These all sound great. Thanks very much for all involved.
On Wed, Aug 25, 2010 at 5:21 AM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
I thought I might draw people's attention to a few new things in
CVS, and encourage testing:
"bundles": we have a new data type, the bundle. This was Jack's
idea, and we worked on it earlier this summer. It's mostly
intended to make life easier for writers of ambitious function
packages. It's documented in the new chapter of the User's Guide
titled "Gretl data types". (That may not the permanent home for
the bundle documentation but it'll do for now.)
filters: I've been corresponding with Stephen (D.S.G.) Pollock and
under the Filters menu for time-series data we now have two items
that he has advocated (and coded in the IDEOLOG software that he
described at last year's gretl conference). That is, the
Butterworth filter and polynomial trend fitting. The Butterworth
filter is described, with an example, in the chapter of the User's
Guide titled "Special functions in genr". Polynomial trend fitting
is, of course, something you could do in gretl before, but
Stephen's algorithm is fast and light on memory, and does not
require you to add powers of t to the dataset. It's also
numerically robust, being based on "orthogonal polynomials" rather
than linear regression on powers of time.
Both of these things are also scriptable, via new functions bwfilt
and polyfit (the first is documented, the second not yet).
A smaller thing is that in the model table you can now select the
precision of the printing in terms of digits after the decimal
point rather than total significant digits if you wish.
A larger thing is in the works and likely to be unveiled before
long, but it's not quite ready yet. Watch this space!
Gretl-devel mailing list
“Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment.” - Benjamin Franklin (1706-1790)