Re: About diagonal matrices

[John W. Eaton]

On 26-Feb-2009, Jaroslav Hajek wrote:

| You mean that the first one is correct? Because it gives sparse(NaN(3,1)) !!!
| If yes, then I 100% agree with you, but it's the same situation as
| with the third one.
| At least, I don't see any difference.
| 
| Please, David, before you start modifying anything, I kindly ask you
| to read the whole conversation and at least try to explain to me what
| *exactly* you would like to change and why the current behaviour is
| wrong.

Right, after the recent discussion, I now agree that filling with NaNs
is probably not a good thing.  As far as I can tell, this only affects
operations that can generate NaNs (correct me if that is not right)
and not generating them is probably more useful behavior.  OTOH, we
should fill with other values (for example, when doing something like
"speye (3) + 1").

jwe