[S] yags (gee solver) revision

Vincent Carey 525-2265 (stvjc@channing.harvard.edu)
Wed, 29 Jul 98 18:44:21 EDT


yags, "Yet Another GEE Solver", version 1.5.1.4, is
now available at

http://biosun1.harvard.edu/~carey/index.ssoft.html

This revision corrects problems noted by
-- Lancelot Renaud (infinite recursion of sum()
when yags attached and plot.gam attempted. Misbehavior
of sum() might have been observed in other circumstances
owing to a flawed definition of sum.default. Both
sum.default() and sum() have been permanently
removed from this distribution. Thanks to Brian Ripley
and Bill Dunlap for helpful suggestions on the resolution
of this problem.)
-- Brian Ripley (defective handling of binomial
response matrices and of weighted fractional binomial
response vectors.)

A "weights" argument has been added. When weights
are present, the vector of nominal variance function
values for the current family is divided elementwise
by the vector of weights, and the result is used in
place of the nominal variance function of the current
family for solution of the GEE and parameter variance
estimation.

[It is noted that the yags.d file refers to version
1.5.1.1. The file is materially correct despite the
flawed version tag. Version synchronization of code
and doc will be improved in subsequent versions.]

[I will not ordinarily announce revisions to yags
to the newsgroup while it is offered through the
URL given above, unless the revisions correct
potentially important problems. Upgrades of
documentation, new family construction, and solver
performance are ongoing, and interested users
should check the URL given above from time to
time to learn of revisions.]

The programs are offered subject to terms of the GPL
(see file COPYING in the distribution.)

-- 
stvjc@gauss.bwh.harvard.edu
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu.  To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message:  unsubscribe s-news