On Mon, Nov 16, 2009 at 7:22 PM, Benoit Jacob
<jacob.benoit.1@xxxxxxxxx> wrote:
It seems like what you want is Gael's proposal for "true array support".
Not really, since I am really required to work on matrices - actually I was working on spectral clustering involving eigenvalue decompositions (so real linear algebra stuff).
vector - scalar is not a standard operation; it doesn't have a
geometric meaning. Actually matrix - scalar is sometimes used as
matrix - scalar*Identity, so that's another reason not to let it mean
coeff-wise substraction.
I totally agree, I was just thinking about consistingly supporting (I do even volunteer to implement it)
.cwise() += | -= | /= | *=
and not in general
MatrixBase::operator += etc.
I just think that some problems (formulated in an algebraic framework) do sometimes need individual rescaling and sometimes even shifting when you need to work e.g. on zero-mean input data.
- Hauke