On Wed, 30 Apr 2008, Stefano Balietti wrote:
I was wrong. The error is not in the matrix operation, but in
the matrix declaration within the loop itself. Here is the test
case...
Thanks, this is now fixed in CVS.
However, I'm playing safe for the moment by defining GENCOMPILE to
0 in monte_carlo.c (the problem that you found arose only in the
experimental speed-up case, with GENCOMPILE = 1).
Our general policy for gretl CVS is that it should hardly ever be
knowingly broken, or overly experimental. But if any brave souls
building from CVS want to try GENCOMPILE, please do so, and report
what you find!
Allin.