3:11 a.m.
On Thu, 5 May 2011, Riccardo (Jack) Lucchetti wrote:
On Wed, 4 May 2011, Olivier Ga wrote:
> I use the arma function under linux and sun solaris. [...]
It seems to me your analysis is correct. Could you try if after changing
the line
for (k=0; k<r-m; k++) {
to
for (k=0; k<r-m-1; k++) {
the problem persists?
I'm confident that Jack's fix is correct. I've now made this
modification in gretl CVS. The mistake actually occurs in two
places in the numerical Hessian code.
This is a cautionary tale: it can be more difficult that you'd
think, converting code from 1-based indexing to 0-based, when you
have nested loops to handle!
Allin Cottrell
7:03 a.m.
New subject: numerical_hessian
On Wed, 4 May 2011, Allin Cottrell wrote:
This is a cautionary tale: it can be more difficult that you'd
think, converting code from 1-based indexing to 0-based, when you
have nested loops to handle!
As I said in my previous message, I find it quite odd that the problem
hasn't surfaced in such a long time. Also mysterious is the failure by
valgrind to detect it. Oh well.
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche
r.lucchetti(a)univpm.it
http://www.econ.univpm.it/lucchetti
12:36 p.m.
New subject: numerical_hessian
Am 05.05.2011 05:11, schrieb Allin Cottrell:
This is a cautionary tale: it can be more difficult that you'd
think, converting code from 1-based indexing to 0-based, when you
have nested loops to handle!
Thinking about it in retrospect, I'm actually surprised that a C-coder
like you chose 1-based indexing for gretl's language. But it probably
helps the average user a lot. I remember that on the python-numpy list
(python is 0-based) this used to be a repeated issue raised by new
users, many coming from Matlab.
-sven
12:39 p.m.
New subject: numerical_hessian
On Thu, 5 May 2011, Sven Schreiber wrote:
Am 05.05.2011 05:11, schrieb Allin Cottrell:
>
> This is a cautionary tale: it can be more difficult that you'd
> think, converting code from 1-based indexing to 0-based, when you
> have nested loops to handle!
Thinking about it in retrospect, I'm actually surprised that a C-coder
like you chose 1-based indexing for gretl's language. But it probably
helps the average user a lot. I remember that on the python-numpy list
(python is 0-based) this used to be a repeated issue raised by new
users, many coming from Matlab.
We use 1-based in userspace (I agree, it's in line with what most
people are likely to expect), but 0-based internally. The problem
with the Hessian came in converting from 1-based code written in
S/R to 0-based C.
Allin
5020
days inactive
5020
days old
3 comments
3 participants
participants (3)
-
Allin Cottrell -
Riccardo (Jack) Lucchetti -
Sven Schreiber