1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57
|
* pstable() bug beta ~= -1, found by Anastasija Sanina, Feb.2016
---> ~/R/MM/Pkg-ex/stabledist/pstable-uniroot-bug.R
* Compare with Jim Lindsey's "stable" package
--> http://www.commanster.eu/rcode.html
Maybe even ask him to make it CRANny
(maybe after transfering maintainership ?)
At SfS ETH: source in /usr/local/app/R/R_local/JimLindsey/stable/
installed in /usr/local/app/R/R_local/JimLindsey/library
* stableMode(): [MM] I'm pretty sure that the 'beta.max' argument is
unneeded and could be eliminated.
* parametrizations -- pm=0, 1, 2 ---
provide functions for the parameter transformation that go from
other to pm=0 -- and back
and then show how to, or use these to to provide transformations
from all pm's to any other.
* [qp] check the lower.tail & log.p argument settings
NB: This check *should* happen in fBasics 'checkdist' ...
* curve(dstable(exp(x), alpha= 0.1, beta=1, pm=1, log=TRUE), -15, 0)
very much depends on zeta.tol ... that should somehow depend on alpha or zeta;
--> tests/dstab-ex.R ("beta = 1")
* As easy alternative to fBasics::stableFit() define one based
on stats4::mle (and possibly also a quantile-matching alternative).
------------------------------------ Done : --------------------------------------
====
* Nolan ("tail.pdf") and other places mention *asymptotic* formulae
--> use these for large |x|, notably for dstable(*, log = TRUE)
* pstable() should also get lower.tail = TRUE, log.p = FALSE
~~~~~~~ arguments
* qstable() ditto
* using doExtras to *not* run some tests when --as-cran , as
From: Prof Brian Ripley <ripley@stats.ox.ac.uk>
Subject: Re: New version of copula uploaded 0.99-2
Date: Tue, 29 May 2012 17:17:05 +0100
.. but the test times
* checking tests ... [173s/174s] OK
Running doRUnit.R [82s/83s]
Running dstab-ex.R [34s/34s]
Running pstab-ex.R [19s/19s]
Running tails.R [39s/39s]
are more than we would like, so please consider reducing substantially
next time around
...
|