Replace zeros with NaN's - when there are already NaN's in the matrix?

Hi

I have a large square matrix which is made up of smaller square matrices which have some NaN values. I used 'blkdiag' to combine the matrices diagonally and all the gaps were filled in with zeros.

I now wish to replace all of the zeros in the large matrix (F) with NaN's.

I tried F(find(~F))=NaN but I get an error saying NaN's cannot be converted to logicals, which makes sense but now I'm not sure how I can get round it. I am fairly new to Matlab.

Any help would be much appreciated.

Answer 1

[Sean de Wolski]

No reason for find and you can't convert nans to logicals.

F(F==0) = nan;

[scour_man]

Great, thanks. Seems obvious now! And yes its to stop zero values showing up in surf and contour plots in between areas of data. Thank you

Answer 2

[Ivan van der Kroon]

Change to double first:

F=double(F);
F(find(~F))=NaN;

Just curious; why would you need this?

[Sean de Wolski]

One use could be to have them show up blank in mesh or surf.

Answer 3

[Andrei Bobrov]

F(~isnan(F))=NaN;

[Sean de Wolski]

That'll switch the whole matrix to nans, not just zero values.

[Andrei Bobrov]

Oops! my bug...

Thanks Sean!