Now tensor multiplications and tensor transpose is immediately evaluated creating temporaries. Instead one could easily generate an expression which is evaluated when encountering the assign operator.
Need to implement lazy evaluation also for allowing einstein summation notation with multiple multiplication such as D = A(_i,_j,_k) * B(_j,_k,_l) * C(_l,_m)
Now tensor multiplications and tensor transpose is immediately evaluated creating temporaries. Instead one could easily generate an expression which is evaluated when encountering the assign operator.