9:55 p.m.
The following produces an error on a post-1.7.2-build:
matrix m1 = zeros(1,5)
matrix m2 = ones(1,5)
mymat = m1 | m2
whereas this version works:
matrix m1 = zeros(1,5)
matrix m2 = ones(1,5)
matrix mymat = m1 | m2
I know the second variant (with explicit matrix declaration) is cleaner,
but shouldn't the first one also work?
-sven
1:45 a.m.
On Sat, 1 Mar 2008, Sven Schreiber wrote:
The following produces an error on a post-1.7.2-build:
matrix m1 = zeros(1,5)
matrix m2 = ones(1,5)
mymat = m1 | m2
whereas this version works:
matrix m1 = zeros(1,5)
matrix m2 = ones(1,5)
matrix mymat = m1 | m2
I know the second variant (with explicit matrix declaration) is
cleaner, but shouldn't the first one also work?
This is the old business of what "|" means: logical OR, or
vertical matrix concatenation?
Maybe for 1.7.4 it's time to insist on "||" for OR.
Allin.
3:05 p.m.
Allin Cottrell schrieb:
On Sat, 1 Mar 2008, Sven Schreiber wrote:
> The following produces an error on a post-1.7.2-build:
>
> matrix m1 = zeros(1,5)
> matrix m2 = ones(1,5)
> mymat = m1 | m2
>
> whereas this version works:
>
> matrix m1 = zeros(1,5)
> matrix m2 = ones(1,5)
> matrix mymat = m1 | m2
>
> I know the second variant (with explicit matrix declaration) is
> cleaner, but shouldn't the first one also work?
This is the old business of what "|" means: logical OR, or
vertical matrix concatenation?
Maybe for 1.7.4 it's time to insist on "||" for OR.
You mean 1.7.3 or 1.7.4? Maybe a big version number is better to
indicate backwards-incompatible changes? (Although gretl's versioning
scheme doesn't work along those lines I guess.)
I can live with the explicit matrix declaration and don't want to push
any potentially code-breaking changes. But if the change is applied, I
think it would be a good idea to make the error message about "|" in
non-matrix contexts informative, for example "Concatenation '|' only
valid with matrices (the logical OR operator is now '||')".
cheers,
sven
4:20 p.m.
Am 03.03.2008 16:05, Sven Schreiber schrieb:
Allin Cottrell schrieb:
> Maybe for 1.7.4 it's time to insist on "||" for
OR.
>
You mean 1.7.3 or 1.7.4?
whoops, I missed the release of 1.7.3 that's why I was confused...
what's the new matrix division stuff that's described in the changelog?
thanks,
sven
5:19 p.m.
On Wed, 5 Mar 2008, Sven Schreiber wrote:
Am 03.03.2008 16:05, Sven Schreiber schrieb:
> Allin Cottrell schrieb:
> > Maybe for 1.7.4 it's time to insist on "||" for OR.
> >
>
> You mean 1.7.3 or 1.7.4?
whoops, I missed the release of 1.7.3 that's why I was
confused...
I tend to assume that everyone on gretl-users is also on
gretl-announce, but I suppose it ain't so.
what's the new matrix division stuff that's described in the
changelog?
See Section 12.3 in the User's Guide:
In matrix ``division'', the statement
matrix C = A/B
is interpreted as a request to find the matrix $C$ that solves
$BC=A$. If $B$ is a square matrix, this is treated as equivalent
to $B^{-1}A$, which fails if $B$ is singular. If $B$ is a $T
\times k$ matrix with $T > k$, then $C$ is the least-squares
solution, $C = (B'B)^{-1}B'A$, which fails if $B'B$ is singular.
Otherwise the operation necessarily fails.
Allin.
6136
days inactive
6140
days old
4 comments
2 participants
participants (2)
-
Allin Cottrell -
Sven Schreiber