Re: Bug/feature in matrix multiply

[Martin Schultz <mgs(at)io.harvard.edu>]

Mark Fardal wrote:
> [...]

> So my question becomes, why does this happen?  If IDL is going to
> treat an 3x1 array differently than a 3-element vector, it shouldn't
> just cavalierly remove the trailing dimension in my opinion.  The
> behavior is probably documented somewhere but I couldn't find it in
> the hyperhelp.  There is this one sentence in "Combining Array
> Subscripts with Others": "As with other subscript operations, trailing
> degenerate dimensions (those with a size of 1) are eliminated."
> 
> I also notice that the behavior is somewhat inconsistent, in that
> converting an expression to one of the same type does _not_ remove
> the trailing dimension:
>[...]

  I agree (somewhat). Generally, I found that IDL is quite "smart" in
removing trailing dimensions so that one doesn't have to worry too much
about them. But when you do encounter a case where you need to (and
this is frequent for any type of matrix manipulation), IDL is just 
too smart and you have to think twice to outsmart it ;-)

   Anyway: you can always make sure you get what you want with

   a = transpose(reform(a))

These statements are not very costly in terms of execution time, because 
it's only messing around with the array descriptor (at least I believe
so).

Regards,
Martin.


-- 
-------------------------------------------------------------------
Dr. Martin Schultz                   
Department for Engineering&Applied Sciences, Harvard University
109 Pierce Hall, 29 Oxford St., Cambridge, MA-02138, USA

phone: (617)-496-8318
fax  : (617)-495-4551

e-mail: mgs@io.harvard.edu
Internet-homepage: http://www-as.harvard.edu/people/staff/mgs/
-------------------------------------------------------------------