Some strange result with the dejong setting
by Paolo Chirico
Hi all,
sorry to bring this up again, but there are a few things I don't understand:
1) If I set diffuse=2 and dejong=0, I expect the default method to be
sequential=1. Instead, into the filter bundle, I find:
/Kalman scalars:
t: 0
diffuse: 1 (not 2?)
exact: 0 (not 1?)
cross: 0
dkvar: 0
sequential: 0 (not 1?)/
If I set diffuse=2 and dejong=1, I find:
/Kalman scalars
t: 0
diffuse: 1 (not 2?)
exact: 0 (not 1?)
cross: 0
dkvar: 0
sequential: 0/
Moreover, if disturbances are not correlated (/cross=0/), settings
(diffuse=1, dejong=1) and (diffuse=2, dejong=1) produce same results as
well as (diffuse=1, dejong=0) and (diffuse=2, dejong=0) produce same
results.
On the other hand, if disturbances are correlated (/cross=1/),
(diffuse=2, dejong=1) provides better results (in terms of AIC, BIC,...)
than (diffuse=1, dejong=1).
However, in no case is diffuse=2 reported in the filter bundle!
Paolo
1 week
gretl4py 0.31 released
by Marcin Błażejowski
Hi All,
today I released gretl4py 0.31 for four leading operating systems:
windows (x86 and arm64), macOS (intel and arm64), and linux (x86). To
install the package just call: 'python(.exe) -m pip install gretl4py'.
The most important change (comparing to 0.30) is that all official gretl
addons are now bundled with gretl4py distribution. So, for example, if
one wants to estimate garch model via gig package, here is a snippet:
<python>
import gretl
gretl.include("gig")
OPTS = gretl.gig_setup( params )
res = gretl.gig_estimate(OPTS)
</python>
Marcin
1 week, 2 days
De Jong's diffuse approach
by Paolo Chirico
Hi all,
I'm not sure which script is correct for De Jong's diffuse approach.
For example, for De Jong's exact diffuse, which of the following is correct?
A) SSMod.diffuse=2
SSMod.dejong=1
B) SSMod.diffuse=2 --dejong=1
Or neither, but which?
Thanks,
Paolo
1 week, 3 days
Error with recent snapshot when using gnuplot command with factors
by Artur T.
Hi all,
I'm encountering an error with the recent gretl snapshot (v2025c)
related to the gnuplot command. The problem arises when using some
options like --dummy and/or --factorized with discrete series.
By the way, I realized that the --dummy option is not documented
anymore(?) (I am pretty sure it was listed in the past). Probably, this
option got superseded by the --factorized option recently (see below).
Below is a minimal example illustrating the issue:
<hansl>
set verbose off
open mrw.gdt --quiet
# "OECD" is a discrete series (see series attributes)
# Note: The following "works" technically, but only 3 data points are shown!
# This used to work in the past:
gnuplot gdpgrow school OECD --dummy --output=display
# This works without error, but again only 3 instead of 121 points are shown
gnuplot gdpgrow school OECD --factorized --output=display
# This returns an error "Invalid argument":
gnuplot gdpgrow school OECD --factorized=OECD --output=display
# After executing the above line, the following also fails with "Invalid
argument"
# even though it worked before:
gnuplot gdpgrow school OECD --factorized --output=display
</hansl>
This may be connected to the recent update in version 2025b according to
the changelog:
<>
2025-06-05 Version 2025b
- "gnuplot" command: generalize the --dummy functionality for factorized
plots; update and improve the help text
</>
I am working under Ubuntu and have gnuplot version 6.0 patchlevel 2
installed
Do you guys face the same error?
Artur
2 weeks, 2 days
Problem running "foreign language=R --send-data" in Ubuntu
by Roland Clarke
Dear Colleagues,
I am having problems running a simple Gretl script to retrieve variables
from the FRED database. The script works perfectly when I run it on a
Mac, but when I run it on my Ubuntu 24.04 it fails on the line:
foreign language=R --send-data
under Ubuntu it generates the following output:
gretl version 2023c
Current session: 2025-10-13 10:15
? foreign language=R --send-data
? print("before loading libraries)
? library(fredr)
? print("loaded fredr)
? library(eFRED)
? print("loaded Fred libraries")
? source("functions/data_management.R")
? print("loaded functions")
? Fd <- Fred_data(gretldata$SLIST)
? Fdata <- Fd[[1]]
? write.csv(Fdata, "temp/FRED_full_from_R.csv", row.names = FALSE)
? end foreign
/home/roland/.gretl/R.msg: No such file or directory
If I create a dummy R.msg then I get the following output:
gretl version 2023c
Current session: 2025-10-13 10:15
? foreign language=R --send-data
? print("before loading libraries)
? library(fredr)
? print("loaded fredr)
? library(eFRED)
? print("loaded Fred libraries")
? source("functions/data_management.R")
? print("loaded functions")
? Fd <- Fred_data(gretldata$SLIST)
? Fdata <- Fd[[1]]
? write.csv(Fdata, "temp/FRED_full_from_R.csv", row.names = FALSE)
? end foreign
4th test
External command failed
If I run the Gretl script on the Mac it retrieves the variables.
I included the print statements for debugging purposes and they seem to
indicate that nothing is executed beyond the foreign language line.
If I run the R script from R on Ubuntu it also works.
The only difference between the Ubuntu and Mac versions is that Ubuntu
uses Gretl 2023c and the Mac uses Gretl 2025c.
Any suggestions would be very welcome.
Thanks,
Roland
2 weeks, 4 days
gretl4py 0.3 released
by Marcin Błażejowski
Hi all,
Today I released a new version of the gretl4py project. This version
includes a major upgrade, with native Python support for gretl bundles
and — most importantly — an interface for calling user-contributed
packages (i.e., all packages available for gretl).
The full changelog is available here:
https://sourceforge.net/p/gretl/gretl4py/ci/v0.30/tree/changelog.txt
A few remarks:
- Linux builds mistakenly include an older OpenBLAS (v0.3.15); this will
be fixed in v0.31.
- macOS users: please allow a couple of days before 0.30 becomes
available due to a certificate issue I need to resolve.
- An updated PDF documentation will be uploaded within a few days.
In the meantime, examples demonstrating how to use gretl4py can be found
here: https://sourceforge.net/p/gretl/gretl4py/ci/v0.30/tree/demo/
Cheers,
Marcin
2 weeks, 5 days
cURL error
by Artur T.
Hi all,
a student of mine has reported an error when trying to install a package
from the gretl server yesterday evening. Initially I thought he has a
problem with a certificate on his WIndows machine because he got the
error code 60.
However, just trying it on my Ubuntu machine returns:
cURL error 28 (Connection timed out after 20002 milliseconds)
Terminated on error
Trying to click through the GUI to the packages in the server, returns:
cURL error 6 (Could not resolve host: gretl.sourceforge.net)
The gretl package site works, however, and also downloading a gfn file
is possible:
https://gretl.sourceforge.net/cgi-bin/gretldata.cgi?opt=SHOW_FUNCS
Best
Artur
4 weeks
Package updates (September 2025)
by Riccardo (Jack) Lucchetti
Dear all,
this message is to inform the community about the activity in our
function package repository: during the month of September 2025, 2
packages were updated to a new version:
"calendar_utils", by Artur Tarassow (Collection of useful date time
related tools)
"yahoo_get", by myself (Downloader of daily financial data from Yahoo)
Download and enjoy!
PS: several people have told me about new function packages that are in
preparation, with a lot of exciting new stuff. Stay tuned.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
4 weeks