The null() returns empty vector, how can I loose it
in [Matlab]
|
Prev: Heading Control
Next: While Loop and Intersect
From: Donya Gozar on 28 May 2010 10:53 Hi everyone, I have Matrix (H) with random complex coefficient, now I need the orthogonal basis for the null space of H. When I use Matlab command null (H), it returns an empty vector which is really not desired because I need to multiple this matrix (the result of null (H)) with something else, so how can I combat this problem? Is there any idea or help? I read today that the reason is that, H is invertible, I don’t know if it is the problem how I can change my matrix and if I do it does it influence my last result? Please help me… Thank in advance Donya
From: Matt J on 28 May 2010 11:05 "Donya Gozar" <dgozar(a)yahoo.com> wrote in message <htolch$sc9$1(a)fred.mathworks.com>... > Hi everyone, > > I have Matrix (H) with random complex coefficient, now I need the orthogonal basis for the null space of H. ==================== But clearly H is nonsingular here and its null space is simply {0}. Such a null space has no basis. You have to decide how your code should behave in this special case.
From: Donya Gozar on 28 May 2010 15:37 "Matt J " <mattjacREMOVE(a)THISieee.spam> You have to decide how your code should behave in this special case. Hi Matt thanks for ur reply, in my case, the orthogonal basis for null spce of matrix (H) should have some vectors!!! because I must multiple it with some other matrix to find out the output variable (Matrix), so I am now total confused how to make this nonsingular matrix to Sigular and if i do that does it influence my final result?I would be really happy if somebody can help me... thanks
From: Matt J on 28 May 2010 15:50 "Donya Gozar" <dgozar(a)yahoo.com> wrote in message <htp611$oeq$1(a)fred.mathworks.com>... > "Matt J " <mattjacREMOVE(a)THISieee.spam> You have to decide how your code should behave in this special case. > Hi Matt thanks for ur reply, in my case, the orthogonal basis for null spce of matrix (H) should have some vectors!!! because I must multiple it with some other matrix to find out the output variable (Matrix), so I am now total confused how to make this nonsingular matrix to Sigular and if i do that does it influence my final result?I would be really happy if somebody can help me... ============= Nobody is going to be able to help you, because you haven't provided enough information for anyone to know what this code is meant to accomplish, or why the input matrix is always expected to be singular. If H is expected to be singular, what has changed so that now all of a sudden you are starting to get nonsingular inputs...?
From: Walter Roberson on 28 May 2010 15:50
Donya Gozar wrote: > "Matt J " <mattjacREMOVE(a)THISieee.spam> You have to decide how your > code should behave in this special case. > Hi Matt thanks for ur reply, in my case, the orthogonal basis for null > spce of matrix (H) should have some vectors!!! because I must multiple > it with some other matrix to find out the output variable (Matrix), so I > am now total confused how to make this nonsingular matrix to Sigular To make a matrix singular, you can make any row or column the same as (or a multiple of) any other row or column (respectively). Or randomly set enough entries to 0, I suppose. > and > if i do that does it influence my final result? Since you are generating your matrix randomly, any random singular matrix is as good as any other matrix, unless you are doing monte carlo calculations (in which case forcing singularity would distort the probabilities.) But of course it will influence your final result. Your final result at present is a crash, so making the crash go away would be a different final result. Will forcing the matrix to be singular give you meaningful results for your purpose? I wouldn't think so. On the other hand, I have my doubts about the validity of any process that requires a random singular matrix and does not specify what you should do if the matrix comes out non-singular. Garbage In, Garbage Out. |