5:37 a.m.
The *)!"$ web frontend ate my message, I'll try again:
From the manual:
12.5 Matrix–scalar operators
...
(Note that the operator ** cannot be used in place of ^ for this purpose because in a
matrix context
it is reserved for the Kronecker product.)
AFAICS it's always ^ in gretl, regardless of the context, no? I don't remember if
that was different before. IMHO it's a bit misleading.
thanks,
sven
-------- Original-Nachricht --------
Datum: Thu, 31 Jul 2008 11:15:43 +0200
Von: "Sven Schreiber" <svetosch(a)gmx.net>
An: Gretl development <gretl-devel(a)lists.wfu.edu>
Betreff: [Gretl-devel] ** documentation glitch
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
5:53 a.m.
BTW, it happened to me that I mistakenly used ** for scalar exponentiation, but gretl
treated the whole thing as a matrix expression with Kronecker product. So for 1**0 I got 0
instead of 1! I'm not necessarily saying this is a bug, but a script definitely can be
hard to debug with this. So I'm not sure the automatic scalar-to-matrix conversion is
good here. What do you think?
-sven
-------- Original-Nachricht --------
Datum: Thu, 31 Jul 2008 11:37:18 +0200
Von: "Sven Schreiber" <svetosch(a)gmx.net>
An: Gretl development <gretl-devel(a)lists.wfu.edu>
Betreff: Re: [Gretl-devel] ** documentation glitch
The *)!"$ web frontend ate my message, I'll try again:
>>From the manual:
12.5 Matrix–scalar operators
...
(Note that the operator ** cannot be used in place of ^ for this purpose
because in a matrix context
it is reserved for the Kronecker product.)
AFAICS it's always ^ in gretl, regardless of the context, no? I don't
remember if that was different before. IMHO it's a bit misleading.
thanks,
sven
-------- Original-Nachricht --------
> Datum: Thu, 31 Jul 2008 11:15:43 +0200
> Von: "Sven Schreiber" <svetosch(a)gmx.net>
> An: Gretl development <gretl-devel(a)lists.wfu.edu>
> Betreff: [Gretl-devel] ** documentation glitch
>
> _______________________________________________
> Gretl-devel mailing list
> Gretl-devel(a)lists.wfu.edu
> http://lists.wfu.edu/mailman/listinfo/gretl-devel
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
8:33 a.m.
On Thu, 31 Jul 2008, Sven Schreiber wrote:
12.5 Matrix?scalar operators
...
(Note that the operator ** cannot be used in place of ^ for this
purpose because in a matrix context it is reserved for the
Kronecker product.)
AFAICS it's always ^ in gretl, regardless of the context, no? I
don't remember if that was different before. IMHO it's a bit
misleading.
Ah yes, that comment is "orphaned". There was a time when we
officially supported "**" for exponentiation (before we had
Kronecker products). But checking the genr help I see that we've
removed any mention of "**" in that role. So that bit of text in
12.5 should be removed.
However, it's a little more complicated. Until recently we
continued to support x**n = x^n unofficially, since it's handy in
dealing with the NIST nonlinear regression test files, where the
functions are written in that way. I wonder if that's now broken?
Anyway, the NIST files can be handled in some other way.
Allin.
8:51 a.m.
On Thu, 31 Jul 2008, Allin Cottrell wrote:
However, it's a little more complicated. Until recently we
continued to support x**n = x^n unofficially, since it's handy in
dealing with the NIST nonlinear regression test files, where the
functions are written in that way. I wonder if that's now broken?
Memory lapse. I'd already fixed the NIST nls test suite so as to
substitute "^" for "**". So, even unofficially, "**" is not
in
any way supported as exponential operator.
I've removed the misleading statement from the matrices chapter of
the Guide in CVS.
Allin.
5953
days inactive
5953
days old
4 comments
2 participants
participants (2)
-
Allin Cottrell -
Sven Schreiber