Re: size and display of N-d matrices (octave-3.4.2)

[Przemek Klosowski]

On 08/10/2011 03:26 PM, Sergei Steshenko wrote:

> octave:1>  a(1,1,1,1) = 1
> octave:2>  a(1,1,1,2) = 2
> a =
>
> ans(:,:,1,1) =  1
> ans(:,:,1,2) =  2

> octave:5>  b(1,1,1,1) = 3
> octave:6>  b(2,1,1,1) = 4
> b =
>
>     3
>     4

> 1) why does 'octave' treat 'a' and 'b' differently ?
> 2) why 'b' is shown to have just two dimensions ?
> 3) why colons when 'a' is displayed ?

1) the issue of style

In the old days, the saying went: 'Punctuality is the courtesy of the 
kings'. Nowadays, careful editing of correspondence seems to have become 
the courtesy of the kings. It is hard to follow a letter that has lots 
of boilerplate quoted or cut-and-paste material, with interesting, 
relevant content hidden at the end. In extreme, it makes a mockery of 
the common recommendation for bottom-posting --- if the quoted material 
is dumped indiscriminately in front of the new content, it would 
actually be less hard to read if it was top-posted, especially on the 
mobile devices that more of us use nowadays. In other words, literate 
persons are expected to weave the trimmed quotes and their responses 
into a coherent message --- that's the true meaning of "bottom-posting".

The point here is that careful editing makes huge difference to the 
reader. Sure, it takes time and effort --- as Seneca said, "I apologize 
for this very long letter but I was out of time" --- but the goal of 
writing is to communicate, so if it is worth doing at all, it is worth 
doing well.

Thank you for the opportunity to get it off my chest. I thought it would 
not be impolite to send it to the list even though most postings  here 
are usefully edited.


2) the issue of your questions

a)
Matlab basic data type started as a 2-D array, with vestiges of 1-D 
linear, column-wise data layout visible behind the curtains. It was then 
extended to multiple dimensions, by allowing more than two indices 
_added_after_ the initial ones. 2-D data is still treated specially, in 
that multidimensional arrays with singular trailing n-2 indices are 
displayed as if those indices didn't exist. Similarly, if you set 
e(2,1)=3 you can call it back as e(2,1) or e(2). Note that you can even 
use e(2,1,1) or e(2,1,1,1,1,1,1), but of course not e(2,1,2).

b)
In the absence of non-trivial 3rd and higher dimensions, the matrices 
are shown as if they were 2-D (see the above explanation)

c)
Colons are a shorthand for the entire range of the index in the 
respective position, so in my example, e(2,:) results in '3', and 
e(:,1) results in the vector [0 3]'.






n order to check out code snippets, one has to cut and paste