just a quick notice that recently two new contributed function packages
have seen the light of the gretl world:
- 'ARprobit', which implements the Kauppi&Saikkonen approach to estimate
a dynamic Probit model including an autoregressive term (in addition to
lags of the dependent dummy).
- 'hamilcycle', which applies the simple but already popular proposal by
Hamilton to do a regression-based filter of a time series to obtain a
cyclical component (based on an idea by Jack a year ago)
As always (sorry for being repetitive) you can get these and other
contributed packages online from within gretl ("fx" button at the
bottom, or through the menu File/Function packages/On server).
I'm writing you on behalf of the Gretl development team.
Since June 2009, we've had a Gretl conference every other year, starting
in Bilbao; the 2019 conference was in Naples. This year, it should have
been London but the pandemic forced us to change our plans. Therefore,
here's what's happening:
* the actual conference is probably going to take place during the winter,
if the COVID situation makes it possible.
* we're having an online event on the 3rd and 4th of June. Since the
majority of the gretl community lives on one side of the Atlantic or the
other, we decided that the event will take place between 1200 and 1800 UTC
(approximately) so that the inconvenience of the time zones is hopefully
The only two fixed events at this stage are
1) a presentation by the project leader (Allin) on the 3rd
2) an invited speech by Mark Steel (Warwick University), Bayesian
extraordinaire, on the 4th.
For the rest of the time, we plan to have
- an open Q&A session in which the community can meet the development team
to ask question, make proposals or simply tell us how much you love us;
- presentations by package authors to show the community some lesser-known
- a few presentations of scientific work done using gretl, or perhaps done
by using some other package but explaining why it would have been
impossible to do in gretl, and what needs to be done to make it
- a virtual meeting of the newly-born gretl association (become a
- (possibly) the usual analysis by me on the trends of the download data
and future perspectives.
Marcin was kind enough to set up a website for the conference
(https://www.gretlconference.org/), so you can visit that page to get the
info as it becomes available.
If you'd like to present something, or have any other comments or
suggestions, feel free to contact me or any other member of the
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
A heads-up for Mac users. We're renaming the gretl packages for Mac
ro reflect the current state of things in the Apple ecosystem. Our
snapshots will now be called:
* For Macs with Intel processor: gretl-macos-intel.pkg
* For Macs with M1 processor: gretl-macos-m1.pkg
When gretl 2021b is released (should happen this week), the releases
will be named:
and so on in future. The "intel" versions should work with macOS
10.6 and higher. M1 versions require macOS 11 or higher.
I am writing an Metropolis-Hastings algorithm for Bayesian estimation of
a model. At least for the moment, I have 10 parameters that will be
under estimation in this way. This means that from each replication I
want to store 20 values, so as to retrieve the "proposal" sequence and
the "accepted" sequence of estimates. Typically this algorithm should
have at least 100,000 replications or even more, like up to 500,000.
These estimates will be stored in matrices. Does it make any difference
in computational speed if I use one matrix of [replications /times/ 20]
dimension, or 20 matrices of [replications /times/ 1] dimension?
I know that matrices become a software's nightmare when both dimensions
grow large, but what happens when the one dimension is large while the
other is small?
Does it pay to make the small dimension as small as possible?
PS: I know that one could also (or alternatively) break the large
dimension in pieces by, I guess, an IF statement inside the loop that
will depend on the loop index value, but I wanted to know if anything
can be improved by keeping the one dimension large while chopping off
Alecos Papadopoulos PhD
Athens University of Economics and Business
After much procrastinating, false-starting and working through
numerous technical issues, I'm very pleased to announce a new video
tutorial for downloading and installing -gretl- using -git- (here, version
2021b-git) which I've just published onto YouTube, available at:
Many grateful thanks to Sven Schreiber and Jack Lucchetti for comments on
an earlier (very poor) video draft sent to them (and Allin Cottrell) back
in January 2020.
Enjoy and tell your friends!
"My colleagues in the social sciences talk a great deal about methodology.
I prefer to call it style." -- Freeman J. Dyson
We're planning to release gretl 2021b in about a week, and there's a
new snapshot in place today which can be considered a "release
We'd be grateful if people could beat on the candidate (run your
favorite scripts, make your favorite GUI moves) and report any
problems. This will help us minimize the bug count in the new
For translators: this would be a good time to update your
translations if need be.
Version 0.41 is very fast again, as fast as using the older API of
dbnomics. Thank you very much for the new version using the new API of
But unfortunately 0.41 produce very often errors.
One common error is:
? list X = dbnomics_bundles_to_list( bs, "series_code" )
File exists*** error in function dbnomics_bundle_get_data, line 28
> outfile tracestr --buffer
called by function dbnomics_bundles_to_list
*** error within loop in function dbnomics_bundles_to_list
> err = dbnomics_bundle_get_data(b, &tmp)
or one not so common error is:
? bundle spec = defbundle("mask",".DEU+FRA..Q")
? bs = dbnomics_get_multiple(provider, database, 1000, 0, spec)
*** error in function sanitize_url, line 2
> ret = strsub(ret, sprintf("\""), "%22")
called by function build_parms
called by function dbnomics_get_multiple
I attache two scipts for examble, running either on gretl 2021a or 2021b
Sometimes there is no error when I run the scipts again after it
produced errors in the first run.
Dear gretl developers,
On my MacOS High Sierra 10.13.6, if I "maximize" an output window - be it a
graphic, a script output or a model window - and "close" it
eventually clicking on "x", gretl (2020e-git) crashes; all windows get
closed including gretl's main window. A long and detailed system error
report is also produced (if it can help I can share it with you).
In a WLS estimation when the weight variable contains zeros gretl stops
executing saying that "Weight variable is not a dummy but contains zeros"
The same worked perfectly before with the observations weighted 0 being
from the estimation.
Here's a basic code for replication
series x = normal()
series y = normal()
series w = uniform()
wls w y x
Dear gretl team,
The reason for my request:
For online teaching, I would like to create individualize exam datasets with multiple modeling outputs as icons, and having individualized exam questions stored in the notes page.
During the exam, I just send each student their individualized datasets and questions for them to work on.
Once they receive the dataset, they just have to open the notes page and see the exam questions stored in there.
This is a way to reduce cheating, for each person’s question and dataset are different. I can’t help if someone else is doing the exam for them, but at least they can’t just simply copy from each other.
I figured out how to create the random sample datasets, the multiple modeling icons by a script.
Saving the exam questions on the notes page is a bit tedious, but doable.
I would like to do:
Save each session that I created using scrip, but I can’t.
It would be great also if I can save text into the notes page using a scrip.