Choose role of NaN when summing two matrices

Dear all

Some time ago, I posted this question: Choosing the role of NaN elements in the sum environment of matrices. That worked great when doing the mean value, but now imagine that I just want to sum the two matrices, given by:

aa=[1 2; 3 NaN];
bb=[NaN 1; 2 NaN];

so that I obtain:

cc=[1 3; 5 NaN];

If I do something like:

cc=sum(cat(3,aa,bb),3,'omitnan')

that gives:

cc=[1 3; 5 0];

Which way would be the best in this case?

Answer 1

[Voss]

Here's one way:

aa=[1 2; 3 NaN];
bb=[NaN 1; 2 NaN];

tmp = cat(3,aa,bb);
cc=sum(tmp,3,'omitnan');
cc(all(isnan(tmp),3)) = NaN

cc = 2×2

1 3 5 NaN

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>