Re: Octave and cargo cult programmingHello

[Sergei Steshenko]

----- Original Message -----
> From: Juan Pablo Carbajal <address@hidden>
> To: Sergei Steshenko <address@hidden>
> Cc: Judd Storrs <address@hidden>; Octave users list <address@hidden>
> Sent: Friday, September 14, 2012 1:04 AM
> Subject: Re: Octave and cargo cult programmingHello
[snip]
> 1d array in simple English (you like lisp, they are called lists
> there!) might be ok, but I prefer some linear algebra language around
> here where x' and x are different, and of course a huge community of
> scientist talking the same language too. You should try it!
> 
[snip]
> -- 
> M. Sc. Juan Pablo Carbajal
> -----
> PhD Student
> University of Zürich
> http://ailab.ifi.uzh.ch/carbajal/
>

No, I should _not_ try it. In what I'm doing I _never_ need a product of two 
vectors that produces a matrix:

"
octave:1> foo = [1 2 3]
foo =

   1   2   3

octave:2> bar = [4 5 6]'
bar =

   4
   5
   6

octave:3> foo .* bar
warning: product: automatic broadcasting operation applied
ans =

    4    8   12
    5   10   15
    6   12   18

octave:4> 
".


And regarding cargo cult linear algebra - row <-> column vector should only be 
taken into account only if not taking it into account can produce ambiguous 
results.


For that matter, since both inner and outer products exist: 
https://en.wikipedia.org/wiki/Matrix_multiplication#The_inner_and_outer_products
 -> https://en.wikipedia.org/wiki/Inner_product -> 
https://en.wikipedia.org/wiki/Outer_product , if they both supported by 
default, none deserves a warning.

So, again, there should be a, say,

linear_algebra_compatibility;

compiler pragma which introduces new laws of dealing with vectors and matrices.


Regards,
  Sergei.