of a hill, they said: "Wrong boulder!"

#2 Опубликовано: 14.02.2012 08:36:24

Radovan Omorjan

325 сообщений из 2052 понравились пользователям.

Группа: Moderator

Hello,

Rather strange! Please look at this picture:

And the associate SMath file rank.sm

Could someone read this SMath file and check it out, please. Compare it to the picture above obtained from my computer.
Why the rank is different for matrix S but the same for matrix A by using different Optimization?

Regards,
Radovan

When Sisyphus climbed to the top of a hill, they said: "Wrong boulder!"

#3 Опубликовано: 14.02.2012 09:46:12

Davide Carpi

1416 сообщений из 2873 понравились пользователям.

Группа: Moderator

Hi,

I've the same problem using your file... the problem seems to be the matrix dimension (S is 6x5 matrix, A is 5x5 matrix... if you enlarge S adding a new column the difference disappears)

regards

w3b5urf3r

If you like my plugins please consider to support the program buying a license; for personal contributions to me: paypal.me/dcprojects

#4 Опубликовано: 14.02.2012 10:41:31

Radovan Omorjan

325 сообщений из 2052 понравились пользователям.

Группа: Moderator

Wrote
I've the same problem using your file... the problem seems to be the matrix dimension (S is 6x5 matrix, A is 5x5 matrix... if you enlarge S adding a new column the difference disappears)

You are right, thank you.

The same happened when we "shrink" the matrix A into the 5x4 matrix. On the other hand rank(A)= 0 only for A equals to zero matrix:

[MATH=eng]rank(mat(0,0,0,0,0,0,2,3))=0[/MATH]

Nevertheless, it should not be this way - the rank of a matrix should not depend on the Optimization applied. There are some other questionable situations where the result is a zero matrix (all zero values in Jacobain matrix and Symbolic optimization, for example- probabelly a bug as well). I suppose this one is a bug too.

Regards,
Radovan

When Sisyphus climbed to the top of a hill, they said: "Wrong boulder!"

#5 Опубликовано: 14.02.2012 12:34:09

Andrey Ivashov

2270 сообщений из 3734 понравились пользователям.

Группа: Super Administrator

Thank you! Fixing...

1 пользователям понравился этот пост

Radovan Omorjan 14.02.2012 12:35:00

#6 Опубликовано: 14.02.2012 17:04:05

Andrey Ivashov

2270 сообщений из 3734 понравились пользователям.

Группа: Super Administrator