sum of matrix always results in a 1x1 matrix

[MDeiml]

If you apply the sum function to a list of nxm matrices you get a 1x1 matrix as result e.g.
sum [M.fromLists [[1,1]], M.fromLists [[2,2]]] == ( 3 )

[Daniel-Diaz]

Right, and I guess this is cause by the sum function actually evaluating your sum as:

M.fromLists [[1,1]] + M.fromLists [[2,2]] + fromInteger 0

Function fromInteger will create the 1x1 matrix ( 0 ), and this is the reason you are getting a 1x1 matrix as a result.

I guess an alternative sum function for matrices will help?

[MDeiml]

Actually M.fromLists [[1,1]] + M.fromLists [[2,2]] + fromInteger 0 crashes with

getElem: Trying to get the (1,2) element from a 1x1 matrix.

What really gets evaluated is fromInteger 0 + fromLists [[1,1]]
Maybe elementwise should be rewritten to accept matrices of different size?

[MDeiml]

This should work:

elementwise f m m' = matrix (max (nrows m) (nrows m')) (max (ncols m) (ncols m')) $
  \k -> f (m ! k) (m' ! k)

[Daste]

Pheraps creating a function like:
elementWiseMonoid :: (Monoid a,Monoid b) => ( a -> b -> c ) -> Matrix a -> Matrix b -> Matrix c elementWiseMonoid f m m' = elementWiseUnsafe f (setSize rows columns m) (setSize rows columns m') where rows = max ( nrows m ) ( nrows m' ) columns = max ( ncols m ) ( ncols m' )
and using it to instantiate Num could help.
I did not test that.