set verbose off
set seed 123456
foo = randgen(N, 10, 2)
freq foo --nbins=3
freq foo --nbins=4
Why I get 5 bins instead of 4 in the second freq call?
just installed 2018a on Windows, and for the fun of it did a check for
updates. What I got is:
"You are running gretl 2018a.
This seems to be a pre-release; the latest version available at
http://ricardo.ecn.wfu.edu/pub/gretl/ is 2017d"
The topic name borrowed from Xavier Sala-i-Martin
(I JUST RAN FOUR MILLION REGRESSIONS )
myfilem4em.zip contains auto_arima output with "-a"
myfilem4em2.zip contains auto_arima output without "-a"
792 series: monthly economic M4 series
In a previous message I wrote how to get this data set
It is possible to look at the results and use
more sophisticated script to obtain some auto-comparisons
A sizable percentage of the series requires some
transformations or large order arima to get rid of
It is interesting to catch whether the percentage
of different optimal IC models depends on whether
the model is misspecified
In current git and snapshots there's a "secret" option for the arma
command, --as197 or just -a, which invokes new code using Melard's
algorithm AS 197 in place of the default Kalman-based code when
estimation is via native exact ML.
The hope -- borne out by our testing to date -- is that AS 197 will
be a good deal faster than the Kalman code without sacrifice of
accuracy. But it would be good to have that confirmed or denied on a
wider range of cases. (However, note that AS 197 will not be used if
there are missing values in the sample range.)
We've also been thinking about revising our arma initialization
code, but at present both Kalman and AS 197 use the "same old"
I might point out that you can expect slightly different
coefficients and/or standard errors from Kalman and AS 197: these
often change a little even when one and the same algorithm is
employed, with updates to the C compiler or blas/lapack. Our concern
is ensuring that we don't get a noticeably inferior maximized
loglikelihood with AS 197, or failure to converge when the Kalman
this is not a new bug / odd behavior, but perhaps it's not on the radar
anymore, because it might be a Windows-only thing. It still happens
"often" (not always), that when you search for a word in the script
editor, the window content is moved such that the next match is just in
the next line below the bottom of the window, so not visible without
This is with recent snapshots and releases, and I think both on Windows
7 and 10.
Maybe tricky to fix, and of course not release-critical.
I think it would be good if in the window with the listing of installed
function packages there were either some indication of the status of the
respective packages (similar to the window with the add-ons: local
status current or not). Or to have a button or clickable link or
something that checks for and update.
The thing is that right now users have to know about the existence and
meaning of the look-on-server button, and then manually have to locate
the package and compare the version number.
The context is for example recently updated packages like getQuandl or
FEP, where the previous versions fail (for different reasons). A regular
user might not guess that a new version is there and fixes the problem,
instead they might conclude that it's simply broken.
The following string is not marked for translation: "Warning:
ignoring". It appears when we give a name for a plot command and try
to save file to disk at the same time:
"my_plot" <- gnuplot PAU --with-lines --time-series --output="my_file.png"
Today's git help entry for max
does not mention matrix argument
set seed 13
matrix a_matrix = mnormal(10^4,10^4)
ma1 = max(vec(a_matrix))
ma2 = maxc(maxr(a_matrix))
I have 1.6487076 sec for max(vec..
and 7.9388368 sec for the second version
And the first version seems to be doomed