Thread Subject: How to judge a large matrix containing a Nan?

I now have a sparse matrix which is computed by me(coming from a discretization of partial differential equation,I have used mex file ).
for example:A is a 10000 by 10000 sparse matrix.
x=A\b (in which b is a 10000 by 1 matrix).
I have found that there is some Nan in x.I guess maybe A also has some nan.How to judge that A(i,j) is Nan?Thank you all

On Jan 1, 12:12 am, "zedong " <zdon...@gmail.com>
>How to judge that A(i,j) is Nan?

The isnan() function will do this for you. In particular,

nnz(isnan(A))

will tell you how many NaN's are in the matrix A, while

isnan(A)

will give you a sparse matrix with 1's in the same positions as the
NaN's in A.

On 1 Jan, 08:12, "zedong " <zdon...@gmail.com> wrote:
> I now have a sparse matrix which is computed by me(coming from a discretization of partial differential equation,I have used mex file ).
> for example:A is a 10000 by 10000 sparse matrix.
> x=A\b (in which b is a 10000 by 1 matrix).
> I have found that there is some Nan in x.I guess maybe A also has some nan.How to judge that A(i,j) is Nan?Thank you all

ISNAN should tell you whether A contains NaNs. If it does, you would
want to have a look at the mex file which generates A to find out
why they occur.

Rune