I have played around a little with OxGauss and I am quite impressed,
very nice piece of work!
That said, I also have some problems (or I wouldn't be writing to the
list ;-) ).
To write code that runs both under Gauss proper and OxGauss it would be
useful to do conditional compilation to implement a work around for a
feature that is missing in OxGauss. E.g to use QuadPack instead of the
intquad1 Gauss function. I expected
#ifdef OX_GAUSS
print "oxgauss";
#else
print "gauss";
#endif
to print "oxgauss" under OxGauss but it prints "gauss" both under
OxGauss and Gauss.
Is there a way to tell in which environment the code is executed?
Another thing that has me stymied is how to actually use QuadPack from
OxGauss. I suppose it is possible to modify g2ox.ox but I would prefer a
portable solution that does not require other people to modify their
g2ox.ox if I give them my code. Any suggestions are welcome.
Finally I think I have stumbled over a few minor bugs in g2ox.ox (Ox
Professional 3.30).
The olsOutput routine seems to work better if the print and println
statements are preceeded by ::.
Changing the test
if (!isstring(dataset))
in the ols routine to
if ( isstring(dataset) && dataset!="" )
and switching the order of the code blocks allows ols to be called like
call ols( "", y, x );
as well as
call ols( 0, y, x );
which according to the Gauss documentation both are valid.
qnewton seems to return the negative of the function value at the
maximum. The Gauss QNewton minimizes and OxGauss maximizes the negative
of the function but fail to reverse this for the returned function value.
/Sune
--
Sune Karlsson | Fax: + 46 8 34 81 61
Stockholm School of Economics | Phone: + 46 8 736 92 39
Box 6501, 113 83 Stockholm, Sweden | http://www.hhs.se/personal/SuneK/
http://econpapers.hhs.se/ | http://swopec.hhs.se/
======================================================================
September 15 to December 19:
University of California, San Diego
Department of Economics, 0508
9500 Gilman Drive
La Jolla, CA 92093-0508
(858) 822-0585
|